declaration of dr richard reiss in support of exponent inc s specialCal. Super. - 1st Dist.October 8, 20211 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 RANDALL W. EDWARDS (S.B. #179053) redwards@omm.com ADAM M. KAPLAN (S.B. #298077) akaplan@omm.com ERIC ORMSBY (S.B. #316956) eormsby@omm.com O'MEL VENY & MYERS LLP Two Embarcadero Center, 28th Floor San Francisco, California 94111-3823 Telephone: (415) 984-8700 Facsimile: (415) 984-8701 RICHARD GOETZ (S.B. # 115666) rgoetz@omm.com O'MEL VENY & MYERS LLP 400 S. Hope Street Los Angeles, CA 90071-2899 Telephone: (213) 430-6000 Facsimile: (213) 430-6407 Attorneys for Defendant Exponent, Inc. SUPERIOR COURT OF THE STATE OF CALIFORNIA FOR THE COUNTY OF SAN FRANCISCO UNLIMITED JURISDICTION-CIVIL PAUL A. GENTOSI, Plaintiff, V. MONSANTO COMPANY ET AL., Defendants. Case No. CGC-19-574219 DECLARATION OF DR. RICHARD REISS IN SUPPORT OF EXPONENT, INC.'S SPECIAL MOTION TO STRIKE CLAIMS AGAINST EXPONENT IN PLAINTIFF'S COMPLAINT PURSUANT TO THE ANTI-SLAPP STATUTE, CALIFORNIA CODE OF CIVIL PROCEDURE § 425.16 Complaint filed: March 1, 2019 REISS DECLARATION ISO EXPONENT ANTI-SLAPP MOTION Case No. CGC-19-574219 ELECTRONICALLY F I L E D Superior Court of California, County of San Francisco 06/20/2019 Clerk of the Court BY: SANDRA SCHIRO Deputy Clerk 1 2 3 DECLARATION OF DR. RICHARD REISS I, Dr. Richard Reiss, declare as follows: 1. I am Group Vice President and a Principal Scientist in the Chemical Regulation & 4 Food Safety at Exponent, Inc. In that role, my responsibilities include, among other things, 5 providing consulting services related to scientific and regulatory issues-including on 6 environmental and chemical risk assessment-and publishing scientific literature on similar 7 issues. I have been involved in consulting for Monsanto Company in connection with their 8 efforts to maintain the registration of glyphosate with the Environmental Protection Agency under 9 federal law. I submit this declaration in support of Exponent's motion to strike the claims against 10 Exponent in the plaintiffs complaint pursuant to the California Anti-SLAPP statute, Cal. Civil 11 Code § 425.16. The facts stated in this declaration are based upon my personal knowledge or on 12 my review of records and information cunently maintained in the ordinary course of business by 13 Exponent. If called as a witness, I could and would testify to the facts stated in this declaration. 14 2. I hold a Doctor of Science in Environmental Health from the Harvard School of 15 Public Health, a Master of Science in Environmental Engineering from Northwestern University, 16 and a Bachelor of Science in Chemical Engineering from the University of California at Santa 17 Barbara. I am a widely recognized expert in the fields of environmental health and risk 18 assessment. I am a Past-President of the Society for Risk Analysis (SRA) and a fellow of the 19 Society. In 2001, I was awarded the Chauncey Stan award from SRA in 2001, which recognizes 20 a risk analyst under 40 that has made significant contributions to the field. In 2018, I was 21 awarded SRA's Outstanding Practitioner Award, which recognizes a scientist with an outstanding 22 risk assessment practice. For nearly a decade, I was the Managing Editor of Risk Analysis: An 23 International Journal, the flagship journal of SRA. Through that experience, I gained significant 24 lmowledge of the scientific publishing process. 25 3. Exponent is a multi-disciplinary engineering and scientific consulting firm that 26 uses its expertise in more than 90 different disciplines to help its clients address science, 27 engineering, regulatory, and business issues. 28 4. For more than 50 years, Exponent has consulted on high-profile matters for a wide REISS DECLARATION ISO 1 EXPONENT ANTI-St.APP MOTION Case No. CGC-19-574219 1 variety of clients including the National Institutes of Health, NASA, and the U.S. departments of 2 Defense, Energy, and Justice, as well as many public and private companies. 3 5. More than 500 Exponent scientists, of approximately 800 professional staff, hold a 4 doctorate in their field of study, and Exponent's research and analysis is often confirmed by 5 entities such as the National Highway Traffic Safety Administration and NASA. 6 6. Exponent does not manufacture, distribute, or sell pesticides, herbicides, or any 7 glyphosate-based products. It is not an affiliate of Monsanto or any other company that 8 manufactures, distributes, or sells those products. 9 7. Exponent provided consulting services to Monsanto Company in connection with 10 Monsanto's efforts to maintain the registration of glyphosate with the Environmental Protection 11 Agency. Exponent was compensated for these services but did not (and does not) receive any 12 direct financial benefit from the sale of any glyphosate-based products. 13 8. Exponent did not itself perfom1 any tests on Roundup or other glyphosate-based 14 products in connection with Monsanto's efforts to maintain the registration of glyphosate with the 15 Environmental Protection Agency. 16 9. Based on its independent research and analyses, Exponent has published scientific 17 literature on specific issues associated with potential health risks of glyphosate and glyphosate- 18 based products. The Exponent-authored articles that the plaintiff, Mr. Gentosi, cites in his 19 complaint include proper attribution, and the conclusions are Exponent's. A copy of those 20 articles, with proper attributions, are attached to this Declaration as Exhibits A-D. The complaint 21 includes a reference to Bus (2016). We are not aware of a relevant paper from Bus in 2016. I 22 have attached papers from Bus in 2015 and 2017 that may be the papers Mr. Gentosi refers to. 23 24 25 26 27 28 2 REISS DECLARATION ISO EXPONENT ANTI-SLAPP MOTION Case No. CGC-19-574219 1 I declare under penalty of pe1jury under the laws of the State of California that the 2 foregoing is true and correct to the best of my knowledge, information, and belief. 3 Executed this 18th day of June, 2019, in Alexandria, Virginia. 4 By: //,~7-:J:i✓ ~ ,.__ -· 5 Dr. Richard Reiss 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 3 REISS DECLARATION ISO EXPONENT ANTI-SLAPP MOTION Case No. CGC-19-574219 EXHIBIT A lable at ScienceDirect Regulatory Toxicology and Pharmacology 73 (2015) 758e764Contents lists avaiRegulatory Toxicology and Pharmacology journal homepage: www.elsevier .com/locate/yrtph CrossMark Analysis of Moms Across America report suggesting bioaccumulation of glyphosate in U.S. mother's breast milk: Implausibility based on inconsistency with available body of glyphosate animal toxicokinetic, human biomonitoring, and physico-chemical data James S. Bus Exponent, Inc., 1800 Diagonal Road, Suite 500, Alexandria, VA, 22314, USAa r t i c l e i n f o Article history: Received 15 May 2015 Received in revised form 22 October 2015 Accepted 23 October 2015 Available online 28 October 2015 Keywords: Glyphosate Breast milk Biomonitoring Bioaccumulation ToxicokineticsE-mail address: jbus@exponent.com. http://dx.doi.org/10.1016/j.yrtph.2015.10.022 0273-2300/© 2015 Elsevier Inc. All rights reserved.a b s t r a c t The non-peer-reviewed biomonitoring report published online by Moms Across America (MAA; Hon- eycutt and Rowlands, 2014) does not support the conclusion that glyphosate concentrations detected in a limited number of urine samples from women, men and children, or breast milk from nursing mothers, pose a health risk to the public, including nursing children. Systemically absorbed doses of glyphosate estimated from the MAA urine biomonitoring data and from other published biomonitoring studies indicate that daily glyphosate doses are substantially below health protective reference standards (ADIs; RfDs) established by regulatory agencies. The MAA report also suggested that detection of relatively high glyphosate concentrations in breast milk in 3 of 10 sampled women raised a concern for bioaccumulation in breast milk. However, the breast milk concentrations reported by MAA are highly implausible when considered in context to low daily systemic doses of glyphosate estimated from human urine bio- monitoring data, and also are inconsistent with animal toxicokinetic data demonstrating no evidence of retention in tissues or milk after single- or multiple-dose glyphosate treatment. In addition, toxicokinetic studies in lactating goats have shown that glyphosate does not partition into milk at concentrations greater than blood, and that only a very small percentage of the total administered dose (<0.03%) is ultimately excreted into milk. The toxicokinetic studies also indicate that human glyphosate exposures estimated from urine biomonitoring fall thousands-of-fold short of external doses capable of producing blood concentrations sufficient to result in the breast milk concentrations described in the MAA report. Finally, in contrast to highly lipophilic compounds with bioaccumulation potential in breast milk, the physico-chemical properties of glyphosate indicate that it is highly hydrophilic (ionized) at physiological pH and unlikely to preferentially distribute into breast milk. © 2015 Elsevier Inc. All rights reserved.1. Introduction In April of 2014, Honeycutt and Rowlands (2014), under spon- sorship of the non-governmental organizations Moms Across America (MAA) and Sustainable Pulse, described the results of a non-peer-reviewed report (Internet posting) in which glyphosate was detected in breast milk of 3 of 10 women located in Florida (166 mg/L), Virginia (76 mg/L) and Oregon (99 mg/L). In addition, glyphosate was detected in 13 of 35 urine samples from women residing in Oregon, California, Washington, Maryland, Colorado, and Hawaii, with the highest urine concentration (18.8 mg/L)detected in a woman from Oregon. The report concluded that the relatively high concentrations of glyphosate detected in breast milk suggested bioaccumulation, and thus posed potential health threats to nursing children. Importantly, however, the report itself acknowledged that the “initial testing” was “not meant to be a full scientific study” and was primarily intended to catalyze further in- depth analyses and understanding of the potential health signifi- cance of glyphosate detection in breast milk. Such a precaution is consistent with cautions from the Centers for Disease Control and Prevention and the National Academy of Science/National Research Council (NAS/NRC) that simple detection of environmental sub- stances in human biomonitoring samples does not necessarily indicate disease-causing potential. Rather, examination of J.S. Bus / Regulatory Toxicology and Pharmacology 73 (2015) 758e764 759biomonitoring findings in the broader context of exposure and toxicity data, as well as established health-protective reference standards, is necessary to understanding when biomonitoring data suggest potential health effects (NAS/NRC, 2006; CDC, 2009; Aylward et al., 2013). A non-peer-reviewed May 2014 analysis of the MAA report by Dr. Ron Kleinman, the physician-in-chief of Massachusetts General Hospital for Children, and Chair of the Department of Pediatrics at Massachusetts General Hospital, has offered two significant methodological concerns associated with the report (Kleinman, 2014). First, the milk analyses relied on an antibody-based assay (ELISA) that was developed originally for qualitativewater analyses, and the report was silent concerning key technical details regarding whether the method was specifically standardized and calibrated for milk assays in the range of glyphosate levels reported in the report. Second, sample quality concerns arise in that limited details were provided as to how the samples were obtained (e.g., details of selection of study participants), shipped, stored, and ul- timately processed by the receiving analytical laboratory. These methodological concerns can be resolved only by generating or providing additional experimental findings and further clarifying the demographic details of sample sources and chain-of-custody protocols. The U.S. EPA also noted these deficiencies and others in a letter sent to MAA (U.S. EPA, 2014). Beyond the methodological concerns, the present analysis demonstrates how an integrated evaluation of animal tox- icokinetics, human biomonitoring and exposure data, and the physico-chemical properties of glyphosate reveals that the con- centrations of glyphosate reported in the three samples of breast milk (76, 99, and 166 mg/L) are implausibly high and not attainable under realistic glyphosate exposure scenarios. This analysis also demonstrates that glyphosate does not bioaccumulate in breast milk. Other integrated evaluations of the overall body of glyphosate toxicology and exposure data have likewise concluded that external daily doses of glyphosate, as determined from human bio- monitoring data or projected from exposure modeling, do not present unacceptable health risks to the public, including nursing children (U.S. EPA, 1993; Williams et al., 2000; McQueen et al., 2012; U.S. EPA, 2013; Niemann et al., 2015). A recent WorldTable 1 Recovery of 14C-glyphosate in urine, feces, and tissues as a percentage of administered d Dose Species Urine Feces Male Female Male Single dose, pob 6.7 mg/kg, 120 ha Rat 14e16 35e43 81e85 10 mg/kg, 24/48 h Rat 17.9/34.0 12.8/12.5 59.3/60.5 10 mg/kg, 72 h Rat 13.0 10.6 88.5 10 mg/kg, 168 h Rat 28.6 22.5 62.4 30 mg/kg, 168 h Rat 29.04 30.71 58.84 1000 mg/kg, 72 h Rat 16.7 17.5 89.6 1000 mg/kg, 168 h Rat 30.55 22.41 53.27 5.7e8.8 mg/kg, 120 h Rabbit 7e11 ND 80e97 Single dose, ipb 2.3e3.6 mg/kg, 120 h Rat 82.90 ND 6e14 Single dose, ivb 10 mg/kg, 168 h Rat 79.0 74.5 4.65 30 mg/kg, 168 h Rat 85.98 84.18 3.42 Single dose, imb 4 mg/animal, 168 h Monkey 89.9 ND ND Repeated dose, pob 10 mg/kg, 72 hc Rat 10.6 10.7 86.6 10 mg/kg, 168 hc Rat 30.9 23.1 61.0 30 mg/kg, 168 hc Rat 34.28 34.63 49.64 400 mg/goat, 120 h Goat ND 9.44 ND a Radioactivity collection times after dosing, hr. b ND ¼ Not Determined; po ¼ oral gavage ip ¼ intraperitoneal; iv ¼ intravenous; im c 14 days of unlabeled glyphosate dosing followed by single radiolabelled dose.Health Organization (WHO) International Agency for Research on Cancer (IARC) decision that glyphosate is a “probable” (IIA) human carcinogen (Guyton et al., 2015) did not include consideration of real-world glyphosate exposures and contradicts decades of in depth regulatory reviews (PMRA, 1991; U.S. EPA, 1993; EC, 2002; APVMA, 2013; U.S. EPA, 2013; PMRA, 2015) as well as previous WHO expert committee reviews that have consistently declared that glyphosate is not carcinogenic (WHO IPCS, 1994; WHO JMPR, 2004; WHO, 2005). Although the MAA report has called for a more in-depth evaluation of potential health concerns as a result of their reported detection of glyphosate in human breast milk and urine, this report illustrates how such concerns can be addressed by an integrated analysis of available animal and human toxicokinetic, biomonitoring, and toxicity data. 2. Animal toxicokinetic (absorption, distribution, metabolism, and elimination; ADME) data do not indicate that glyphosate has bioaccumulation potential in human breast milk The toxicokinetics and metabolism of glyphosate have been extensively characterized and reported in series of single- and multiple-dose studies conducted in rats and other animal species (Brewster et al., 1991; reviewed in Williams et al., 2000; WHO JMPR, 2004; Anadon et al., 2009). As shown in the various data- sets summarized in Table 1 (modified from WHO JMPR, 2004), these studies consistently demonstrate that approximately 20e30 percent of orally administered glyphosate (as 14C-glyphosate) is absorbed at doses between 6.7 and 1000 mg/kg, with the remainder of the ingested dose excreted in feces. Recovery of total administered radioactivity in urine and feces is largely complete by 48 h after dosing, indicating rapid clearance of absorbed (urine) and non-absorbed (fecal) glyphosate. Importantly, Howe et al. (1988, reviewed in WHO JMPR, 2004), using highly sensitive and specific chromatographic (cation-exchange and ion-pair HPLC) and spec- troscopic (1H-NMR, 13P-NMR and mass spectrometry) techniques, confirmed that glyphosate experienced very limited overall meta- bolism following oral 10 mg/kg doses in male and female rats. Glyphosate represented 98.5 to 99.3 percent of the totalose following oral or parenteral administration (modified from WHO JMPR, 2004). Tissues Reference Female Male Female 49e55 0.14e0.65 0.83e1.02 Colvin and Miller (1973) 80.3/91.2 NDb ND Davies (1996a) 88.7 0.59 0.49 Davies (1996a) 69.4 0.44 0.31 Ridley and Mirly (1988) 56.53 0.62 0.64 Powles (1992) 84.5 0.52 0.58 Davies (1996b) 60.37 0.47 0.40 Powles (1992) ND 0.1e1.2 ND Colvin and Miller (1973) ND 0.53e1.00 ND Colvin and Miller (1973) 8.3 1.27 1.09 Ridley and Mirly (1988) 1.48 1.35 1.09 Powles (1992) ND ND ND Maibach (1983) 90.7 0.46 0.41 Davies (1996c) 70.9 0.54 0.35 Ridley and Mirly (1988) 46.73 0.96 0.83 Powles (1992) 78.16 ND ND Powles (1994) ¼ intramuscular. J.S. Bus / Regulatory Toxicology and Pharmacology 73 (2015) 758e764760administered 14C-dose recovered in urine and feces; trace amounts of the metabolite aminomethylphosphonic acid (AMPA) were detected. The limited metabolism of glyphosate was attributed to gut microflora metabolism, because AMPA was not observed in urine following intravenous administration of glyphosate (Howe et al., 1988). An unpublished ADME study in bile-cannulated rats found that less than 0.08% of an oral 1 mg/kg glyphosate dose in rats was recovered in bile, indicating that excretion of glyphosate in the feces was due to a lack of absorption and not to enterohepatic elimination (Knowles and Mookherjee, 1996). The findings of these unpublished studies have been confirmed by Brewster et al. (1991), in which a similar oral versus fecal excretion pattern (36% urine, 51% feces), rapid clearance in urine and feces, and limited meta- bolism to AMPA (<0.1%) were observed following a single oral 10 mg/kg dose of glyphosate to rats. More recently, Anadon et al. (2009) reported an oral bioavailability of 23.2% in rats treated with 100 mg/kg glyphosate. A significant consideration affecting glyphosate systemic toxicokinetics is that glyphosate was shown to rapidly distribute into water-rich body compartments, after which flux was reversed as rapid excretion proceeded. This dynamic is clearly distinguished from bioaccumulation, where flux would not be rapidly reversed, and complete and rapid levels of excretion would not be overtly evident. The extensive and rapid clearance of unmetabolized glyphosate into urine significantly facilitates inter- pretation of human biomonitoring studies, in that 24-h urine col- lections are representative of total systemic daily doses of glyphosate. The glyphosate toxicokinetic profile described above is entirely consistent with a compound having no bioaccumulation potential. Repeated-dose (15-day) toxicokinetic studies conducted at 10 or 30 mg/kg/day glyphosate (Table 1) also demonstrate an absorption and excretion profile essentially identical to that of single-dose administration. This further reinforces the conclusion that glyph- osate lacks bioaccumulation potential. In addition, residual radio- activity remaining in the tissues was very low and equivalent in each tissue following either single- or repeated-dose administra- tion of 14C-glyphosate. The lack of glyphosate bioaccumulation potential in breast milk, a lipid (fat)-rich tissue, is supported by the representative single- or repeated-dose tissue distribution data obtained from the Ridley and Mirly (1988) toxicokinetic study in rats (Table 2; reviewed inWHO JMPR, 2004). Only trace amounts of radioactivity were detected in a spectrum of tissues collected 168 h after oral dosing of 14C-glyphosate. Residual radioactivity in blood,Table 2 Mean tissue concentrations of radioactivity (ppm) determined 168 h after single or repeated oral gavage administration of 10 or 1000 mg/kg14C-glyphosate in rats (modified from WHO JMPR, 2004)a. 10 mg/kg single dose 10 mg/kg repeated doseb 1000mg/kg single dose Tissue Male Female Male Female Male Female Whole blood 0.004 0.003 0.005 0.003 0.328 0.166 Liver 0.030 0.014 0.041 0.026 1.91 1.37 Brain 0.007 0.006 0.014 0.011 0.750 0.556 Kidney 0.022 0.013 0.033 0.020 1.94 1.35 Lung 0.015 0.012 0.012 0.017 1.54 1.13 Testes/Ovaries 0.003 0.003 0.005 0.008 0.363 0.572 Stomach 0.008 0.004 0.038 0.024 2.38 2.36 Colon 0.034 0.016 0.043 0.030 11.0 9.2 Bone 0.552 0.313 0.748 0.462 30.6 19.7 Muscle 0.002 0.002 0.003 0.002 0.262 0.214 Fat 0.004 0.003 0.006 0.006 0.418 0.457 Carcass 0.106 0.087 0.157 0.101 8.27 7.74 a Selected tissue data from Ridley and Mirly (1988); tissue concentrations (ppm) are rounded values. b 14 days of unlabeled daily gavage administration of glyphosate dosing followed by a single radiolabeled dose.the distribution compartment of glyphosate to distal systemic tis- sues/fluids, including breast milk, was generally low and equivalent to levels in all tissues, including abdominal fat. The only exception to this general observation was that residual radioactivity in bone was slightly higher than that seen in other tissues. An absence of glyphosate bioaccumulation in milk has been demonstrated directly in a distribution and mass balance investi- gation conducted in two lactating goats given repeated oral (capsule gavage) daily doses of approximately 8 and 6 mg/kg, respectively, of 14C-glyphosate for 5 and 3 consecutive days (un- published report of Powles 1994; reviewed in WHO JMPR, 2004). The goats were milked twice daily, and milk was pooled for determination of total daily glyphosatemilk excretion. Total excreta were collected over 24 h intervals after the first dose; the first goat was euthanized 23.5 h after the last dose, and the second goat was euthanized 8 h after the final dose. The overall excretion profile of radioactivity was similar to the results from rats, with approxi- mately 78% of dosed radioactivity recovered in feces and 12% in urine and cage-wash/debris in the first goat. Clearance of radioac- tivity from tissues after termination of dosing was rapid; only 0.05% of the administered dose was recovered in tissues 23.5 h after the last dose to the first goat. Again paralleling rat data, glyphosate represented 94e96 percent of the total radioactivity recovered in urine and feces, as confirmed by HPLC and Fourier-transform infrared spectroscopy; thin-layer chromatography tentatively identified AMPA as a minor urine and fecal metabolite. Importantly, total radioactivity recovered in milk over the course of the three to five daily treatments represented only 0.03% of the total adminis- tered dose in both goats. Peak concentrations of radioactivity in milk were 72 and 86 mg (glyphosate)/L (milk) for each respective goat and were less than the peak blood concentrations of 102e101 mg (glyphosate)/L (blood) observed 6e8 h after the last dose in the second goat. These data demonstrate that glyphosate is not selectively taken up and/or retained in breast milk relative to blood concentrations or other body tissues. The findings of Powles (1994) were replicated in a similar study in which three lactating goats were administered a combination of 14C/13C-radiolabeled glyphosate and AMPA (glyphosate average dose of 5 mg/kg/day; AMPA 0.5 mg/kg/day) for 5 consecutive days (Bodden 1988; reviewed in Williams et al., 2000). Peak total radioactivity con- centrations in pooled daily milk collections were the equivalent of 35, 38, and 86 mg glyphosate/L, and reached steady state by the second day of dosing. Using very precise radiolabel analysis, it was demonstrated that glyphosate collected in milk represented less than 0.01% of the total administered dose of glyphosate in all three goats, and the concentration of glyphosate decreased rapidly, from 38 mg/L to 6 mg/L during a 5-day post-treatment period in one goat. The goat data directly show that only small amounts of glyph- osate are distributed into breast milk, which, when coupled with rapid urinary clearance following termination of dosing in both rats and goats, further demonstrate that glyphosate does not bio- accumulate in tissues or breast milk. 3. Glyphosate breast milk concentrations reported in 3 of 10 women in the MAA report (Honeycutt and Rowlands, 2014) are biologically implausible when compared to systemic and external doses estimated from urine biomonitoring findings, or from blood concentrations resulting from such doses Systemic and/or external doses of glyphosate associated with occupational and incidental/dietary exposures to glyphosate have been investigated in two U.S.-based biomonitoring studies (Acquavella et al., 2004; Curwin et al., 2007; reviewed in Niemann et al., 2015). Niemann and coworkers (2015) summarized the daily average or maximum urine concentrations from these J.S. Bus / Regulatory Toxicology and Pharmacology 73 (2015) 758e764 761biomonitoring studies. They then developed associated estimates of mean and/or maximum doses of systemic (orally absorbed) and/ or total external dietary doses of glyphosate, and compared these dose estimates to those projected from the maximum urine con- centration. (18.8 mg/L) reported in the limited MAA report (Table 3). Using urine concentrations detected in 48 farmer/applicators in Minnesota and South Carolina during the day of, and the immediate 3 days after engaging in glyphosate agricultural spraying opera- tions (Acquavella et al., 2004), Niemann and coworkers (2015) calculated mean and maximum glyphosate systemic doses of 0.11 and 8.3 mg/kg/day, respectively (Table 3). Although the geometric mean systemic dose in farmer/applicators was low, the maximum dose in the Acquavella et al. (2004) study was measured in a single farmer who engaged in lengthy repairs to the boom sprayer, had evidence of spills while mixing and loading, and was observed not using protective gloves. The maximum urine concentration and systemic dose observed in a “child” (29 ppm; 0.8 mg/kg/day) was measured in the teenage sonwho assisted the same farmer (father). Acquavella and coworkers (2004) also reported urinary glyphosate detection in two spouses (4% of those measured) on the day of application, and calculated a maximum systemic dose of 0.04 mg/ kg/day. Both spouses and children lived near the glyphosate mix- ing/loading and spraying activities, so these individuals likely represented a high-exposure scenario relative to the general pop- ulation. This study also found that urine concentrations declined substantially during the immediate 3 day post-application period, which was consistent with findings in animal studies demon- strating rapid and complete urinary excretion of glyphosate. Curwin et al. (2007) measured glyphosate urine concentrations in Iowa farm and non-farm families within 1e5 days of a pesticide application operation. Urine concentrations were generally in the same range as those reported in Acquavella et al. (2004), andTable 3 Glyphosate concentrations in human urine samples (mean and maximum) and calculated are well below RfDs established as health protective, including to children: EPA (2003 (modified from Niemann et al., 2015). Study Study Participants Group Acquavella et al. (2004) 48 male farmers, spouses, 79 children from farm families Farmer Child Spouse Curwin et al. (2007) 47 farmers, 48 mothers, 117 children from farm and non-farm families in Iowa All Groups Mothers (a - Farm - Non-Far McQueen et al. (2012) 20 pregnant non-farm, Australia Pregnant w (non-biom composite Honeycutt and Rowlands (2014) 35 women, men, children, across US Across all Honeycutt and Rowlands (2014) 10 women, across US Infants, 5 a Systemic doses of glyphosate calculated by or using formula of Niemann et al. (2015 b Total external dietary daily doses of glyphosate calculated by Niemann et al. (2015) a dose of glyphosate was systemically absorbed. c Systemic dose calculations from Acquavella et al. (2004) which used bodyweights of i of application, and all but 1 was present at or participated in mixing/loading operations; m applicator. Glyphosate was detected only in 2 spouses. d Not Determined. e Not applicable. f External doses estimated from analysis of glyphosate in composite diets of the individ 2015) ¼ 0.2 mg/kg/day (mean) and 1.0 mg/kg/day (maximum). g Maximum 166 mg/L breast milk concentration is highly implausible based on glyphos (see text); estimated maximum dose of 33 mg/kg/day is highly implausible, yet still welltranslated to essentially the same highest mean systemic dose of 0.1 mg/kg/day and a highest mean external dietary dose of 0.5 mg/ kg/day (Table 3). Glyphosate urine concentrations were essentially the same in farm and non-farm families, and were not affected by whether the father engaged directly in glyphosate spraying oper- ations. For farmmothers, urinary glyphosate concentrations did not correlate with farm size, number of acres or amount applied, or the number of days since the last applications, possibly consistent with rapid clearance from the body and the 1e5 day interval windows between glyphosate applications and urine sampling. The maximum urine concentration described in the MAA report (18.8 mg/L; Honeycutt and Rowlands, 2014) was in the range of the maximumvalues reported for the other two biomonitoring studies, and was converted to a maximum external daily dietary dose of 3.3 mg/kg/day and a 0.66 mg/kg/day systemic dose (Table 3). The geographic location data (state and zip code) described in the MAA report indicate that none of the three women with detectable glyphosate in breast milk were simultaneously evaluated for glyphosate in urine. However, given the glyphosate toxicokinetic and physico-chemical property considerations outlined below, it is highly unlikely that the three women with detected glyphosate in breast milk could have had urine concentrations correlating to blood concentrations sufficient to result in the reported breast milk concentrations. In addition, the 18.8 mg/L maximum urine con- centration reported by MAA likely represents a reasonable peak general-population urine value, in that it is in the range of maximum concentrations seen in the other two more compre- hensive biomonitoring studies, and is substantially higher than geometric mean urine glyphosate concentrations reported for exposure scenarios encompassing direct engagement in glyphosate application operations or families living in proximity to such ac- tivities (Table 3).estimates of systemic and external doses. All estimated systemic and external doses ) ¼ 1,750 mg/kg/day; European Food Safety Authority (proposed) ¼ 500 mg/kg/day Urine concentrations (mg/L) Estimated systemic dosea or external doseb (mg/kg/day) Geometric mean Maximum Mean Maximum 3.2 233 0.11a 8.3a (4.0c) NDd 29c NDd 0.8c NDd 3c NDd 0.04c 1.1e2.7 18 NDd 0.1a (adult) ll) 1.1e1.5 NDd 0.5b (adult) 1.1e1.5 11 NDd 0.17a m 1.2 5 NDd 0.36a omen onitored; food analysis) NAe NAe 1f 5f groups NDd 18.8 NDd 0.66a 3.3b Breast Milk Concentration (mg/L) External Dose to 5 kg Infant (mg/kg/day) kg body weight NDd 166g NDd 33g ); assume 2 L of urine per day and 60 kg body weight. ssuming, using data from animal toxicokinetic studies, that 20% of the external daily ndividual study participants. 9 of 78 Children had detectable urine glyphosate on day aximum urine concentration observed in a single teenage child who assisted father/ ual women; systemic dose assumes 20% absorption of external dose (Niemann et al., ate animal toxicokinetic, physico-chemical property and human biomonitoring data below any country's acceptable daily intake for humans including nursing children. J.S. Bus / Regulatory Toxicology and Pharmacology 73 (2015) 758e764762Importantly, integration of toxicokinetic data from goat and rat studies with urine biomonitoring data reported by MAA reveals that the concentrations of glyphosate reported in breastmilk (76e166 mg/ L) of 3 of 10 women in the MAA report are highly implausible. The 6 and8mg/kgdosesused in thegoat toxicokinetic study (Powles1994), despite being 1,818- and 2,424-fold higher than the maximum external glyphosate dose of 3.3 mg/kg/day estimated from the MAA urine biomonitoring data (Table 3), resulted in peak milk concen- trations (72 and 86 mg/L) approximating those described in the MAA report. In a rat toxicokinetic study (Brewster et al., 1991), glyphosate blood concentration peaked at 0.38% of the administered dose of 10 mg/kg 2 h after dosing, which results in an estimated blood con- centration of 0.475 mg/mL (10 mg/kg dose to 0.125 kg rats¼ 1.25 mg administered dose; 0.38% of dose¼ 0.00475mg¼ 4.75 mg; rat blood volume ¼ 8% body weight ¼ 10 mL; blood concentration ¼ 4.75 mg/ 10 mL ¼ 0.475 mg/mL ¼ 475 mg/L). The 475 mg/L peak blood concen- tration in rats, which is an expectedly higher blood concentration than theapproximate100mg/Lpeakobserved inagoatdosedat6mg/ kg, converts to an estimated milk concentration in rats of about 380 mg/L (assuming a parallel ratio of peak blood:milk concentration observed for goats of 1:0.8). Similarly to goats, the rat dose of 10mg/ kg (10,000 mg/kg), which is 3,030-fold higher than the maximum external dose of 3.3 mg/kg/day described in the MAA report, would result in only 2- to 5-fold estimated higher milk concentrations (380 mg/L vs 76e166 mg/L). Assuming similar oral absorption kinetics and blood:milk distribution ratios between goats, rats and humans, the goat and rat toxicokinetic data indicate the milk concentrations reported in theMAA studywould have required external glyphosate doses improbably higher (e.g., greater than 1,000-fold) than the low mg/kg/day glyphosate doses identified in the MAA and other urine biomonitoring datasets (Table 3). The above conclusions derived from comparisons of the animal toxicokinetic and human biomonitoring data also are consistent with findings from a dietary exposure investigation in pregnant women. Glyphosate residues in composites of foods directly consumed by the women translated to an estimated dietary daily dose of 0.001 mg/kg/day (1 mg/kg/day), in close agreement with exposures predicted from biomonitoring studies (McQueen et al., 2012, Table 3). This estimated human dose to pregnant women is approximately 6,000-fold lower than doses in lactating goats (5e8 mg/kg/day) that resulted in milk concentrations approxi- mately equal to the three reported breast milk concentrations in the MAA report. 4. Glyphosate physico-chemical properties are not consistent with a hypothesis that glyphosate bioaccumulates in breast milk The MAA report (Honeycutt and Rowlands, 2014) speculates that the concentrations of glyphosate detected in human breast milk are consistent with the behavior of certain lipophilic mole- cules that bioaccumulate in milk. However, such speculation is unwarranted when considered in context to glyphosate tox- icokinetics and physico-chemico properties. As noted in the sec- tions above, numerous toxicokinetic studies demonstrate that glyphosate does not exhibit bioaccumulation potential in the body or in breast milk. Whether administered in single or repeat doses, systemically absorbed glyphosate is rapidly and completely excreted into urine, generally within 48 h after termination of dosing (Tables 1 and 2). Studies in lactating goats confirm that only 0.01 to 0.03 percent of total administered glyphosate doses are recovered in breast milk. The lack of bioaccumulation demonstrated in animal tox- icokinetic studies, in large part, can be attributed to the fact that glyphosate is not lipophilic, distinguishing it from typicalbioaccumulative compounds. The pKa of the carboxylate moiety of glyphosate is 2.3, indicating that glyphosate will be essentially completely ionized at physiological pH. It is a well-established toxicokinetic principle that highly ionized substances are unlikely to distribute into lipophilic environments such as breast milk at concentrations greater than those in blood. In addition, the octa- nolewater partition coefficient (log Pow) for glyphosate is 3.2 at a pH of 5e9, indicating high hydrophilicity (partitions to water, not lipids). In contrast, a log Pow value of 6.41 has been reported for 2,2ʹ,4,4ʹ,5-pentachlorobiphenyl, a prototypical polychlorinated biphenyl (PCB) compound representative of extremely lipophilic molecules typically associated with bioaccumulation (Brodsky and Ballschmiter, 1988). 5. High-quality biomonitoring studies indicate that both systemically absorbed and external doses of glyphosate are well below U.S. and European reference-dose standards (e.g., acceptable daily intakes (ADIs) or reference doses (RfDs) established as protective of public health, including nursing children) The U.S. EPA has established a glyphosate RfD of 1.75 mg/kg/day based on maternal toxicity observed in rabbit developmental toxicity studies at a no-observed-effect level (NOEL) dose of 175 mg/kg/day (total uncertainty factor of 100; U.S. EPA, 1993). The European Food Safety Authority (EFSA) is currently re-evaluating glyphosate and has recommended an Acceptable Daily Intake (ADI) of 0.5 mg/kg/day (500 mg/kg/day) based on a rabbit maternal toxicity NOEL of 50mg/kg/day and application of a total uncertainty factor of 100 (Niemann et al., 2015). The biomonitoring data described in Table 3 indicate that all systemic and external doses of glyphosate estimated from glyph- osate urine concentrations are well below the established RfD and proposed ADI (reviewed in Niemann et al., 2015; for the EFSA ADI). Both the mean and maximum systemic doses of glyphosate calculated from Acquavella et al. (2004; Table 3), 0.11 or 8.3 mg/kg/ day, respectively, are well below the proposed EFSA ADI of 500 mg/ kg/day and the EPA RfD of 1,750 mg/kg. The highest mean external dietary dose of 0.5 mg/kg/day and highest mean systemic dose of 0.1 mg/kg/day calculated from the Curwin et al. (2007) data are likewise well below EPA and EFSA reference safety standards, as are respective external and systemic doses of 1.0 and 0.2 mg/kg/day estimated from food consumption data in pregnant women (Table 3; McQueen et al., 2012). Importantly, even the maximum external dose of 3.3 mg/kg/day and the systemic dose of 0.66 mg/kg/ day calculated from the MAA report remain well below both con- tinents' regulatory accepted or recommended safety standards. In addition, the 3.3 mg/kg/daymaximum external dose estimated from the MAA data is based on the assumption that glyphosate absorp- tion is limited to 20 percent based on oral absorption estimates obtained from animal gavage toxicokinetic studies (Niemann et al., 2015, Table 3). If glyphosate were to exhibit a higher percentage absorption in humans due to dietary intake as compared to gavage dosing in animals, e.g., 100 percent, the human external dose es- timate would be equivalent to the 0.66 mg/kg/day daily dose esti- mated from the urine biomonitoring sample. Finally, even if the highly implausible breast milk concentra- tions reported in the MAA report are accepted as real, the resulting doses to nursing children would still not exceed safety standards intended as health protective for infants. Thus, a 5-kg nursing in- fant consuming 1 L of breast milk per day containing 166 mg/L glyphosate would receive a daily glyphosate dose of 33 mg/kg/day, which is below both EPA and proposed EFSA reference safety doses. More realistically, given the analyses presented in the preceding sections, infant exposures are likely to be substantially lower. J.S. Bus / Regulatory Toxicology and Pharmacology 73 (2015) 758e764 7636. Conclusions The non-peer-reviewed and self-admitted preliminary MAA biomonitoring report of Honeycutt and Rowlands (2014) does not support the conclusion that glyphosate concentrations, detected in either urine or breast milk samples obtained from a small number of individuals across the U.S., present a health risk to the public, including nursing children. In contrast, daily external and/or sys- temically absorbed doses of glyphosate estimated from two more robust biomonitoring studies (Acquavella et al., 2004; Curwin et al., 2007) indicate that daily glyphosate doses are substantially below health-protective reference standards (ADIs; RfDs), as were all glyphosate doses estimated from the limited MAA urine bio- monitoring dataset. The MAA report also suggested that glyphosate bioaccumulates in breast milk, based on detection of relatively high glyphosate concentrations in breast milk samples from 3 of 10 women. How- ever, the concentrations of glyphosate reported in breast milk are highly implausible and are inconsistent with animal toxicokinetic data demonstrating that glyphosate is not preferentially distributed to or bioaccumulated in milk, and is rapidly cleared from the body after either single- or multiple-dose administration. The tox- icokinetic studies also indicate that glyphosate exposures esti- mated from urine biomonitoring fall thousands-of-fold short of external doses necessary to plausibly produce blood concentrations sufficient to result in the breast milk concentrations described in the MAA report. Finally, the physico-chemical properties of glyphosate are highly consistent with toxicokinetic evidence demonstrating a low distribution to, and no bioaccumulation in, breast milk. In conclusion, integration of an array of animal toxicokinetic, human biomonitoring/exposure, and physico-chemical property data for glyphosate demonstrates that glyphosate does not bio- accumulate in human breast milk, and that the high concentrations of glyphosate reported in 3 of 10 women in the MAA report are highly biologically implausible. Acknowledgments The author (JSB) received funding from the Joint Glyphosate Task Force, Raleigh, North Carolina, to conduct this analysis. Drafts of the manuscript were reviewed by members of the Joint Glyph- osate Task Force; however, the analyses and conclusions presented remain those of the author. The Joint Glyphosate Task Force is composed ofmember companies holding technical registrations for glyphosate, and was formed for the purpose of generating data in response to requirements from the US Environmental Protection Agency and Health Canada's Pest Management Regulatory Authority. Transparency document Transparency document related to this article can be found online at Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.yrtph.2015.10.022. References Acquavella, J.F., Alexander, B.H., Mandel, J.S., Gustin, C., Baker, B., Chapman, P., Bleeke, M., 2004. Glyphosate biomonitoring for farmers and their families: results from the farm family exposure study. Environ. Health Persp 112 (3), 321e326. Anadon, A., Martinez-Larranage, M.R., Martinez, M.A., Castellano, V.J., Martinez, M., Martin, M.T., Nozal, M.J., Bernal, J.L., 2009. Toxicokinetics of glyphosate and its metabolite aminomethyl phosphonic acid in rats. Toxicol. Lett. 190, 91e95. APVMA, 2013. A Review of the Earth Open Source (EOS) Report “Roundup and Birth Defects: Is the Public Being Kept in the Dark”. Australian Pesticides andVeterinary Medicines Authority. http://archive.apvma.gov.au/news_media/ docs/glyphosate_scitox_review_july_2013.pdf. Aylward, L.L., Kirman, C.R., Schoeny, R., Portier, C.J., Hays, S.M., 2013. Evaluation of biomonitoring data from the CDC national exposure report in a risk assessment context: perspectives across chemicals. Environ. Health Perspect. 121, 287e294. Bodden, R.M., 1988. Metabolism Study of Synthetic 13C/14C-labeled Glyphosate and Aminomethylphosphonic Acid in Lactating Goats. Unpublished report. Hazleton Laboratories America, Inc, Madison, WI. Brewster, D.W., Warren, J., Hopkins, W.E., 1991. Metabolism of glyphosate in Spra- gueeDawley rats: tissue distribution, identification, and quantification of glyphosate-derived materials following a single oral dose. Fundam. Appl. Tox- icol. 17, 43e51. Brodsky, J., Ballschmiter, K., 1988. Reversed phase liquid chromatography of PCBs as a basis for the calculation of water solubility and log Kow for poly- chlorobiphenyls. Fresenius Z. fȕr Anal. Chem. 331, 295e301. CDC, 2009. Fourth National Report on Human Exposure to Environmental Chem- icals. Department of Health and Human Services, Centers for Disease Control and Prevention. Colvin, L.B., Miller, J.A., 1973. CP 67573 Residue and Metabolism. Part 8: the Gross Metabolism of N-phosphonomethylglycine-14C (CP 67573e14C) in the Labo- ratory Rat Following a Single Dose. Unpublished report No. 297, study No. 9-23- 760.06-7863, dated 15 June 1973, from Monsanto Commercial Products Co, Agricultural Division, Research Department. Submitted to WHO by Monsanto Int. Services SA, Brussels, Belgium. Curwin, B.D., Hein, M.J., Sanderson, W.T., Striley, S., Heederik, D., Kromhout, H., Reynolds, S.J., Alavanja, M.C., 2007. Urinary pesticide concentrations among children, mothers and fathers living in farm and non-farm households in Iowa. Ann. Occup. Hyg. 51 (1), 53e65. Davies, D.J., 1996a. Glyphosate Acid: Excretion and Tissue Retention of a Single Oral Dose (10mg/kg) in the Rat. Unpublished report No. CTL/P/4940, dated 26 April 1996, from Zeneca Agrochemicals, Central Toxicology Laboratory, Alderley Park, Macclesfield, Cheshire, England. Submitted to WHO by Syngenta Crop Protec- tion AG, Basel, Switzerland. Davies, D.J., 1996b. Glyphosate Acid: Excretion and Tissue Retention of a Single Oral Dose (1000 Mg/kg) in the Rat. Unpublished report No. CTL/P/4942, dated 19 June 1996, from Zeneca Agrochemicals, Central Toxicology Laboratory, Alderley Park, Macclesfield, Cheshire, England. Submitted to WHO by Syngenta Crop Protection AG, Basel, Switzerland. Davies, D.J., 1996c. Glyphosate Acid: Excretion and Tissue Retention of a Single Oral Dose (10 Mg/kg) in the Rat Following Repeat Dosing. Unpublished report No. CTL/P/4944, dated 22 May 1996, from Zeneca Agrochemicals, Central Toxi- cology Laboratory, Alderley Park, Macclesfield, Cheshire, England. Submitted to WHO by Syngenta Crop Protection AG, Basel, Switzerland. EC, 2002. Review Report for the Active Substance Glyphosate. European Commis- sion., Directorate E e Food Safety: Plant Health, Animal Health and Welfare, International Questions. E1-Plant health. Guyton, K.Z., Loomis, D., Grosse, Y., El Ghissassi, F., Benbrahim-Tallaa, L., Guha, N., Scoccianti, C., Mattock, H., Straif, K., 2015. Carcinogenicity of tetrachlorvinphos, parathion, malathion, diazinon, and glyphosate. Lancet Oncol. 16, 490e491. Honeycutt, Z., Rowlands, H., 2014. Glyphosate Testing Report: Findings in American Mothers' Breast Milk, Urine and Water. Unpublished report, dated 7 April 2014, available from: the websites of “Moms Across America” and “Sustainable Pulse.”. Howe, R.K., Chott, R.C., McClanahan, R.H., 1988. The Metabolism of Glyphosate in Sprague-Dawley Rats. Part II. Identification, Characterisation and Quantitation of Glyphosate and its Metabolites after Intravenous and Oral Administration. Unpublished report No. MSL-7206, study No. 206300, dated February 1988, from Monsanto Environmental Health Laboratory, St. Louis, Missouri, and Monsanto Life Sciences Research Center, Chesterfield, Missouri, USA. Submitted to WHO by Monsanto Int. Services SA, Brussels, Belgium. Kleinman, R., May 6, 2014. Mass General Pediactrics Chief Says Glyphosate, Used with Some GM Crops, No Danger in Breast Milk. Genetic Literacy Project. http:// www.geneticliteracyproject.org/2014/05/06/mass-general-pediatrics-chief- says-glyphosate-poses-no-danger-in-breast-milk/. Knowles, S.L., Mookherjee, C.R., October, 1996. 14C-Glyphosate: Absorption, Distri- bution, Metabolism and Excretion Following Oral Administration to the Rat. Unpublished report prepared for Mastra Industries Sdn Bhd, Report number 1413/2e1011. Maibach, H.I., 1983. (a) Elimination of 14C-glyphosate in Rhesus Monkeys Following a Single Parenteral Dose, (b) Percutaneous Absorption of 14C-glyphosate in Roundup Formulation in Rhesus Monkeys Following a Single Topical Dose. Unpublished report No. MA-81e349, dated 1 April 1983, from University of California. School of Medicine; San Francisco, California, USA. Submitted to WHO by Monsanto Int Services SA, Brussels, Belgium. McQueen, H., Callan, A.C., Hinwood, A.L., 2012. Estimating maternal and prenatal exposure to glyphosate in the community setting. Int. J. Hyg. Environ. Health 215, 570e576. NAS/NRC, 2006. Human Biomonitoring for Environmental Chemicals. National Academy of Sciences/National Research Council. National Academies Press, Washington DC. Niemann, L., Sielke, C., Pfeil, R., Solecki, R., 2015. A critical review of glyphosate finding in human urine samples and comparison with exposures of operators and consumers. J. Consum. Prot. Fd. Saf. doi 10:1007/s00003-14-0927-3. PMRA, 1991. Discussion Document Preharvest Application of Glyphosate (Roundup®) Herbicide. Pest Management Regulatory Authority, Health Canada. J.S. Bus / Regulatory Toxicology and Pharmacology 73 (2015) 758e764764PMRA, 13 April, 2015. Proposed Re-evaluation Decision PRVD2015-01 Glyphosate. Pest management regulatory authority, Health Canada. Powles, P., 1992. 14C-Glyphosate: Absorption and Distribution in the Ratdprelimi- nary Study. Unpublished report No. 6365e676/1, dated 10 June 1992, from Hazleton UK, Harrogate, England. Submitted to WHO by Cheminova A/S, Lemvig, Denmark. Powles, P., 1994. 14C-Glyphosate: Absorption, Distribution, Metabolism and Excre- tion Following Repeated Oral Administration to the Dairy Goat. Unpublished report No. 676/9e1011, dated 7 November 1994, from Hazleton Europe, Har- rogate, England. Submitted to WHO by Cheminova A/S, Lemvig, Denmark. Ridley, W.P., Mirly, K., 1988. The Metabolism of Glyphosate in Sprague-Dawley Rats. Part I. Excretion and Tissue Distribution of Glyphosate and its Metabolites Following Intravenous and Oral Administration. Unpublished report, study No. 86139, project No. ML-86e438, dated March 1988, from Monsanto Environ- mental Health Laboratory, St. Louis, Missouri, USA. Submitted to WHO by Monsanto Int. Services SA, Brussels, Belgium. U.S. EPA, 1993. Registration Eligibility Decision (RED) Glyphosate. U.S. EPA, 2013. III. Aggregate risk assessment and determination of safety. glypho- sate; pesticide tolerances. Fed. Regist./Vol. 78, No. 84/Wednesday, May 1, 2013,p. 25397e25399. U.S. EPA, 2014. Agency Response to Moms Across America Breast Milk Study and Binder. Letter to Zen L. Honeycutt (MAA) From Neil Anderson, Branch Chief, Risk Management and Implementation Branch I. Pesticide Re-evaluation Divi- sion, EPA. June 26, 2014. WHO IPCS, 1994. International programme on chemical safety. Environ. Health Criter. 159. Glyphosate. http://www.inchem.org/documents/ehc/ehc/ehc159. htm. WHO JMPR, 2004. Joint Meeting of the FAO Panel of Experts on Pesticide Residues in Food and the Environment and the WHO Core Assessment Group, Rome, Italy, 20-29 September 2004 (2004) Glyphosate, pp. 95e169. World Health Organization, WHO Press, Geneva 2006. WHO, 2005. Glyphosate and AMPA in Drinking-water. Background Document for Development of WHO Guidelines for Drinking-water Quality. http://www.who. int/water_sanitation_health/dwq/chemicals/glyphosateampa290605.pdf?ua¼1. Williams, G.M., Kroes, R., Munro, I.C., 2000. Safety evaluation and risk assessment of the herbicide Roundup and its active ingredient, glyphosate, for humans. Regul. Toxicol. Pharmacol. 31, 117e165. EXHIBIT B lable at ScienceDirect Regulatory Toxicology and Pharmacology 86 (2017) 157e166Contents lists avaiRegulatory Toxicology and Pharmacology journal homepage: www.elsevier .com/locate/yrtph ,~. R.-,gulatory Toxicology and l'hannacology ;.:---===-~ ~I CrossMark IARC use of oxidative stress as key mode of action characteristic for facilitating cancer classification: Glyphosate case example illustrating a lack of robustness in interpretative implementation James S. Bus Exponent, Inc., 1800 Diagonal Road, Suite 500, Alexandria, VA 22314, United Statesa r t i c l e i n f o Article history: Received 27 January 2017 Received in revised form 28 February 2017 Accepted 2 March 2017 Available online 6 March 2017 Keywords: Glyphosate IARC Oxidative stress Mechanism of action Mode of action Cancer classificationE-mail address: jbus@exponent.com. http://dx.doi.org/10.1016/j.yrtph.2017.03.004 0273-2300/© 2017 Elsevier Inc. All rights reserved.a b s t r a c t The International Agency for Research on Cancer (IARC) has formulated 10 key characteristics of human carcinogens to incorporate mechanistic data into cancer hazard classifications. The analysis used glyphosate as a case example to examine the robustness of IARC's determination of oxidative stress as “strong” evidence supporting a plausible cancer mechanism in humans. The IARC analysis primarily relied on 14 human/mammalian studies; 19 non-mammalian studies were uninformative of human cancer given the broad spectrum of test species and extensive use of formulations and aquatic testing. The mammalian studies had substantial experimental limitations for informing cancer mechanism including use of: single doses and time points; cytotoxic/toxic test doses; tissues not identified as po- tential cancer targets; glyphosate formulations or mixtures; technically limited oxidative stress bio- markers. The doses were many orders of magnitude higher than human exposures determined in human biomonitoring studies. The glyphosate case example reveals that the IARC evaluation fell substantially short of “strong” supporting evidence of oxidative stress as a plausible human cancer mechanism, and suggests that other IARC monographs relying on the 10 key characteristics approach should be similarly examined for a lack of robust data integration fundamental to reasonable mode of action evaluations. © 2017 Elsevier Inc. All rights reserved.1. Introduction The International Agency for Research on Cancer (IARC), a unit of the World Health Organization, has prepared over 900 Mono- graphs using expert panel reviews to evaluate specific environ- mental chemical and physical agents, mixtures, occupational scenarios, and lifestyle factors for their potential to be human cancer hazards (IARC, 2015a). As emphasized in its Preamble (IARC, 2006), the Monographs do not characterize human cancer risks but rather are focused on classifying cancer hazards. Cancer risk, which considers cancer potential under conditions of defined exposures, is differentiated from cancer hazard which considers whether an agent is “capable of causing cancer under some circumstances”. Despite the articulation of the key functions, practices and value of risk-based approaches for health based decision-making beginning in the early 1980's (NAS, 1983), and with substantial expansion and refinement of this science to the current day, IARC has continued with hazard-based evaluations in its Monograph series under therationale that “new uses or unforeseen exposures could engender risks that are significantly higher [than those adequately addressed by existing risk-based evaluations] (IARC, 2006).” It is important to note that justification of hazard-based assessments directed at highly regulated compounds such as pesticides is of particularly questionable value because exposures resulting from new and/or expanded uses of these agents are highly controlled based on long- established risk-based regulatory practices. Thus, as a group, pes- ticides are unlikely to present the unforeseen higher exposure scenarios of concern intended to be addressed by IARC hazard- based evaluations. The IARC Preamble (IARC, 2006) outlines a high-level descrip- tion of the processes guiding preparation and generation of cancer hazard classifications. The classifications are positioned within 4 major groups: Group 1, “The agent is carcinogenic to humans”; Group 2A, “The agent is probably carcinogenic to humans; Group 2B, “The agent is possibly carcinogenic to humans”; Group 3, “The agent is not classifiable as to its carcinogenicity to humans”; and Group 4, “The agent is probably not carcinogenic to humans”. Mechanism or mode of action (MoA) information represents an increasingly pivotal data consideration within the IARC Monographs as a means J.S. Bus / Regulatory Toxicology and Pharmacology 86 (2017) 157e166158for informing the biological/toxicological extrapolation and plau- sibility of animal cancer findings to potential human cancer haz- ards. IARC requires use of MoA data, if available, as a means for possible upgrade or downgrade decisions within the classification groups that would otherwise be based on epidemiological or ani- mal cancer findings alone. IARC has recognized that use of MoA information in hazard evaluations has been substantially challenged in recent years by the growing complexity and volume of these data, and thus has developed proposed processes intended to facilitate more trans- parent, consistent, objective, focused and rapid evaluation of this information. The primary IARC MoA evaluation strategy has been articulated in a recent review by Smith et al. (2016), and is centered on IARC's development of 10 key MoA characteristics of human carcinogens. Those characteristics are then used to systematically organize and summarize MoA data for intended subsequent conceptualization and integration into MoA hypotheses examining the potential operational plausibility of the MoA(s) in humans. It has become increasingly clear that one of the 10 key characteristics, oxidative stress, has emerged as a characteristic providing “strong” evidence of humanMoA plausibility across several pesticide cancer hazard evaluations. Thus, Monographs 112, 113 and 115 have attributed oxidative stress to glyphosate, malathion, DDT, diazinon, 2,4-dichlorophenoxyacetic acid (2,4-D), N,N-dimethyformamide, hydrazine, and tetrabromobisphenol a (Guyton et al., 2015; Loomis et al., 2015; Grosse et al., 2016). The objective of this analysis is to examine how IARC has implemented the MoA evaluation expectations associated with use of the 10 key characteristics as described in Smith et al. (2016). The glyphosate Monograph 112 provides a specific case example illus- trating how the 10 key characteristics were considered in actual practice, and whether the conclusion that glyphosate exhibited “strong” evidence of oxidative stress is justified by the MoA data presented. A review of the glyphosate MoA analysis thus provides general insight into the validity of other IARC MoA conclusions in which oxidative stress was asserted as “strong” evidence of a plausible activity in humans. 1.1. IARC approach to development of the 10 key characteristics of human carcinogens as applied to oxidative stress Smith et al. (2016) summarized the conclusions of two work- shops conducted by IARC in 2012 in which the participants “extensive debated” potential mechanisms associated with sub- stances identified as IARC Group 1 carcinogens. The workshop discussions ultimately condensed the spectrum of MoA properties that the Group 1 carcinogensmight exhibit to 10 key characteristics of human carcinogens, one of which was “oxidative stress”. The other nine characteristics of human carcinogens were identified as: is electrophilic or metabolically activated to such; genotoxicity; alters DNA repair or genomic stability; induces epigenetic alter- ations; causes chronic inflammation; is immunosuppressive; modulates receptor-mediated effects; causes immortalization; and alters cell proliferation, death or nutrient supply. Interestingly, however, Smith et al. (2016) noted that oxidative stress is differ- entiated from the other 9 characteristics in that “… oxidative stress is not unique to cancer induction and is associatedwith a number of chronic diseases and pathological conditions [that are not associ- ated with cancer outcomes].” This acknowledgment raises the caution that mere identification of chemical-specific oxidative stress is not sufficient evidence per se to justify a cancer outcome, and that such data must be integrated into reasonable cancer MoA hypotheses consistent and coherent with plausible human mech- anisms of cancer. IARC (Smith et al., 2016) importantly emphasized that the 10 keycharacteristics were “not mechanisms in and of themselves nor are they adverse outcome pathways”, but rather were intended to facilitate the identification, organization and summarization of MoA data that must be subsequently integrated into human plau- sible MoA constructs. The latter presumption was specifically illustrated with benzene and PCBs case examples in which oxida- tive stress was integrated and positioned within overall sets of key events inclusive of other of the 10 characteristics. The overall data integration thus identified data-based hypotheses for examining the plausibility of potential human cancer outcomes. The case ex- amples presented in Smith et al. (2016) made it abundantly clear that simple collating and summarizing of pertinent MoA data into any of the 10 key characteristics bins was only a first step in a MoA analysis, and that the mere presence of studies in any given MoA bin is not sufficient to flag any single characteristic as plausible evidence of human carcinogenicity. Importantly, Smith et al. (2016) also identified some critical MoA considerations that are necessary to form rational MoA hypotheses from data integrated within and across the individual 10 charac- teristic bins. Thus, these authors emphasized that dose and species relevance, temporality of the MoA events to cancer outcome, and other MoA elements long-representing the foundations of MoA framework analyses (Sonich-Mullin et al., 2001; Boobis et al., 2006; Meek et al., 2013) are essential as evidentiary support for the plausibility of human-function of a hypothesized MoA. The recog- nition that dose is an intrinsic mechanistic consideration distinctly differentiated MoA evaluations from IARC's overall operational philosophy that hazard-based assessments are generally per- formed independent of substantive dose/exposure considerations. Indeed, Smith et al. (2016) specifically emphasized this differenti- ation by stating that it was important “… to consider whether key [MoA] characteristics of carcinogens are apparent upon exposures that are relevant to human health” (citing Thomas et al., 2013; emphasis added). The Smith et al. (2016) report cites Kushman et al. (2013; sharing two co-authors with Smith) as an example of a prototype data se- lection and MoA binning process in which the large and complex literature database for diethylhexylphthalate was winnowed to a dataset manageable for MoA evaluations. That selection process, which is similar in concept to that proposed by IARC (Smith et al., 2016), proactively screened out studies from MoA consideration if a study did not: 1) specifically study the chemical of interest; 2) was not conducted in the species or target organ(s)/tissue of interest, i.e., that exhibiting the cancer response; or 3) did not employ an appropriate dose/concentration design, e.g., was a single dose study. Not specifically addressed by either Kushman et al. (2013) or Smith et al. (2016) is exclusion based on studies employing only a single time point, a consideration emphasized as particularly important for investigations of oxidative stress (Halliwell and Whiteman, 2004). Both single dose and/or single time point studies cannot rule out if oxidative stress is secondary to an excessively high dose and/or single measurement time point that is outside of the dose and/or time ranges of alternative MoA(s) serving as the true drivers of the cancer outcomes. Smith et al. (2016) also note that their proposed data selection and analysis approach “may be difficult to translate to agents with limited mechanistic evidence,” in part because the availability of data for just a few of the 10 characteristics excludes the possibility that endpoints not specifically examined might otherwise offer alternative MoA considerations. Although the IARC Preamble (2006) notes that monographs are not intended to present comprehensive reviews of the literature, the glyphosate mono- graph does not provide any description of the outputs of literature searches aimed at identifying the totality of mechanistic database, thus making it uncertain as to the potential number, or quality, of J.S. Bus / Regulatory Toxicology and Pharmacology 86 (2017) 157e166 159citations addressing oxidative stress. However, it is reasonable to assume that the limited number of human cell and non-human mammalian studies identified in the glyphosate monograph rep- resented those most informative of an oxidative stress mechanism. There are several additional considerations that suggest caution in universal attribution of oxidative stress as a plausible primary hypothesis accounting for cancer in humans. There is ample liter- ature demonstrating oxidative stress as associated with biological pathways related to potential cancer outcomes (Halliwell, 2007), although as previously observed by Smith et al. (2016), oxidative stress is commonly associated with non-cancer disease outcomes as well. However, the actual literature cited in Smith et al. (2016) in support of oxidative stress as a key characteristic of human car- cinogens is primarily limited to two review papers (Klaunig et al., 2011; Berquist and Wilson, 2012). In fact, Klaunig et al. (2011) pointed out that, dependent on dose, oxidative stress can serve as both an anti-carcinogenic MoA (due to induction of oxidative repair defense mechanisms) or a possible carcinogenic MoA (due to excessive DNA damage, cell replication, etc.). Furthermore, the Smith et al. analysis was limited to evaluation of IARC Group 1 carcinogens, but did not examine if oxidative stress also correlated with lesser, and particularly Group 3, carcinogens. A quick literature survey of these agents indicates that oxidative stress responses similar to that reported for the glyphosate case examined herein are also associated with these lesser cancer classification groups. The likelihood that oxidative stress and 6 other of the 10 key characteristics may not be adequate to differentiate a true carcin- ogen from a likely non-carcinogen has been further evaluated in an analysis of chemicals classified by the US Environmental Protection Agency as “probable” or “not likely to be” human carcinogens and for which high-throughput-screening MoA data (ToxCast™) consistent with 7 of the 10 IARC key characteristics also was available (Becker et al., manuscript submitted, 2017). Modeling of these data demonstrated that the 7 characteristics were no better than chance for predicting positive human cancer outcomes (3 characteristics were not examined because ToxCast™ datawere not available: alters DNA repair or genomic stability; is immunosup- pressive; and causes cell immortalization). Finally, the IARC anal- ysis did not consider counterfactual data directly challenging the conclusion that oxidative stress alone is a MoA sufficient to induce tumorigenicity. Such counterfactuals include the bipyridylium herbicides paraquat and diquat. Both agents represent prime ex- amples of an oxidative stress mechanism in that their primary in vivo metabolism consists of redox cycling, i.e., spinning offTable 1 Summary of IARC citations used as evidence of “strong” evidence of oxidative stress. IARC citations (type) Citations (number) Tested as formulation or mixture (number) Oxidative stress limited to formulation or mixture Human (in vitro) 7 4c 4 Non-human (mammalian) 7 7 5 Non-mammalian 19 19e 13f a IARC (2015a,b) concluded “sufficient evidence” of cancer in animals based primarily o in 1 of 2 studies for each tumor type); in addition, rats exhibited increased pancreatic isl adenoma and thyroid c-cell adenoma (females, 1/4 relied on studies, no progression to car et al. (2016) concluded no evidence of glyphosate-related tumorigenicity (included stud b Only a single method used to determine evidence of oxidative stress and/or inclu dichlorodihydrofluorescein for detection of H2O2). c Human in vitro: 2 of 4 also included glyphosate alone; non-human (mammalian): 2 d Includes studies using intraperitoneal administration. e IARC (2015a,b) describes 21 citations; 1 was review, and 1 was erratum to another c f No or equivocal evidence of oxidative stress reported in 6 of the cited studies. g Includes one mixture study of multiple pesticides, making attribution of oxidative stsuperoxide radicals through repetitive single electron cellular reduction/oxidation reactions of the parent paraquat and diquat compounds (Stancliffe and Pirie, 1971; Bus and Gibson, 1984). Yet, neither of these agents is a rodent carcinogen (EC, 1996; EPA, 1997; EPA, 1995). While the preceding addresses the importance of experimental design and data integration elements in effective MoA analyses, it also must not be overlooked that interpretation of oxidative stress datasets are highly dependent on the methodology used to assess it. For example, the glyphosate Monograph points out that because of experimental limitations associated with use of dichlorodihyrofluorescein-diacetate as a biomarker of oxidative stress (Halliwell and Whiteman, 2004), studies examining this response alone should be interpreted with caution. The same cautionary note has been sounded regarding sole use of a single method to examine oxidative damage to DNA, and particularly if measured by 8-hydroxydeoxyguanosine (8OHdG) adduct forma- tion or COMET assays modified with enzyme co-treatments to specifically cleave DNA at sites of oxidized purines or pyrimidines (Halliwell and Whiteman, 2004; Cemeli et al., 2009; Speit et al., 2015). Speit et al. (2015) have further emphasized that enzyme- modified COMET assay findings cannot be concluded as positive evidence of oxidative stress if observed at cytotoxic and/or necrotic test concentrations/doses, and thus dose-response data excluding such toxicity are necessary to rule out this confounder. Cemeli et al. (2009) also have cautioned that in vitro Total Antioxidant Capacity (TAC) assays do not extrapolate well to similar outcomes in in vivo treatments, and thus have questionable human relevance. Impor- tantly, these reviews of oxidative stress methods emphasize the experimental challenges and pitfalls associated with measurement of oxidative stress, and that there is no single reliable biomarker of oxidative stress and use of complementary endpoints are needed for effective assessment of this MoA. 1.2. Glyphosate case example: Has IARC followed its own guidance recommendations in concluding “strong” evidence of oxidative stress? The IARC evidence for oxidative stress is organized into 3 sub- bins consisting of studies in in vitro human cells, and non-human mammalian and non-mammalian studies. This case example is primarily focused on the first two considerations (Tables 1e3). However, the totality of the studies described in the 3 oxidative stress sub-bins illustrates that the studies are limited in numberSingle dose Single time point Non-relevant tissue or speciesa Single and/or limited method(s) for detection of oxidative stressb 4 7 7 7 6 6 5 7d 7 10 NA 8g n renal tubule adenoma/carcinoma and hemangiosarcomas in male mice (response et cell adenoma (males, 2/4 relied on studies, no dose-response) and hepatocellular cinomas). EPA (2013); EFSA (2016); Greim et al. (2015); Tarone (2016); andWilliams ies not evaluated by IARC). ded method(s) with recognized limitations for detection of oxidative stress (e.g., of 7 included glyphosate alone. ited paper. ress to glyphosate impossible. Table 2 Summary of IARC citations offered as evidence oxidative stress in human (in vitro) studies. Effects Comments Citations: human in vitro Test systema Dose parametersb ROSc detection LPOc detection Anti- oxidantc enzymes Anti-oxidantc impact Gehin et al. (2005) HaCaT skin; vitamins E and/or C pre- and co-incubation; G or F 24 h G IC50 ¼ 22 mM ¼ 3718 mg/ml 24 h F IC50 ¼ 19.5 mM ¼ 3295 mg/ml ND ND ND Minor attenuation of F LC50 by vitamin C þ E; no change in G LC50 Unrealistic dose relative to animal dosesb; toxicity at LC50 not informative of oxidant stress Elie-Caille et al. (2010) HaCaT skin; G only 0.5 h G LC50 ¼ 50 mM ¼ 8450 mg/ml [DCFH-DA ND ND ND Unrealistic dose relative to animal dosesb; toxicity examined at LC50 George and Shukla (2013) HaCaT skin F only 24 h F 0.01e0.1 mM ¼ 1.69e16.9 mg/ml [DCFH-DA ND YSOD YROS with NAC Single ROS method; YSOD inconsistent with other citations Chaufan et al. (2014) HepG2 liver; G or F 24 h G LC20 ¼ 900 mg/ml 24 h F LC50 ¼ 40 mg/ml G (-)DCFH- DA F [DCFH-DA ND G (-) SOD, CAT F [SOD, (-)CAT G (-)GSH F [GSH No effects with G, even at unrealistic dose relative to animalsb; F effects at LC50; IARC did not identify as DCFH-DA ROS method Coalova et al. (2014) HepG2 liver; F only 24 h F LC50 ¼ 376.4 mg/ml [DCFH-DA ND (-)SOD [CAT [GSH Unrealistic dose relative to animal dosesb; F effects at LC50; differing F SOD and CAT responses compared to Chaufan et al. (same lab); IARC did not identify as DCFH-DA ROS method Mladinic et al. (2009) Primary Lymphocytes ± S9; G only 4 h G 0.5e580 mg/ml (-) -S9 COMET [þS9 COMET at 580 mg/ml [ TBARS ± S9 at 580 mg/ml ND [TAC ± S9 ¼ 580 mg/ml Unrealistic dose relative to animal dosesb; COMET ROS responses inconsistent with TBARS; 580 mg/ml [ cell necrosis Kwiatkowska et al. (2014) Erythrocytes; NAC pre- Incubation; G only 1 h G 0.01e5 mM ¼ 1.69e845 mg/ml [R123 at 0.25 mM ¼ 42 mg/ml ND ND YROS with 30m preincubation 0.1 mM NAC Unrealistic dose relative to animal dosesb a None of test systems equivalent to target organs identified by IARC in animals (kidney and pancreas); G ¼ glyphosate; F ¼ glyphosate formulation. b For comparative purposes, Anadon et al. (2009) reported mean plasma Cmax of 4.6 mg/ml in rats orally administered 400 mg/kg glyphosate. c ROS ¼ Reactive Oxygen Species; (DCFH-DA ¼ dichlorodihydrofluorescein diacetate; COMET ¼ enzyme-modified assay for base oxidation; R123 ¼ dihydrorhodamine123); LPO ¼ Lipid Peroxidation; TBARS ¼ thiobarbituric acid reacting substances; Anti-oxidant enzymes: SOD ¼ superoxide dismutase; CAT ¼ catalase; Anti-oxidant impact: NAC ¼ n-acetylcysteine; GSH ¼ reduced glutathione; TAC ¼ Total Antioxidant Capacity measured as Ferric Reducing Ability of Plasma (FRAP); [ ¼ increased; Y ¼ decreased; (-) ¼ no change; ND ¼ not determined. J.S.Bus / Regulatory Toxicology and Pharm acology 86 (2017) 157 e 166 160 Table 3 Summary of IARC citations offered as evidence oxidative stress in non-human mammalian studies. Effectsc Comments Citations: non-human mammalian In vivo Test systema Dose parametersb ROS detection LPO detection Anti-oxidant enzymes Anti-oxidant impact Astiz et al. (2009) Rats; G or M of: (Zþ G), (DþG), (ZþGþD); Tissues: L,B,K,P G ¼ 10 mg/kg/day ip, 3X/wk, 5 wks [B,P NOx (-)L,K NOx [L,B,K,P TBARS (-)P prot. carbonyls; L: YSOD, (-) CAT, GPx, GR; B: YSOD, [CAT; (-) GPx, GR; K: (-)SOD, CAT, GPx, GR TGSH: [P, (-)B,L,K; GSH: [P, (-)B; TAC: YP; Vitamin E: YL,B Mixture findings not described due to lack of G-specific relevance; Not all Effects determined in all tissues; Variable Effects, [,Y,(-) across tissues and Effects Bolognesi et al. (1997) Mice; G or F Tissues: L,K G ¼ 300 mg/kg ip; F ¼ 270 mg/kg G equivalent, ip; Single dose, 8 and 24 h times G: 8OHdG, [L, 24 h; (-) K; F: 8OHdG, (-)L, [K, 8, 24 h ND ND ND Single biomarker of ROS; Inappropriate route of dosing; No effect of G in kidney; IARC erroneously reports G increase in kidney; see (-)K for G but [K for F George et al. (2010) Mice; F only; Tissue: skin F ¼ 50 mg/kg, single dose; Skin application ND ND YSOD; [PRXII ND Proteomic expression only; G systemic skin absorption ~ 1% Cavuşoglu et al., 2011 Mice; F only; Tissues: L,K F ¼ 50 mg/kg, ip, single dose; with 50 or 150 mg/kg Ginkgo biloba extract, 1X/day/5d before G dose and 3d after ND Malondialdehyde: L: [F only; [FþG.b. (both G.b. doses) K: [F only; [FþG.b. (both G.b. doses) ND GSH: L: YF only; YFþG.b. (both G.b. doses) K: YF only; YFþG.b. (both G.b. doses) F alone caused severe histological effects in L and K 3d after dosing, confounding LPO detection; G.biloba minimally attenuated F L and K effects; Jasper et al., 2012) Mice (M/F); F only; Tissue: L F ¼ 50 or 500 mg/kg/day, oral, 15d ND TBARS: L: [ M/F, both doses ND GSH: L: Y M, both doses; YF, high dose only F only; F caused significantly decreased (> 10%) body weights in M/F mice Cattani et al. (2014) Rat pups; F only; Tissue: pup B F ¼ 1% drinking water (0.38% G), dams GD5- LD15; LD15 pup brain incubated with 0.01% F, 30m ND TBARS: B[ G6PD: BY GSH: BY F caused approximately 20% decrease in body weights; Pup brain tissue incubated directly with 0.01% F after drinking water treatment of dams from GD5-LD15 Astiz et al. (2013) Rats; M only: G þ Z þ D; Tissue: testes, P M: G ¼ 10 mg/kg/day, ip, 5X/wk/5wk; Lipoic acid or vitamin E, ip, co-treatment ND TBARS: Testes: [ P: [ ND Attenuation of LPO in plasma and testes with Lipoic acid and vitamin E co-treatment Mixture only treatment e not interpretable for G-specific toxicity; Inappropriate route of dosing a G ¼ glyphosate only; F ¼ glyphosate formulation; L ¼ liver, B ¼ brain, K ¼ kidney, P ¼ plasma; M ¼ mixture, Z ¼ zineb, D ¼ dimethoate. b For context to animal doses, estimated human daily doses (from urine biomonitoring) for farmers directly using glyphosate: maximum ¼ 4 mg/kg/day, geometric mean ¼ 0.1 mg/kg/day; for farm families: maximum (spouses) ¼ 0.04 mg/kg/day, maximum (children) ¼ 0.8 mg/kg/day (Acquavella et al., 2004). Other studies have estimated maximum daily external glyphosate doses of 0.1e5 mg/kg/day (reviewed in Bus, 2015); ip ¼ intraperitoneal; GD ¼ gestation day, LD ¼ lactation day. c [¼ increase, Y¼ decrease, (-)¼ no change; ROS¼ Reactive Oxygen Species, NOx¼ reactive nitrogen species, 8OHdG¼ 8-hydroxydeoxyguanosine; LPO¼ Lipid Peroxidation, TBARS¼ thiobarbituric acid reacting substances; SOD ¼ superoxide dismutase, CAT ¼ catalase, GPx ¼ glutathione peroxidase, GR ¼ glutathione reductase, PRXII ¼ peroxyredoxin 2, G6PD ¼ glucose-6-phosphate dehydrogenase; TGSH ¼ total glutathione, GSH ¼ reduced glutathione, TAC ¼ Total Antioxidant Capacity measured as FRAP (ferric reducing ability of plasma); ND ¼ not determined. J.S.Bus / Regulatory Toxicology and Pharm acology 86 (2017) 157 e 166 161 J.S. Bus / Regulatory Toxicology and Pharmacology 86 (2017) 157e166162and suffer substantial experimental deficiencies relative to expec- tations of reasonably designed MoA studies inclusive of oxidative stress. The IARC conclusion of “sufficient” evidence of animal carcino- genicity is primarily based on its determination of positive trends in the incidence of renal tubule carcinomas in male mice in one early bioassay and in hemangiosarcomas in male mice in a second study. IARC also noted as supporting evidence an increased incidence of pancreatic islet cell adenomas and hepatocellular adenomas in male rats and a positive trend of thyroid C-cell adenomas in female rats (IARC, 2015b). However, IARC's interpretation of these cancer findings has been substantively challenged in the literature. Mul- tiple weight-of-evidence analyses have concluded that glyphosate is not an animal carcinogenwhen studies reviewed by IARC as well as additional unpublished high quality studies not reviewed by IARC were evaluated (Williams et al., 2000; EPA, 2013; EFSA, 2016; Greim et al., 2015; JMPR, 2016; Tarone, 2016; Williams et al., 2016). The lack of glyphosate carcinogenicity is also consistent with other reviews concluding that glyphosate is not genotoxic (Kier and Kirkland, 2013; Kier, 2015; Brusick et al., 2016), and with the IARC conclusion that the epidemiological evidence (non-Hodgkin's lymphoma) is “limited” (another weight-of-evidence analysis of the epidemiological data concluded there was no causal association for this cancer; Acquavella et al., 2016). Thus, a significant amount of caution must be raised as to the possibility that glyphosate MoA investigations represent an example of a mechanism(s) in search of an outcome, rather than the expected path of using mechanisms to ascertain the human relevance of cancer findings unequivocally demonstrated in animals. This latter point is not to say, however, that experimentally well-designed MoA studies, particularly when coupled to assessments of real-world human exposures, cannot be used to proactively screen chemicals not tested in conventional lifetime rodent bioassays for potential carcinogenic potential (Thomas et al., 2013). 1.2.1. In vitro studies of oxidative stress in human cells (Tables 1 and 2) Of the seven human tissue studies, 4 used a single test con- centration and all used a single time point. All were conducted in tissues/organs in which tumors were not observed in animals: 3 in HaCaT keratinocyte (skin) cells; 2 in a Hep2G liver cell line; 1 in primary human lymphocytes; and 1 in human erythrocytes (particularly disparate in biological handling of oxygen relative to other somatic/germ cells). Hep2G liver cells are not regarded as relevant to the IARC identified targets of mouse hemangiosarcomas or rat liver adenomas. Liver adenomas were identified in only 1 of 4 rat studies judged as having adequate design by IARC, and the positive interpretation of that study has been challenged (Williams et al., 2016). In addition, liver tumors were not observed in other unpublished high-quality rat studies not considered in the IARC review (described in reviews cited above); hemangiosarcomas originate from blood vessel endothelial cells, not hepatocytes. It should also be noted that 3 of the 7 studies used immortalized HaCaT cells that have been shown to contain mutated p53 (Lehman et al., 1993). The p53 gene controls anti- or pro-oxidant responses to oxidative stress, depending on intensity (Halliwell, 2007; Liu and Xu, 2011), cautioning that measurement of oxidative stress in such a cell line may not reflect oxidative stress responses in normal human cells. Of the 4 studies examining glyphosate formulations, 2 tested formulations only (Table 2). One of those, George and Shukla (2013), produced oxidative stress at the lowest test concentra- tions in the human tissue series (1.69e16.9 mg/ml in HaCaT cells). However, testing of formulations is particularly problematic and artifactual for in vitro studies given direct cell contact tosurfactants and other unidentified substances composing the formulations. In addition, the in vivo blood/organ dosimetry of the various formulation components is very unlikely to be a direct ratio to that of the formulation itself because of differential environmental fate and human toxicokinetics of formulation components, including glyphosate. The results of the George and Shukla (2013) study also suggest that the formulation tested by these investigators was particularly toxic in that the LC50 values of glyphosate and/or another formulation evaluated in HaCaT cells were substantially higher (Gehin et al., 2005; Elie-Caille et al., 2010). Such a conclusion is tempered, however, by the lack of dose-response analyses conducted for oxidative stress in these studies. The in vitro studies also consisted of a mix of oxidative stress methods, none of which were consistently examined across the studies, and for those that did, inconsistent results were found. For example, differing responses in superoxide dismutase or catalase activities in HepG2 studies were observed in studies originating from the same laboratory (Table 2; Chaufan et al., 2014; Coalova et al., 2014). In addition, 4 of the7 studies useddichlorodihydrofluorescein- diacetate, a questionable biomarker for quantitation of oxidative stress. Importantly, with the exception of one study testing only a formulation (George and Shukla, 2013), all of the remaining studies used excessively high concentrations of glyphosate itself or in for- mulations that are uninformative of an oxidative stress MoA in humans. Four of the 7 studies examined oxidative stress only at LC50 test concentrations. The use of such excessively toxic con- centrations renders any interpretative value of these studies essentially moot given the substantial disruption of cell functions occurring under conditions of 50% cell death. In addition, attribu- tion of changes in enzyme activity and/or cell antioxidant levels specifically to oxidative stress is highly questionable when no other biomarkers reflecting generalized cell toxicity were simultaneously examined in these studies. In other studies of this series, glyphosate increased an enzyme-modified-COMET response in primary lym- phocytes, but again only at a necrosis-inducing concentration of 580 mg/ml (Mladinic et al., 2009). Only one study (Kwiatkowska et al., 2014) demonstrated oxidative stress at a non-toxic dose of glyphosate alone, although this response was noted only in non- target erythrocytes. In contrast to these findings, however, oxida- tive stress was not observed when glyphosate alone was tested in HepG2 cells at an approximate LC20 dose of 900 mg/ml (Chaufan et al., 2014). The use of excessively high and mechanistically uninformative concentrations are further revealed when compared to glyphosate rodent toxicokinetic data in which a glyphosate plasma Cmax con- centration of 4.6 mg/ml was reported after a single 400 mg/kg gavage dose (Anadon et al., 2009). With the sole exception of the formulation-only study of George and Shukla (2013), concentra- tions eliciting biomarkers of oxidative stress ranged from 9 to 820- fold higher than this in vivo plasma concentration (Table 2). The excessively high concentrations are further illustrated by exam- ining the context of the lowest concentration of glyphosate alone resulting in oxidative stress, 42 mg/ml in erythrocytes (Kwiatkowska et al., 2014), to that of the rat plasma Cmax reported in Anadon et al. (2009). If the oral toxicokinetics of glyphosate are assumed as linear, the in vitro erythrocyte concentration would be equivalent to an approximate oral glyphosate dose of 3640 mg/kg (42 mg/ml is 9.1X > 4.6 mg/ml; 9.1 400 ¼ 3640 mg/kg). The correspondingly higher glyphosate concentrations inducing oxidative stress in 5 of 7 of the remaining tests would translate to massively higher oral dose equivalents of glyphosate (e.g., approximately 735 g/kg in Elie-Caille et al., 2010). J.S. Bus / Regulatory Toxicology and Pharmacology 86 (2017) 157e166 1631.2.2. In vivo studies in non-human mammalian species (Tables 1 and 3) The in vivo non-humanmammalian studies cited by IARC reflect many of the same substantive experimental deficiencies as those from the human in vitro series. All 7 of the cited studies included testing of either formulations (5) or mixtures (2), with only 2 studies also including evaluations of glyphosate alone (Astiz et al., 2009; Bolognesi et al., 1997). It is impossible to attribute oxida- tive stress specifically to glyphosate based on themixture studies in which glyphosatewas combined with zineb and dimethoate in that separate mixtures of zineb and dimethoate resulted in oxidative stress similar to that of the 3-chemical mixture (Astiz et al., 2009, 2013). Oxidative stress resulting from direct formulation treat- ment also is of limited value in supporting a glyphosate-specific MoA in humans. Other than possible occupational dermal expo- sures, chronic general population exposures consist almost entirely of dietary residues and would not involve systemic co-exposures to formulant agents at doses similar to those used in the experimental studies due to differential fate in the environment and disposition in humans. Five of 7 of the in vivo studies in this series were conducted by non-relevant intraperitoneal (ip; 4) or dermal (1) routes of administration. These dose routes are particularly problematic for investigation of human relevant MoAs of systemic glyphosate toxicity because its oral and dermal absorption is less than 30% and 1%, respectively (Williams et al., 2000). Thus, ip administration results in artifactually higher systemic doses relative to oral treat- ment, while systemic doses resulting from dermal applications will be substantially less than equivalent oral or ip doses. In addition, 6 of the 7 studies in this series are severely limited for MoA assess- ment purposes by use of single doses and/or time points, and 5 were conducted in cancer non-target tissues of liver, plasma, testes and brain. An examination of oxidative stress in developing rat brain (Cattani et al., 2014) also included an unrealistic post-sacrifice incubation of isolated brain tissuewith a 0.01% formulation dilution in addition to in-life treatment. Interpretation of oxidative stress in 3 of the formulation studies was confounded by the co-presence of significant toxicity at the time of evaluation. Cavuşoglu et al. (2011) reported severe histo- logical injury in both liver and kidney in mice at the time of oxidative stress measurement, while body weight losses of greater than 10%, exceeding a Maximum Tolerated Dose (MTD), were seen in the studies of Jasper et al. (2012) Bolognesi et al. (1997) was the only study which evaluated oxidative stress of glyphosate alone (albeit at a single dose) in a species and tissue (mouse kidney) identified as a carcinogen target by IARC. However, this study found increased 8OHdG in liver but not in the postulated target, kidney. Treatment with a formulation delivering approximately the same dose used in the glyphosate- alone experiment resulted in opposite responses (no change in liver; increase in kidney), thus demonstrating a lack of consistency relative to treatment with glyphosate dose. In addition to the substantive experimental limitations of the in vivo studies described above, a comparison of real-world human exposures to the experimental doses used in the IARC-cited studies indicates oxidative stress is not a reasonable cancer MoA hypoth- esis in humans based on exposure considerations alone (and particularly since, with the exception of IARC, multiple weight-of- evidence reviews have concluded glyphosate is not an animal or human carcinogen; see above). Biomonitoring studies of farm families during periods of active glyphosate applications reported maximum daily systemically absorbed doses of 4 mg/kg/day in farmers (geometric mean dose ¼ 0.1 mg/kg/day), 0.04 mg/kg/day in spouses and 0.8 mg/kg/day in children (Acquavella et al., 2004). The maximum systemic dose of 4 mg/kg/day in farmworkers is 2500-fold less than the daily 10 mg/kg/day intraperitoneal (systemic) dose reported in Astiz et al. (2009; Table 3). It is important to note, however, that a systemically-administered 10 mg/kg ip dose of glyphosate is approximately equivalent to a 50 mg/kg oral dose in that glyphosate oral absorption is estimated at approximately 20%; thus, general population oral dietary doses are substantially more differentiated from the ip dose used in the Astiz et al. study. Solomon (2016) has summarized systemic general population glyphosate dosesmeasured by urine biomonitoring as ranging from 0.0005 to 0.63 mg/kg/day, and operator/applicator doses from 0.013 to 4.6 mg/kg/day. These doses are very substantially differentiated from the doses used in the studies described in Table 3, e.g., the 300 mg/kg ip systemic dose of glyphosate alone reported in Bolognesi et al. (1997) is 468750- and 46875-higher than the maximum in range biomonitored doses from the general popula- tion and operators/applicators, respectively. 1.2.3. Non-mammalian (environmental) species data The IARC glyphosate Monograph (2015b) provides 21 citations describing oxidative stress in non-mammalian species, although only 19 included original experimental data (Slaninova et al., 2009; is a review; Marques et al., 2015; is an erratum to Marques et al., 2014). Glyphosate was tested as a formulation in 18 of the studies (Lushchak et al., 2009; Ferreira et al., 2010; Guilherme et al., 2010, 2012a, b; 2014a; Modesto and Martinez, 2010a,b; Cattaneo et al., 2011; Glusczak et al., 2011; de Menezes et al., 2011; Ortiz- Ordo~nez et al., 2011; Nwani et al., 2013; Marques et al., 2014, 2015; Sinhorin et al., 2014; Uren Webster et al., 2014; Costa et al., 2008); only 2 of the studies also included glyphosate alone (Guilherme et al., 2012a; Uren Webster et al., 2014). One study (Guilherme et al., 2014b) only evaluated an environmental metabolite of glyphosate, aminomethylphosphonic acid (AMPA). One study was a mixture only study of 8 pesticides including glyphosate in oysters (Geret et al., 2013), and thus provides no in- formation specific to glyphosate. Thus, the extensive use of for- mulations and highly varied test species across these studies, as well as use of a route of administration (water dose delivery to aquatic species) of unknown dosimetric relevance to real-world human oral exposures, renders these studies of very limited interpretative value in supporting plausible evidence of an oxida- tive stress cancer MoA in humans. Because of the substantive interpretative limitations of these non-mammalian studies relative to supporting the plausibility of oxidative stress as cancer MoA in humans, the details of these studies are not summarized in a tabular format. However, some general observations are useful regarding some of the findings reported in these studies. Of the 19 data-based studies, six originated from the same team of investigators, and most of these studies relied on enzyme- modified COMET assays in a native-captured European eel species to examine potential oxidative stress damage to purine and py- rimidine DNA bases (Guilherme et al., 2012a, b; 2014a, b; Marques et al., 2014). In contrast to the IARC (2015b) inference that these studies constituted supportive evidence of an oxidative stress MoA, the actual experimental details of this series of studies reveals that, although glyphosate formulations generally increased non-specific DNA damage in eels, evidence of oxidative stress-derived DNA damage was either not observed or was equivocal (e.g., effects limited to a single dose or time point). One formulation-only study (Guilherme et al., 2010) used a COMET assay that was not enzyme- modified for detection of oxidative-specific DNA damage, but due to an additional general lack of responsiveness of antioxidant de- fenses to glyphosate treatment, led to the conclusion that “… oxidative stress was not severe.” In another study which included testing of glyphosate alone, evidence of oxidative stress-induced DNA damage was unaffected or equivocal in erythrocytes (no J.S. Bus / Regulatory Toxicology and Pharmacology 86 (2017) 157e166164change in DNA oxidative damage at 2 time points in one enzyme- modified COMET assay; no change at 1 time point in another (different) enzyme-modified COMET assay, with increased damage noted only at the high dose at a second time point). The authors concluded that “… DNA oxidationwas not perceived as a dominant mechanism of damage” (Guilherme et al., 2012a). Interestingly, another study from this research team (Marques et al., 2014) demonstrated that a glyphosate formulation decreased oxidative damage to pyrimidine bases in eel liver using an enzyme-modified COMET assay, and these data were used to infer possible compen- satory induction of oxidant defense mechanisms by the formula- tion exposure (however, oxidant defenses were not experimentally measured). Evidence of oxidative stress from glyphosate-treatment alone in a zebrafish study (UrenWebster et al., 2014) was limited to a single- dose and single time point exposure to 10 mg/L glyphosate alone. This study reported contrasting findings in potential oxidant de- fense responses, with superoxide dismutase and catalase gene expression respectively down- and up-regulated in testes (a non- relevant target to IARC-postulated mammalian cancer-sensitive organs). The very limited experimental design of this study, how- ever, does not allow for the differential responses of these two oxidant-responsive genes to be interpreted as conclusive evidence of an oxidative stressMoA in this organ, and evenmore remotely, as an IARC-postulated plausible cancer MoA in humans. 2. Conclusions IARC has proposed use of the 10 key characteristics of human carcinogens as a means to effectively identify, organize and sum- marize mechanistic literature for subsequent integration into MoA analyses that inform their plausibility in mediating human cancer outcomes (Smith et al., 2016). Although IARC has emphasized that its cancer classifications are hazard focused (IARC, 2006), IARC ef- forts to refine incorporation of MoA data into its cancer hazard classifications have also noted the critical role of dose, species, temporality, and other considerations as intrinsic to the MoA evaluations (Smith et al., 2016). The 10 key characteristics of human carcinogens, therefore, are not intended to represent bins for collection of studies that can then be assumed as complete and meaningful MoA hypotheses absent further integrated analyses. Rather, the studies within and across the characteristic bins must be subsequently molded into hypotheses providing plausible evidentiary support of human carcinogenicity. In its review of the carcinogenic hazard of glyphosate (IARC, 2015b), IARC concluded that one of the 10 key characteristics of human carcinogens, oxidative stress, constituted “strong” evidence that this endpoint could be plausibly operational in humans and thus provided supporting evidence in the classification of glypho- sate as a “probable” (Group 2A) human carcinogen. However, the current review of studies collected in the oxidative stress charac- teristic bin indicates that the IARC glyphosatemonograph has fallen substantially short of performing an integrated MoA analysis necessary to develop and support a defensible evidentiary basis of potential oxidative stress function in humans. The IARC glyphosate monograph, which serves as an early case example of how IARC has implemented its systematic MoA review process in practice, in- dicates that the conclusion of “strong” evidence oxidative stress apparently was reached simply from the basic observation that a small number of oxidative stress studies were identified for that characteristic bin. There was no evidence that the monograph ful- filled the integrated analysis expectations outlined in Smith et al. (2016). Studies collected within and across characteristic bins were not evolved into integrated analyses addressing consider- ations intrinsic to all MoA assessments such as dose-exposurerelationships, temporality, coherence, consistency, target organ relevance, etc. In this regard, the monograph did not supply evi- dence table summaries of the cited oxidative stress literature, a critical component of the systemic literature review process necessary to facilitate meaningful visualization and integration of the proffered MoA datasets. In fact, appropriate tabulation of the oxidative stress data reviewed by IARC would have immediately revealed, as demonstrated in the current analysis, that few, if any, of the data would have fulfilled criteria (e.g., dose-response, tempo- rality, target organ relevance, consistency and coherence) neces- sary for assembly of plausible MoA hypotheses. It is clear from the analysis of this review that the oxidative stress literature cited in the glyphosate monograph was dramati- cally deficient in its ability to fulfill even basic assumptions of sound MoA assessments. For example, of the 7 in vitro human and 7 in vivo non-human mammalian studies cited as evidence of oxidative stress, almost all were single-dose and single time point evalua- tions, which rule them ineffectual in MoA evaluations (Table 1). These specific and serious design limitations provided no ability to differentiate whether oxidative stress was a true MoA event plau- sibly accounting for a human cancer outcome, or rather repre- sented only secondary sequelae to use of excessively high test doses and/or cellular events associated with late-stage toxicities and resulting cell/organism dysfunction. The likelihood of such sec- ondary events was highly probable in the literature citations offered in the glyphosate monograph in that almost all of the studies used doses eliciting substantial toxicity (e.g., conducted at LC50 concentrations or doses producing severe toxicity; Tables 2 and 3). In addition, almost all of studies did not examine oxida- tive stress in relevant tissues/organs identified by IARC as possible cancer targets. It should be emphasized, however, that the cancer target organs identified by IARC have been challenged as true glyphosate cancer endpoints, i.e., multiple expert reviews have concluded that glyphosate is not an animal carcinogenicity hazard and/or risk (EPA, 2013; EFSA, 2016; Greim et al., 2015; Williams et al., 2000, 2016; JMPR, 2016). The conclusions of these reviews raise the important question of whether the IARC oxidative stress MoA assessment represents a mechanism in search of on outcome, versus offering meaningful human-relevant mechanistic insights into a solidly established animal or human cancer responses. It also needs to be noted that a high proportion of the IARC-cited oxidative stress studies tested only formulations or mixtures, substantially confounding specific attribution of oxidative stress to glyphosate alone under conditions of human exposures. Finally, extension of the IARC-cited studies as evidence of a human plausible MoA was substantially compromised by the fact that the doses eliciting oxidative stress responses were separated from real-world glyphosate exposures by many orders of magni- tude. Thus, there are no unforeseen circumstances under which human exposuresmight even remotely approach the doses eliciting oxidative stress in the IARC-cited literature, and thus unanticipated and realistic cancer hazards are very unlikely to be mediated by the postulated oxidative stress MoA. This concern is particularly real- ized for pesticides, in which any new or expanded uses are proac- tively and exhaustively scrutinized by regulatory authorities for potential changes in exposure profiles relative to those already established as protective of cancer and other health risks. In conclusion, the glyphosate oxidative stress case example il- lustrates that IARC has not provided adequate guidance criteria allowing its Working Group review panels to effectively and effi- ciently differentiate and integrate “weak”, “moderate” or “strong” evidence of any of the 10 key characteristics of human carcinogens into human-plausible cancer MoA hypotheses. Other commen- taries have likewise recommended that IARC should provide its Working Groups with improved guidance for conducting clear and J.S. Bus / Regulatory Toxicology and Pharmacology 86 (2017) 157e166 165consistent evaluations of mechanistic evidence, and particularly to utilize existing MoA frameworks explicitly constructed to uni- formly identify and evaluate the strengths and limitations (including consideration of alternative hypotheses) of human, an- imal and MoA carcinogenicity data for the purpose of establishing human risk relevance (Boobis et al., 2016; Goodman and Lynch, 2017). As emphasized in the IARC report outlining the formulation of the 10 key characteristics of human carcinogens (Smith et al., 2016), and as further demonstrated in the current case example, funda- mental MoA evaluation criteria such as dose-exposure relation- ships, temporality of response to outcome, species and target organ relevance, etc. were obviously not robustly considered in the IARC review of glyphosate oxidative stress MoA (IARC, 2015b). Had such expectations been implemented, the conclusion of “strong” evi- dence for plausible oxidative stress function in humans would have been readily determined as “weak” evidence (the lowest evidence category for IARC MoA evaluations and regarded as non-supportive of a MoA). The glyphosate case example also importantly suggests that other recent IARC MoA analyses relying on the 10 key char- acteristics of human carcinogens approach should be similarly evaluated for evidence of robust data integration necessary to support conclusions of plausible human cancer mechanisms. Declaration of interests Support for this work was supplied by a contract from Crop Life America with the author's employer, Exponent, Inc. The author had sole responsibility for the writing and content of the paper, and the views and opinions expressed do not necessarily reflect the views of the paper's sponsor. In addition, the author is a retiree of The Dow Chemical Company, and his current employer, Exponent, Inc., has received contracts from Monsanto; both companies have commercial interests in glyphosate. Transparency document Transparency document related to this article can be found online at http://dx.doi.org/10.1016/j.yrtph.2017.03.004. References Acquavella, J.F., Alexander, B.H., Mandel, J.S., Gustin, C., Baker, B., Chapman, P., Bleeke, M., 2004. Glyphosate biomonitoring for farmers and their families: results from the farm family exposure study. Environ. Hlth Perspect. 112, 321e326. Acquavella, J., Garabrant, D., Marsh, G., Sorahan, T., Weed, D.L., 2016. Glyphosate epidemiology expert panel review: a weight of evidence systematic review of the relationship between glyphosate exposure and non-Hodgkin’s lymphoma or multiple myeloma. Crit. Rev. Toxicol. 46 (S1), 28e43. Anadon, A., Martinez-Larranaga, M.R., Martinez, M.A., Castellano, V.J., Martinez, M., Martin, M.T., et al., 2009. Toxicokinetics of glyphosate and its metabolite ami- nomethyl phosphonic acid in rats. Toxicol. Lett. 190, 91e95. Astiz, M., de Alaniz, M.J., Marra, C.A., 2009. Antioxidant defense system in rats simultaneously intoxicated with agrochemicals. Environ. Toxicol. Pharmacol. 28, 465e473. Astiz, M., Hurtado de Catalfo, G.E., García, M.N., Galletti, S.M., Errecalde, A.L., de Alaniz, M.J., et al., 2013. Pesticide-induced decrease in rat testicular steroido- genesis is differentially prevented by lipoate and tocopherol. Ecotoxicol. Envi- ron. Saf. 91, 129e138. Becker, R.A., Dreier, D.A., Manibusan, M.K., Cox, L.A. (Tony), Simon, T.W., Bus, J.S., 2017. How well can carcinogenicity be predicted by high throughput “charac- teristics of carcinogens” mechanistic data? Regul. Toxicol. Pharmacol. submitted. Berquist, B.R., Wilson III, D.M., 2012. Pathways for repairing and tolerating the spectrum of oxidative DNA lesions. Cancer Lett. 327, 61e72. Bolognesi, C., Bonatti, S., Degan, P., Gallerani, E., Peluso, M., Rabboni, R., et al., 1997. Genotoxic activity of glyphosate and its technical formulation Roundup. J. Agric. Food Chem. 45, 1957e1962. Boobis, A.R., Cohen, S.M., Dellarco, V., McGregor, D., Meek, M.E., Vickers, C., Willcocks, D., Farland, W., 2006. IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit. Rev. Toxicol. 36, 781e792.Boobis, A.R., Cohen, S.M., Dellarco, V.L., Doe, J.E., Fenner-Crisp, P.A., Moretto, A., et al., 2016. Classification schemes for carcinogenicity based on hazard- identification have become outmoded and serve neither science nor society. Regul. Toxicol. Pharmacol. 82, 158e166. Brusick, D., Aardema, M., Kier, L., Kirkland, D., Williams, G., 2016. Genotoxicity Expert Panel review. Weight-of-evidence evaluation of the genotoxicity of glyphosate, glyphosate-based formulations and aminomethylphosphonic acid. Crit. Rev. Toxicol. 46 (S1), 56e74. Bus, J.S., Gibson, J.E., 1984. Paraquat: model for oxidant-initiated toxicity. Environ. Hlth Perspect. 55, 37e46. Bus, J.S., 2015. Analysis of Moms across America report suggesting bioaccumulation of glyphosate in U.S. mother's breast milk: implausibility based on inconsis- tency with available body of glyphosate animal toxicokinetic, human bio- monitoring, and physico-chemical data. Regul. Toxicol. Pharmacol. 73, 758e764. Cattaneo, R., Clasen, B., Loro, V.L., de Menezes, C.C., Pretto, A., Baldisserotto, B., et al., 2011. Toxicological responses of Cyprinus carpio exposed to a commercial formulation containing glyphosate. Bull. Environ. Contam. Toxicol. 87, 597e602. Cattani, D., de Liz Oliveira Cavalli, V.L., Heinz Rieg, C.E., Domingues, J.T., Dal-Cim, T., Tasca, C.I., et al., 2014. Mechanisms underlying the neurotoxicity induced by glyphosate-based herbicide in immature rat hippocampus: involvement of glutamate excitotoxicity. Toxicology 320, 34e45. Cavuşoglu, K., Yapar, K., Oruç, E., Yalçın, E., 2011. Protective effect of Ginkgo biloba L. leaf extract against glyphosate toxicity in Swiss albino mice. J. Med. Food 14, 1263e1272. Chaufan, G., Coalova, I., Ríos de Molina, M.C., 2014. Glyphosate commercial formulation causes cytotoxicity, oxidative effects, and apoptosis on human cells: differences with its active ingredient. Int. J. Toxicol. 33, 29e38. Cemeli, E., Baumgartner, A., Anderson, D., 2009. Antioxidants and the comet assay. Mut. Res/Rev Mut. Res. 681, 51e67. Coalova, I., Ríos de Molina, M.C., Chaufan, G., 2014. Influence of the spray adjuvant on the toxicity effects of a glyphosate formulation. Toxicol Vitro 28, 1306e1311. Costa, M.J., Monteiro, D.A., Oliveira-Neto, A.L., Rantin, F.T., Kalinin, A.L., 2008. Oxidative stress biomarkers and heart function in bullfrog tadpoles exposed to Roundup Original. Ecotoxicology 17, 153e163. de Menezes, C.C., da Fonseca, M.B., Loro, V.L., Santi, A., Cattaneo, R., Clasen, B., et al., 2011. Roundup effects on oxidative stress parameters and recovery pattern of Rhamdia quelen. Arch. Environ. Contam. Toxicol. 60, 665e671. Elie-Caille, C., Heu, C., Guyon, C., Nicod, L., 2010. Morphological damages of a glyphosate-treated human keratinocyte cell line revealed by a micro- to nanoscale microscopic investigation. Cell Biol. Toxicol. 26, 331e339. EC (European Commission) (1996). Council Directive 91/414/EEC Regulation 3600/ 92. Paraquat. Vol. 1. Report and proposed decision of the United Kingdom made to the European Commission under Article 7(1) of Regulation 3600/92, September, 1996. EFSA (European Food Safety Authority), 2016. Open Letter: Review of Carcinoge- nicity of Glyphosate by EFSA and BrR. Letter Addressed to Prof. C Portier, 13, January, 2016. Ref. BU/JK/JR/aa (2016) e out-15124233. EPA (US Environmental Protection Agency), 1995. Reregistration Eligibility Docu- ment (RED). Diquat Dibromide. Prevention, Pesticides and Toxic Substances. EPA 738-R-95e016, July 1995. EPA (US Environmental Protection Agency), 1997. Reregistration Eligibility Decision (RED). Paraquat Dichloride. Prevention, Pesticides and Toxic Substances. EPA 738-F-96e018, August 1997. EPA ((US Environmental Protection Agency), 2013. Glyphosate pesticide tolerances; Final rule (40 CFR Part 180) [EPAeHQeOPPe2012e0132; FRLe9384e3]. Fed. Regist. (US) 78, 25396e25401. Available from: http://www.regulations.gov/#% 2621documentDetail;D¼EPA-HQ-OPP-2012-0132-0009. Ferreira, D., da Motta, A.C., Kreutz, L.C., Toni, C., Loro, V.L., Barcellos, L.J., 2010. Assessment of oxidative stress in Rhamdia quelen exposed to agrichemicals. Chemosphere 79, 914e921. Gehin, A., Guillaume, Y.C., Millet, J., Guyon, C., Nicod, L., 2005. Vitamins C and E reverse effect of herbicide-induced toxicity on human epidermal cells HaCaT: a biochemometric approach. Int. J. Pharm. 288, 219e226. George, J., Prasad, S., Mahmood, Z., Shukla, Y., 2010. Studies on glyphosate-induced carcinogenicity in mouse skin: a proteomic approach. J. Proteomics 73, 951e964. George, J., Shukla, Y., 2013. Emptying of intracellular calcium pool and oxidative stress imbalance are associated with the glyphosate-induced proliferation in human skin keratinocytes HaCaT cells. ISRN Dermatol 2013, 825180. Geret, F., Burgeot, T., Haure, J., Gagnaire, B., Renault, T., Communal, P.Y., et al., 2013. Effects of low-dose exposure to pesticide mixture on physiological responses of the Pacific oyster, Crassostrea gigas. Environ. Toxicol. 28, 689e699. Glusczak, L., Loro, V.L., Pretto, A., Moraes, B.S., Raabe, A., Duarte, M.F., et al., 2011. Acute exposure to glyphosate herbicide affects oxidative parameters in piava (Leporinus obtusidens). Arch. Environ. Contam. Toxicol. 61, 624e630. Goodman, J., Lynch, H., 2017. Improving the international agency for research on Cancer's consideration of mechanistic evidence. Toxicol. Appl. Pharmacol. 319, 39e46. Greim, H., Saltmiras, D., Moster, V., Strupp, C., 2015. Evaluation of carcinogenic potential of the herbicide glyphosate, drawing on tumor incidence data from fourteen chronic/carcinogenicity rodent studies. Crit. Rev. Toxicol. 45, 185e208. Grosse, Y., Loomis, D., Guyton, K.Z., El Ghissassi, F., Bouvard, V., Benbrahim-Tallaa, L., Mattock, H., Straif, K., 2016. Carcinogenicity of some industrial chemicals. Lancet Oncol. 17, 419e420. J.S. Bus / Regulatory Toxicology and Pharmacology 86 (2017) 157e166166Guilherme, S., Gaiv~ao, I., Santos, M.A., Pacheco, M., 2010. European eel (Anguilla anguilla) genotoxic and pro-oxidant responses following short-term exposure to Roundupea glyphosate-based herbicide. Mutagenesis 25, 523e530. Guilherme, S., Gaiv~ao, I., Santos, M.A., Pacheco, M., 2012a. DNA damage in fish (Anguilla anguilla) exposed to a glyphosate-based herbicide e elucidation of organ-specificity and the role of oxidative stress. Mutat. Res. 743, 1e9. Guilherme, S., Santos, M.A., Barroso, C., Gaiv~ao, I., Pacheco, M., 2012b. Differential genotoxicity of Roundup(®) formulation and its constituents in blood cells of fish (Anguilla anguilla): considerations on chemical interactions and DNA damaging mechanisms. Ecotoxicology 21, 1381e1390. Guilherme, S., Santos, M.A., Gaiv~ao, I., Pacheco, M., 2014a. Are DNA-damaging ef- fects induced by herbicide formulations (Roundup® and Garlon®) in fish tran- sient and reversible upon cessation of exposure? Aquat. Toxicol. 155, 213e221. Guilherme, S., Santos, M.A., Gaiv~ao, I., Pacheco, M., 2014b. DNA and chromosomal damage induced in fish (Anguilla anguilla L.) by aminomethylphosphonic acid (AMPA)ethe major environmental breakdown product of glyphosate. Environ. Sci. Pollut. Res. Int. 21, 8730e8739. Guyton, K.Z., Loomis, D., Grosse, Y., El Ghissassi, F., Benbrahim-Tallaa, L., Guha, N., Scoccianti, C., Mattock, H., Straif, K., 2015. Carcinogenicity of tetrachlorvinphos, parathion, malathion, diazinon, and glyphosate. Lancet Oncol. 16, 490e491. Halliwell, B., Whiteman, M., 2004. Measuring reactive species and oxidative dam- age in vivo and in cell culture: how should you do it and what do the results mean? Brit J. Pharmacol. 142, 231e255. Halliwell, B., 2007. Oxidative stress and cancer: have we moved forward? Biochem. J. 401, 1e11. IARC, 2006. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Preamble. IARC, Lyon, France. IARC, 2015a. IARC: International Agency for Research on Cancer: the First 50 Years. R Saracci and CP Wild, IARC, Lyon, France. IARC, 2015b. Glyphosate Monograph. http://monographs.iarc.fr/ENG/Monographs/ vol112/mono112-09.pdf. Jasper, R., Locatelli, G.O., Pilati, C., Locatelli, C., 2012. Evaluation of biochemical, hematological and oxidative parameters in mice exposed to the herbicide glyphosate-Roundup(®). Interdiscip. Toxicol. 5, 133e140. JMPR (Joint FAO/WHO Meeting on Pesticide Residues), 2016. Glyphosate (158). In: pesticide residues in food 2016. In: Special Session of the Joint FAO/WHO Meeting on Pesticide Residues, Geneva, 9 to 13 May 2016. Rome: Food and Agriculture Organization of the United Nations/Geneva, vol. 45. World Health Organization (WHO) (FAO Plant Production and Protection Paper No. 227), p. 19e28, pp. 72e82. Available from: http://www.fao.org/3/a-i5693e.pdf. Kier, L.D., Kirkland, D.J., 2013. Review of genotoxicity studies of glyphosate and glyphosate-based formulations. Crit. Rev. Toxico 43, 283e315. Kier, L.D., 2015. Review of genotoxicity biomonitoring studies of glyphosate-based formulations. Crit. Rev. Toxicol. 45, 209e918. Klaunig, J.E., Wang, Z., Pu, X., Zhou, S., 2011. Oxidative stress and oxidative damage in chemical carcinogenesis. Toxicol. Appl. Pharmacol. 254, 86e99. Kushman, M.E., Kraft, A.D., Guyton, K.Z., Chiu, W.A., Makris, S.L., Rusyn, I., 2013. A systematic approach for identifying and presenting mechanistic evidence in human health assessments. Regul. Toxicol. Pharmacol. 67, 266e277. Kwiatkowska, M., Huras, B., Bukowska, B., 2014. The effect of metabolites and im- purities of glyphosate on human erythrocytes (in vitro). Pestic. Biochem. Physiol. 109, 34e43. Lehman, T.A., Modali, R., Boukamp, P., Stanek, J., Bennett, W.P., Welsh, J.A., Metcalf, R.A., Stampfer, M.R., Fusenig, N., Rogan, E.M., Harris, C.C., 1993. p53 mutations in human immortalized epithelial cell lines. Carcinogenesis 14, 833e839. Liu, D., Xu, Y., 2011. p53, oxidative stress, and aging. Antioxidants Redox Signal. 15, 1669e1678. Loomis, D., Guyton, K., Grosse, Y., El Ghissasi, F., Bouvard, V., Benbrahim-Tallaa, L., Guha, N., Mattock, H., Straif, K., 2015. Carcinogenicity of lindane, DDT, and 2,4- dichlorophenoxyacetic acid. Lancet Oncol. 16, 891e892. Lushchak, O.V., Kubrak, O.I., Storey, J.M., Storey, K.B., Lushchak, V.I., 2009. Low toxic herbicide Roundup induces mild oxidative stress in goldfish tissues. Chemo- sphere 76, 932e937. Marques, A., Guilherme, S., Gaiv~ao, I., Santos, M.A., Pacheco, M., 2014. Progression of DNA damage induced by a glyphosate-based herbicide in fish (Anguilla anguilla) upon exposure and post-exposure periodseinsights into the mechanisms ofgenotoxicity and DNA repair. Comp. Biochem. Physiol. C Toxicol. Pharmacol. 166, 126e133. Marques, A., Guilherme, S., Gaiv~ao, I., Santos, M.A., Pacheco, M., 2015. Erratum to: “Progression of DNA damage induced by a glyphosate-based herbicide in fish (Anguilla anguilla) upon exposure and post-exposure periods - insights into the mechanisms of genotoxicity and DNA repair” [Comp. Biochem. Physiol. C 166 (2014) 126e133]. Comp. Biochem. Physiol. C Toxicol. Pharmacol. 168C, 1. Meek, M.E., Boobis, A., Cote, I., Dellarco, V., Fotakis, G., Munn, S., Seed, J., Vickers, C., 2013. New developments in the evolution and application of the WHO/IPCS framework on mode of action/species concordance analysis. J. Appl. Toxicol. 34, 1e18. Mladinic, M., Berend, S., Vrdoljak, A.L., Kopjar, N., Radic, B., Zeljezic, D., 2009. Evaluation of genome damage and its relation to oxidative stress induced by glyphosate in human lymphocytes in vitro. Environ. Mol. Mutagen 50, 800e807. Modesto, K.A., Martinez, C.B., 2010a. Effects of Roundup Transorb on fish: hema- tology, antioxidant defenses and acetylcholinesterase activity. Chemosphere 81, 781e787. Modesto, K.A., Martinez, C.B., 2010b. Roundup causes oxidative stress in liver and inhibits acetylcholinesterase in muscle and brain of the fish Prochilodus lineatus. Chemosphere 78, 294e299. NAS (National Academy of Sciences/National Research Council), 1983. Risk Assess- ment in the Federal Government: Managing the Process. National Academy Press, Washington, DC. Nwani, C.D., Nagpure, N.S., Kumar, R., Kushwaha, B., Lakra, W.S., 2013. DNA damage and oxidative stress modulatory effects of glyphosate-based herbicide in freshwater fish. Channa Punctatus. Environ. Toxicol. Pharmacol. 36, 539e547. Ortiz-Ordo~nez, E., Uría-Galicia, E., Ruiz-Picos, R.A., Duran, A.G., Trejo, Y.H., Sede~no- Díaz, J.E., et al., 2011. Effect of Yerbimat herbicide on lipid peroxidation, catalase activity, and histological damage in gills and liver of the freshwater fish Goodea atripinnis. Arch. Environ. Contam. Toxicol. 6, 443e452. Sinhorin, V.D., Sinhorin, A.P., Teixeira, J.M., Mileski, K.M., Hansen, P.C., Moreira, P.S., et al., 2014. Effects of the acute exposition to glyphosate-based herbicide on oxidative stress parameters and antioxidant responses in a hybrid Amazon fish surubim (Pseudoplatystoma sp). Ecotoxicol. Environ. Saf. 106, 181e187. Slaninova, A., Smutna, M., Modra, H., Svobodova, Z., 2009. A review: oxidative stress in fish induced by pesticides. Neuro Endocrinol. Lett. 30:Suppl (1), 2e12. Smith, M.T., Guyton, K.Z., Gibbons, C.F., Fritz, J.M., Portier, C.J., Rusyn, I., et al., 2016. Key characteristics of carcinogens as a basis for organizing data on mechanisms of carcinogenesis. Environ. Hlth Perspect. 124, 713e721. Solomon, K.R., 2016. Glyphosate in the general population and in applicators: a critical review of studies on exposures. Crit. Rev. Toxicol. 46 (S1), 21e27. Sonich-Mullin, C., Fielder, R., Wiltse, J., Baetcke, K., Dempsey, J., Fenner-Crisp, P., Grant, D., et al., 2001. International Programme on Chemical Safety. 2001. IPCS conceptual framework for evaluating a mode of action for chemical carcino- genesis. Regul. Toxicol. Pharmacol. 34, 146e152. Speit, G., Kojima, H., Burlinson, B., Collins, A.R., Kasper, P., Plappert-Helbig, U., et al., 2015. Critical issues with the in vivo comet assay: a report of the comet assay working group in the 6th International Workshop on Genotoxicity Testing (IWGT). Mut. Res/Genet Toxicol. Environ. Mut. 783, 6e12. Stancliffe, T.C., Pirie, A., 1971. The production of superoxide radicals in reactions of the herbicide diquat. FEBS Lett. 17, 297e299. Tarone, R.E., 2016. On the International Agency for Research on Cancer classification of glyphosate as a probable human carcinogen. Eur. J. Canc Prev. Epub ahead of print, August 22, 2016, http://dx.doi.org/10.1097/CEJ.0000000000000289. Thomas, R.S., Philbert, M.A., Auerbach, S.S., Wetmore, B.A., Devito, M.J., Cote, I., et al., 2013. Incorporating new technologies into toxicity testing and risk assessment: moving from 21st century vision to a data driven framework. Toxicol. Sci. 136, 4e18. Uren Webster, T.M., Laing, L.V., Florance, H., Santos, E.M., 2014. Effects of glyphosate and its formulation, Roundup, on reproduction in zebrafish (Danio rerio). En- viron. Sci. Technol. 48, 1271e1279. Williams, G.M., Kroes, R., Munro, I.C., 2000. Safety evaluation and risk assessment of the herbicide Roundup and its active ingredient, glyphosate, for humans. Regul. Toxicol. Pharmacol. 31, 117e165. Williams, G.M., Berry, C., Burns, M., de Camargo, J.L.V., Greim, H., 2016. Glyphosate rodent carcinogenicity bioassay expert panel review. Crit. Rev. Toxicol. 46 (S1), 44e55. EXHIBIT C Systematic review and meta-analysis of glyphosate exposure and risk of lymphohematopoietic cancers Ellen T. Changa,b and Elizabeth Delzella aCenter for Epidemiology and Computational Biology, Health Sciences Practice, Exponent, Inc., Menlo Park, California and Alexandria, Virginia, USA; bDivision of Epidemiology, Department of Health Research and Policy, Stanford University School of Medicine, Stanford, California, USA ARTICLE HISTORY Received 28 November 2013 ABSTRACT This systematic review and meta-analysis rigorously examines the relationship between glyphosate exposure and risk of lymphohematopoietic cancer (LHC) including NHL, Hodgkin lymphoma (HL), multiple myeloma (MM), and leukemia. Meta-relative risks (meta-RRs) were positive and marginally statistically significant for the association between any versus no use of glyphosate and risk of NHL (meta-RR D 1.3, 95% confidence interval (CI) D 1.0-1.6, based on six studies) and MM (meta-RR D 1.4, 95% CI D 1.0-1.9; four studies). Associations were statistically null for HL (meta-RR D 1.1, 95% CI D 0.7-1.6; two studies), leukemia (meta-RR D 1.0, 95% CI D 0.6-1.5; three studies), and NHL subtypes except B-cell lymphoma (two studies each). Bias and confounding may account for observed associations. Meta-analysis is constrained by few studies and a crude exposure metric, while the overall body of literature is methodologically limited and findings are not strong or consistent. Thus, a causal relationship has not been established between glyphosate exposure and risk of any type of LHC. KEYWORDS Glyphosate; non-Hodgkin lymphoma; Hodgkin lymphoma; multiple myeloma; leukemia; hematologic malignancies; herbicides; meta-analysis Introduction The broad-spectrum herbicide glyphosate (N-(phosphono- methyl)glycine), as a constituent of more than 750 products for agricultural, forestry, urban, and residential applications, is the most commonly used herbicide in the world. Therefore, under- standing its potential human carcinogenicity has major impli- cations for public health and risk assessment. In 2014, the German Federal Institute for Risk Assessment (BfR), on behalf of the European Union, reviewed all toxicolog- ical studies of glyphosate in laboratory animals, as well as over 30 epidemiological studies in humans, and concluded that “the available data do not show carcinogenic or mutagenic proper- ties of glyphosate” and “there is no validated or significant rela- tionship between exposure to glyphosate and an increased risk of non-Hodgkin lymphoma or other types of cancer.”[1,2] This conclusion was consistent with those previously reached by the United States Environmental Protection Agency (U.S. EPA) and the Joint Meeting on Pesticide Residues (JMPR), sponsored by the Food and Agriculture Organization of the United Nations and the World Health Organization (WHO), which concluded that glyphosate was unlikely to be carcinogenic to humans.[3-5] By contrast, the International Agency for Research on Can- cer (IARC) in 2015 classified glyphosate as “probably carcino- genic to humans” (Group 2A). In arriving at this classification, IARC characterized evidence of carcinogenicity in humans as “limited,” based on the data available for non-Hodgkin lym- phoma (NHL).[6] IARC considered the evidence of carcinoge- nicity in experimental animals as “sufficient.” The latter determination was based on the occurrence of renal tubule carcinoma, hemangiosarcoma, and pancreatic islet-cell ade- noma in rodents, as well as mechanistic evidence. To incorporate the IARC classification into the European Union review of glyphosate, BfR was commissioned by the Ger- man government and the European Food Safety Authority (EFSA) to review the IARC assessment.[7] In its subsequent revised assessment report, BfR reached the conclusion that “no carcinogenic risk to humans is to be expected from glyphosate if it is used in the proper manner for the intended purpose.”[8] This assessment was supported by all European Union member states except one (Sweden) and by EFSA.[9] The WHO also has established an expert taskforce to re-evaluate the available data on glyphosate and report its findings to JMPR.[10] In summarizing the epidemiological evidence, IARC stated that “case-control studies in the USA, Canada, and Sweden reported increased risks for NHL associated with exposure to glyphosate. The increased risk persisted in the studies that adjusted for exposure to other pesticides. The [Agricultural Health Study] cohort did not show an excess of NHL. The Working Group noted that there were excesses reported for multiple myeloma in three studies; however, they did not weight this evidence as strongly as that of NHL because of the possibility that chance could not be excluded; none of the risk estimates were statistically significant nor were they adjusted for other pesticide exposures.”[6] A recent meta-analysis con- ducted by investigators from IARC[11] found a statistically sig- nificant positive association between glyphosate use and NHL risk (meta-relative risk [RR] D 1.5, 95% confidence interval [CI] D 1.1-2.0), based on six studies.[12-17] The same meta- analysis also found a significant positive association between CONTACT Ellen T. Chang echang@exponent.com Health Sciences Practice, Exponent, Inc., 149 Commonwealth Drive, Menlo Park, CA 94025, USA. © 2016 Exponent, Inc. Published with license by Taylor & Francis. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way. JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 2016, VOL. 51, NO. 6, 402-428 http://dx.doi.org/10.1080/03601234.2016.1142748 C\ Taylor & Francis ~ Taylor&FrancisGroup @ OPEN ACCESS glyphosate use and risk of B-cell NHL, based on two studies.[14,18] Although Schinasi and Leon[11] stated that in their meta- analysis, “[i]n an effort to use the most unbiased estimate, [they] extracted the most adjusted effect estimate,” two or argu- ably three of the RR estimates that they selected for inclusion were not the most highly adjusted estimates reported by the original authors.[13-15] Instead, in a personal communication (11 August 2015), Dr. Schinasi indicated that other estimates were selected based on considerations of consistency of esti- mates across meta-analyses of other pesticides, secondary anal- yses, and statistical modeling approach. Meta-analyses are not intended to identify, validate, or dis- pute causal relationships. Although they can be useful in pro- viding a summary measure of association and identifying heterogeneity among research results, they can obscure impor- tant differences in methods and results among studies that can be more thoroughly evaluated in a detailed qualitative review. Schinasi and Leon[11] did not assess study quality and did not specifically address the potential impact of study limitations on the findings for glyphosate, nor did they discuss whether the apparent association between glyphosate and NHL risk is likely to be causal. On the other hand, Mink et al.[19] conducted a qualitative systematic review, without a meta-analysis, of epide- miologic studies of glyphosate and various cancers, including NHL. Taking into account potential sources of error, including selection bias, confounding, and especially exposure misclassifi- cation, the authors concluded that they “found no consistent pattern of positive associations indicating a causal relationship between total cancer (in adults or children) or any site-specific cancer and exposure to glyphosate.” Given the conflicting findings surrounding this issue, we conducted this systematic review and meta-analysis to examine more rigorously the relationship between exposure to glypho- sate and risk of NHL, as well as major histopathological sub- types of NHL, in human epidemiologic studies. Because NHL is often considered alongside other lymphohematopoietic can- cers (LHC), whose ever-changing classification systems now characterize some leukemias and multiple myeloma (MM) as NHL subtypes,[20] we also included Hodgkin lymphoma (HL), MM, and leukemia in this review. Despite the limitations of quantitative meta-analysis for observational epidemiology,[21,22] we conducted a meta-analysis largely to determine the impact of using RR estimates not used in the meta-analysis by Schinasi and Leon.[11] In addition, we conducted a qualitative evaluation of potential for error and bias. Thus, this article goes beyond previous work by examining all types of LHC, conducting a new meta-analysis, providing a detailed evaluation of study quality and potential for bias, and synthesizing the overall epi- demiologic evidence for a causal association between glypho- sate and LHC risk. Methods Literature search Sources eligible for inclusion in the meta-analysis were original articles describing epidemiological studies that provided numeric point estimates of the RR (i.e., odds ratio, rate ratio, or prevalence ratio) of LHC, including NHL, HL, MM, leukemia, and any subtypes of these disease entities, associated with indi- vidual-level glyphosate exposure, along with corresponding interval estimates (e.g., 95% confidence intervals [CI]) or suffi- cient raw data to calculate RRs and CIs. Reviews, commentar- ies, letters to the editor without original data, and non-human studies were excluded, as were articles that did not report quan- titative measures of association between glyphosate exposure (e.g., those assessing broadly defined categories of pesticides or herbicides) and risk of LHC (e.g., those assessing other cancers or all malignancies combined). To identify all potentially relevant articles, we searched MEDLINE via PubMed (Supplementary methods), with addi- tional targeted searches in Web of Science and Google Scholar, along with a review of the bibliographies of recent review articles. Based on a review of titles and abstracts to exclude articles without pertinent information, followed by a review of the full text of relevant articles, 19 articles (as well as one letter to the editor[23] that contained additional results from a study described in another one of the included articles,[24] and one abstract[25] that preceded a full-length article[26]) were ulti- mately deemed eligible for inclusion (Appendix Fig. A1). Two authors independently reviewed and agreed upon the list of eli- gible articles. Of the 19 articles reporting on the association between glyph- osate and risk of specific forms of LHC, 12 pertained to NHL or its subtypes (including hairy-cell leukemia, which is a subtype of B-cell NHL),[12-18,24,27-30] 2 pertained to HL,[17,31] 6 pertained to MM,[12,17,26,32-34] and 3 pertained to leukemia.[12,35,36] Evaluation of study characteristics and quality From each eligible study, we extracted the following informa- tion: first author, publication year, study location, study design, study years, source population, number of subjects, proportion of proxy respondents, exposure assessment method, outcome assessment method, confounders adjusted, number of subjects in each exposure category, and RR estimates with CIs. In addition to summarizing study characteristics, we qualita- tively evaluated the methodological quality of each study in terms of its potential for selection bias, information bias/expo- sure misclassification, confounding, reporting bias, and other issues affecting validity. Potential for bias was evaluated based on subject identification strategy, participation rates, investiga- tor blinding, assessment methods for exposures, outcomes, and potential confounders, statistical approach, reporting of results, and other considerations.[37-39] Selection of data for meta-analysis From each publication, we selected an RR point estimate for inclusion in the meta-analysis based on a set of rules specified a priori. First, if unadjusted and adjusted RRs were reported in a publication or across multiple publications from the same study population, the most fully adjusted RR was selected for inclu- sion. The most fully adjusted RR was defined as the RR estimate that took into consideration, by restriction or statistical adjust- ment, the most covariates that appeared to be confounders. The rationale for choosing the most fully adjusted RR was JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 403 based on the assumption that the adjusted covariates were found by the authors to act as confounders by altering the esti- mate of association (either directly or by acting as a surrogate for another, unmeasured confounder); however, some authors did not explain how confounders were selected, so this assump- tion may not hold for all studies. If an adjusted RR was not reported, the unadjusted (crude) RR was included as reported by the authors or as calculated from available raw data. Second, if multiple eligible publications were derived from the same study population, the RR from the most recent publication was selected for inclusion unless it was based on a subset of the overall eligible study population, in which case the RR based on the most complete study population was included. Third, sub- ject to the first two rules, the RR for dichotomous exposure with the largest number of exposed cases was selected for inclu- sion in the meta-analysis. In a few instances where another RR from a given study nearly met these inclusion criteria but was superseded by a more fully adjusted, more recent, or more robust RR, the alternative RR was considered in secondary analyses. RRs for multiple categories of exposure also were extracted to enable qualitative evaluation of exposure-response trends (based on the assumption, discussed later, that studies were able to distinguish among exposure levels). However, because no two studies used the same set of three or more categories to classify glyphosate exposure, these estimates could not be com- bined in meta-analysis. Statistical approach For associations with at least two independent RR estimates from different study populations, we estimated both fixed- effects and random-effects meta-RRs with 95% CIs. We used comparison of meta-RR estimates from fixed-effects and random-effects models as one approach to the evalua- tion of the impact of between-study heterogeneity on the meta-RRs. As a quantitative measure of between-study het- erogeneity, we calculated I2, which represents the percentage of between-study variance in RRs that is attributable to study heterogeneity (as opposed to chance).[40] We also tested for statistically significant between-study heterogene- ity based on Cochran’s Q statistic,[41] although this test has low power to detect modest heterogeneity across a limited number of studies.[42] In the absence of statistically significant heterogeneity, the presence of at least one statistically significant association, I2 < 50%, and at least four contributing studies, we evaluated evi- dence of publication bias (i.e., non-random selection of studies for publication, with a tendency toward submission and publi- cation of studies that report larger, statistically significant asso- ciations[43]) by using the linear regression approach of Egger et al.,[44] which measures the degree of funnel plot asymmetry. We also estimated meta-RRs corrected for publication bias by imputing results for missing studies using the trim-and-fill pro- cedure developed by Duval and Tweedie,[45] which iteratively trims asymmetric studies from the overbalanced side of a fun- nel plot to locate the unbiased effect, and then fills the plot by re-inserting the trimmed studies on the original side of the mean effect, along with their imputed counterparts on the opposite side. Again, we used these approaches with the under- standing that they have limited power to detect publication bias based on few studies.[42] The meta-analysis was conducted using Comprehensive Meta-Analysis Software (Biostat, Inc., Englewood, NJ, USA). All calculated meta-RRs and 95% CIs were confirmed using Episheet (www.krothman.org/episheet.xls). Sensitivity analysis To evaluate the robustness of results to various potential sour- ces of heterogeneity, we planned a priori to conduct a sensitiv- ity analysis with stratification of studies by study design (case- control vs. cohort), source of controls (population-based vs. hospital-based), gender (males only vs. males and females), geographic region (North America vs. Europe), and time period of cancer diagnosis (1980s, 1990s, or 2000s, with studies con- tributing to a given stratum if any part of the case diagnosis period was in a given decade). Overall evaluation To guide a qualitative assessment of the combined epidemio- logic evidence for a causal relationship between glyphosate exposure and risk of LHC, we used Sir Austin Bradford Hill’s “viewpoints” as a general framework.[46] Because this review is restricted to the epidemiologic literature, our consideration of the biological plausibility of the association and the coherence of the human, animal, and mechanistic evidence was limited. Results Study characteristics and overlap Studies of NHL and subtypes Twelve studies from seven independent study populations, including eleven case-control studies and one prospective Figure 1. Forest plots of relative risk (RR) estimates and 95% confidence intervals (CIs) for the association between glyphosate exposure and risk of non-Hodgkin lym- phoma. Meta-RRs were identical in random-effects and fixed-effects models. 404 E. T. CHANG AND E. DELZELL@ Authors Year RR 95% CI Relative weight (%) De~ ct al. IBJ 2003 1.6 0.9- 2.8 16.2 Dc ~ ctal. 112J 2005 l.l 0.7- 1.9 21.0 Eriksson ct al. 114J 2008 1.51 0.77- 2.94 11.6 fimWl ct al. (I SJ 2002 1.85 0.55- 6.20 3.6 McDuffie ct al. (16J 2001 1.20 0.83- l. 74 38.l Qrn ctal. (1 J 2009 LO 0.5- 2.2 9.5 Meta-RR 1.3 l.0- 1.6 OJ 1.0 10 cohort study, evaluated the relationship between glyphosate use and risk of NHL and/or its histopathological sub- types.[12-18,24,27-30] Characteristics of these studies are sum- marized in Table 1. All of the studies considered glyphosate use in agricultural operations or settings, and most evaluated overall NHL as an outcome. The exceptions were Cocco et al.,[18] which analyzed B-cell lymphoma and other NHL subtypes, but not overall NHL, and Nordstrom et al.,[30] which included only hairy-cell leukemia. Eriksson et al.[14] presented results for B-cell lymphoma and other NHL sub- types, as well as for overall NHL, while Orsi et al.[17] included results for overall NHL and several specific NHL subtypes. De Roos et al.[13] combined data from Cantor et al.[24] with data from two other studies that did not independently report associations between glyphosate use and NHL risk;[47,48] there- fore, we did not further consider Cantor et al.[24] as a separate study. Lee et al.[29] was based on Cantor et al.[24] and Hoar Zahm et al.,[48] but not Hoar et al.,[47] and stratified results by asthma status (with no apparent interaction between glypho- sate exposure and asthma); therefore, results from De Roos et al.[13] took precedence in our analysis over those from Lee et al.[29] The study by Hardell et al.[15] pooled data from two other studies that reported on glyphosate use and NHL risk.[27,30] Consequently, the latter two studies were not consid- ered further with respect to NHL, although Nordstrom et al.[30] was evaluated separately with respect to hairy-cell leukemia. Based on the same study population as McDuffie et al.[16] (except for four fewer cases excluded after pathology review), Hohenadel et al.[28] reported associations with use of glyphosate with or without malathion, but not glyphosate overall; there- fore, the results from McDuffie et al.[16] were prioritized in our analysis. The seven independent studies ranged markedly in size with respect to the number of NHL cases classified as exposed to glyphosate (based on reported use): Cocco et al.,[18] 4 B-cell lymphoma cases exposed; Hardell et al.,[15] 8 exposed; Orsi et al.,[17] 12 exposed; Eriksson et al.,[14] 29 exposed; De Roos et al.,[13] 36 exposed; McDuffie et al.,[16] 51 exposed; De Roos et al.,[12] 71 exposed in the total eligible cohort. Four studies were based in Europe[14,15,17,18] and three in North Amer- ica[12,13,16] (Table 1). Four of the case-control studies were pop- ulation-based,[13-16] one was hospital-based,[17] and one included a mixture of population-based and hospital-based cases and controls.[18] Four studies were restricted to males,[13,15-17] while the rest included males and females. Two studies conducted at least some case ascertainment during the 1980s,[13,15] five during the 1990s,[12,14-16,18] and four during the 2000s[12,14,17,18] (categories are overlapping). For reference, glyphosate entered the U.S. and European commercial markets in 1974.[49] Studies of HL Two case-control studies estimated the OR between glyphosate use and risk of HL.[17,31] Characteristics of these studies are summarized in Table 1. The study by Karunanayake et al.[31] used the same methods and source population as McDuffie et al.,[16] but focused on HL rather than NHL. As described in the section on NHL studies, Orsi et al.[17] was a hospital-based case-control study set in Europe (France), restricted to males, with case ascertainment in the 2000s, par- ticipation rates > 90%, and no proxy respondents. This study classified six HL cases as exposed to glyphosate. Karunanayake et al.[31] was a population-based case-control study set in North America (Canada), restricted to males, with case ascertainment in the 1990s, participation rates of 68% for cases and 48% for controls, and an unspecified proportion of proxy respondents. In this study, 38 HL cases were classified as glyphosate-exposed. Studies of MM Six studies from four independent study populations, including four case-control studies and two prospective cohort studies, evaluated the association between glyphosate use and risk of MM.[12,17,26,32-34] These studies are described in Table 1. A cross-sectional analysis within a subset of the Agricultural Health Study Cohort examined the association between glypho- sate use and risk of monoclonal gammopathy of unknown sig- nificance (MGUS), an MM precursor;[50] this study was not included in the present review. The studies by De Roos et al.[12] and Sorahan[26] were based on virtually identical datasets from the Agricultural Health Study cohort (except that the dataset used by Sorahan was stripped of data on race, state of residence, and applicator type due to privacy concerns; these differences should not have affected the results substantively). Because the Sorahan[26] study included all eligible cohort members, whereas the De Roos et al.[12] study was based on a restricted subset of the cohort with complete data,[51] the Sorahan[26] results were pri- oritized in our analysis of MM. Brown et al.[32] employed the same methods and source population as Cantor et al.,[24] which was included in the pooled analysis of NHL by De Roos et al.[13] Pahwa et al.[34] and Kachuri et al.[33] conducted over- lapping analyses in the same Canadian source population as McDuffie et al.,[16] Hohenadel et al.,[28] and Karunanayake et al.[31] Pahwa et al.[34] included more controls in their analy- sis, but these controls were excluded from Kachuri et al.[33] because they were younger than any enrolled MM cases (29 years) and thus did not contribute meaningfully to the analysis. Kachuri et al.[33] also controlled for more confounders, and therefore was prioritized in our analysis. With respect to glyphosate use, the four independent studies of MM included, respectively, 5 exposed cases,[17] 11 exposed cases,[32] 24 exposed cases,[26] and 32 exposed cases.[33] All but one study, which was based in France,[17] were conducted in North America, and all except one[26] were restricted to males. One of the two case-control studies was population-based[32] and the other was hospital-based.[17] Case ascertainment took place during the early 1980s in one study,[32] at least partly dur- ing the 1990s in two studies,[26,33] and at least partly during the 2000s in two studies.[17,26] Studies of leukemia Two case-control studies and one prospective cohort study inves- tigated the relationship between glyphosate use and risk of leuke- mia.[12,35,36] Key characteristics of these studies are provided in Table 1. The study by Brown et al.[35] used the same methods JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 405 Ta bl e 1. D es ig n ch ar ac te ris tic s of st ud ie s of gl yp ho sa te ex po su re an d ris k of ly m ph oh em at op oi et ic ca nc er (L H C) ,i nc lu di ng no n- H od gk in ly m ph om a (N H L) ,N H L su bt yp es ,H od gk in ly m ph om a (H L) ,m ul tip le m ye lo m a (M M ), an d le uk em ia . Au th or s Ye ar O ut co m es st ud ie d St ud y lo ca tio n St ud y de si gn St ud y ye ar s So ur ce po pu la tio n Su bj ec ti de nt ifi ca tio n Su bj ec tp ar tic ip at io n Su bj ec ts (n ) Pr ox y re sp on de nt s Br ow n et al .[3 5] 19 90 Le uk em ia (in cl ud in g m ye lo dy sp la si as ) U ni te d St at es (Io w a an d M in ne so ta ) Po pu la tio n- ba se d ca se -c on tr ol 19 80 -1 98 3 W hi te m en ag ed 30 ye ar s in Io w a an d M in ne so ta ,e xc lu di ng M in ne ap ol is ,S t. Pa ul , D ul ut h, an d Ro ch es te r Ca se s: Io w a Tu m or Re gi st ry an d sp ec ia ls ur ve ill an ce of M in ne so ta ho sp ita la nd pa th ol og y la bo ra to ry re co rd s Co nt ro ls :r an do m -d ig it di al in g if ag ed < 65 ye ar s, M ed ic ar e fi le s if ag ed 65 ye ar s, st at e de at h ce rt ifi ca te fi le s if de ce as ed Ca se s: 86 % Co nt ro ls :7 7% ra nd om di gi t di al in g, 79 % M ed ic ar e, 77 % pr ox ie s fo rd ec ea se d Su pp le m en ta li nt er vi ew :9 3% ca se s, 96 % co nt ro ls Ca se s: 57 8 Co nt ro ls :1 ,2 45 Su pp le m en ta li nt er vi ew :8 6 ca se s, 20 3 co nt ro ls Ca se s: 23 8 (4 1% ) Co nt ro ls :4 25 (3 4% ) Su pp le m en ta l in te rv ie w ,6 3 (7 3% ) ca se s, 57 (2 8% ) co nt ro ls Br ow n et al .[3 2] 19 93 M M U ni te d St at es (Io w a) Po pu la tio n- ba se d ca se -c on tr ol 19 81 -1 98 4 W hi te m en ag ed 30 ye ar s in Io w a Ca se s: Io w a H ea lth Re gi st ry Co nt ro ls :r an do m -d ig it di al in g if ag ed < 65 ye ar s, M ed ic ar e fi le s if ag ed 65 ye ar s, st at e de at h ce rt ifi ca te s if de ce as ed Ca se s: 84 % Co nt ro ls :7 8% ov er al l Ca se s: 17 3 Co nt ro ls :6 50 Ca se s: 72 (4 2% ) Co nt ro ls :1 98 (3 0% ) Ca nt or et al .[2 4] 19 92 N H L U ni te d St at es (Io w a an d M in ne so ta ) Po pu la tio n- ba se d ca se -c on tr ol 19 80 -1 98 3 W hi te m en ag ed 30 ye ar s in Io w a an d M in ne so ta ,e xc lu di ng M in ne ap ol is ,S t. Pa ul , D ul ut h, an d Ro ch es te r Ca se s: Io w a St at e H ea lth Re gi st ry an d sp ec ia ls ur ve ill an ce of M in ne so ta ho sp ita la nd pa th ol og y la bo ra to ry re co rd s Co nt ro ls :r an do m -d ig it di al in g if ag ed < 65 ye ar s, M ed ic ar e fi le s if ag ed 65 ye ar s, st at e de at h ce rt ifi ca te fi le s if de ce as ed Ca se s: 89 % Co nt ro ls :7 7% ra nd om -d ig it di al in g, 79 % M ed ic ar e, 77 % pr ox ie s fo rd ec ea se d Ca se s: 62 2 Co nt ro ls :1 24 5 Ca se s: 18 4 (3 0% ) Co nt ro ls :4 25 (3 4% ) Co cc o et al .[1 8] 20 13 B- ce ll N H L Eu ro pe (C ze ch Re pu bl ic ,F ra nc e, G er m an y, Ire la nd , Ita ly ,a nd Sp ai n) Po pu la tio n- an d ho sp ita l-b as ed ca se -c on tr ol 19 98 -2 00 4 Pe rs on s ag ed 17 ye ar s in G er m an y an d Ita ly ge ne ra lp op ul at io ns , an d in re fe rr al ar ea s of pa rt ic ip at in g ho sp ita ls in Cz ec h Re pu bl ic , Fr an ce ,I re la nd ,a nd Sp ai n Ca se s: N R Co nt ro ls :r an do m sa m pl in g of po pu la tio n re gi st er s in G er m an y an d Ita ly ;r ec ru itm en tf ro m ho sp ita ld ep ar tm en ts fo r in fe ct io us an d pa ra si tic (1 7. 6% ), m en ta la nd ne rv ou s (1 4. 6% ), ci rc ul at or y (8 .7 % ), di ge st iv e (7 .1 % ), en do cr in e an d m et ab ol ic (4 .1 % ), re sp ira to ry (3 .9 % ), an d se ve ra lo th er co nd iti on s (3 3. 2% ), ex cl ud in g ca nc er ,i n Cz ec h Re pu bl ic ,F ra nc e, Ire la nd ,a nd Sp ai n Ca se s: 88 % ov er al l; 90 % Cz ec h Re pu bl ic ,9 1% Fr an ce ,8 7% G er m an y, 90 % Ire la nd ,9 3% Ita ly , 82 % Sp ai n Co nt ro ls :6 9% ov er al l, 81 % ho sp ita l-b as ed ,5 2% po pu la tio n- ba se d; 60 % Cz ec h Re pu bl ic ,7 4% Fr an ce ,4 4% G er m an y, 75 % Ire la nd ,6 6% Ita ly ,9 6% Sp ai n Ca se s: 23 48 Co nt ro ls :2 46 2 N on e D e Ro os et al .[1 3] 20 03 N H L U ni te d St at es (N eb ra sk a, Io w a, M in ne so ta ,a nd Ka ns as ) Po pu la tio n- ba se d ca se -c on tr ol (p oo le d an al ys is of 3 st ud ie s) 19 79 -1 98 6 W hi te m en ag ed 21 ye ar s in on e of th e 66 co un tie s of ea st er n N eb ra sk a; w hi te m en ag ed 30 ye ar s in Io w a an d M in ne so ta , ex cl ud in g M in ne ap ol is , St .P au l, D ul ut h, an d Ro ch es te r; w hi te m en ag ed 21 ye ar s in Ka ns as Ca se s: N eb ra sk a Ly m ph om a St ud y G ro up an d ar ea ho sp ita ls ;I ow a St at e H ea lth Re gi st ry ;s pe ci al su rv ei lla nc e of M in ne so ta ho sp ita la nd pa th ol og y la bo ra to ry re co rd s; U ni ve rs ity of Ka ns as Ca nc er D at a Se rv ic e re gi st ry Co nt ro ls :r an do m -d ig it di al in g if ag ed < 65 ye ar s, M ed ic ar e fi le s if ag ed 65 ye ar s, st at e de at h ce rt ifi ca te fi le s if de ce as ed Ca se s: 91 % N eb ra sk a (9 3% liv in g, 89 % de ce as ed ); 89 % Io w a an d M in ne so ta ;9 6% Ka ns as Co nt ro ls :8 5% N eb ra sk a; 77 % ra nd om -d ig it di al in g, 79 % M ed ic ar e, 77 % de ce as ed (p ro xi es )I ow a an d M in ne so ta ; 93 % Ka ns as An al ys is re st ric te d to su bj ec ts w ho liv ed or w or ke d on a fa rm be fo re 18 ye ar s of ag e (% N R) ;a na ly si s of m ul tip le pe st ic id es re st ric te d to su bj ec ts w ith no n- m is si ng da ta (7 5% ca se s, 75 % co nt ro ls ) Ca se s: 65 0 (in an al ys es of m ul tip le pe st ic id es ) Co nt ro ls :1 93 3 (in an al ys es of m ul tip le pe st ic id es ) Ca se s: 20 1 (3 0. 9% )( in an al ys es of m ul tip le pe st ic id es ) Co nt ro ls :7 67 (3 9. 7% ) (in an al ys es of m ul tip le pe st ic id es ) D e Ro os et al .[1 2] 20 05 LH C, N H L, M M ,l eu ke m ia U ni te d St at es (Io w a an d N or th Ca ro lin a) Pr os pe ct iv e co ho rt 19 93 -1 99 7 th ro ug h 20 01 M ed ia n D 6. 7 ye ar s Pr iv at e an d co m m er ci al pe st ic id e ap pl ic at or s in Io w a an d N or th Ca ro lin a w ho w er e lic en se d to ap pl y re st ric te d- us e pe st ic id es Pe st ic id e ap pl ic at or s id en tifi ed w he n se ek in g a st at e- is su ed re st ric te d- us e pe st ic id e lic en se ; in vi te d to co m pl et e th e en ro llm en tq ue st io nn ai re at th e lic en si ng fa ci lit y 29 8 su bj ec ts (0 .5 % )l os tt o fo llo w - up or w ith no pe rs on -t im e co nt rib ut ed > 80 % of el ig ib le pe st ic id e ap pl ic at or s en ro lle d in st ud y by co m pl et in g on -s ite qu es tio nn ai re 44 % of ap pl ic at or s co m pl et ed ta ke -h om e qu es tio nn ai re El ig ib le co ho rt :3 6, 50 9- 49 ,2 11 in an al ys es ad ju st ed fo r de m og ra ph ic s an d lif es ty le 30 ,6 13 -4 0, 71 9 in an al ys es ad di tio na lly ad ju st ed fo r ot he rp es tic id es N on e Er ik ss on et al .[1 4] 20 08 N H L, B- ce ll N H L, SL L/ CL L, FL gr ad es I-I II, D LB CL , ot he rs pe ci fi ed B- ce ll N H L, un sp ec ifi ed B- ce ll N H L, T- ce ll N H L, un sp ec ifi ed N H L Eu ro pe (S w ed en ) Po pu la tio n- ba se d ca se -c on tr ol 19 99 -2 00 2 Ad ul ts ag ed 18 -7 4 ye ar s in 4 of 7 he al th se rv ic e re gi on s in Sw ed en as so ci at ed w ith un iv er si ty ho sp ita ls in Lu nd ,L in k€ o pi ng , € O re br o, an d U m ea Ca se s: co nt ac tw ith tr ea tin g ph ys ic ia ns an d pa th ol og is ts Co nt ro ls :n at io na lp op ul at io n re gi st ry Ca se s: 81 % Co nt ro ls :6 5% (9 2% of in iti al ly en ro lle d co nt ro ls w ith 71 % pa rt ic ip at io n) Ca se s: 99 5 Co nt ro ls :1 01 6 N on e 406 Ta bl e 1. (C on tin ue d ) Au th or s Ye ar O ut co m es st ud ie d St ud y lo ca tio n St ud y de si gn St ud y ye ar s So ur ce po pu la tio n Su bj ec ti de nt ifi ca tio n Su bj ec tp ar tic ip at io n Su bj ec ts (n ) Pr ox y re sp on de nt s H ar de ll an d Er ik ss on [2 7] 19 99 N H L Eu ro pe (S w ed en ) Po pu la tio n- ba se d ca se -c on tr ol 19 87 -1 99 0 M en ag ed 25 ye ar s in th e fo ur no rt he rn m os t co un tie s of Sw ed en an d th re e co un tie s in m id - Sw ed en Ca se s: re gi on al ca nc er re gi st rie s Co nt ro ls :n at io na lp op ul at io n re gi st ry if liv in g, na tio na lr eg is tr y fo rc au se s of de at h if de ce as ed Ca se s: 91 % (9 1% liv in g, 92 % de ce as ed ) Co nt ro ls :8 4% (8 3% liv in g, 85 % de ce as ed ) Ca se s: 40 4 Co nt ro ls :7 41 Ca se s: 17 7 (4 4% ) Co nt ro ls :N R (» 44 % ; m at ch ed to ca se s) H ar de ll et al .[1 5] 20 02 N H L in cl ud in g ha iry -c el l le uk em ia Eu ro pe (S w ed en ) Po pu la tio n- ba se d ca se -c on tr ol 19 87 -1 99 0 M en ag ed 25 ye ar s in th e fo ur no rt he rn m os t co un tie s of Sw ed en an d th re e co un tie s in m id - Sw ed en (fo rN H L) or in th e en tir e co un tr y of Sw ed en (fo rh ai ry -c el l le uk em ia ) Ca se s: re gi on al ca nc er re gi st rie s fo rN H L, na tio na lc an ce rr eg is tr y fo rh ai ry -c el ll eu ke m ia Co nt ro ls :n at io na lp op ul at io n re gi st ry ,n at io na lr eg is tr y fo r ca us es of de at h if de ce as ed Ca se s: 91 % Co nt ro ls :8 4% Ca se s: 51 5 Co nt ro ls :1 14 1 Ca se s: »3 5% (N R) Co nt ro ls :» 29 % (N R) H oh en ad el et al .[2 8] 20 11 N H L Ca na da (A lb er ta , Br iti sh Co lu m bi a, M an ito ba ,O nt ar io , Q ue be c, an d Sa sk at ch ew an ) Po pu la tio n- ba se d ca se -c on tr ol 19 91 -1 99 4 M en ag ed 19 ye ar s in Al be rt a, Br iti sh Co lu m bi a, M an ito ba , O nt ar io ,Q ue be c, an d Sa sk at ch ew an Ca se s: ho sp ita lr ec or ds in Q ue be c, ca nc er re gi st rie s in al lo th er pr ov in ce s Co nt ro ls :p ro vi nc ia lh ea lth in su ra nc e re co rd s in Al be rt a, Sa sk at ch ew an ,M an ito ba ,a nd Q ue be c; co m pu te riz ed te le ph on e lis tin gs in O nt ar io ; vo te rl is ts in Br iti sh Co lu m bi a Ca se s: 67 % Co nt ro ls :4 8% Ba se d on po st al co de s, re sp on de nt s w er e no tm or e or le ss lik el y th an no n- re sp on de nt s to liv e in a ru ra la re a. Ca se s: 51 3 Co nt ro ls :1 50 6 Ca se s: 11 0 (2 1% ) Co nt ro ls :2 20 (1 5% ) Ka ch ur ie ta l.[ 33 ] 20 13 M M Ca na da (A lb er ta , Br iti sh Co lu m bi a, M an ito ba ,O nt ar io , Q ue be c, an d Sa sk at ch ew an ) Po pu la tio n- ba se d ca se -c on tr ol 19 91 -1 99 4 M en ag ed 19 ye ar s ( 30 ye ar s in an al ys is )i n Al be rt a, Br iti sh Co lu m bi a, M an ito ba , O nt ar io ,Q ue be c, an d Sa sk at ch ew an Ca se s: ho sp ita lr ec or ds in Q ue be c, ca nc er re gi st rie s in al lo th er pr ov in ce s Co nt ro ls :p ro vi nc ia lh ea lth in su ra nc e re co rd s in Al be rt a, Sa sk at ch ew an ,M an ito ba ,a nd Q ue be c; co m pu te riz ed te le ph on e lis tin gs in O nt ar io ; vo te rl is ts in Br iti sh Co lu m bi a Ca se s: 58 % Co nt ro ls :4 8% Ba se d on po st al co de s, re sp on de nt s w er e no tm or e or le ss lik el y th an no n- re sp on de nt s to liv e in a ru ra la re a. Ca se s: 34 2 Co nt ro ls :1 35 7 Ca se s: 10 3 (3 0% ) Co nt ro ls :2 02 (1 5% ) Ka ru na na ya ke et al .[3 1] 20 12 H L Ca na da (A lb er ta , Br iti sh Co lu m bi a, M an ito ba ,O nt ar io , Q ue be c, an d Sa sk at ch ew an ) Po pu la tio n- ba se d ca se -c on tr ol 19 91 -1 99 4 M en ag ed 19 ye ar s in Al be rt a, Br iti sh Co lu m bi a, M an ito ba , O nt ar io ,Q ue be c, an d Sa sk at ch ew an Ca se s: ho sp ita lr ec or ds in Q ue be c, ca nc er re gi st rie s in al lo th er pr ov in ce s Co nt ro ls :p ro vi nc ia lh ea lth in su ra nc e re co rd s in Al be rt a, Sa sk at ch ew an ,M an ito ba ,a nd Q ue be c; co m pu te riz ed te le ph on e lis tin gs in O nt ar io ; vo te rl is ts in Br iti sh Co lu m bi a Ca se s: 68 % Co nt ro ls :4 8% Ba se d on po st al co de s, re sp on de nt s w er e no tm or e or le ss lik el y th an no n- re sp on de nt s to liv e in a ru ra la re a. Ca se s: 31 6 Co nt ro ls :1 50 6 Ca se s: N R Co nt ro ls :2 20 (1 5% ) Ka uf m an et al .[3 6] 20 09 Le uk em ia Ba ng ko k, Th ai la nd H os pi ta l-b as ed ca se -c on tr ol 19 97 -2 00 3 Pa tie nt s ag ed 18 ye ar s re si di ng in Ba ng ko k pr op er an d su bu rb s of N on th ab ur i, N ak or np at ho m , Pa tu m th an i, Sa m ut pr ak ar n, an d Sa m us ak or n, ad m itt ed to Si rir aj H os pi ta lo r D ho nb ur iH os pi ta l Ca se s: ho sp ita lr ec or ds Co nt ro ls :h os pi ta lr ec or ds fo ra cu te in fe ct io n or in fl am m at io n (3 3% ), tr au m a (2 2% ), ac ut e ab do m in al em er ge nc ie s su ch as ap pe nd ic iti s (2 7% ), or va rio us ot he rd ia gn os es w ith el ec tiv e ad m is si on ,s uc h as ca ta ra ct , he rn ia re pa ir, or co sm et ic su rg er y (1 7% ), ex cl ud in g he ad tr au m a w ith lo ss of co ns ci ou sn es s or ca nc er ;c on tr ol s at D ho nb ur iH os pi ta l( a ne ar by pr iv at e ho sp ita l) m at ch ed to 21 ca se s ad m itt ed to pr iv at e w ar ds fo rw ea lth y pa tie nt s Ca se s: 10 0% Co nt ro ls :1 00 % Ca se s: 18 0 Co nt ro ls :7 56 N on e Le e et al .[2 9] 20 04 N H L U ni te d St at es (N eb ra sk a, Io w a, an d M in ne so ta ) Po pu la tio n- ba se d ca se -c on tr ol (p oo le d an al ys is of 2 st ud ie s) 19 80 -1 98 6 W hi te m en an d w om en ag ed 21 ye ar s in on e of 45 co un tie s in ea st er n N eb ra sk a; w hi te m en ag ed 30 ye ar s in Io w a an d M in ne so ta , ex cl ud in g M in ne ap ol is , St .P au l, D ul ut h, an d Ro ch es te r Ca se s: N eb ra sk a Ly m ph om a St ud y G ro up an d ar ea ho sp ita ls ;I ow a St at e H ea lth Re gi st ry ;s pe ci al su rv ei lla nc e of M in ne so ta ho sp ita la nd pa th ol og y la bo ra to ry re co rd s Co nt ro ls :r an do m -d ig it di al in g if ag ed < 65 ye ar s, M ed ic ar e fi le s if ag ed 65 ye ar s, st at e de at h ce rt ifi ca te fi le s if de ce as ed Ca se s: 91 % N eb ra sk a, 89 % Io w a an d M in ne so ta Co nt ro ls :8 5% N eb ra sk a, 78 % Io w a an d M in ne so ta Ca se s: 87 2 Co nt ro ls :2 33 6 Ca se s: 26 6 (3 1% ) Co nt ro ls :7 79 (3 3% ) (C on tin ue d on ne xt pa ge ) 407 Ta bl e 1. (C on tin ue d ) Au th or s Ye ar O ut co m es st ud ie d St ud y lo ca tio n St ud y de si gn St ud y ye ar s So ur ce po pu la tio n Su bj ec ti de nt ifi ca tio n Su bj ec tp ar tic ip at io n Su bj ec ts (n ) Pr ox y re sp on de nt s M cD uf fi e et al .[1 6] 20 01 N H L Ca na da (A lb er ta , Br iti sh Co lu m bi a, M an ito ba ,O nt ar io , Q ue be c, an d Sa sk at ch ew an ) Po pu la tio n- ba se d ca se -c on tr ol 19 91 -1 99 4 M en ag ed 19 ye ar s in Al be rt a, Br iti sh Co lu m bi a, M an ito ba , O nt ar io ,Q ue be c, an d Sa sk at ch ew an Ca se s: ho sp ita lr ec or ds in Q ue be c, ca nc er re gi st rie s in al lo th er pr ov in ce s Co nt ro ls :p ro vi nc ia lh ea lth in su ra nc e re co rd s in Al be rt a, Sa sk at ch ew an ,M an ito ba ,a nd Q ue be c; co m pu te riz ed te le ph on e lis tin gs in O nt ar io ; vo te rl is ts in Br iti sh Co lu m bi a Ca se s: 67 % Co nt ro ls :4 8% Ba se d on po st al co de s, re sp on de nt s w er e no tm or e or le ss lik el y th an no n- re sp on de nt s to liv e in a ru ra la re a. Ca se s: 51 7 Co nt ro ls :1 50 6 Ca se s: »2 1% (N R) Co nt ro ls :2 20 (1 5% ) N or ds tr € om et al .[3 0] 19 98 H ai ry -c el ll eu ke m ia Eu ro pe (S w ed en ) Po pu la tio n- ba se d ca se -c on tr ol 19 87 -1 99 2 (1 99 3 fo r on e ca se ) M en liv in g in Sw ed en Ca se s: na tio na lc an ce rr eg is tr y Co nt ro ls :n at io na lp op ul at io n re gi st ry Ca se s: 91 % Co nt ro ls :8 3% Ca se s: 11 1 Co nt ro ls :4 00 Ca se s: 4 (4 % ) Co nt ro ls :5 (1 % ) O rs ie ta l.[ 17 ] 20 09 LH C, N H L, D LB CL ,F L, LP S, CL L, ha iry -c el ll eu ke m ia , H L, M M Eu ro pe (F ra nc e) H os pi ta l-b as ed ca se - co nt ro l 20 00 -2 00 4 M en ag ed 20 -7 5 ye ar s liv in g in th e ca tc hm en t ar ea s of th e m ai n ho sp ita ls in Br es t, Ca en , N an te s, Li lle ,T ou lo us e, an d Bo rd ea ux ,w ith no hi st or y of im m un os up pr es si on or ta ki ng im m un os up pr es sa nt dr ug s Ca se s: ho sp ita lr ec or ds Co nt ro ls :h os pi ta lr ec or ds fo r or th op ed ic or rh eu m at ol og ic al co nd iti on s (8 9. 3% ), ga st ro in te st in al or ge ni to ur in ar y tr ac td is ea se s (4 .8 % ), ca rd io va sc ul ar di se as es (1 .1 % ), sk in an d su bc ut an eo us tis su e di se as e (1 .8 % ), an d in fe ct io ns (3 .0 % ), ex cl ud in g pa tie nt s ad m itt ed fo rc an ce ro ra di se as e di re ct ly re la te d to oc cu pa tio n, sm ok in g, or al co ho la bu se Ca se s: 95 .7 % Co nt ro ls :9 1. 2% Ca se s: 49 1 LH C, 24 4 N H L, 10 4 LP S, 87 H L, 56 M M Co nt ro ls :4 56 N on e Pa hw a et al .[3 4] 20 12 M M Ca na da (A lb er ta , Br iti sh Co lu m bi a, M an ito ba ,O nt ar io , Q ue be c, an d Sa sk at ch ew an ) Po pu la tio n- ba se d ca se -c on tr ol 19 91 -1 99 4 M en ag ed 19 ye ar s in Al be rt a, Br iti sh Co lu m bi a, M an ito ba , O nt ar io ,Q ue be c, an d Sa sk at ch ew an Ca se s: ho sp ita lr ec or ds in Q ue be c, ca nc er re gi st rie s in al lo th er pr ov in ce s Co nt ro ls :p ro vi nc ia lh ea lth in su ra nc e re co rd s in Al be rt a, Sa sk at ch ew an ,M an ito ba ,a nd Q ue be c; co m pu te riz ed te le ph on e lis tin gs in O nt ar io ; vo te rl is ts in Br iti sh Co lu m bi a Ca se s: 58 % Co nt ro ls :4 8% Ba se d on po st al co de s, re sp on de nt s w er e no tm or e or le ss lik el y th an no n- re sp on de nt s to liv e in a ru ra la re a. Ca se s: 34 2 Co nt ro ls :1 50 6 Ca se s: 10 3 (3 0% ) Co nt ro ls :2 20 (1 5% ) So ra ha n[ 26 ] 20 15 M M U ni te d St at es (Io w a an d N or th Ca ro lin a) Pr os pe ct iv e co ho rt 19 93 -1 99 7 th ro ug h 20 01 M ed ia n D 6. 7 ye ar s Pr iv at e an d co m m er ci al pe st ic id e ap pl ic at or s in Io w a an d N or th Ca ro lin a w ho w er e lic en se d to ap pl y re st ric te d- us e pe st ic id es Pe st ic id e ap pl ic at or s id en tifi ed w he n se ek in g a st at e- is su ed re st ric te d- us e pe st ic id e lic en se ; in vi te d to co m pl et e th e en ro llm en tq ue st io nn ai re at th e lic en si ng fa ci lit y 29 8 su bj ec ts (0 .5 % )l os tt o fo llo w - up or w ith no pe rs on -t im e co nt rib ut ed > 80 % of el ig ib le pe st ic id e ap pl ic at or s en ro lle d in st ud y by co m pl et in g on -s ite qu es tio nn ai re 44 % of ap pl ic at or s co m pl et ed ta ke -h om e qu es tio nn ai re El ig ib le co ho rt (1 ): 54 ,3 15 ex cl ud in g su bj ec ts w ith ca nc er be fo re en ro llm en t, lo ss to fo llo w -u p, m is si ng ag e at en ro llm en t, or m is si ng gl yp ho sa te us e 49 ,2 11 al so ex cl ud in g m is si ng ed uc at io n, sm ok in g, or al co ho l 40 ,7 19 ex cl ud in g m is si ng ot he rp es tic id es El ig ib le co ho rt (2 ): 53 ,6 56 ex cl ud in g su bj ec ts w ith ca nc er be fo re en ro llm en t, lo ss to fo llo w -u p, m is si ng ag e at en ro llm en t, m is si ng gl yp ho sa te us e, or m is si ng cu m ul at iv e ex po su re da ys of gl yp ho sa te us e 53 ,3 04 al so ex cl ud in g m is si ng in te ns ity of gl yp ho sa te us e El ig ib le co ho rt (3 ): 55 ,9 34 ex cl ud in g su bj ec ts w ith ca nc er be fo re en ro llm en t, lo ss to fo llo w -u p, or m is si ng ag e at en ro llm en t N on e (C on tin ue d on ne xt pa ge ) 408 Ta bl e 1. Co nt in ue d (a dd iti on al co lu m ns ). Au th or s Ye ar Ex po su re as se ss m en t O ut co m e as se ss m en t In ve st ig at or bl in di ng Co nf ou nd er s co ns id er ed or ad ju st ed Fu nd in g so ur ce O ve rla p Br ow n et al .[3 5] 19 90 In -p er so n st ru ct ur ed in te rv ie w ,i nc lu di ng de ta ile d fa rm in g an d pe st ic id e us e hi st or y Fo re ac h pe st ic id e, ev al ua te d ev er us e, fi rs ta nd la st ye ar of us e, an d pe rs on al ap pl yi ng /m ix in g/ ha nd lin g In 19 87 ,s up pl em en ta lt el ep ho ne in te rv ie w to ev al ua te us ua ln um be ro fd ay s of pe st ic id e us e pe ry ea ra m on g Io w a su bj ec ts w ho ha d re po rt ed ag ric ul tu ra lu se of sp ec ifi c pe st ic id es D ia gn os tic co nfi rm at io n by re gi on al pa th ol og is ts ;s pe ci al re vi ew of m ye lo dy sp la si as by on e pa th ol og is tc o- au th or N o Ad ju st ed :v ita ls ta tu s, ag e, st at e, ev er us ed to ba cc o da ily ,fi rs t- de gr ee fa m ily hi st or y of LH C, no n- fa rm in g jo b re la te d to le uk em ia ris k in th is st ud y, ex po su re to su bs ta nc es (b en ze ne ,n ap ht ha ,h ai r dy es )r el at ed to le uk em ia ris k in th is st ud y Pa rt ia ls up po rt fr om N at io na lI ns tit ut e of En vi ro nm en ta lH ea lth Sc ie nc es Br ow n et al .[3 2] ;C an to re ta l.[ 24 ] ; D e Ro os et al .[1 3] ;L ee et al .[2 9] Br ow n et al .[3 2] 19 93 In -p er so n st ru ct ur ed in te rv ie w ,i nc lu di ng de ta ile d fa rm in g an d pe st ic id e us e hi st or y Fo re ac h pe st ic id e, ev al ua te d ev er us e, fi rs ta nd la st ye ar of us e, pe rs on al ap pl yi ng /m ix in g/ ha nd lin g, an d us e of pr ot ec tiv e eq ui pm en t D ia gn os tic co nfi rm at io n by an ex pe rt pa th ol og is t N o Ad ju st ed :v ita ls ta tu s, ag e Co ns id er ed :s m ok in g, ed uc at io n, ot he rf ac to rs fo un d no tt o be co nf ou nd er s of ag ric ul tu ra lr is k fa ct or s Pa rt ia ls up po rt fr om N at io na lI ns tit ut e of En vi ro nm en ta lH ea lth Sc ie nc es Br ow n et al .[3 5] ;C an to re ta l.[ 24 ] ; D e Ro os et al .[1 3] ;L ee et al .[2 9] Ca nt or et al .[2 4] 19 92 In -p er so n st ru ct ur ed in te rv ie w ,i nc lu di ng de ta ile d fa rm in g an d pe st ic id e us e hi st or y of al ls ub je ct s w ho ha d w or ke d on a fa rm fo r 6 m on th s si nc e ag e 18 ye ar s Fo re ac h pe st ic id e, ev al ua te d ev er us e, fi rs ta nd la st ye ar of us e, m et ho d of ap pl ic at io n, pe rs on al ap pl yi ng /m ix in g/ ha nd lin g, an d us e of pr ot ec tiv e eq ui pm en t D ia gn os tic co nfi rm at io n an d m or ph ol og ic al cl as si fi ca tio n by pa ne lo f4 ex pe rie nc ed re gi on al pa th ol og is ts N o Ad ju st ed :v ita ls ta tu s, st at e, ag e, ci ga re tt e sm ok in g st at us ,fi rs t- de gr ee fa m ily hi st or y of LH C, no n- fa rm in g jo b re la te d to N H L ris k in th is st ud y, ex po su re to ha ir dy es ,e xp os ur e to ot he r su bs ta nc es as so ci at ed w ith N H L ris k in th is st ud y Co ns id er ed :p es tic id es be lo ng in g to ot he rc he m ic al fa m ili es Pa rt ia ls up po rt fr om N at io na lI ns tit ut e of En vi ro nm en ta lH ea lth Sc ie nc es Br ow n et al .[3 5] ;B ro w n et al .[3 2] ; D e Ro os et al .[1 3] ;L ee et al .[2 9] Co cc o et al .[1 8] 20 13 In -p er so n st ru ct ur ed in te rv ie w ,i nc lu di ng de ta ile d fa rm in g an d pe st ic id e us e hi st or y fo ra ll su bj ec ts w ho re po rt ed ha vi ng w or ke d in ag ric ul tu re Fo re ac h ag ric ul tu ra lj ob ,r ep or te d ta sk s, cr op s, si ze of cu lti va te d ar ea ,p es ts tr ea te d, pe st ic id es us ed , cr op tr ea tm en tp ro ce du re s, us e of pe rs on al pr ot ec tiv e eq ui pm en t, re -e nt ry af te rt re at m en t, an d fr eq ue nc y of tr ea tm en ti n da ys pe ry ea r H is to lo gi ca lly or cy to lo gi ca lly co nfi rm ed ca se s w ith ce nt ra l re vi ew of sl id es of »2 0% by an in te rn at io na lt ea m of pa th ol og is ts N o Ad ju st ed :a ge ,g en de r, ed uc at io n, st ud y ce nt er Eu ro pe an Co m m is si on ,5 th an d 6t h Fr am ew or k Pr og ra m m es ;S pa ni sh M in is tr y of H ea lth ;G er m an Fe de ra lO ffi ce fo rR ad ia tio n Pr ot ec tio n; La Fo nd at io n de Fr an ce ;I ta lia n M in is tr y fo rE du ca tio n, U ni ve rs ity an d Re se ar ch ;I ta lia n As so ci at io n fo rC an ce rR es ea rc h N on e D e Ro os et al .[1 3] 20 03 Te le ph on e in te rv ie w in N eb ra sk a an d Ka ns as ;i n- pe rs on st ru ct ur ed in te rv ie w in Io w a an d M in ne so ta N eb ra sk a: Q ue st io n ab ou tu se of an y pe st ic id e, fo llo w ed by pr om pt in g fo rs pe ci fi c se le ct ed pe st ic id es ,i nc lu di ng ye ar s of us e an d av er ag e da ys pe ry ea r Io w a an d M in ne so ta :D ire ct qu es tio n ab ou ta se le ct ed us e of sp ec ifi c pe st ic id es ,i nc lu di ng fi rs t an d la st ye ar s of us e Ka ns as :O pe n- en de d qu es tio n ab ou tu se of pe st ic id es ,f ol lo w ed by qu es tio ns on du ra tio n of us e an d da ys pe ry ea rf or gr ou ps of pe st ic id es bu t no ti nd iv id ua lp es tic id es (w ith va lid at io n st ud y) N eb ra sk a: Pa th ol og y re vi ew w ith hi st ol og ic al co nfi rm at io n an d cl as si fi ca tio n in cl ud in g im m un ol og ic ph en ot yp in g Io w a an d M in ne so ta :D ia gn os tic co nfi rm at io n an d m or ph ol og ic al cl as si fi ca tio n by pa ne lo f4 ex pe rie nc ed re gi on al pa th ol og is ts Ka ns as :D ia gn os tic co nfi rm at io n an d cl as si fi ca tio n by pa ne lo f3 pa th ol og is ts Ye s in N eb ra sk a; no in Io w a, M in ne so ta , an d Ka ns as Ad ju st ed :a ge ,s tu dy si te ,o th er in di vi du al pe st ic id es w ith 20 us er s in fu ll st ud y Co ns id er ed :fi rs t- de gr ee fa m ily hi st or y of LH C, ed uc at io n, sm ok in g N R; as su m e N at io na lC an ce rI ns tit ut e Br ow n et al .[3 5] ;B ro w n et al .[3 2] ; Ca nt or et al .[2 4] ;L ee et al .[2 9] (a ls o H oa re ta l.[ 47 ] ; H oa rZ ah m et al .[4 8] ) D e Ro os et al .[1 2] 20 05 Se lf- ad m in is te re d w rit te n qu es tio nn ai re (w ith va lid at io n st ud y) ev al ua tin g de ta ile d us e of 22 pe st ic id es fo rp riv at e ap pl ic at or s, 28 pe st ic id es fo r co m m er ci al ap pl ic at or s (e ve r/ ne ve ru se , fr eq ue nc y, du ra tio n, an d in te ns ity of us e, de ca de of fi rs tu se ), an d ev er /n ev er us e fo ra dd iti on al pe st ic id es up to to ta lo f5 0, w ith ge ne ra l in fo rm at io n on pe st ic id e ap pl ic at io n m et ho ds , pe rs on al pr ot ec tiv e eq ui pm en t, pe st ic id e m ix in g, an d eq ui pm en tr ep ai r Ad di tio na ls el f-a dm in is te re d ta ke -h om e qu es tio nn ai re w ith fu rt he rq ue st io ns on oc cu pa tio na le xp os ur es an d lif es ty le fa ct or s Li nk ag e to st at e ca nc er re gi st ry fi le s, st at e de at h re gi st rie s, an d N at io na lD ea th In de x N on e Ad ju st ed :a ge at en ro llm en t, ed uc at io n, ci ga re tt e sm ok in g pa ck -y ea rs ,a lc oh ol co ns um pt io n in pa st ye ar ,fi rs t- de gr ee fa m ily hi st or y of ca nc er ,s ta te of re si de nc e Co ns id er ed (a dj us te d fo rM M on ly ): 5 pe st ic id es fo r w hi ch cu m ul at iv e ex po su re -d ay s w er e m os t hi gh ly as so ci at ed w ith th os e fo rg ly ph os at e (i. e. , 2, 4- di ch lo ro ph en ox ya ce tic ac id ,a la ch lo r, at ra zi ne , m et ol ac hl or ,t rifl ur al in ), 5 pe st ic id es fo rw hi ch ev er /n ev er us e w as m os th ig hl y as so ci at ed w ith th at fo rg ly ph os at e (i. e. ,b en om yl ,m an eb , pa ra qu at ,c ar ba ry l, di az in on ) N at io na lC an ce rI ns tit ut e, N at io na lI ns tit ut e of En vi ro nm en ta lH ea lth Sc ie nc es , En vi ro nm en ta lP ro te ct io n Ag en cy ,a nd N at io na lI ns tit ut e fo rO cc up at io na lS af et y an d H ea lth So ra ha n[ 26 ] Er ik ss on et al .[1 4] 20 08 Se lf- ad m in is te re d m ai le d qu es tio nn ai re w ith ad di tio na lt el ep ho ne in te rv ie w fo rm is si ng or un cl ea ra ns w er s; ev al ua te d oc cu pa tio na le xp os ur e to in di vi du al pe st ic id es ,i nc lu di ng nu m be ro f ye ar s, nu m be ro fd ay s pe ry ea r, an d ap pr ox im at e le ng th of ex po su re pe rd ay D ia gn os tic pa th ol og ic al sp ec im en s ex am in ed an d cl as si fi ed by 1 of 5 Sw ed is h ex pe rt ly m ph om a re fe re nc e pa th ol og is ts ,i fn ot al re ad y in iti al ly re vi ew ed by on e of th em ;p an el re vi ew if cl as si fi ca tio n di ffe re d fr om or ig in al re po rt Ye s Ad ju st ed :a ge ,s ex ,a nd ye ar of di ag no si s or en ro llm en t; ot he ra ss oc ia te d ag en ts (4 -c hl or o- 2- m et hy lp he no xy ac et ic ac id ,2 ,4 - di ch lo ro ph en ox ya ce tic ac id an d/ or 2, 4, 5- tr ic hl or op he no xy ac et ic ac id ,m er cu ria ls ee d dr es si ng ,a rs en ic ,c re os ot e, ta r) fo rN H L on ly Sw ed is h Co un ci lf or W or ki ng Li fe an d So ci al Re se ar ch ;C an ce ra nd Al le rg y Fu nd ;K ey Fu nd ;€ O re br o U ni ve rs ity H os pi ta lC an ce r Fu nd N on e (C on tin ue d on ne xt pa ge ) 409 Ta bl e 1. (C on tin ue d ) Au th or s Ye ar Ex po su re as se ss m en t O ut co m e as se ss m en t In ve st ig at or bl in di ng Co nf ou nd er s co ns id er ed or ad ju st ed Fu nd in g so ur ce O ve rla p H ar de ll an d Er ik ss on [2 7] 19 99 Se lf- ad m in is te re d m ai le d qu es tio nn ai re w ith su pp le m en ta lt el ep ho ne in te rv ie w fo ru nc le ar an sw er s; as se ss ed us e of pe st ic id es w ith in di ffe re nt oc cu pa tio ns ,w et co nt ac ti fn ot ha nd lin g th e sp ra ye r, br an d na m es of pe st ic id es ,y ea rs of ex po su re ,a nd cu m ul at iv e da ys of ex po su re Ex po su re ex cl ud ed 1 ye ar pr io rt o di ag no si s or in de x ye ar H is to pa th ol og ic al di ag no si s of N H L re po rt ed to re gi on al ca nc er re gi st rie s, co nfi rm ed by re vi ew of pa th ol og y re po rt s Ye s Ad ju st ed :a ge ,c ou nt y, vi ta ls ta tu s, ye ar of de at h if de ce as ed ,u se of ph en ox ya ce tic ac id s Sw ed is h W or k En vi ro nm en tF un d, Sw ed is h M ed ic al Re se ar ch Co un ci l, € O re br o Co un ty Co un ci lR es ea rc h Co m m itt ee ,€ O re br o M ed ic al Ce nt er Re se ar ch Fo un da tio n H ar de ll et al .[1 5] H ar de ll et al .[1 5] 20 02 Se lf- ad m in is te re d m ai le d qu es tio nn ai re w ith su pp le m en ta lt el ep ho ne in te rv ie w fo ru nc le ar an sw er s; as se ss ed ye ar s an d to ta ln um be ro fd ay s of oc cu pa tio na le xp os ur e to va rio us ag en ts an d na m es of ag en ts Ex po su re de fi ne d as 1 w or ki ng da y w ith in du ct io n pe rio d of 1 ye ar H is to lo gi ca lly ve rifi ed N H L; co nfi rm at io n of ha iry -c el l le uk em ia N R Ye s Ad ju st ed :s tu dy ,s tu dy ar ea ,v ita ls ta tu s, ot he r as so ci at ed pe st ic id es (4 -c hl or o- 2- m et hy l ph en ox ya ce tic ac id ,2 ,4 -d ic hl or op he no xy ac et ic ac id C 2, 4, 5- tr ic hl or op he no xy ac et ic ac id ,o th er he rb ic id es ) Sw ed is h Ca nc er Re se ar ch Fu nd ,S w ed is h M ed ic al Re se ar ch Co un ci l, € O re br o Co un ty Co un ci lR es ea rc h Co m m itt ee ,€ O re br o M ed ic al Ce nt re Re se ar ch Fo un da tio n H ar de ll an d Er ik ss on [2 7] N or ds tr € om et al .[3 0] H oh en ad el et al .[2 8] 20 11 Te le ph on e in te rv ie w fo rd et ai le d in fo rm at io n on pe st ic id e us e in su bj ec ts w ho re po rt ed in a se lf- ad m in is te re d m ai lq ue st io nn ai re th at th ey ha d 10 ho ur s of pe st ic id e us e du rin g th ei r lif et im e, pl us 15 % ra nd om sa m pl e of su bj ec ts w ith < 10 ho ur s Pe st ic id e in te rv ie w (w ith va lid at io n st ud y) in cl ud ed a pr e- m ai le d lis to fs pe ci fi c pe st ic id es (c he m ic al an d tr ad e na m es )w ith nu m be ro fd ay s us ed an d nu m be ro fh ou rs pe rd ay at ho m e or w or k fo r ea ch pe st ic id e D ia gn os tic co nfi rm at io n ba se d on in fo rm at io n, in cl ud in g pa th ol og y re po rt s, fr om ca nc er re gi st rie s an d ho sp ita ls ;p at ho lo gi ca l m at er ia lr ev ie w ed an d cl as si fi ed by a re fe re nc e pa th ol og is t; su bj ec ts w ith un av ai la bl e pa th ol og ic al m at er ia lr et ai ne d in st ud y N o Ad ju st ed :a ge ,p ro vi nc e, us e of a pr ox y re sp on de nt Co ns id er ed :d ie se le xh au st ,u ltr av io le tr ad ia tio n, fa rm an im al s, ch em ic al s su ch as be nz en e, fi rs t- de gr ee fa m ily hi st or y of ca nc er H ea lth Ca na da ,B rit is h Co lu m bi a H ea lth Re se ar ch Fo un da tio n, Ce nt re fo r Ag ric ul tu ra lM ed ic in e at U ni ve rs ity of Sa sk at ch ew an Ka ch ur ie ta l.[ 33 ] ; Ka ru na na ya ke et al .[3 1] ; M cD uf fi e et al .[1 6] ; Pa hw a et al .[3 4] Ka ch ur ie ta l.[ 33 ] 20 13 Te le ph on e in te rv ie w fo rd et ai le d in fo rm at io n on pe st ic id e us e in su bj ec ts w ho re po rt ed in a se lf- ad m in is te re d m ai lq ue st io nn ai re th at th ey ha d 10 ho ur s of pe st ic id e us e du rin g th ei r lif et im e, pl us 15 % ra nd om sa m pl e of su bj ec ts w ith < 10 ho ur s Pe st ic id e in te rv ie w (w ith va lid at io n st ud y) in cl ud ed a pr e- m ai le d lis to fs pe ci fi c pe st ic id es (c he m ic al an d tr ad e na m es )w ith nu m be ro fd ay s us ed an d nu m be ro fh ou rs pe rd ay at ho m e or w or k fo r ea ch pe st ic id e D ia gn os tic co nfi rm at io n ba se d on in fo rm at io n, in cl ud in g pa th ol og y re po rt s, fr om ca nc er re gi st rie s an d ho sp ita ls ;p at ho lo gi ca l m at er ia lr ev ie w ed an d cl as si fi ed by a re fe re nc e pa th ol og is t (in cl ud in g pa th ol og y an d tu m or tis su e sl id es fo r1 25 [3 7% ]o f3 42 ca se s) ;s ub je ct s w ith un av ai la bl e pa th ol og ic al m at er ia lr et ai ne d in st ud y N o Ad ju st ed :a ge ,p ro vi nc e, us e of a pr ox y re sp on de nt , sm ok in g st at us ,p er so na lh is to ry of rh eu m at oi d ar th rit is ,a lle rg ie s, m ea sl es ,s hi ng le s, or ca nc er , fa m ily hi st or y of ca nc er O cc up at io na lC an ce rR es ea rc h Ce nt re ; Ca nc er Ca re O nt ar io ;O nt ar io W or kp la ce Sa fe ty an d In su ra nc e Bo ar d; Ca na di an Ca nc er So ci et y, O nt ar io D iv is io n, M ita cs - Ac ce le ra te G ra du at e Re se ar ch In te rn sh ip Pr og ra m H oh en ad el et al .[2 8] ; Ka ru na na ya ke et al .[3 1] ; M cD uf fi e et al .[1 6] ; Pa hw a et al .[3 4] Ka ru na na ya ke et al .[3 1] 20 12 Te le ph on e in te rv ie w fo rd et ai le d in fo rm at io n on pe st ic id e us e in su bj ec ts w ho re po rt ed in a se lf- ad m in is te re d m ai lq ue st io nn ai re th at th ey ha d 10 ho ur s/ ye ar of cu m ul at iv e ex po su re to an y co m bi na tio n of he rb ic id es ,i ns ec tic id es , fu ng ic id es ,f um ig an ts ,a nd al gi ci de s Pe st ic id e in te rv ie w co lle ct ed in fo rm at io n on ex po su re to in di vi du al pe st ic id es ,p la ce of pe st ic id e us e, ye ar of fi rs tu se ,fi rs ty ea ro n m ar ke t, nu m be ro fy ea rs of us e, an d da ys pe ry ea ro fu se [N ot e di ffe re nc es fr om re la te d st ud ie s] In iti al di ag no si s ba se d on in fo rm at io n fr om ca nc er re gi st rie s an d ho sp ita ls ; pa th ol og y an d tu m or tis su e sl id es fo r1 55 of 31 6 ca se s re vi ew ed by a re fe re nc e pa th ol og is tw ho co nfi rm ed H L in 15 0/ 15 5 ca se s, pl us 7 ca se s or ig in al ly cl as si fi ed as N H L; su bj ec ts w ith un av ai la bl e pa th ol og ic al m at er ia lr et ai ne d in st ud y N o Ad ju st ed :a ge ,p ro vi nc e, pe rs on al hi st or y of m ea sl es , ac ne ,h ay fe ve r, or sh in gl es ,fi rs t- de gr ee fa m ily hi st or y of ca nc er N R; as su m e sa m e as in re la te d st ud ie s H oh en ad el et al .[2 8] ; Ka ch ur ie ta l.[ 33 ] ; M cD uf fi e et al .[1 6] ; Pa hw a et al .[3 4] Ka uf m an et al .[3 6] 20 09 In te rv ie w w ith nu rs e to as se ss oc cu pa tio na la nd no n- oc cu pa tio na le xp os ur e to pe st ic id es an d ot he rp ot en tia lr is k fa ct or s H is to lo gi ca lly co nfi rm ed le uk em ia di ag no se d w ith in 6 m on th s be fo re cu rr en th os pi ta l at te nd an ce or ad m is si on N o Co ns id er ed :a ge ,s ex ,i nc om e, us e of ce llu la r te le ph on es ,b en ze ne an d ot he rs ol ve nt ex po su re , oc cu pa tio na la nd no n- oc cu pa tio na lp es tic id e ex po su re ,p es tic id es us ed ne ar ho m e, w or ki ng w ith po w er lin es ,l iv in g ne ar po w er lin es , ex po su re to X- ra ys ,e xp os ur e to ce rt ai n ty pe s of el ec tr om ag ne tic fi el ds ,u se of ha ir dy es Th ai la nd Re se ar ch Fu nd an d Co m m is si on on H ig he rE du ca tio n N on e Le e et al .[2 9] 20 04 Te le ph on e in te rv ie w in N eb ra sk a; in -p er so n st ru ct ur ed in te rv ie w in Io w a an d M in ne so ta Q ue st io ns in cl ud ed pe rs on al ha nd lin g of gr ou ps of pe st ic id es an d in di vi du al pe st ic id es us ed on cr op s or an im al s, w ith ye ar s of fi rs ta nd la st us e N eb ra sk a: Pa th ol og y re vi ew w ith hi st ol og ic al co nfi rm at io n an d cl as si fi ca tio n in cl ud in g im m un ol og ic ph en ot yp in g Io w a an d M in ne so ta :D ia gn os tic co nfi rm at io n an d m or ph ol og ic al cl as si fi ca tio n by pa ne lo f4 ex pe rie nc ed re gi on al pa th ol og is ts Ye s in N eb ra sk a; no in Io w a an d M in ne so ta Ad ju st ed :a ge ,s ta te ,v ita ls ta tu s Co ns id er ed :g en de r, sm ok in g, fi rs t- de gr ee fa m ily hi st or y of LH C, ev er ha vi ng a jo b co rr el at ed w ith ris k of LH C (e .g ., pa in tin g or w el di ng ), us e of pr ot ec tiv e eq ui pm en t N R; as su m e N at io na lC an ce rI ns tit ut e Br ow n et al .[3 5] ; Br ow n et al .[3 2] ; Ca nt or et al .[2 4] ; D e Ro os et al .[1 3] (a ls o H oa rZ ah m et al .[4 8] ) 410 Ta bl e 1. (C on tin ue d ) Au th or s Ye ar Ex po su re as se ss m en t O ut co m e as se ss m en t In ve st ig at or bl in di ng Co nf ou nd er s co ns id er ed or ad ju st ed Fu nd in g so ur ce O ve rla p M cD uf fi e et al .[1 6] 20 01 Te le ph on e in te rv ie w fo rd et ai le d in fo rm at io n on pe st ic id e us e in su bj ec ts w ho re po rt ed in a se lf- ad m in is te re d m ai lq ue st io nn ai re th at th ey ha d 10 ho ur s of pe st ic id e us e du rin g th ei rl ife tim e, pl us 15 % ra nd om sa m pl e of su bj ec ts w ith < 10 ho ur s (t ot al D 17 9 ca se s, 45 6 co nt ro ls w ith te le ph on e in te rv ie w ) Pe st ic id e in te rv ie w (w ith va lid at io n st ud y) in cl ud ed a pr e- m ai le d lis to fs pe ci fi c pe st ic id es (c he m ic al an d tr ad e na m es )w ith nu m be ro fd ay s us ed an d nu m be ro fh ou rs pe rd ay at ho m e or w or k fo r ea ch pe st ic id e D ia gn os tic co nfi rm at io n fr om ca nc er re gi st rie s an d ho sp ita ls ; pa th ol og ic al m at er ia lr ev ie w ed an d cl as si fi ed by a re fe re nc e pa th ol og is t; su bj ec ts w ith un av ai la bl e pa th ol og ic al m at er ia lr et ai ne d in st ud y N o Ad ju st ed :a ge ,p ro vi nc e, pe rs on al hi st or y of m ea sl es , m um ps ,c an ce r, or al le rg y de se ns iti za tio n sh ot s, fi rs t- de gr ee fa m ily hi st or y of ca nc er Co ns id er ed :p es tic id e ex po su re ,s m ok in g hi st or y H ea lth Ca na da ,B rit is h Co lu m bi a H ea lth Re se ar ch Fo un da tio n, Ce nt re fo r Ag ric ul tu ra lM ed ic in e at U ni ve rs ity of Sa sk at ch ew an H oh en ad el et al .[2 8] ; Ka ch ur ie ta l.[ 33 ] ; Ka ru na na ya ke et al .[3 1] ; Pa hw a et al .[3 4] N or ds tr € om et al .[3 0] 19 98 Se lf- ad m in is te re d m ai le d qu es tio nn ai re w ith su pp le m en ta lt el ep ho ne in te rv ie w fo ru nc le ar or m is si ng an sw er s; as se ss ed to ta ln um be ro fd ay s of oc cu pa tio na le xp os ur e to va rio us ag en ts Ex po su re de fi ne d as 1 w or ki ng da y w ith in du ct io n pe rio d of 1 ye ar Re po rt ed to na tio na lc an ce r re gi st ry ;f ur th er co nfi rm at io n no t de sc rib ed Ye s Ad ju st ed :a ge Co ns id er ed :e xp os ur e to an im al s, he rb ic id es , in se ct ic id es ,f un gi ci de s, im pr eg na tin g ag en ts , or ga ni c so lv en ts ,e xh au st s, or ul tr av io le tl ig ht Sw ed is h W or k En vi ro nm en tF un d, € O re br o Co un ty Co un ci lR es ea rc h Co m m itt ee , € O re br o M ed ic al Ce nt re Re se ar ch Fo un da tio n. H ar de ll et al .[1 5] O rs ie ta l.[ 17 ] 20 09 Se lf- ad m in is te re d w rit te n qu es tio nn ai re w ith lif et im e oc cu pa tio na lh is to ry ,f ol lo w ed by in - pe rs on st ru ct ur ed in te rv ie w ev al ua tin g no n- oc cu pa tio na le xp os ur e to pe st ic id es an d ag ric ul tu ra lq ue st io nn ai re fo rs ub je ct s w ho ha d w or ke d as a fa rm er or ga rd en er fo r 6 m on th s du rin g lif et im e Ag ric ul tu ra lq ue st io nn ai re co lle ct ed da ta on lo ca tio n of al lf ar m s w he re su bj ec th ad w or ke d fo r 6 m on th s, pe rio d of oc cu pa tio n an d ar ea ,f ar m er ’s st at us at ea ch fa rm ,c ro ps an d an im al hu sb an dr y w ith m ea n si ze s, al lp es tic id es us ed on ea ch cr op du rin g a gi ve n pe rio d, w he th er su bj ec th ad pe rs on al ly pr ep ar ed ,m ix ed ,o rs pr ay ed th e pe st ic id e, ch em ic al us ed ,b ra nd na m e, m ai n us e, ty pe of sp ra yi ng eq ui pm en tu se d, an nu al nu m be r an d du ra tio n of ap pl ic at io ns ,a nd us e of pe st ic id es in fa rm bu ild in gs fo ra ni m al s, gr ai n, ha y or st ra w , or to cl ea rl an es an d ya rd s Al lq ue st io nn ai re s re vi ew ed by an oc cu pa tio na l hy gi en is ta nd an ag ro no m is t; re pe at te le ph on e in te rv ie w s co nd uc te d to cl ar ify in fo rm at io n fr om 95 (5 6. 8% )o f1 58 su bj ec ts w ho co m pl et ed th e ag ric ul tu ra lq ue st io nn ai re ,n ot co m pl et ed by 35 (2 0. 8% )w ho re fu se d (n D 15 ), di ed /w er e in po or he al th (n D 10 ), or co ul d no tb e co nt ac te d (n D 15 ); al lc he m ic al s co de d us in g ad ho c sy st em an d cl as si fi ed as de fi ni te or po ss ib le ex po su re Al ld ia gn os es cy to lo gi ca lly or hi st ol og ic al ly co nfi rm ed an d re vi ew ed by a pa ne lo f pa th ol og is ts an d he m at ol og is ts Ye s Ad ju st ed :a ge ,s tu dy ce nt er ,s oc io ec on om ic ca te go ry Co ns id er ed :a ll co m bi na tio ns of pe st ic id e fa m ili es as so ci at ed w ith th e LH C su bt yp e co ns id er ed w ith a p- va lu e 0. 10 ,r ur al /u rb an st at us ,t yp e of ho us in g, ed uc at io na ll ev el ,h is to ry of m on on uc le os is ,h is to ry of in fl ue nz a im m un iz at io n, fa m ily hi st or y of ca nc er ,s ki n ch ar ac te ris tic s, sm ok in g st at us ,a nd al co ho ld rin ki ng st at us As so ci at io n po ur la Re ch er ch e co nt re le Ca nc er ,F on da tio n de Fr an ce ,A FS SE T, Fa be rg e em pl oy ee s (d on at io n) N on e Pa hw a et al .[3 4] 20 12 Te le ph on e in te rv ie w fo rd et ai le d in fo rm at io n on pe st ic id e us e in su bj ec ts w ho re po rt ed in a se lf- ad m in is te re d m ai lq ue st io nn ai re th at th ey ha d 10 h of pe st ic id e us e du rin g th ei rl ife tim e, pl us 15 % ra nd om sa m pl e of su bj ec ts w ith < 10 h Pe st ic id e in te rv ie w (w ith va lid at io n st ud y) in cl ud ed a pr e- m ai le d lis to fs pe ci fi c pe st ic id es (c he m ic al an d tr ad e na m es )w ith nu m be ro fd ay s us ed an d nu m be ro fh ou rs pe rd ay at ho m e or w or k fo r ea ch pe st ic id e D ia gn os tic co nfi rm at io n ba se d on in fo rm at io n, in cl ud in g pa th ol og y re po rt s, fr om ca nc er re gi st rie s an d ho sp ita ls ;p at ho lo gi ca l m at er ia lr ev ie w ed an d cl as si fi ed by a re fe re nc e pa th ol og is t (in cl ud in g pa th ol og y an d tu m or tis su e sl id es fo r1 25 [3 7% ]o f3 42 ca se s) ;s ub je ct s w ith un av ai la bl e pa th ol og ic al m at er ia lr et ai ne d in st ud y N o Ad ju st ed :a ge ,p ro vi nc e, pe rs on al hi st or y of m ea sl es , m um ps ,a lle rg ie s, ar th rit is ,o rs hi ng le s, fi rs t- de gr ee fa m ily hi st or y of ca nc er O cc up at io na lC an ce rR es ea rc h Ce nt re ; Ca nc er Ca re O nt ar io ;O nt ar io W or kp la ce Sa fe ty an d In su ra nc e Bo ar d; Ca na di an Ca nc er So ci et y H oh en ad el et al .[2 8] ; Ka ch ur ie ta l.[ 33 ] ; Ka ru na na ya ke et al .[3 1] ; M cD uf fi e et al .[1 6] (C on tin ue d on ne xt pa ge ) 411 Ta bl e 1. (C on tin ue d ) Au th or s Ye ar Ex po su re as se ss m en t O ut co m e as se ss m en t In ve st ig at or bl in di ng Co nf ou nd er s co ns id er ed or ad ju st ed Fu nd in g so ur ce O ve rla p So ra ha n[ 26 ] 20 15 Se lf- ad m in is te re d w rit te n qu es tio nn ai re (w ith va lid at io n st ud y) ev al ua tin g de ta ile d us e of 22 pe st ic id es fo rp riv at e ap pl ic at or s, 28 pe st ic id es fo r co m m er ci al ap pl ic at or s (e ve r/ ne ve ru se , fr eq ue nc y, du ra tio n, an d in te ns ity of us e, de ca de of fi rs tu se ), an d ev er /n ev er us e fo ra dd iti on al pe st ic id es up to to ta lo f5 0, w ith ge ne ra l in fo rm at io n on pe st ic id e ap pl ic at io n m et ho ds , pe rs on al pr ot ec tiv e eq ui pm en t, pe st ic id e m ix in g, an d eq ui pm en tr ep ai r Ad di tio na ls el f-a dm in is te re d ta ke -h om e qu es tio nn ai re w ith fu rt he rq ue st io ns on oc cu pa tio na le xp os ur es an d lif es ty le fa ct or s M is si ng da ta cl as si fi ed in to “n ot kn ow n/ m is si ng ” ca te go ry ,w ith un kn ow n us e of 2, 4- di ch lo ro ph en ox ya ce tic ac id cl as si fi ed w ith no us e an d un kn ow n ed uc at io n cl as si fi ed w ith no ed uc at io n be yo nd hi gh sc ho ol du e to la ck of M M ca se s in un kn ow n ca te go rie s Li nk ag e to st at e ca nc er re gi st ry fi le s, st at e de at h re gi st rie s, an d N at io na lD ea th In de x N on e Fu lly ad ju st ed :a ge ,g en de r, sm ok in g pa ck -y ea rs , al co ho lu se in ye ar be fo re en ro llm en t, fi rs t- de gr ee fa m ily hi st or y of ca nc er ,e du ca tio n, us e of 2, 4- di ch lo ro ph en ox ya ce tic ac id ,a la ch lo r, at ra zi ne , m et ol ac hl or ,o rt rifl ur al in ,e ve ru se of be no m yl , m an eb ,p ar aq ua t, ca rb ar yl ,o rd ia zi no n In te rm ed ia te ad ju st ed :a ge ,g en de r, sm ok in g, al co ho l, fa m ily hi st or y of ca nc er ,e du ca tio n Ad ju st ed in fu ll co ho rt :a ge ,g en de r, fa m ily hi st or y of ca nc er ,e du ca tio n M on sa nt o Eu ro pe SA /N V D e Ro os et al .[1 2] CI :c on fi de nc e in te rv al ;C LL :c hr on ic ly m ph oc yt ic le uk em ia ;D LB CL :d iff us e la rg e B- ce ll ly m ph om a; FL :f ol lic ul ar ly m ph om a; H L: H od gk in ly m ph om a; LH C: ly m ph oh em at op oi et ic ca nc er ;L PS :l ym ph op ro lif er at iv e sy nd ro m e; M M :m ul tip le m ye lo m a; N H L: no n- H od gk in ly m ph om a; N R: no tr ep or te d; O R: od ds ra tio ;S LL :s m al ll ym ph oc yt ic ly m ph om a. 412 and source population as Brown et al.,[32] which was described in the section on MM, and Cantor et al.,[24] which was included as part of De Roos et al.[13] in a pooled analysis of NHL. As described earlier, De Roos et al.,[12] the only prospective cohort study included, was based in North America (Iowa and North Carolina), enrolled both males and females, ascertained cancer incidence in the 1990s and 2000s, and had a 99.5% fol- low-up rate through 2001. In the total eligible cohort, 43 leuke- mia cases occurred among glyphosate users. Brown et al.[35] was a population-based case-control study set in North Amer- ica (Iowa and Minnesota), restricted to white males, with cases identified in 1980-1983, participation rates of 86% for cases and 77-79% for controls, and proxy respondent rates of 41% for cases and 34% for controls. Fifteen leukemia cases in this study were classified as having used glyphosate. The other case- control study of leukemia, by Kaufman et al.,[36] was a hospital- based study set in Asia (Thailand), with males and females, case ascertainment in the 1990s and 2000s, participation rates of 100%, and no proxy respondents for cases or controls. Meta-analysis NHL All relevant RRs and 95% CIs for the association between reported glyphosate use and risk of overall NHL, including those not used in the meta-analysis, such as estimates within subgroups, minimally adjusted estimates, and esti- mates of exposure-response patterns, are provided in Table 2. The estimates selected from each independent study population for inclusion in the meta-analysis, accord- ing to the rules specified in the methods section, are pro- vided in Table 3. As shown in Table 3 and Fig. 1, the combined meta-RR for overall NHL in association with any use of glyphosate, based on six studies,[12-17] was 1.3 (95% CI D 1.0-1.6). The results were identical in the random-effects and fixed- effects models, suggesting limited between-study heteroge- neity in the association. Little heterogeneity also was indi- cated by the I2 value of 0.0% and the highly non- significant P-value of 0.84 for Cochran’s Q. Given the lack of heterogeneity and at least one statistically significant association, we tested for publication bias using Egger’s linear regression approach to evaluating funnel plot asym- metry, and found no significant asymmetry (one-tailed P- value D 0.20). Using Duval and Tweedie’s trim-and-fill approach to adjust for publication bias, the imputed meta- RR for both the random-effects and fixed-effects models was 1.2 (95% CI D 1.0-1.6). In secondary analyses, we replaced the RR estimated by De Roos et al.[13] using a hierarchical (i.e., multistage) regression model with the RR estimated using a more traditional logistic regression model (Table 3). (The hierarchical regression RR was selected for the primary analysis because, as stated by the authors, hierarchical regression models can yield “increased pre- cision and accuracy for the ensemble of estimates” when model- ing multiple pesticides simultaneously, and the more conservative prior assumptions specified in these models “seemed appropriate in a largely exploratory analysis of multiple exposures for which there is little prior knowledge about how pesticide exposures interact in relation to the risk of NHL.”) Using the logistic regression RR did not appreciably affect the results of the meta-analysis (meta-RR D 1.3, 95% CI D 1.0-1.6; identical for random-effects and fixed-effects models). In another secondary analysis, we replaced the RR reported by McDuffie et al.[16] with the results reported by Hohenadel et al.[28] in the same study population (minus four previously misclassified NHL cases) (Table 3). Because Hohenadel et al.[28] reported two estimates for glyphosate use-one in the absence of malathion use and one in the presence of malathion use-we combined these two estimates into a single estimate (RR D 1.40, 95% CI D 0.62-3.15) using random-effects meta-analysis. Using this alternative estimate also did not appreciably affect the meta-RR (1.3, 95% CI D 1.0-1.7; identical for random- effects and fixed-effects models). Finally, using both the logistic regression RR instead of the hierarchical regression RR from De Roos et al.[13] and the combined RR from Hohenadel et al.[28] instead of the RR from McDuffie et al.[16] slightly but non-significantly increased the meta-RR to 1.4 (95% CI D 1.0- 1.8; identical for random-effects and fixed-effects models) (Table 3). As noted earlier, in their meta-analysis of the association between glyphosate use and NHL risk, Schinasi and Leon[11] included RR estimates from Eriksson et al.[14] and Hardell et al.[15] that were not the most highly adjusted esti- mates reported by the authors (shown in Table 2 as univari- ate odds ratios). They also used the logistic regression estimate from De Roos et al.[13] that arguably was not as highly adjusted as the hierarchical regression estimate. When we included these estimates in the meta-analysis, along with the same estimates from De Roos et al.,[13] McDuffie et al.,[16] and Orsi et al.[17] as included in our main meta-analysis, we obtained the same results as reported by Schinasi and Leon:[11] random-effects meta-RR D 1.5, 95% CI D 1.1-2.0 (I2 D 32.7%, pheterogeneity D 0.19). The fixed-effects meta-RR based on these estimates (not reported by Schinasi and Leon[11]) was 1.4 (95% CI D 1.1- 1.8). NHL subtypes All reported RRs and 95% CIs for the association between glyphosate use and risk of various NHL subtypes are shown in Table 2. The estimates included in meta-analyses, which were conducted for B-cell lymphoma, diffuse large B-cell lymphoma, chronic lymphocytic leukemia/small lymphocytic lymphoma, follicular lymphoma, and hairy-cell leukemia (i.e., all NHL sub- types for which at least two estimates from independent studies were available), are shown in Table 3. Too few studies of any given NHL subtype were conducted to justify testing for publi- cation bias. The meta-RR for the association between any use of glyphosate and risk of B-cell lymphoma, based on two stud- ies,[14,18] was 2.0 (95% CI D 1.1-3.6) according to both the random-effects and the fixed-effects model (I2 D 0.0%, phe- terogeneity D 0.58) (Table 3). These results are the same as reported by Schinasi and Leon.[11] The four B-cell lym- phoma cases who were classified by Cocco et al.[18] as hav- ing used glyphosate consisted of one patient with diffuse large B-cell lymphoma, one with chronic lymphocytic JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 413 Ta bl e 2. Es tim at ed as so ci at io ns be tw ee n gl yp ho sa te ex po su re an d ris k of ly m ph oh em at op oi et ic ca nc er (L H C) , in cl ud in g no n- H od gk in ly m ph om a (N H L) , N H L su bt yp es , H od gk in ly m ph om a (H L) , m ul tip le m ye lo m a (M M ), an d le uk em ia . Au th or s Ye ar Ex po su re gr ou ps an d nu m be ro fs ub je ct s Re la tiv e ris k 95 % CI Br ow n et al .[3 5] 19 90 N on -fa rm er s: 24 3 ca se s, 54 7 co nt ro ls Ev er m ix ed ,h an dl ed ,o ra pp lie d gl yp ho sa te :1 5 ca se s, 49 co nt ro ls Le uk em ia O R D 0. 9 Le uk em ia 95 % CI D 0. 5- 1. 6 Br ow n et al .[3 2] 19 93 N on -fa rm er s: 62 ca se s, 27 2 co nt ro ls Ev er m ix ed ,h an dl ed ,o ra pp lie d gl yp ho sa te :1 1 ca se s, 40 co nt ro ls M M O R D 1. 7 Am on g th os e w ho di d no tu se pr ot ec tiv e eq ui pm en t, M M O R D 1. 9 M M 95 % CI D 0. 8- 3. 6 Am on g th os e w ho di d no tu se pr ot ec tiv e eq ui pm en t, M M 95 % CI D N R Ca nt or et al .[2 4] 19 92 N on -fa rm er s: 22 6 ca se s, 54 7 co nt ro ls Ev er ha nd le d, m ix ed ,o ra pp lie d gl yp ho sa te :2 6 ca se s, 49 co nt ro ls N H L O R D 1. 1 N H L 95 % CI D 0. 7- 1. 9 Co cc o et al .[1 8] 20 13 U ne xp os ed to an y pe st ic id es :N R ca se s, 22 62 co nt ro ls O cc up at io na lly ex po se d to gl yp ho sa te :4 ca se s (1 D LB CL ,1 CL L, 1 M M ,1 un sp ec ifi ed B- ce ll N H L) ,2 co nt ro ls B- ce ll N H L O R D 3. 1 B- ce ll N H L 95 % CI D 0. 6- 17 .1 D e Ro os et al .[1 3] 20 03 U ne xp os ed to gl yp ho sa te :6 14 ca se s, 18 92 co nt ro ls Ex po se d to gl yp ho sa te :3 6 ca se s, 61 co nt ro ls H ie ra rc hi ca lr eg re ss io n N H L O R D 1. 6 Lo gi st ic re gr es si on N H L O R D 2. 1 H ie ra rc hi ca lr eg re ss io n N H L 95 % CI D 0. 9- 2. 8 Lo gi st ic re gr es si on N H L 95 % CI D 1. 1- 4. 0 D e Ro os et al .[1 2] 20 05 N ev er us ed gl yp ho sa te :4 7 LH C, 21 N H L, 8 M M ,1 4 le uk em ia ; 13 ,2 80 co ho rt m em be rs Ev er us ed gl yp ho sa te :1 43 LH C, 71 N H L, 24 M M ,4 3 le uk em ia ; 41 ,0 35 co ho rt m em be rs Fu lly ad ju st ed LH C RR D 1. 1 Ag e- ad ju st ed LH C RR D 1. 1 Fu lly ad ju st ed N H L RR D 1. 1 Ag e- ad ju st ed N H L RR D 1. 2 Fu lly ad ju st ed M M RR D 2. 6 (2 .6 in Io w a, 2. 7 in N or th Ca ro lin a) Ag e- ad ju st ed M M RR D 1. 1 Fu lly ad ju st ed le uk em ia RR D 1. 0 Ag e- ad ju st ed le uk em ia RR D 1. 1 Fu lly ad ju st ed LH C 95 % CI D 0. 8- 1. 6 Ag e- ad ju st ed LH C 95 % CI D 0. 8- 1. 5 Fu lly ad ju st ed N H L 95 % CI D 0. 7- 1. 9 Ag e- ad ju st ed N H L 95 % CI D 0. 7- 1. 9 Fu lly ad ju st ed M M 95 % CI D 0. 7- 9. 4 Ag e- ad ju st ed M M 95 % CI D 0. 5- 2. 4 Fu lly ad ju st ed le uk em ia 95 % CI D 0. 5- 1. 9 Ag e- ad ju st ed le uk em ia 95 % CI D 0. 6- 2. 0 1- 20 gl yp ho sa te ex po su re da ys :4 8 LH C, 29 N H L, 8 M M ,9 le uk em ia 21 -5 6 gl yp ho sa te ex po su re da ys :3 8 LH C, 15 N H L, 5 M M ,1 4 le uk em ia 57 -2 ,6 78 gl yp ho sa te ex po su re da ys :3 6 LH C, 17 N H L, 6 M M ,9 le uk em ia 0. 1- 79 .5 in te ns ity -w ei gh te d gl yp ho sa te ex po su re da ys :3 8 LH C, 24 N H L, 5 M M ,7 le uk em ia 79 .6 -3 37 .1 in te ns ity -w ei gh te d gl yp ho sa te ex po su re da ys :4 0 LH C, 15 N H L, 6 M M ,1 7 le uk em ia 33 7. 2- 18 ,2 41 in te ns ity -w ei gh te d gl yp ho sa te ex po su re da ys :4 3 LH C, 22 N H L, 8 M M ,8 le uk em ia Cu m ul at iv e ex po su re da ys ,t er til es 2 an d 3 vs .1 LH C RR s D 1. 2, 1. 2; p- tr en d D 0. 69 N H L RR s D 0. 7, 0. 9; p- tr en d D 0. 73 M M RR s D 1. 1, 1. 9; p- tr en d D 0. 27 Le uk em ia RR s D 1. 9, 1. 0; p- tr en d D 0. 61 > 10 8 vs .> 0- 9 ex po su re da ys ,N H L RR D 0. 9 In te ns ity -w ei gh te d ex po su re da ys ,t er til es 2 an d 3 vs .1 LH C RR s D 1. 0, 1. 0; p- tr en d D 0. 90 N H L RR s D 0. 6, 0. 8; p- tr en d D 0. 99 M M RR s D 1. 2, 2. 1; p- tr en d D 0. 17 Le uk em ia RR s D 1. 9, 0. 7; p- tr en d D 0. 11 In te ns ity te rt ile 3 vs .1 M M RR D 0. 6 Cu m ul at iv e ex po su re da ys ,t er til es 1, 2, an d 3 vs .n ev er M M RR s D 2. 3, 2. 6, 4. 4; p- tr en d D 0. 09 Cu m ul at iv e ex po su re da ys ,q ua rt ile 4 vs .n ev er M M RR D 6. 6; p- tr en d D 0. 01 Cu m ul at iv e ex po su re da ys ,t er til es 2 an d 3 vs .1 LH C 95 % CI s D 0. 8- 1. 8, 0. 8- 1. 8 N H L 95 % CI s D 0. 4- 1. 4, 0. 5- 1. 6 M M 95 % CI s D 0. 4- 3. 5, 0. 6- 6. 3 Le uk em ia 95 % CI s D 0. 8- 4. 5, 0. 4- 2. 9 > 10 8 vs .> 0- 9 ex po su re da ys ,N H L 95 % CI D 0. 4- 2. 1 In te ns ity -w ei gh te d ex po su re da ys ,t er til es 2 an d 3 vs .1 LH C 95 % CI s D 0. 6- 1. 5, 0. 7- 1. 6 N H L 95 % CI s D 0. 3- 1. 1, 0. 5- 1. 4 M M 95 % CI s D 0. 4- 3. 8, 0. 6- 7. 0 Le uk em ia 95 % CI s D 0. 8- 4. 7, 0. 2- 2. 1 In te ns ity te rt ile 3 vs .1 M M 95 % CI D 0. 2- 1. 8 Cu m ul at iv e ex po su re da ys ,t er til es 1, 2, an d 3 vs .n ev er M M 95 % CI s D 0. 6- 8. 9, 0. 6- 11 .5 ,1 .0 -2 0. 2 Cu m ul at iv e ex po su re da ys ,q ua rt ile 4 vs .n ev er M M 95 % CI D 1. 4- 30 .6 Er ik ss on et al .[1 4] 20 08 N o pe st ic id e ex po su re :N R G ly ph os at e ex po su re fo r 1 fu ll w or ki ng da y, 1 ca le nd ar ye ar pr io rt o ye ar of di ag no si s or en ro llm en t: 29 N H L ca se s, 18 co nt ro ls (N H L su bt yp es N R) G ly ph os at e ex po su re fo r1 to 10 da ys :1 2 N H L ca se s, 9 co nt ro ls G ly ph os at e ex po su re fo r> 10 da ys :1 7 N H L ca se s, 9 co nt ro ls N H L O R, an y gl yp ho sa te ,m ul tiv ar ia te D 1. 51 N H L O R, an y gl yp ho sa te ,u ni va ria te D 2. 02 N H L O R, gl yp ho sa te 1 to 10 da ys D 1. 69 N H L O R, gl yp ho sa te > 10 da ys D 2. 36 N H L O R, an y gl yp ho sa te ,l at en cy 1- 10 ye ar s D 1. 11 N H L O R, an y gl yp ho sa te ,l at en cy > 10 ye ar s D 2. 26 N H L 95 % CI ,a ny gl yp ho sa te ,m ul tiv ar ia te D 0. 77 -2 .9 4 N H L 95 % CI ,a ny gl yp ho sa te ,u ni va ria te D 1. 10 -3 .7 1 N H L 95 % CI ,g ly ph os at e 1 to 10 da ys D 0. 70 -4 .0 7 N H L 95 % CI ,g ly ph os at e > 10 da ys D 1. 04 -5 .3 7 N H L 95 % CI ,a ny gl yp ho sa te ,l at en cy 1- 10 ye ar s D 0. 24 -5 .0 8 N H L 95 % CI ,a ny gl yp ho sa te ,l at en cy > 10 ye ar s D 1. 16 -4 .4 0 B- ce ll N H L O R, an y gl yp ho sa te D 1. 87 SL L/ CL L O R, an y gl yp ho sa te D 3. 35 FL gr ad es I- III O R, an y gl yp ho sa te D 1. 89 D LB CL O R, an y gl yp ho sa te D 1. 22 O th er sp ec ifi ed B- ce ll N H L O R, an y gl yp ho sa te D 1. 63 U ns pe ci fi ed B- ce ll N H L O R, an y gl yp ho sa te D 1. 47 T- ce ll N H L O R, an y gl yp ho sa te D 2. 29 U ns pe ci fi ed N H L O R, an y gl yp ho sa te D 5. 63 B- ce ll N H L 95 % CI ,a ny gl yp ho sa te D 0. 99 8- 3. 51 SL L/ CL L 95 % CI ,a ny gl yp ho sa te D 1. 42 -7 .8 9 FL gr ad es I- III 95 % CI ,a ny gl yp ho sa te D 0. 62 -5 .7 9 D LB CL 95 % CI ,a ny gl yp ho sa te D 0. 44 -3 .3 5 O th er sp ec ifi ed B- ce ll N H L 95 % CI ,a ny gl yp ho sa te D 0. 53 -4 .9 6 U ns pe ci fi ed B- ce ll N H L 95 % CI ,a ny gl yp ho sa te D 0. 33 -6 .6 1 T- ce ll N H L 95 % CI ,a ny gl yp ho sa te D 0. 51 -1 0. 4 U ns pe ci fi ed N H L 95 % CI ,a ny gl yp ho sa te D 1. 44 -2 2. 0 H ar de ll an d Er ik ss on [2 7] 19 99 N o pe st ic id e ex po su re G ly ph os at e ex po su re 1 ye ar pr io rt o di ag no si s or co nt ro li nd ex ye ar :4 ca se s, 3 co nt ro ls N H L O R ad ju st ed fo rp he no xy ac et ic ac id s D 5. 8 N H L O R un ad ju st ed fo rp he no xy ac et ic ac id s D 2. 3 N H L 95 % CI ad ju st ed fo rp he no xy ac et ic ac id s D 0. 6- 54 N H L 95 % CI un ad ju st ed fo rp he no xy ac et ic ac id s D 0. 4- 13 H ar de ll et al .[1 5] 20 02 N o pe st ic id e ex po su re :N R G ly ph os at e ex po su re fo r 1 w or ki ng da y, 1 ye ar pr io rt o di ag no si s or co nt ro li nd ex da te :8 ca se s, 8 co nt ro ls M ul tiv ar ia te N H L O R D 1. 85 U ni va ria te N H L O R D 3. 04 M ul tiv ar ia te N H L 95 % CI D 0. 55 -6 .2 0 U ni va ria te N H L 95 % CI D 1. 08 -8 .5 2 H oh en ad el et al .[2 8] 20 11 U se of ne ith er gl yp ho sa te no rm al at hi on :4 22 ca se s, 13 01 co nt ro ls U se of gl yp ho sa te on ly :1 9 ca se s, 78 co nt ro ls U se of m al at hi on on ly :4 1 ca se s, 72 co nt ro ls U se of gl yp ho sa te an d m al at hi on :3 1 ca se s, 55 co nt ro ls N H L O R, gl yp ho sa te on ly D 0. 92 N H L O R, m al at hi on on ly D 1. 95 N H L O R, gl yp ho sa te an d m al at hi on D 2. 10 In te ra ct io n co nt ra st ra tio D 0. 23 ,P -in te ra ct io n D 0. 69 N H L 95 % CI ,g ly ph os at e on ly D 0. 54 -1 .5 5 N H L 95 % CI ,m al at hi on on ly D 1. 29 -2 .9 3 N H L 95 % CI ,g ly ph os at e an d m al at hi on D 1. 31 -3 .3 7 414 Ta bl e 2. (C on tin ue d ) Au th or s Ye ar Ex po su re gr ou ps an d nu m be ro fs ub je ct s Re la tiv e ris k 95 % CI Ka ch ur ie ta l.[ 33 ] 20 13 N ev er us ed gl yp ho sa te :3 10 ca se s, 12 36 co nt ro ls (2 16 ca se s, 10 47 co nt ro ls w ith ou tp ro xy ) Ev er us ed gl yp ho sa te :3 2 ca se s, 12 1 co nt ro ls (2 3 ca se s, 10 8 co nt ro ls w ith ou tp ro xy ) U se d gl yp ho sa te fo r> 0 to 2 da ys pe ry ea r: 15 ca se s, 88 co nt ro ls (1 1 ca se s, 78 co nt ro ls w ith ou tp ro xy ) U se d gl yp ho sa te fo r> 2 da ys pe ry ea r: 12 ca se s, 29 co nt ro ls (1 0 ca se s, 26 co nt ro ls w ith ou tp ro xy ) M M O R, ev er gl yp ho sa te D 1. 19 M M O R, ev er gl yp ho sa te ,n o pr ox ie s D 1. 11 M M O R, gl yp ho sa te > 0 to 2 da ys pe ry ea rD 0. 72 M M O R, gl yp ho sa te > 0 to 2 da ys pe ry ea r, no pr ox ie s D 0. 70 M M O R, gl yp ho sa te > 2 da ys pe ry ea rD 2. 04 M M O R, gl yp ho sa te > 2 da ys pe ry ea r, no pr ox ie s D 2. 11 M M 95 % CI ,e ve rg ly ph os at e D 0. 76 -1 .8 7 M M 95 % CI ,e ve rg ly ph os at e, no pr ox ie s D 0. 66 -1 .8 6 M M 95 % CI ,g ly ph os at e > 0 to 2 da ys pe ry ea rD 0. 39 -1 .3 2 M M 95 % CI ,g ly ph os at e > 0 to 2 da ys pe ry ea r, no pr ox ie s D 0. 35 -1 .4 0 M M 95 % CI ,g ly ph os at e > 2 da ys pe ry ea rD 0. 98 -4 .2 3 M M 95 % CI ,g ly ph os at e > 2 da ys pe ry ea r, no pr ox ie s D 0. 95 -4 .7 0 Ka ru na na ya ke et al .[3 1] 20 12 N ev er us ed gl yp ho sa te :2 78 ca se s, 13 73 co nt ro ls Ev er us ed gl yp ho sa te :3 8 ca se s, 13 3 co nt ro ls Fu lly ad ju st ed H L O R D 0. 99 M in im al ly ad ju st ed (a ge ,p ro vi nc e) H L O R D 1. 14 Fu lly ad ju st ed H L 95 % CI D 0. 62 -1 .5 6 M in im al ly ad ju st ed (a ge ,p ro vi nc e) H L 95 % CI D 0. 74 -1 .7 6 Ka uf m an et al .[3 6] 20 09 N o gl yp ho sa te us e: 17 9 ca se s, 75 3 co nt ro ls G ly ph os at e: 1 ca se ,3 co nt ro ls Cr ud e le uk em ia O R D 1. 40 Cr ud e le uk em ia 95 % CI D 0. 15 -1 3. 56 Le e et al .[2 9] 20 04 N on -fa rm er s, no n- as th m at ic s: 25 9 ca se s, 68 4 co nt ro ls N on -fa rm er s, as th m at ic s: 9 ca se s, 37 co nt ro ls Ex po se d to gl yp ho sa te ,n on -a st hm at ic s: 53 ca se s, 91 co nt ro ls Ex po se d to gl yp ho sa te ,a st hm at ic s: 6 ca se s, 12 co nt ro ls N H L O R, no n- fa rm er s, as th m at ic s D 0. 6 N H L O R, gl yp ho sa te ,n on -a st hm at ic s D 1. 4 N H L O R, gl yp ho sa te ,a st hm at ic s D 1. 2 N H L 95 % CI ,n on -fa rm er s, as th m at ic s D 0. 3- 1. 4 N H L 95 % CI ,g ly ph os at e, no n- as th m at ic s D 0. 98 -2 .1 N H L 95 % CI ,g ly ph os at e, as th m at ic s D 0. 4- 3. 3 M cD uf fi e et al .[1 6] 20 01 N ev er us ed gl yp ho sa te :4 66 ca se s, 13 73 co nt ro ls Ev er us ed gl yp ho sa te :5 1 ca se s, 15 06 co nt ro ls G ly ph os at e us e fo r> 0 to 2 da ys pe ry ea r G ly ph os at e us e fo r> 2 da ys pe ry ea r Fu lly ad ju st ed N H L O R, ev er gl yp ho sa te D 1. 20 M in im al ly ad ju st ed (a ge ,p ro vi nc e) N H L O R, ev er gl yp ho sa te D 1. 26 M in im al ly ad ju st ed N H L O R, gl yp ho sa te > 0 to 2 da ys pe ry ea rD 1. 00 M in im al ly ad ju st ed N H L O R, gl yp ho sa te > 2 da ys pe ry ea rD 2. 12 Fu lly ad ju st ed N H L 95 % CI ,e ve rg ly ph os at e D 0. 83 -1 .7 4 M in im al ly ad ju st ed (a ge ,p ro vi nc e) N H L 95 % CI ,e ve rg ly ph os at e D 0. 87 -1 .8 0 M in im al ly ad ju st ed N H L 95 % CI ,g ly ph os at e > 0 to 2 da ys pe ry ea rD 0. 63 -1 .5 7 M in im al ly ad ju st ed N H L 95 % CI ,g ly ph os at e > 2 da ys pe ry ea rD 1. 20 -3 .7 3 N or ds tr € om et al .[3 0] 19 98 N o gl yp ho sa te ex po su re :1 07 ca se s, 39 5 co nt ro ls G ly ph os at e ex po su re fo r 1 w or ki ng da y, 1 ye ar pr io rt o di ag no si s or co nt ro li nd ex da te :4 ca se s, 5 co nt ro ls H ai ry -c el ll eu ke m ia O R D 3. 1 H ai ry -c el ll eu ke m ia 95 % CI D 0. 8- 12 O rs ie ta l.[ 17 ] 20 09 N ev er ex po se d to gl yp ho sa te :4 64 LH C, 23 2 N H L, 10 2 D LB CL ,4 7 FL ,1 00 LP S, 75 CL L, 25 ha iry -c el ll eu ke m ia 81 H L, 51 M M ,4 32 co nt ro ls Ev er ex po se d to gl yp ho sa te :2 7 LH C, 12 N H L, 5 D LB CL ,3 FL ,4 LP S, 2 CL L, 2 ha iry -c el ll eu ke m ia ,6 H L, 5 M M ,2 4 co nt ro ls LH C O R D 1. 2 N H L O R D 1. 0 D LB CL O R D 1. 0 FL O R D 1. 4 LP S O R D 0. 6 CL L O R D 0. 4 H ai ry -c el ll eu ke m ia O R D 1. 8 H L O R D 1. 7 M M O R D 2. 4 LH C 95 % CI D 0. 6- 2. 1 N H L 95 % CI D 0. 5- 2. 2 D LB CL 95 % CI D 0. 3- 2. 7 FL 95 % CI D 0. 4- 5. 2 LP S 95 % CI D 0. 2- 2. 1 CL L 95 % CI D 0. 1- 1. 8 H ai ry -c el ll eu ke m ia 95 % CI D 0. 3- 9. 3 H L 95 % CI D 0. 6- 5. 0 M M 95 % CI D 0. 8- 7. 3 Pa hw a et al .[3 4] 20 12 N ev er us ed gl yp ho sa te :3 10 ca se s, 13 73 co nt ro ls Ev er us ed gl yp ho sa te :3 2 ca se s, 13 3 co nt ro ls M M O R D 1. 22 M M 95 % CI D 0. 77 -1 .9 3 So ra ha n[ 26 ] 20 15 N ev er us ed gl yp ho sa te :8 ca se s, 13 ,2 80 co ho rt m em be rs (o f 54 ,3 15 ); 4 ca se s, 11 ,8 81 co ho rt m em be rs (o f4 9, 21 1) ;3 ca se s, 98 09 co ho rt m em be rs (o f4 0, 71 9) Ev er us ed gl yp ho sa te :2 4 ca se s, 41 ,0 35 co ho rt m em be rs (o f 54 ,3 15 ); 22 ca se s, 37 ,3 30 co ho rt m em be rs (o f4 9, 21 1) ; 19 ca se s, 30 ,9 10 co ho rt m em be rs (o f4 0, 71 9) Fu lly ad ju st ed M M RR ,c oh or to f5 4, 31 5 D 1. 24 Ag e- an d se x- ad ju st ed M M RR ,c oh or to f5 4, 31 5 D 1. 12 Ag e- ad ju st ed M M RR ,c oh or to f5 4, 31 5 D 1. 08 Ag e- ad ju st ed M M RR ,c oh or to f4 9, 21 1 D 1. 91 In te rm ed ia te ad ju st ed M M RR ,c oh or to f4 9, 21 1 D 2. 07 Ag e- ad ju st ed M M RR ,c oh or to f4 0, 71 9 D 2. 21 Fu lly ad ju st ed M M RR ,c oh or to f4 0, 71 9 D 2. 79 Fu lly ad ju st ed M M 95 % CI ,c oh or to f5 4, 31 5 D 0. 52 -2 .9 4 Ag e- an d se x- ad ju st ed M M 95 % CI ,c oh or to f5 4, 31 5 D 0. 50 -2 .4 9 Ag e- ad ju st ed M M 95 % CI ,c oh or to f5 4, 31 5 D 0. 48 -2 .4 1 Ag e- ad ju st ed M M 95 % CI ,c oh or to f4 9, 21 1 D 0. 66 -5 .5 3 In te rm ed ia te ad ju st ed M M 95 % CI ,c oh or to f4 9, 21 1 D 0. 71 -6 .0 4 Ag e- ad ju st ed M M 95 % CI ,c oh or to f4 0, 71 9 D 0. 65 -7 .4 8 Fu lly ad ju st ed M M 95 % CI ,c oh or to f4 0, 71 9 D 0. 78 -9 .9 6 1- 20 gl yp ho sa te ex po su re da ys :1 0 ca se s 21 -5 6 gl yp ho sa te ex po su re da ys :8 ca se s 57 -2 67 8 gl yp ho sa te ex po su re da ys :6 ca se s 0. 1- 79 .5 in te ns ity -w ei gh te d gl yp ho sa te ex po su re da ys :6 ca se s 79 .6 -3 37 .1 in te ns ity -w ei gh te d gl yp ho sa te ex po su re da ys :8 ca se s 33 7. 2- 18 ,2 41 in te ns ity -w ei gh te d gl yp ho sa te ex po su re da ys :1 0 ca se s Cu m ul at iv e ex po su re da ys ,t er til es 1, 2, an d 3 vs .n ev er Fu lly ad ju st ed M M RR s D 1. 14 ,1 .5 2, 1. 38 ;p -t re nd D 0. 48 us in g sc or es ,> 0. 50 us in g m ea ns In te rm ed ia te ad ju st ed M M RR s D 1. 13 ,1 .5 0, 1. 23 ;p -t re nd > 0. 50 us in g sc or es or m ea ns Ag e- an d se x- ad ju st ed M M RR s D 1. 06 ,1 .3 4, 1. 08 ;p -t re nd > 0. 50 us in g sc or es or m ea ns In te ns ity -w ei gh te d ex po su re da ys ,t er til es 1, 2, an d 3 vs .n ev er Fu lly ad ju st ed M M RR s D 1. 00 ,1 .2 7, 1. 87 ;p -t re nd D 0. 22 us in g sc or es ,0 .1 8 us in g m ea ns In te rm ed ia te ad ju st ed M M RR s D 0. 99 ,1 .2 2, 1. 65 ;p -t re nd D 0. 27 us in g sc or es ,0 .2 4 us in g m ea ns Ag e- an d se x- ad ju st ed M M RR s D 0. 91 ,1 .1 2, 1. 44 ;p -t re nd D 0. 39 us in g sc or es ,0 .3 3 us in g m ea ns Cu m ul at iv e ex po su re da ys ,t er til es 1, 2, an d 3 vs .n ev er Fu lly ad ju st ed M M 95 % CI s D 0. 43 -3 .0 3, 0. 54 -4 .3 4, 0. 42 -4 .4 5 In te rm ed ia te ad ju st ed M M 95 % CI s D 0. 44 -2 .8 8, 0. 56 -4 .0 5, 0. 42 -3 .5 8 Ag e- an d se x- ad ju st ed M M 95 % CI s D 0. 42 -2 .7 0, 0. 50 -3 .5 8, 0. 37 -3 .1 1 In te ns ity -w ei gh te d ex po su re da ys ,t er til es 1, 2, an d 3 vs .n ev er Fu lly ad ju st ed M M 95 % CI s D 0. 33 -3 .0 0, 0. 45 -3 .5 6, 0. 67 -5 .2 7 In te rm ed ia te ad ju st ed M M 95 % CI s D 0. 34 -2 .8 6, 0. 45 -3 .2 8, 0. 64 -4 .2 4 Ag e- an d se x- ad ju st ed M M 95 % CI s D 0. 31 -2 .6 2, 0. 42 -3 .0 0, 0. 57 -3 .6 7 N ev er us ed gl yp ho sa te :8 ca se s Ev er us ed gl yp ho sa te :2 4 U nk no w n gl yp ho sa te us e: 2 ca se s M M RR ,e ve rg ly ph os at e D 1. 18 M M RR ,u nk no w n gl yp ho sa te D 1. 71 Cu m ul at iv e ex po su re da ys ,t er til es 1, 2, 3, an d un kn ow n vs .n ev er M M RR s D 1. 11 ,1 .4 5, 1. 17 ,1 .1 9; p- tr en d > 0. 50 us in g sc or es or m ea ns ,e xc lu di ng un kn ow n In te ns ity -w ei gh te d ex po su re da ys ,t er til es 1, 2, 3, an d un kn ow n vs .n ev er M M RR s D 0. 95 ,1 .1 9, 1. 58 ,1 .0 4; p- tr en d D 0. 30 us in g sc or es ,0 .2 6 us in g m ea ns ,e xc lu di ng un kn ow n M M 95 % CI ,e ve rg ly ph os at e D 0. 53 -2 .6 5 M M 95 % CI ,u nk no w n gl yp ho sa te D 0. 36 -8 .2 0 Cu m ul at iv e ex po su re da ys ,t er til es 1, 2, 3, an d un kn ow n vs .n ev er M M 95 % CI s D 0. 44 -2 .8 3, 0. 54 -3 .8 8, 0. 40 -3 .4 1, 0. 25 -5 .6 5 In te ns ity -w ei gh te d ex po su re da ys ,t er til es 1, 2, 3, an d un kn ow n vs .n ev er M M 95 % CI s D 0. 33 -2 .7 5, 0. 44 -3 .1 9, 0. 62 -4 .0 5, 0. 22 -4 .9 2 CI :c on fi de nc e in te rv al ;C LL :c hr on ic ly m ph oc yt ic le uk em ia ;D LB CL :d iff us e la rg e B- ce ll ly m ph om a; FL :f ol lic ul ar ly m ph om a; H L: H od gk in ly m ph om a; LH C: ly m ph oh em at op oi et ic ca nc er ;L PS :l ym ph op ro lif er at iv e sy nd ro m e; M M :m ul tip le m ye lo m a; N H L: no n- H od gk in ly m ph om a; N R: no tr ep or te d; O R: od ds ra tio ;R R: re la tiv e ris k; SL L: sm al ll ym ph oc yt ic ly m ph om a. 415 leukemia, one with unspecified B-cell lymphoma, and one with MM. Eriksson et al.[14] did not report the number of exposed cases, but overall the B-cell lymphomas in their study comprised 29% diffuse large B-cell lymphoma, 24% chronic lymphocytic leukemia/small lymphocytic lym- phoma, 20% follicular lymphoma grades I-III, 16% other specified B-cell lymphoma, and 11% unspecified B-cell lym- phoma; MM cases were not included. The meta-RR for the association between any use of glypho- sate and risk of diffuse large B-cell lymphoma, based on two studies,[14,17] was 1.1 (95% CI D 0.5-2.3) using both the ran- dom-effects and the fixed-effects models (I2 D 0.0%, pheterogeneity D 0.79) (Table 3). Based on the same two studies,[14,17] the meta-RR for the association between any use of glyphosate and risk of chronic lymphocytic leukemia/small lymphocytic lymphoma was 1.3 (95% CI D 0.2-10.0) according to the random- effects model and 1.9 (95% CI D 0.9-4.0) according to the fixed-effects model, with significant heterogeneity between the two included estimates (I2 D 83.7%, pheterogeneity D 0.01) (Table 3). Results for follicular lymphoma from these two studies,[14,17] by contrast, were not significantly heterogeneous (I2 D 0.0%, pheterogeneity D 0.73), with a meta-RR of 1.7 (95% CI D 0.7-3.9) in both the random-effects and the fixed-effects models (Table 3). Table 3. Selected estimates included in meta-analyses and calculated meta-analysis relative risks (meta-RRs) of the association between glyphosate exposure and risk of (LHC), including non-Hodgkin lymphoma (NHL), NHL subtypes, Hodgkin lymphoma (HL), multiple myeloma (MM), and leukemia. Study # Authors Year Outcome Number of exposed subjects RR 95% CI 1 De Roos et al.[13] 2003 Non-Hodgkin lymphoma 36 cases, 61 controls a. 1.6 (hierarchical regression) b. 2.1 (logistic regression) a. 0.9-2.8 (hierarchical regression) b. 1.1-4.0 (logistic regression) 2 De Roos et al.[12] 2005 Non-Hodgkin lymphoma 71 cases 1.1 0.7-1.9 3 Eriksson et al.[14] 2008 Non-Hodgkin lymphoma 29 cases, 18 controls 1.51 0.77-2.94 4 Hardell et al.[15] 2002 Non-Hodgkin lymphoma 8 cases, 8 controls 1.85 0.55-6.20 5 Hohenadel et al.[28] 2011 Non-Hodgkin lymphoma 50 cases, 133 controls 1.40 (random effects meta-RR) 0.62-3.15 (random effects meta-CI) 6 McDuffie et al.[16] 2001 Non-Hodgkin lymphoma 51 cases, 133 controls 1.2 0.83-1.74 7 Orsi et al.[17] 2009 Non-Hodgkin lymphoma 12 cases, 24 controls 1.0 0.5-2.2 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 Non-Hodgkin lymphoma 1a, 2, 3, 4, 6, 7 1.3 1.0-1.6 0.0% 0.84 Model 2 Non-Hodgkin lymphoma 1b, 2, 3, 4, 6, 7 1.3 1.0-1.6 0.0% 0.59 Model 3 Non-Hodgkin lymphoma 1a, 2, 3, 4, 5, 7 1.3 1.0-1.7 0.0% 0.85 Model 4 Non-Hodgkin lymphoma 1b, 2, 3, 4, 5, 7 1.4 1.0-1.8 0.0% 0.63 3 Eriksson et al.[14] 2008 B-cell lymphoma Not reported 1.87 0.998-3.51 8 Cocco et al.[18] 2013 B-cell lymphoma 4 cases, 2 controls 3.1 0.6-17.1 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 B-cell lymphoma 3, 8 2.0 1.1-3.6 0.0% 0.58 3 Eriksson et al.[14] 2008 Diffuse large B-cell lymphoma Not reported 1.22 0.44-3.35 7 Orsi et al.[17] 2009 Diffuse large B-cell lymphoma 5 cases, 24 controls 1.0 0.3-2.7 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 Diffuse large B-cell lymphoma 3, 7 1.1 0.5-2.3 0.0% 0.79 3 Eriksson et al.[14] 2008 CLL/SLL Not reported 3.35 1.42-7.89 7 Orsi et al.[17] 2009 CLL/SLL 2 cases, 18 controls 0.4 0.1-1.8 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1, random effects CLL/SLL 3, 7 1.3 0.2-10.0 83.7% 0.01 Model 1, fixed effects CLL/SLL 3, 7 1.9 0.9-4.0 3 Eriksson et al.[14] 2008 Follicular lymphoma Not reported 1.89 0.62-5.79 7 Orsi et al.[17] 2009 Follicular lymphoma 3 cases, 24 controls 1.4 0.4-5.2 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 Follicular lymphoma 3, 7 1.7 0.7-3.9 0.0% 0.73 7 Orsi et al.[17] 2009 Hairy-cell leukemia 2 cases, 18 controls 1.8 0.3-9.3 9 Nordstr€om et al.[30] 1998 Hairy-cell leukemia 4 cases, 5 controls 3.1 0.8-12 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 Hairy-cell leukemia 7, 9 2.5 0.9-7.3 0.0% 0.63 7 Orsi et al.[17] 2009 Hodgkin lymphoma 6 cases, 24 controls 1.7 0.6-5.0 10 Karunanayake et al.[31] 2012 Hodgkin lymphoma 38 cases, 133 controls 0.99 0.62-1.56 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 Hodgkin lymphoma 7, 10 1.1 0.7-1.6 0.0% 0.36 2 De Roos et al.[12] 2005 Multiple myeloma 19 casesy 2.6 0.7-9.4 7 Orsi et al.[17] 2009 Multiple myeloma 5 cases, 24 controls 2.4 0.8-7.3 11 Brown et al.[32] 1993 Multiple myeloma 11 cases, 40 controls 1.7 0.8-3.6 12 Kachuri et al.[33] 2013 Multiple myeloma 32 cases, 121 controls a. 1.19 (with proxies) b. 1.11 (without proxies) a. 0.76-1.87 (with proxies) b. 0.66-1.86 (without proxies) 13 Pahwa et al.[34] 2012 Multiple myeloma 32 cases, 133 controls 1.22 0.77-1.93 14 Sorahan[26] 2015 Multiple myeloma 24 cases 1.24 0.52-2.94 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 Multiple myeloma 7, 11, 12a, 14 1.4 1.0-1.9 0.0% 0.63 Model 2 Multiple myeloma 2, 7, 11, 12a 1.5 1.0-2.1 0.0% 0.48 Model 3 Multiple myeloma 7, 11, 12b, 14 1.4 0.9-1.9 0.0% 0.58 Model 4 Multiple myeloma 7, 11, 13, 14 1.4 1.0-2.0 0.0% 0.66 Model 5 Multiple myeloma 2, 7, 11, 13 1.5 1.0-2.1 0.0% 0.52 2 De Roos et al.[12] 2005 Leukemia 43 cases 1.0 0.5-1.9 16 Brown et al.[35] 1990 Leukemia 15 cases, 49 controls 0.9 0.5-1.6 17 Kaufman et al.[36] 2009 Leukemia 1 case, 3 controls 1.4 0.15-13.56 Meta-analysis model Outcome Studies included Meta-RR 95% CI I2 Pheterogeneity Model 1 Leukemia 2, 16, 17 1.0 0.6-1.5 0.0% 0.92 Number of exposed cases is provided for the total cohort of 54,315 subjects; the number of exposed cases in the analytic cohort of 49,211 subjects is not stated. yNumber of exposed cases is provided for the analytic cohort of 40,719 subjects, as reported by Sorahan.[26] CI: confidence interval; CLL: chronic lymphocytic leukemia; RR: relative risk; SLL: small lymphocytic lymphoma. 416 E. T. CHANG AND E. DELZELL Finally, the two studies that reported associations between any glyphosate use and risk of hairy-cell leukemia[17,30] yielded a meta-RR of 2.5 (95% CI D 0.9-7.3) in the random-effects and fixed-effects models (I2 D 0.0%, pheterogeneity D 0.63) (Table 3). HL Both of the published, fully adjusted RRs and 95% CIs for the association between any glyphosate use and HL risk (Table 2) were included in the meta-analysis (Table 3). Based on two studies,[17,31] the meta-RR was 1.1 (95% CI D 0.7-1.6) in both the random-effects and the fixed-effects models, with I2 D 0.0% and pheterogeneity D 0.36 (Table 3). Publication bias was not eval- uated due to the availability of only two studies of HL. MM All relevant RRs and 95% CIs for the association between glyphosate use and risk of MM, including estimates that did not contribute to the meta-analysis, are shown in Table 2. The independent estimates selected for inclusion in the meta-analy- sis are shown in Table 3. The combined meta-RR for the association between any glyphosate use and risk of MM, based on four stud- ies,[17,26,32,33] was 1.4 (95% CI D 1.0-1.9) according to the random-effects and fixed-effects models (Table 3, Fig. 2). On the basis of the I2 value of 0.0% and the P-value of 0.63 for Cochran’s Q statistic, between-study heterogeneity was not evident. Egger’s linear regression approach yielded no significant evidence of publication bias (one-tailed P-value for asymmetry D 0.10), while the imputed meta-RR using the trim-and-fill procedure to adjust for publication bias was 1.3 (95% CI D 0.9-1.8). Several secondary analyses were conducted for MM by replacing RRs in the primary meta-analysis with alternative esti- mates (Table 3). When the RR reported by De Roos et al.,[12] who excluded cohort members with missing data from their analysis, was substituted for the one reported by Sorahan,[26] who included such subjects by creating a separate category for missing or unknown data, the meta-RR was slightly increased to 1.5 (95% CI D 1.0-2.1) and was the same for random-effects and fixed-effects models. When the main RR from Kachuri et al.[33] was replaced with the RR from the same study after exclusion of data reported by proxy respondents, the meta-RR was not appreciably different from the original estimate (alter- native meta-RR D 1.4, 95% CI D 0.9-1.9 in random-effects and fixed-effects models). Another secondary analysis included the RR reported by Pahwa et al.,[34] who adjusted for a slightly dif- ferent (and smaller) set of confounders than Kachuri et al.[33] and also retained controls who were too young to have any age- matched MM cases in this Canadian study. This change had minimal impact on the meta-RR (1.4, 95% CI D 1.0-2.0; same for random-effects and fixed-effects models). When both the De Roos et al.[12] and the Pahwa et al.[34] substitutions were made, the resultant meta-RR was the same as that when only De Roos et al.[12] was used (meta-RR D 1.5, 95% CI D 1.0-1.2 in ran- dom-effects and fixed-effects models). Leukemia Of the four published RRs and 95% CIs for the association between any use of glyphosate and risk of leukemia (Table 2), three (excluding one age-adjusted RR in favor of a more fully adjusted RR from De Roos et al.[12]) were included in the meta- analysis (Table 3). The meta-RR based on three studies[12,35,36] was 1.0 (95% CI D 0.6-1.5) using the random-effects model and the fixed-effects model (I2 D 0.0%, pheterogeneity D 0.92) (Table 3). Publication bias was not assessed because only three studies of leukemia were available. Sensitivity analysis A sensitivity analysis was conducted for overall NHL only (Table 4), because other outcomes had an insufficient num- ber of studies for stratification. In all strata, the random- effects and fixed-effects meta-RRs were identical and I2 was 0.0%. Results did not differ substantially from the main meta-RR (1.3, 95% CI D 1.0-1.6) when the analysis was restricted to case-control studies (meta-RR D 1.3, 95% CI D 1.0-1.7) or those with population-based controls (meta-RR D 1.4, 95% CI D 1.0-1.8). Meta-analysis could not be con- ducted for cohort studies or studies with hospital-based Figure 2. Forest plots of relative risk (RR) estimates and 95% confidence intervals (CIs) for the association between glyphosate exposure and risk of multiple myeloma. Meta-RRs were identical in random-effects and fixed-effects models. Table 4. Sensitivity analysis of the association between glyphosate exposure and risk of non-Hodgkin lymphoma (NHL). Stratum Number of studies Meta-RR 95% CI All 6 1.3 1.0-1.6 Case-control 5 1.3 1.0-1.7 Cohort 1 NR Population controls 4 1.4 1.0-1.8 Hospital controls 1 NR Males only 4 1.3 1.0-1.7 Males and females 2 1.2 0.8-1.8 North America 3 1.2 1.0-1.6 Europe 3 1.3 0.8-2.1 Sweden 2 1.6 0.9-2.8 Cases in 1980s 2 1.6 1.0-2.7 Cases in 1990s 4 1.2 1.0-1.6 Cases in 2000s 3 1.2 0.8-1.7 All meta-RRs were identical in random-effects and fixed-effects models. CI: confidence interval; meta-RR: meta-analysis relative risk; NR: not reported, when only one study was available. JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 417@ Authors Year RR 95% CI f- Relative weight(%) 1993 1.7 0.8- 3.6 20.0 2013 1.19 0.76- 1.87 55.7 2009 2.4 0.8- 7.3 9.2 2015 1.24 0.52- 2.94 15. l 1.4 1.0- 1.9 0.1 1.0 10 controls because only one of each of these study types was available. No major differences were detected between stud- ies restricted to males (meta-RR D 1.3, 95% CI D 1.0-1.7) and those that included males and females (meta-RR D 1.2, 95% CI D 0.8-1.8) or between those conducted in North America (meta-RR D 1.2, 95% CI D 1.0-1.6) and those con- ducted in Europe (meta-RR D 1.3, 95% CI D 0.8-2.1). Prompted by Schinasi and Leon,[11] we also conducted a stratified meta-analysis of the two studies conducted in Swe- den[14,15] and found a stronger, albeit statistically non-signif- icant, association in these particular studies (meta-RR D 1.6, 95% CI D 0.9-2.8). The estimated meta-RR declined some- what from studies that ascertained cases in the 1980s (meta- RR D 1.6, 95% CI D 1.0-2.7) to those conducted in the 1990s (meta-RR D 1.2, 95% CI D 1.0-1.6) to those con- ducted in the 2000s (meta-RR D 1.2, 95% CI D 0.8-1.7). Exposure-response trends NHL and subtypes. Three studies evaluated exposure-response trends between glyphosate use and NHL risk, with exposure classified as cumulative lifetime[12,14] or annual[16] days of glyphosate use (Table 2). Two studies detected some evidence of a positive exposure-response trend (statistical significance not reported),[14,16] whereas the other did not.[12] All of these studies relied wholly or in part on evaluating days of glyphosate use in an attempt to quantify exposure; however, this metric has been shown to be a poor indicator of actual glyphosate dose received.[52] In a model adjusted for age, sex, and year of diagnosis or enrollment, Eriksson et al.[14] found that the RR of NHL was higher with > 10 days of lifetime glyphosate use (RR D 2.36, 95% CI D 1.04-5.37) than with 10 days (RR D 1.69, 95% CI D 0.70-4.07), compared with no pesticide use. Also, the RR of NHL was higher after more than 10 years since first use of glyphosate (RR D 2.26, 95% CI D 1.16-4.40) than after 1- 10 years (RR D 1.11, 95% CI D 0.24-5.08). Statistical tests for trend were not performed, and exposure-response analyses adjusted for other potential confounders (i.e., 2-methyl-4- chlorophenoxyacetic acid (MCPA), 2,4,5-trichlorophenoxyace- tic acid (2,4,5-T) and/or 2,4-dichlorophenoxyacetic acid (2,4- D), mercurial seed dressing, arsenic, creosote, and tar) were not presented, even though adjustment for these characteristics attenuated the RR for overall glyphosate use from 2.02 to 1.51. McDuffie et al.[16] reported that the RR for more than two days of glyphosate use per year (RR = 2.12, 95% CI D 1.20- 3.73) was higher than that for up to two days per year (RR D 1.00, 95% CI D 0.63-1.57), compared with never use, adjusting for age and province of residence. Tests for a significant expo- sure-response trend were not performed, and results were not reported after adjustment for other potential confounders (i.e., personal medical history and family history of cancer; adjust- ment for these characteristics attenuated the RR for overall glyphosate use from 1.26 to 1.20) or significantly associated pesticides (i.e., aldrin, dicamba, and mecoprop) in this study population. The most detailed analysis of glyphosate-NHL exposure- response trends was performed by De Roos et al.,[12] who exam- ined tertiles of cumulative lifetime days of glyphosate use (1-20, 21-56, or 57-2,678 days) and tertiles of intensity-weighted cumulative days of use (i.e., years of use £ days per year £ intensity level, where intensity was defined as (mixing status C application method C equipment repair status) £ personal protective equipment use). In analyses adjusted for age, educa- tion, smoking, alcohol, family history of cancer, and state of residence, no significant trend was detected for NHL risk in association with increasing cumulative days of glyphosate use (RRs for tertiles 1, 2, and 3, respectively D 1.0 (referent), 0.7 (95% CI D 0.4-1.4), and 0.9 (95% CI D 0.5-1.6); ptrend D 0.73) or intensity-weighted cumulative exposure days (RRs D 1.0 (referent), 0.6 (95% CI D 0.3-1.1), and 0.8 (95% CI D 0.5-1.4); ptrend D 0.99). Exposure-response trends between glyphosate use and risk of specific NHL subtypes were not evaluated in any of the included studies. HL. No studies assessed exposure-response trends between glyphosate use and risk of HL. MM. Three studies reported exposure-response trends between glyphosate use and MM risk, including the two analy- ses based on the same Agricultural Health Study cohort data- set[12,26] and the Canadian case-control study[33] (Table 2). The case-control study found mixed evidence of a positive trend (statistical significance not reported), while a positive trend was detected in one analysis of the cohort data[12] but not the other.[25] The Canadian case-control study found a lower risk of MM among those who used glyphosate for up to two days per year than those who had never used glyphosate (RR D 0.72, 95% CI D 0.39-1.32).[33] However, risk was higher in those with more than two days of glyphosate use per year (RR D 2.04, 95% CI D 0.98-4.23), adjusting for age, province of residence, proxy status, smoking, personal medical history, and family history of cancer. Results were similar after exclusion of data reported by proxy subjects. The authors did not conduct statistical tests for exposure-response trends. Based on the 55% of Agricultural Health Study cohort mem- bers who had available exposure and covariate data, De Roos et al.[12] reported a positive, albeit statistically non-significant, trend between MM risk and increasing tertiles of cumulative days of glyphosate use (RRs for tertiles 1, 2, and 3, respectively D 1.0 (referent), 1.1 (95% CI D 0.4-3.5), and 1.9 (95% CI D 0.6-6.3); ptrend D 0.27) or intensity-weighted cumulative days of use (RRs D 1.0 (referent), 1.2 (95% CI D 0.4-3.8), and 2.1 (95% CI D 0.6-7.0); ptrend D 0.17). These estimates were adjusted for age, education, smoking, alcohol, family history of cancer, state of residence, the five pesticides for which cumula- tive-use variables were most highly associated with glyphosate cumulative use days (i.e., 2,4-D, alachlor, atrazine, metolachlor, and trifluralin), and the five pesticides that were most highly associated with ever use of glyphosate (i.e., benomyl, maneb, paraquat, carbaryl, and diazinon). When intensity alone was analyzed in association with MM risk, the RR for the highest versus the lowest tertile was 0.6 (95% CI D 0.2-1.8), indicating that the suggested trend was due only to total days of use. When subjects who never used glyphosate were set as the refer- ence group, the RRs for tertiles 1, 2, and 3 of cumulative days 418 E. T. CHANG AND E. DELZELL of use were 2.3 (95% CI D 0.6-8.9), 2.6 (95% CI D 0.6-11.5), and 4.4 (95% CI D 1.0-20.2); ptrend D 0.09. When cumulative use was categorized into quartiles, the RR for the highest quar- tile versus never use was 6.6 (95% CI D 1.4-30.6); ptrend D 0.01. In contrast to De Roos et al.,[12] Sorahan[26] included more than 53,000 eligible cohort members in the analysis (excluding only those with a history of cancer before enrollment, loss to follow-up, missing data on age at enrollment, or missing data on glyphosate use) by creating separate categories for missing or unknown exposure and covariate data. Adjusting for age, sex, education, smoking, alcohol, family history of cancer, and the same 10 pesticides as De Roos et al.,[12] the RRs for each tertile of cumulative days of glyphosate use, compared with never use, were 1.14 (95% CI D 0.43-3.03), 1.52 (95% CI D 0.54-4.34), and 1.38 (95% CI D 0.42-4.45); ptrend D 0.48 using category scores of 1-4, ptrend > 0.50 using mean expo- sures within categories. RRs for increasing tertiles of intensity- weighted days of use versus never use were 1.00 (95% CI D 0.33-3.00), 1.27 (95% CI D 0.45-3.56), and 1.87 (95% CI D 0.67-5.27); ptrend D 0.22 using scores, ptrend D 0.18 using means. When Sorahan[26] expanded the eligible cohort to 55,934 subjects to include those with unknown use of glypho- sate, he again detected no significant exposure-response trends with respect to either cumulative days of use (for tertiles 1, 2, and 3 and unknown use versus never use, respectively, RRs D 1.11 (95% CI D 0.44-2.83), 1.45 (95% CI D 0.54-3.88), 1.17 (95% CI D 0.40-3.41), and 1.19 (95% CI D 0.25-5.65); ptrend > 0.50 across categories of known use using scores or means, excluding unknown) or intensity-weighted cumulative days of use (RRs D 0.95 (95% CI D 0.33-2.75), 1.19 (95% CI D 0.44-3.19), 1.58 (95% CI D 0.62-4.05), and 1.04 (95% CI D 0.22-4.92); ptrend D 0.30 using scores, ptrend D 0.26 using means, excluding unknown). Leukemia. The De Roos et al.[12] study based on the Agri- cultural Health Study cohort was the only study that reported exposure-response trends between glyphosate use and risk of leukemia (Table 2). No significant trend was observed between increasing tertiles of cumulative days of glyphosate use (RRs D 1.0 (referent), 1.9 (95% CI D 0.8- 4.5), and 1.0 (95% CI D 0.4-2.9) for tertiles 1, 2, and 3, respectively; ptrend D 0.61) or intensity-weighted cumulative days of use (RRs D 1.0 (referent), 1.9 (95% CI D 0.8-4.7), and 0.7 (95% CI D 0.2-2.1); ptrend D 0.11), adjusting for demographic and lifestyle factors as well as other pesticides. Evaluation of bias Selection bias All studies of the association between glyphosate exposure and risk of LHC were case-control studies except for the Agricul- tural Health Study, the prospective cohort study that served as the basis for the studies by De Roos et al.[12] and Sorahan.[26] In case-control studies, differences in participation patterns between cases and controls can result in selection bias if partici- pation is related to the exposure of interest. In cohort studies, selection bias can occur if loss to follow-up is related to the exposure and outcome of interest or, less commonly, if baseline participation differs by exposure status and risk of developing the outcome of interest in the future (e.g., based on having a positive family history of an outcome with a genetic susceptibil- ity component). Selection bias in any study also can occur if inclusion in the data analysis, e.g., predicated on data complete- ness, differs by exposure and outcome status. In general, lower participation, follow-up, or data completeness and large differ- ences in participation between groups increase the potential magnitude of selection bias. Table 1 shows the reported participation and follow-up pro- portions in all reviewed studies. Most studies did not report data completeness. The substantial differences in participation between cases and controls in the European multi-center study,[18] the most recent Swedish study,[14] and the Canadian study, which also had relatively low absolute participation pro- portions of <70% for cases and <50% for controls,[16,28,31,33,34] are of particular concern. However, the smaller discrepancies between case and control participation in other studies also could have produced selection bias. Moreover, even identical participation by cases and controls can obscure differences in reasons for study participation that could result in bias. Given that several case-control studies were originally designed to evaluate associations between pesticides and risk of LHC,[13-16,28,31-35] it is plausible that cases with a history of agricultural pesticide use were more likely than controls to par- ticipate, thereby biasing results toward a positive association for glyphosate as well as other pesticides. It is also possible that certain sources of controls in some of these studies (e.g., resi- dential telephone calls and voter lists) were more likely to iden- tify individuals who were not farmers, again biasing results toward a positive association. Investigators from the Canadian study[16,28,31,33,34] reported that an analysis of postal codes showed that respondents and non-respondents did not differ significantly in terms of rural versus urban residence, but they could not examine differences in occupation or pesticide use. Although the initial follow-up completion of >99% in the Agricultural Health Study was high,[12,25] the sizeable propor- tions of subjects with missing data raise concerns about selec- tion bias. Specifically, 88% of the eligible cohort (excluding those who were diagnosed with cancer before enrollment or were lost to follow-up) provided usable data on ever use of glyphosate and key demographic and lifestyle covariates, 73% additionally provided data on use of other pesticides, 65-66% contributed to analyses of cumulative days of glyphosate use (with or without intensity weighting), and 55% contributed to analyses of cumulative use additionally adjusted for other pesti- cides. Questionnaire completion could conceivably have varied by demographic and lifestyle factors that are associated with LHC risk, thereby producing bias. Neither analysis accounted for missing data using methods such as multiple imputation or inverse probability weighting. Differential data completeness by disease status is more likely to occur in case-control studies, such as the pooled Mid- western U.S. study conducted by De Roos et al.[13] In this study, the analysis of multiple pesticides excluded 25% of cases and 25% of controls who lacked complete data. Although the overall frequency of missing data was the same between cases and con- trols, this exclusion could have led to selection bias if subjects’ JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 419 reasons for providing complete data, and thus being included in the analysis, differed by disease status and were related to glyphosate exposure status. The authors also excluded subjects who had lived or worked on a farm before age 18 years. If glyphosate use was more common in such subjects, then RR estimates would have been biased upward if a childhood farm environment was inversely associated with NHL risk[53] and biased downward if the association was positive.[54] Exposure misclassification All of the included studies assessed use of glyphosate and other pesticides based on self-reported information (Table 1), which is prone to various types of error, such as better recall by cases than controls and by subjects than proxies, inaccurate recall of specific pesticides and amounts used, and a lack of the best measure of biological dose received.[55] Thus, probable expo- sure misclassification is a key limitation of all of these studies. The degree of misclassification may vary by mode of data col- lection, for example, by written questionnaire, telephone inter- view, or in-person interview.[56] The extent of misclassification also may depend on questionnaire structure, for example, whether subjects were asked in an open-ended manner to report use of any pesticides or whether they were prompted to report use of specific pesticides based on a prepared list.[57] Some authors did not clearly describe the structure of their study’s questions on pesticide use. Of the eight independent study populations included in this review (seven studies of NHL with or without other types of LHC and one study of leukemia), three provided information on validation of their exposure assessment methods: the Cana- dian case-control study,[16,28,31,33,34] the Agricultural Health Study,[12,26] and the Kansas case-control study[47] that contrib- uted to the pooled Midwestern U.S. study by De Roos et al.[13] Overall, these studies do not establish the validity of self- reported information on glyphosate use; rather, the limited results suggest considerable error and inconsistency in such data. Specifically, in the Canadian study, Dosman et al.[58] reported on the results of a validation pilot study of 21 volunteer farmers whose self-reported pesticide use was compared with written records of pesticide purchases through their local agrochemical supplier. Of the 21 farmers, 17 (81%) had a supplier who had retained written records; the remaining four transactions were conducted with cash. Based on the written records, 146 (65%) of 226 chemicals reported by farmers were verified; 50 of the unverified reports were potentially explained by aerial applica- tions, home and garden use, use more than five years in the past (i.e., during 1958-1984), or use outside of Canada. In 32 instan- ces (for 25 chemicals) the suppliers’ records indicated a pur- chase of chemicals that was unreported by the farmer; 2 of these were for glyphosate. Detailed self-reported exposure (e.g., fre- quency, intensity, and duration of use of specific pesticides) could not be validated in this pilot study. Likewise, Hoar et al.[47] reported that suppliers for 110 sub- jects in the Kansas study (out of 130 sought) were located and provided information on the subjects’ crops and herbicide and insecticide purchases as “corroborative evidence” of self- reported pesticide use. The authors observed that suppliers usu- ally reported less pesticide use than subjects; that agreement on specific years of use was better for insecticide use than herbicide use; that the differences between agreement for cases and con- trols were not consistent; and that agreement between suppliers and subjects was better for pesticide use within the last 10 years than for earlier use. Quantitative results on concordance were not provided by Hoar et al.,[47] but in a summary of this study shared with Dosman et al.[58] the authors stated that reports on herbicide use agreed 59% of the time, with little variation by crop type, and that reports on insecticide use also agreed 59% of the time, but differed by crop type. In the Agricultural Health Study, the reliability of the ques- tion on ever having mixed or applied glyphosate was evaluated by comparing responses to two questionnaires completed one year apart by 3,763 pesticide applicators.[59] Agreement on a pos- itive response to the question was 82%, and the kappa statistic value for inter-rater agreement was moderate (0.54, 95% CI D 0.52-0.58). For more detailed questions about glyphosate use, including years mixed or applied, days per year mixed or applied, and decade first applied, the percentage with exact agreement ranged from 52% to 62% and kappa ranged from 0.37 to 0.71. These metrics evaluated only the reliability (i.e., reproducibility) of self-reported glyphosate use, not its accuracy. Subsequent exposure validation studies for other pesticides in the Agricultural Health Study, based on comparisons between exposure intensity estimated from an expert-derived algorithm using self-reported or directly observed exposure data and pesticide biomarker levels measured in urine, yielded Spearman correlation coefficients between 0.4 and 0.8, depend- ing on the type of pesticide.[60,61] Correlations with urinary biomarker levels were poorer for self-reported determinants of pesticide exposure such as kilograms of active ingredient, hours spent mixing and applying, and number of acres treated, with correlation coefficients of ¡0.4 to 0.2, but application method and use of personal protective equipment were found to be important determinants of exposure intensity. However, the latter factors were evaluated in the study questionnaire only for pesticides or pesticide classes in general, not for glyph- osate or other individual pesticides;[62] thus, limitations remain in the assessment of specific pesticide exposures. Several studies included a sizeable proportion of surveys that were completed by proxy respondents for deceased or other- wise unavailable cases and controls (Table 1). The use of expo- sure data reported by surrogates most likely resulted in even poorer accuracy of exposure information in these studies. Although some exposure misclassification may have been non- differential by disease status, such error does not inevitably result in underestimated exposure-disease associations unless additional strict conditions are met, such as independence from other classification errors.[63,64] Furthermore, differential exposure misclassification in case- control studies can readily result in overestimated associations. Reasonable scenarios include more accurate and/or detailed recollection of past exposures by cases, who are more motivated than controls to try to understand the potential causes of their disease; false recollection by cases, who are more aware of sci- entific hypotheses or media reports that a certain exposure has been linked to their disease; and unconscious influence by study investigators who are aware of causal hypotheses and subjects’ case-control status. Only the authors of the Swedish 420 E. T. CHANG AND E. DELZELL studies,[14,15] the French study,[17] and the Nebraska compo- nent of the pooled Midwestern U.S. study[48] specifically stated that investigators were blinded to case-control status. In reality, such blinding is often difficult to achieve in studies that collect interview data. Others have discussed in detail the problems of estimating individual subjects’ exposure to glyphosate from responses to interviews and questionnaires asking about days of use, mixing and application procedures, use of personal protective equip- ment, and other work practices.[19,52] Acquavella et al.[52] reported that any given day of pesticide use can entail highly variable amounts of pesticides used and numbers of mixing operations, and that urine concentrations of glyphosate were poorly correlated with lifetime average exposure intensity scores derived from data self-reported by farmers using this agent. Although recall bias between cases and controls generally might be anticipated to affect all specific pesticides (including glyphosate) equally, variation in the degree of misclassification due to these and other factors affecting usage and exposure could result in different pesticide-specific associations. Most of the case-control studies did not use procedures to exclude glyphosate exposure that might have occurred after dis- ease onset. The Swedish studies omitted glyphosate use within one year prior to diagnosis or the index date in controls,[15,30] or within the same calendar year or the year before.[14] In some cases, however, these restrictions may not have been sufficient to exclude exposure that occurred during the latency period between disease onset and diagnosis. Inclusion of any such post-disease exposure would have led to misclassification. Finally, exposure misclassification resulting from the crude dichotomization of glyphosate use as ever versus never is an important limitation of most of the included studies. This clas- sification conflates individuals with considerably different fre- quencies, intensities, and durations of glyphosate use, and precludes potentially informative analyses of any gradient in LHC risk with increasing glyphosate exposure. As described earlier in the section on exposure-response trends, only three independent studies reported on glyphosate use in more than two (ever vs. never) categories, and only the Agricultural Health Study evaluated more than three exposure categories. Confounding As shown in Table 1, the degree of control for confounding var- ied widely among the reviewed studies. Although several stud- ies considered potential confounding by other pesticides or pesticide families, only a minority[12-15,26,28] reported RR esti- mates for the association between glyphosate use and LHC risk adjusted for use of other pesticides. Given that Schinasi and Leon[11] found significant associations between NHL risk and several other types of pesticides, including carbamate insecti- cides, organophosphorus insecticides, lindane, and MCPA, and numerous other associations of specific pesticides with LHC risk have been reported in the literature (e.g.,[65,66])-and because most people who use pesticides occupationally are exposed to multiple pesticides-it is important to control for confounding, whether direct or indirect (if pesticides are surro- gates for other risk factors), by these agents. None of the studies controlled for potential confounding by agricultural exposures other than pesticides, such as other agricultural chemicals, farm animals, allergens, and infectious agents. These exposures have been hypothesized, and in some studies shown, to be associated with risk of NHL, HL, MM, or leukemia,[67-73] and they are probably correlated with glyphosate use, making them potential confounders of associations between glyphosate and LHC risk. Medical history, certain infections, diet, alcohol consumption, and obesity also may be associated with risk of these malignancies[74-77] and could vary by glypho- sate use, again making them possible confounders. Even in stud- ies where numerous confounders were included in multivariable regression models, crude categorization or other misclassification of confounders could have enabled residual confounding of observed associations. The direction and magnitude of con- founding depend on the relationships of each factor with glyph- osate use and LHC risk, and are therefore difficult to predict. Other issues Additional issues related to the design, conduct, and reporting of the included studies also could have affected study results and their interpretation. For instance, Hardell et al.[15] enrolled some prevalent rather than incident cases, since eligible NHL cases were diagnosed in 1987-1990 but interviewed in 1993- 1995.[27] The relatively long time interval between diagnosis and interview may have hampered recollection of past expo- sures, thereby undermining the accuracy of self-reported expo- sure data in this study. The delay between diagnosis and interview also almost certainly increased the proportion of cases and matched controls who were deceased (43%) and had proxy interviews, leading to further exposure misclassification. In the studies by De Roos et al.[13] and Brown et al.,[32,35] LHC cases were diagnosed in 1979-1986, 1980-1983, and 1980-1984, respectively. With glyphosate having come to mar- ket in 1974, the cases in these studies would have had a rela- tively short potential induction time since first use of glyphosate. However, few studies to date have considered the issue of induction time. The Agricultural Health Study collected information on decade of first use of glyphosate in the baseline questionnaire for private pesticide applicators,[62] but did not use this information in the published analysis.[12] If glyphosate is a cause of LHC, the actual induction time is unknown because the mechanism of carcinogenesis is not established. Orsi et al.,[17] Kaufman et al.,[36] and four of the six study centers included in Cocco et al.[18] enrolled hospital-based rather than population-based cases and controls. Given that farmers have lower hospitalization rates than non-farmers,[78] hospital-based controls may be less likely than population- based controls to report agricultural occupational exposures, including pesticides, thereby resulting in overestimated RRs for pesticide use. On the other hand, occupational injuries are more common in agriculture than in general private indus- try,[79] possibly leading to oversampling of farmers from hospi- tal trauma/emergency and orthopedics departments, which might result in underestimated RRs. We did not observe any meaningful change in the meta-RR after restriction to popula- tion-based case-control studies. As noted in Table 1, many possible analyses were not con- ducted or not reported by authors. De Roos et al.[13] specifically acknowledged that they did not report results for pesticide combinations that were analyzed but yielded statistically null JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 421 associations for joint effects, and Hohenadel et al.[28] likewise did not show results for pesticide combinations without evi- dence of joint effects. Most other authors did not explicitly state when null results were not reported, but the Methods sections of several papers suggested that certain analyses were per- formed, yet not shown. Given the widespread predilection for emphasizing statistically significant associations in published research articles,[80] unreported results probably are usually sta- tistically null. The omission of null results is a form of reporting bias that favors positive associations. Other evidence suggests that statistically null associations between glyphosate and LHC risk have been underreported in the epidemiologic literature. For example, two of the studies that contributed to the pooled analysis conducted by De Roos et al.[13] apparently collected information on glyphosate use, yet associations between glyphosate and NHL risk were not reported in the original publications.[47,48] In an analysis of interactions between pesticide use and asthma, allergies, or hay fever diagnosis in relation to NHL risk in the Canadian case- control study,[81] results were reported for several specific pesti- cides, but not glyphosate, even though information was avail- able for glyphosate use. The most probable scenario in each of these cases is that no significant association was detected between glyphosate use and NHL risk. The omission of such results from the published literature represents a distortion of the body of epidemiologic evidence. The largest number of studies included in any of the meta- analyses described here was six (in the analysis of NHL), and the majority of meta-analyses (of HL, B-cell lymphoma, diffuse large B-cell lymphoma, chronic lymphocytic leukemia/small lymphocytic lymphoma, follicular lymphoma, and hairy-cell leukemia) included only two studies. The small number of available studies limits the robustness of the estimated meta- RRs, as well as the ability to perform informative sensitivity analysis and evaluation of heterogeneity and publication bias. Even with 10 contributing studies (which we lacked), the statis- tical power to detect modest heterogeneity using Cochran’s Q statistic is “low.”[42] The small number of studies also provides little opportunity to qualitatively investigate possible sources of heterogeneity by subject characteristics or study design. Thus, the results of the meta-analyses and related statistical tests reported here should be interpreted cautiously in light of the sparse and possibly selectively published literature, as well as the high potential for bias and confounding in most of the available studies. Overall evaluation The validity of the meta-RRs for glyphosate use and LHC risk reported here and by others[11] is uncertain because systematic error due to bias and confounding cannot reasonably be ruled out as explanations for the observed associations (including both positive and null associations). In addition, an evaluation of the association between glyphosate exposure and risk of LHC based on the Bradford Hill viewpoints[46] does not favor a causal relationship with NHL, any NHL subtype, HL, MM, or leukemia. These nine viewpoints are strength, consistency, specificity, temporality, biological gradient, plausibility, coher- ence, experiment, and analogy. To evaluate the strength of the association between glypho- sate use and risk of each type of LHC, we considered the mag- nitude of study-specific RRs and the corresponding meta-RRs. In individual studies, estimates of the association between glyphosate use and risk of NHL ranged between 1.0 and 2.1, and estimates of the association with NHL subtypes ranged between 0.4 and 3.35 (Table 3). For HL, the two estimates of association were 0.99 and 1.7. For MM, RRs ranged between 1.0 and 2.4, and those for leukemia ranged between 0.9 and 1.40. Most study-specific estimates were between 1.0 and 1.5. The estimated meta-RRs for all LHC outcomes, including those calculated in secondary and sensitivity analyses, ranged between 1.0 (for leukemia) and 2.5 (for hairy-cell leukemia). The meta-RRs calculated based on at least four studies ranged between 1.3 and 1.4. These associations are not of sufficient magnitude to exclude modest bias or confounding as reason- able explanations for the observed results. Results were not consistent between case-control studies of NHL and the one prospective cohort study of NHL, which reported no association.[12] Even among the six studies that con- tributed to the meta-analysis of NHL, RR point estimates varied by more than two-fold, only one statistically significant positive association was observed, and results from some studies were internally inconsistent (Table 3). Another, arguably more appro- priately adjusted RR (from a hierarchical regression model) that was 24% lower and statistically non-significant was reported in the same study that found a significant association.[13] The lack of statistically significant heterogeneity among studies of NHL, based on an underpowered statistical test, does not indicate con- sistency of results. For NHL subtypes, RR estimates also were variable, except for diffuse large B-cell lymphoma, for which both estimates were close to 1.0. Only one statistically significant positive association was detected (for chronic lymphocytic leuke- mia/small lymphocytic lymphoma),[14] and this result was con- tradicted by a non-significant inverse association in the other study of this outcome.[17] No significant associations with ever use of glyphosate were detected for HL, MM, or leukemia, and for MM the RR point estimates varied by more than two-fold. Results for MM in the Agricultural Health Study were internally inconsistent;[12,26] and the positive association with cumulative glyphosate exposure probably was due largely to selection bias. Numerous associations have been hypothesized between glyphosate exposure and diverse health outcomes, and between various exposures and risk of NHL, NHL subtypes, HL, MM, or leukemia. Thus, the putative associations are not specific to either the exposure or any of the outcomes. As noted by Brad- ford Hill,[46] “diseases may have more than one cause” and “one-to-one relationships are not frequent”; therefore, a lack of specificity does not detract from a causal hypothesis. In case-control studies, where exposure assessment was retrospective, a temporal sequence was not definitively estab- lished with glyphosate use preceding the time of disease onset. Although some studies attempted to exclude use close to the time of case diagnosis (or enrollment, for con- trols),[14,15,30] in practice individuals may not accurately recall the timing of use. Only the prospective Agricultural Health Study[12,26] was designed to collect information on glyphosate use prior to cancer ascertainment. However, the authors did not exclude malignancies diagnosed close to 422 E. T. CHANG AND E. DELZELL (e.g., within one year of) study enrollment, nor did they report the distribution of diagnoses with respect to time since first use of glyphosate. Thus, some preclinical cancers may have existed prior to study entry and, possibly, prior to at least some reported glyphosate use. As discussed in detail earlier, in the three studies of NHL with information on frequency, intensity, and/or duration of glyphosate use,[12,14,16] a positive biological gradient was not consistently demonstrated and was notably lacking in the Agri- cultural Health Study,[12] which had the most detailed exposure information (Table 2). One case-control study[33] and one pro- spective cohort study[12] of MM reported results suggesting a positive biological gradient with glyphosate use, but the alterna- tive analysis of the Agricultural Health Study data[26] did not demonstrate such a trend. No data were available to evaluate exposure-response trends between glyphosate and risk of NHL subtypes or HL, and the single study with such data for leuke- mia found no apparent trend.[12] Inhalation exposure to glyphosate from agricultural or resi- dential uses is likely to be slight due to glyphosate’s extremely low vapor pressure.[82] Although dermal contact can be consid- erable, the very low skin penetrability of glyphosate[83] should result in minimal, if any, biologically absorbed dose. A study of farm families with a lower limit of detection of 0.001 mg/mL (1 ppb) found that 40% of glyphosate applicators had undetect- able urinary glyphosate, which reflects all routes of exposure (dermal, inhalation, and oral).[84] Among those with detectable urinary glyphosate, the distribution of concentrations was right skewed, with a peak geometric mean concentration of 0.0032 mg/mL (3.2 ppb) on the day of application and declining thereafter. A review of seven human biomonitoring studies of glyphosate (including[84]) yielded the conclusion that “no health concern was revealed because the resulting exposure estimates were by magnitudes lower” than the science-based acceptable daily intake and the acceptable operator exposure level proposed by EFSA.[85] Glyphosate is usually applied in agricultural operations only a few days per year. Given the low biological dose of glyphosate that is expected to be sustained, along with the lack of information on the mechanism of carci- nogenesis that may exist in humans, the biological plausibility of LHC development due to typical glyphosate exposure has not been established. IARC recently determined based on their process that there is “sufficient” evidence of carcinogenicity of glyphosate in experimental animals and mechanistic evidence of genotoxicity and oxidative stress.[6] By contrast, U.S. EPA,[86] JMPR,[3] BfR,[1] EFSA,[9] and others[87,88] concluded that glyphosate does not have genotoxic, mutagenic, or carcinogenic effects in in vivo animal and in vitro studies, and that the negative find- ings constitute evidence against carcinogenicity. Given these widely divergent opinions, one cannot unambiguously con- clude whether the scientific evidence is coherent with the hypothesis that glyphosate causes any or all LHC. No true experimental evidence exists regarding the associa- tion between glyphosate exposure and risk of LHC in humans. However, positive associations between farming and risk of LHC were detected prior to 1974, when glyphosate was first commercially marketed.[89,90] Thus, if the apparent associations between farming and risk of LHC are due to causal agricultural exposures, they cannot be explained only by glyphosate expo- sure. Likewise, the recent worldwide increase (followed by a plateau or decline) in NHL incidence began before the 1970s[91,92]-although any impact of glyphosate on NHL inci- dence trends might be obscured by stronger risk factors. No marked increase in the incidence of HL, MM, or leukemia has been observed in parallel with the introduction and expansion of glyphosate use.[93-96] Finally, numerous analogies exist to support or oppose the hypothesis of a causal link between glyphosate exposure and risk of LHC. On balance, such analogies do not strengthen or weaken a conclusion of causality. In summary, although none of the Bradford Hill viewpoints can establish or disprove causality, we did not find compelling evidence in support of causality based on any of the nine view- points. Thus, on balance, the existing epidemiologic evidence does not favor a causal effect of glyphosate on NHL, HL, MM, leukemia, or any subtype of these malignancies. Discussion Our meta-analysis yielded borderline significant RRs of 1.3 and 1.4 between glyphosate use and risk of NHL and MM, respec- tively, and no significant association with risk of HL or leuke- mia. Based on more fully adjusted RRs, our NHL meta-RR of 1.3 (95% CI D 1.0-1.6) was weaker than that reported by Schi- nasi and Leon[11] (RR D 1.5, 95% CI D 1.1-2.0). The largest meta-RR of 2.5 (for hairy-cell leukemia) and the only meta-RR with a lower 95% confidence limit that excluded 1.0 (for B-cell lymphoma) were based on only two studies each, and the maxi- mum number of studies contributing to any meta-analysis was six. The few studies with available data did not consistently detect positive exposure-response trends between quantitative measures of glyphosate use and risk of any LHC. Consideration of the available epidemiologic evidence in light of the Bradford Hill viewpoints does not substantiate a causal relationship between glyphosate exposure and risk of any type of LHC. A conclusion in favor of causality also is undermined by the studies’ methodological limitations, which could reasonably account for at least part of the observed associations. These limi- tations include exposure misclassification (which may differ by outcome status especially in case-control studies, which consti- tute nearly all available studies), selection bias (due to differential enrollment, follow-up, or data completeness), poor adjustment for confounding (by other agricultural exposures, for instance), small numbers (which lead to low statistical power as well as a higher probability that a statistically significant finding is false[97]), and potential reporting and publication bias. Although underpowered statistical tests did not formally detect publication bias, we identified several examples of studies with available data that did not report associations between glyphosate use and LHC risk, and these unreported associations were most likely null. Underpowered statistical tests also generally did not detect heterogeneity of results among studies, except for chronic lym- phocytic leukemia/small lymphocytic lymphoma and MM. Nevertheless, our sensitivity analysis revealed some evidence of stronger associations with NHL risk in studies based in Sweden and those that ascertained cases in the 1980s, whereas the meta-RRs for studies that ascertained cases in the 2000s were JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 423 close to the null and statistically non-significant The stronger association with NHL diagnosed in the 1980s raises questions about whether glyphosate, an agent first introduced in 1974 in the United States and Europe, could plausibly cause lymphoma less than a decade later. However, deliberation on the potential induction time requires an understanding of the presumed mechanism of carcinogenesis, which is unknown for glypho- sate. The classification system for lymphoid tumors underwent major changes in 1994 and 2001,[20] such that the definition of NHL as a disease entity is not entirely comparable between recent studies and those conducted in the 1980s. Study quality also may have improved over time, for example, due to refine- ments in survey design, interviewing techniques, data manage- ment, and other methods to augment data integrity. The stronger association in Swedish studies probably is not explained by geographical differences in glyphosate use or effect modifiers related to NHL risk. One possible explanation is that of the six NHL studies, only the two Swedish stud- ies[14,15] compared subjects who used glyphosate with those who did not use any pesticides as the reference group, whereas the other studies defined the reference group as those who did not use glyphosate in particular. Comparisons with subjects who do not use any pesticides are more likely to be confounded by other pesticides and agricultural exposures. Meta-analysis can be problematic when applied to observa- tional epidemiology.[21,22] Meta-analysis increases statistical precision by combining results from studies that may differ substantially in terms of source population, exposure and out- come assessment and classification, control for confounding, and other key characteristics. In the presence of such heteroge- neity, even if not detectable using formal statistical tests, a sin- gle summary estimate may not be scientifically meaningful. Additionally, even when studies are statistically homogeneous, meta-analysis may not yield valid results, since this technique cannot overcome problems in the design and conduct of the underlying studies. Instead, given that bias can seldom be ruled out and unmeasured and uncontrolled confounding can never be eliminated from observational epidemiologic studies, modest meta-RRs detected across multiple studies may simply be due to shared biases, rather than a true association.[21] As stated earlier, the purpose of meta-analysis is not to evaluate whether associations are causal. We conducted a meta-analysis primar- ily for comparison with published findings. Considering the shortcomings of the existing literature, what can be done to shed further light on whether glypho- sate causes LHC in humans? Perhaps the foremost need is better exposure assessment. Self-reported information on use of specific pesticides, unless validated by comparison with sales records (which most likely would need to be col- lected prospectively, and might not be closely correlated with pesticide use) or other objective documentation, is not sufficiently accurate and reliable to yield credible estimates of association, especially exposure-response trends. Urinary glyphosate levels would provide more accurate and quanti- tatively detailed information on biological dose of glypho- sate received, but would probably have to be measured repeatedly to reflect long-term exposure. Information about temporal aspects of glyphosate expo- sure, such as the putative induction time since first use of glyphosate, duration of use, and time since last use, could help to shed light on the exposure-outcome relationship. Results from additional prospective cohort studies are nec- essary to alleviate concerns about selection and reporting bias in case-control studies. More specific outcome classification also is needed. Only two studies[14,17] examined associations between glyphosate use and more than one histological subtype of NHL, despite growing evi- dence of important etiologic heterogeneity among NHL sub- types.[74] Information on NHL subtypes also is available in the Agricultural Health Study,[66] and publication of risk associa- tions with glyphosate is anticipated. Risk factors for HL and leu- kemia also are known to differ by subtype,[76,77] yet no studies estimated associations with glyphosate separately for subtypes of these tumors. (Chronic lymphocytic leukemia and hairy-cell leu- kemia, which were analyzed as distinct outcomes, are classified as NHL subtypes.[20]) Large, probably pooled studies with histo- pathological data can determine whether associations with spe- cific tumor subtypes might be obscured by analyzing overall NHL, HL, MM, or leukemia as a single disease entity. Conclusion In conclusion, we found marginally significant positive meta- RRs for the association between glyphosate use and risk of NHL and MM, and statistically null associations with HL and leukemia. A statistically significant positive meta-RR for B-cell lymphoma, but not other NHL subtypes, was calculated based on only two studies. Combining these results with recognition of the methodological weaknesses of the small number of exist- ing studies and an overall body of literature that is not strong, consistent, temporally unambiguous, or indicative of a positive biological gradient, we determined that no causal relationship has been established between glyphosate exposure and risk of NHL, HL, MM, leukemia, or any subtype of LHC. Acknowledgments The authors thank John Acquavella and Thomas Sorahan for their thoughtful comments on earlier drafts of this manuscript, and Bernard Beckerman for his technical review of the tables. Funding This work was supported by Monsanto Company, the original producer and marketer of glyphosate formulations. Disclosure statement The sponsors were provided the opportunity to review the manuscript prior to journal submission, but inclusion of their suggestions was left to the discretion of the authors, who retained sole control of the manuscript content and the findings. Statements in this paper are those of the authors and not those of the authors’ employer or the sponsors. The authors are employed by Exponent, a scientific research and consulting firm that pro- vides services for private and governmental clients, including on projects concerning glyphosate and other pesticides. In the past five years, Ellen Chang has provided consulting services through Exponent on behalf of Monsanto Company on other issues, and she also has provided consulting services on other pesticides and lymphohematopoietic cancers for other clients. 424 E. T. CHANG AND E. DELZELL References [1] BfR. The BfR has finalised its draft report for the re-evaluation of glyphosate. Press release. Bundesinstitut f€ur Risikobewertung (BfR): Berlin, 2014. Available at http://www.bfr.bund.de/en/the_bfr_has_fi nalised_its_draft_report_for_the_re_evaluation_of_glyphosate- 188632.html (accessed Nov 2015). [2] BfR. Does glyphosate cause cancer? BfR Communication No 007/ 2015, 23 March 2015. Bundesinstitut f€ur Risikobewertung (BfR): Ber- lin, 2015. Available at http://www.bfr.bund.de/cm/349/does-glypho sate-cause-cancer.pdf (accessed Nov 2015). [3] JMPR. Pesticide residues in food - 2004. Joint FAO/WHO Meeting on Pesticide Residues. Evaluations 2004. Part II - Toxicological. World Health Organization: Geneva, 2006. [4] U.S. EPA. Reregistration Eligibility Decision (RED): Glyphosate. United States Environmental Protection Agency (U.S. EPA), Office of Prevention, Pesticides and Toxic Substances: Washington, DC, 1993. [5] U.S. EPA. Glyphosate; Pesticide Tolerances. 40 CFR Part 180 [EPA- HQ-OPP-2012-0132; FRL-9384-3]. Final Rule. Federal Register. 2013, 78, 25396-25401. [6] IARC. Glyphosate. In IARC Monographs on the Evaluation of Carci- nogenic Risks to Humans. Volume 112. Some Organophosphate Insec- ticides and Herbicides: Diazinon, Glyphosate, Malathion, Parathion, and Tetrachlorvinphos; International Agency for Research on Cancer (IARC): Lyon, France, 2015. [7] BfR. Frequently asked questions on the procedure for the re-assess- ment of glyphosate within the framework of the EU active substance review. Available at http://www.bfr.bund.de/en/frequently_asked_ questions_on_the_procedure_for_the_re_assessment_of_glyphosa te_within_the_framework_of_the_eu_active_substance_review- 195637.html (accessed Nov 2015). Bundesinstitut f€ur Risikobewer- tung (BfR): Berlin, 2015. [8] BfR. Frequently asked questions on the assessment of the health risk of glyphosate. Available at http://www.bfr.bund.de/en/frequently_as ked_questions_on_the_assessment_of_the_health_risk_of_glypho sate-127871.html#topic_195669 (accessed Nov 2015). Bundesinstitut f€ur Risikobewertung (BfR): Berlin, 2015. [9] EFSA. Conclusion on the peer review of the pesticide risk assessment of the active substance glyphosate. EFSA (European Food Safety Authority) Journal. 2015, 13, 4302. [10] JMPR. World Health Organization. Food safety. Joint FAO/WHO Meeting on Pesticide Residues (JMPR). Expert Taskforce on Diazinon, Glyphosate and Malathion. Available at http://www.who.int/foodsaf ety/areas_work/chemical-risks/jmpr/en/ (accessed Nov 2015.). [11] Schinasi, L.; Leon, M.E. Non-Hodgkin lymphoma and occupational exposure to agricultural pesticide chemical groups and active ingre- dients: a systematic review and meta-analysis. Int. J. Environ. Res. Public Health. 2014, 11, 4449-4527. [12] De Roos, A.J.; Blair, A.; Rusiecki, J.A.; Hoppin, J.A.; Svec, M.; Dose- meci, M.; Sandler, D.P.; Alavanja, M.C. Cancer incidence among glyphosate-exposed pesticide applicators in the Agricultural Health Study. Environ. Health Perspect. 2005, 113, 49-54. [13] De Roos, A.J.; Zahm, S.H.; Cantor, K.P.; Weisenburger, D.D.; Holmes, F.F.; Burmeister, L.F.; Blair, A. Integrative assessment of multiple pesticides as risk factors for non-Hodgkin’s lymphoma among men. Occup. Environ. Med. 2003, 60, E11. [14] Eriksson, M.; Hardell, L.; Carlberg, M.; Akerman, M. Pesticide expo- sure as risk factor for non-Hodgkin lymphoma including histopatho- logical subgroup analysis. Int. J. Cancer. 2008, 123, 1657-1663. [15] Hardell, L.; Eriksson, M.; Nordstrom, M. Exposure to pesticides as risk factor for non-Hodgkin’s lymphoma and hairy cell leukemia: pooled analysis of two Swedish case-control studies. Leuk. Lym- phoma. 2002, 43, 1043-1049. [16] McDuffie, H.H.; Pahwa, P.; McLaughlin, J.R.; Spinelli, J.J.; Fincham, S.; Dosman, J.A.; Robson, D.; Skinnider, L.F.; Choi, N.W. Non-Hodg- kin’s lymphoma and specific pesticide exposures in men: cross-Can- ada study of pesticides and health. Cancer Epidemiol. Biomarkers Prev. 2001, 10, 1155-1163. [17] Orsi, L.; Delabre, L.; Monnereau, A.; Delval, P.; Berthou, C.; Fenaux, P.; Marit, G.; Soubeyran, P.; Huguet, F.; Milpied, N.; Leporrier, M.; Hemon, D.; Troussard, X.; Clavel, J. Occupational exposure to pesti- cides and lymphoid neoplasms among men: results of a French case- control study. Occup. Environ. Med. 2009, 66, 291-298. [18] Cocco, P.; Satta, G.; Dubois, S.; Pili, C.; Pilleri, M.; Zucca, M.; t Mannetje, A.M.; Becker, N.; Benavente, Y.; de Sanjose, S.; Foretova, L.; Staines, A.; Maynadie, M.; Nieters, A.; Brennan, P.; Miligi, L.; Ennas, M.G.; Boffetta, P. Lymphoma risk and occupational exposure to pesticides: results of the Epilymph study. Occup. Environ. Med. 2013, 70, 91-98. [19] Mink, P.J.; Mandel, J.S.; Sceurman, B.K.; Lundin, J.I. Epidemiologic studies of glyphosate and cancer: a review. Regul. Toxicol. Pharma- col. 2012, 63, 440-452. [20] Swerdlow, S.H.; Campo, E.H., N.L.; Jaffe, E.S.; Pileri, S.A.; Stein, H.; Thiele, J.; Vardiman, J.W.WHO Classification of Tumours of Haema- topoietic and Lymphoid Tissues, 4th Ed. World Health Organization (WHO): Geneva, 2008. [21] Shapiro, S. Meta-analysis/Shmeta-analysis. Am. J. Epidemiol. 1994, 140, 771-778. [22] Weed, D.L. Meta-analysis and causal inference: a case study of benzene and non-Hodgkin lymphoma. Ann. Epidemiol. 2010, 20, 347-355. [23] Cantor, K.P.; Blair, A.; Brown, L.M.; Burmeister, L.F.; Everett, G. Pes- ticides and other agricultural risk factors for non-Hodgkin’s lym- phoma among men in Iowa and Minnesota. Cancer Res. 1992, 52, 2447-2455. [24] Cantor, K.P.; Blair, A.; Everett, G.; Gibson, R.; Burmeister, L.F.; Brown, L.M.; Schuman, L.; Dick, F.R. Pesticides and other agricul- tural risk factors for non-Hodgkin’s lymphoma among men in Iowa and Minnesota. Cancer Res. 1992, 52, 2447-2455. [25] Sorahan, T. Multiple myeloma and glyphosate use: a re-analysis of US Agricultural Health Study data. Toxicol. Lett. 2012, 211, S169-S169. [26] Sorahan, T. Multiple myeloma and glyphosate use: a re-analysis of US Agricultural Health Study (AHS) data. Int. J. Environ. Res. Public Health. 2015, 12, 1548-1559. [27] Hardell, L.; Eriksson, M. A case-control study of non-Hodgkin lym- phoma and exposure to pesticides. Cancer. 1999, 85, 1353-1360. [28] Hohenadel, K.; Harris, S.A.; McLaughlin, J.R.; Spinelli, J.J.; Pahwa, P.; Dosman, J.A.; Demers, P.A.; Blair, A. Exposure to multiple pesticides and risk of non-Hodgkin lymphoma in men from six Canadian provinces. Int. J. Environ. Res. Public Health. 2011, 8, 2320-2330. [29] Lee, W.J.; Cantor, K.P.; Berzofsky, J.A.; Zahm, S.H.; Blair, A. Non- Hodgkin’s lymphoma among asthmatics exposed to pesticides. Int. J. Cancer. 2004, 111, 298-302. [30] Nordstr€om, M.; Hardell, L.; Magnuson, A.; Hagberg, H.; Rask- Andersen, A. Occupational exposures, animal exposure and smoking as risk factors for hairy cell leukaemia evaluated in a case-control study. Br. J. Cancer. 1998, 77, 2048-2052. [31] Karunanayake, C.P.; Spinelli, J.J.; McLaughlin, J.R.; Dosman, J.A.; Pahwa, P.; McDuffie, H.H. Hodgkin lymphoma and pesticides exposure in men: a Canadian case-control study. J. Agromedicine. 2012, 17, 30-39. [32] Brown, L.M.; Burmeister, L.F.; Everett, G.D.; Blair, A. Pesticide expo- sures and multiple myeloma in Iowa men. Cancer Causes Control. 1993, 4, 153-156. [33] Kachuri, L.; Demers, P.A.; Blair, A.; Spinelli, J.J.; Pahwa, M.; McLaughlin, J.R.; Pahwa, P.; Dosman, J.A.; Harris, S.A. Multiple pes- ticide exposures and the risk of multiple myeloma in Canadian men. Int. J. Cancer. 2013, 133, 1846-1858. [34] Pahwa, P.; Karunanayake, C.P.; Dosman, J.A.; Spinelli, J.J.; McDuffie, H.H.; McLaughlin, J.R. Multiple myeloma and expo- sure to pesticides: a Canadian case-control study. J Agromedi- cine. 2012, 17, 40-50. [35] Brown, L.M.; Blair, A.; Gibson, R.; Everett, G.D.; Cantor, K.P.; Schu- man, L.M.; Burmeister, L.F.; Van Lier, S.F.; Dick, F. Pesticide expo- sures and other agricultural risk factors for leukemia among men in Iowa and Minnesota. Cancer Res. 1990, 50, 6585-6591. [36] Kaufman, D.W.; Anderson, T.E.; Issaragrisil, S. Risk factors for leuke- mia in Thailand. Ann. Hematol. 2009, 88, 1079-1088. JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 425 [37] Woodruff, T.J.; Sutton, P. The Navigation Guide systematic review methodology: a rigorous and transparent method for translating environmental health science into better health outcomes. Environ. Health Perspect. 2014, 122, 1007-1014. [38] Higgins, J.P.T.; Altman, D.G.; Sterne, J.A.C. Chapter 8: Assessing risk of bias in included studies. In Cochrane Handbook for Systematic Reviews of Interventions; Higgins, J.P.T.; Green, S. Eds.; The Cochrane Collaboration, 2011. Version 5.1.0 [updated March 2011]. Available at http://handbook.cochrane.org/ (accessed Nov 2015). [39] Guyatt, G.H.; Oxman, A.D.; Vist, G.; Kunz, R.; Brozek, J.; Alonso- Coello, P.; Montori, V.; Akl, E.A.; Djulbegovic, B.; Falck-Ytter, Y.; Norris, S.L.; Williams, J.W., Jr.; Atkins, D.; Meerpohl, J.; Schunemann, H.J. GRADE guidelines: 4. Rating the quality of evidence-study limita- tions (risk of bias). J. Clin. Epidemiol. 2011, 64, 407-415. [40] Higgins, J.P.; Thompson, S.G.; Deeks, J.J.; Altman, D.G. Measuring inconsistency in meta-analyses. BMJ. 2003, 327, 557-560. [41] Cochran, W.G. The combination of estimates from different experi- ments. Biometrics. 1954, 10, 101-129. [42] Ioannidis, J.P. Interpretation of tests of heterogeneity and bias in meta-analysis. J. Eval. Clin. Pract. 2008, 14, 951-957. [43] Dickersin, K.; Min, Y.I. Publication bias: the problem that won’t go away. Ann. N. Y. Acad. Sci. 1993, 703, 135-146; discussion 146-138. [44] Egger, M.; Davey Smith, G.; Schneider, M.; Minder, C. Bias in meta- analysis detected by a simple, graphical test. BMJ. 1997, 315, 629-634. [45] Duval, S.; Tweedie, R. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics. 2000, 56, 455-463. [46] Hill, A.B. The Environment and Disease: Association or Causation? Proc. R. Soc. Med. 1965, 58, 295-300. [47] Hoar, S.K.; Blair, A.; Holmes, F.F.; Boysen, C.D.; Robel, R.J.; Hoover, R.; Fraumeni, J.F., Jr. Agricultural herbicide use and risk of lymphoma and soft-tissue sarcoma. JAMA. 1986, 256, 1141-1147. [48] Hoar Zahm, S.; Weisenburger, D.D.; Babbitt, P.A.; Saal, R.C.; Vaught, J.B.; Cantor, K.P.; Blair, A. A case-control study of non-Hodgkin’s lymphoma and the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) in eastern Nebraska. Epidemiol. 1990, 1, 349-356. [49] Glyphosate Task Force. History of glyphosate. 2013. Available at http://www.glyphosate.eu/glyphosate-basics/history-glyphosate. (accessed Nov 2015). [50] Landgren, O.; Kyle, R.A.; Hoppin, J.A.; Beane Freeman, L.E.; Cerhan, J.R.; Katzmann, J.A.; Rajkumar, S.V.; Alavanja, M.C. Pesticide expo- sure and risk of monoclonal gammopathy of undetermined signifi- cance in the Agricultural Health Study. Blood. 2009, 113, 6386-6391. [51] Lash, T.L. Bias analysis applied to Agricultural Health Study publica- tions to estimate non-random sources of uncertainty. J. Occup. Med. Toxicol. 2007, 2, 15. [52] Acquavella, J.F.; Alexander, B.H.; Mandel, J.S.; Burns, C.J.; Gustin, C. Exposure misclassification in studies of agricultural pesticides: insights from biomonitoring. Epidemiol. 2006, 17, 69-74. [53] Rudant, J.; Orsi, L.; Monnereau, A.; Patte, C.; Pacquement, H.; Land- man-Parker, J.; Bergeron, C.; Robert, A.; Michel, G.; Lambilliotte, A.; Aladjidi, N.; Gandemer, V.; Lutz, P.; Margueritte, G.; Plantaz, D.; Mechi- naud, F.; Hemon, D.; Clavel, J. Childhood Hodgkin’s lymphoma, non- Hodgkin’s lymphoma and factors related to the immune system: the Escale Study (SFCE). Int. J. Cancer. 2011, 129, 2236-2247. [54] Hofmann, J.N.; Hoppin, J.A.; Lynch, C.F.; Poole, J.A.; Purdue, M.P.; Blair, A.; Alavanja, M.C.; Beane Freeman, L.E. Farm characteristics, allergy symptoms, and risk of non-Hodgkin lymphoid neoplasms in the agricultural health study. Cancer Epidemiol. Biomarkers Prev. 2015, 24, 587-594. [55] Blair, A.; Zahm, S.H. Methodologic issues in exposure assessment for case-control studies of cancer and herbicides. Am. J. Ind. Med. 1990, 18, 285-293. [56] Bowling, A. Mode of questionnaire administration can have serious effects on data quality. J. Public Health (Oxf). 2005, 27, 281-291. [57] Griffith, L.E.; Cook, D.J.; Guyatt, G.H.; Charles, C.A. Comparison of open and closed questionnaire formats in obtaining demographic information from Canadian general internists. J. Clin. Epidemiol. 1999, 52, 997-1005. [58] Dosman, J.A.; McDuffie, H.H.; Pahwa, P.; Fincham, S.; McLaugh- lin, J.R.; Robson, D.; Theriault, G. Pesticides, Soft Tissue Sar- coma, Lymphoma, and Multiple Myeloma. A Case Control Study in Three Regions of Canada. Report to Health and Welfare Can- ada on Project 6008-1223; University of Saskatchewan: Saska- toon, Canada, 1990. [59] Blair, A.; Tarone, R.; Sandler, D.; Lynch, C.F.; Rowland, A.; Winters- teen, W.; Steen, W.C.; Samanic, C.; Dosemeci, M.; Alavanja, M.C. Reliability of reporting on life-style and agricultural factors by a sam- ple of participants in the Agricultural Health Study from Iowa. Epi- demiol. 2002, 13, 94-99. [60] Blair, A.; Thomas, K.; Coble, J.; Sandler, D.P.; Hines, C.J.; Lynch, C. F.; Knott, C.; Purdue, M.P.; Zahm, S.H.; Alavanja, M.C.; Dosemeci, M.; Kamel, F.; Hoppin, J.A.; Freeman, L.B.; Lubin, J.H. Impact of pes- ticide exposure misclassification on estimates of relative risks in the Agricultural Health Study. Occup. Environ. Med. 2011, 68, 537-541. [61] Coble, J.; Thomas, K.W.; Hines, C.J.; Hoppin, J.A.; Dosemeci, M.; Curwin, B.; Lubin, J.H.; Beane Freeman, L.E.; Blair, A.; Sandler, D.P.; Alavanja, M.C. An updated algorithm for estimation of pesticide exposure intensity in the agricultural health study. Int. J. Environ. Res. Public Health. 2011, 8, 4608-4622. [62] Agricultural Health Study. Agricultural Health Study Questionnaires & Study Data. Available at http://aghealth.nih.gov/collaboration/ques tionnaires.html (accessed 28 November 2015). 1996. [63] Jurek, A.M.; Greenland, S.; Maldonado, G. How far from non-differential does exposure or disease misclassification have to be to bias measures of association away from the null? Int. J. Epidemiol. 2008, 37, 382-385. [64] Jurek, A.M.; Greenland, S.; Maldonado, G.; Church, T.R. Proper inter- pretation of non-differential misclassification effects: expectations vs observations. Int. J. Epidemiol. 2005, 34, 680-687. [65] Weichenthal, S.; Moase, C.; Chan, P. A review of pesticide exposure and cancer incidence in the Agricultural Health Study cohort. Envi- ron. Health Perspect. 2010, 118, 1117-1125. [66] Alavanja, M.C.; Hofmann, J.N.; Lynch, C.F.; Hines, C.J.; Barry, K.H.; Barker, J.; Buckman, D.W.; Thomas, K.; Sandler, D.P.; Hoppin, J.A.; Kou- tros, S.; Andreotti, G.; Lubin, J.H.; Blair, A.; Beane Freeman, L.E. Non- Hodgkin lymphoma risk and insecticide, fungicide and fumigant use in the Agricultural Health Study. PLoSOne. 2014, 9, e109332. [67] Pearce, N.; McLean, D. Agricultural exposures and non-Hodgkin’s lymphoma. Scand. J. Work Environ. Health. 2005, 31(Suppl 1), 18- 25; discussion 15-17. [68] McDuffie, H.H.; Pahwa, P.; Spinelli, J.J.; McLaughlin, J.R.; Fin- cham, S.; Robson, D.; Dosman, J.A.; Hu, J. Canadian male farm residents, pesticide safety handling practices, exposure to animals and non-Hodgkin’s lymphoma (NHL). Am. J. Ind. Med. 2002, Suppl 2, 54-61. [69] Fritschi, L.; Johnson, K.C.; Kliewer, E.V.; Fry, R.; Canadian Cancer Registries Epidemiology Research, G. Animal-related occupations and the risk of leukemia, myeloma, and non-Hodgkin’s lymphoma in Canada. Cancer Causes Control. 2002, 13, 563-571. [70] Khuder, S.A.; Mutgi, A.B.; Schaub, E.A.; Tano, B.D. Meta-analysis of Hodgkin’s disease among farmers. Scand. J. Work Environ. Health. 1999, 25, 436-441. [71] Pahwa, P.; McDuffie, H.H.; Dosman, J.A.; Robson, D.; McLaughlin, J.R.; Spinelli, J.J.; Fincham, S. Exposure to animals and selected risk factors among Canadian farm residents with Hodgkin’s disease, multiple mye- loma, or soft tissue sarcoma. J. Occup. Environ.Med. 2003, 45, 857-868. [72] Perrotta, C.; Staines, A.; Cocco, P. Multiple myeloma and farming. A systematic review of 30 years of research. Where next? J. Occup. Med. Toxicol. 2008, 3, 27. [73] Keller-Byrne, J.E.; Khuder, S.A.; Schaub, E.A. Meta-analysis of leuke- mia and farming. Environ. Res. 1995, 71, 1-10. [74] Morton, L.M.; Slager, S.L.; Cerhan, J.R.; Wang, S.S.; Vajdic, C.M.; Ski- bola, C.F.; Bracci, P.M.; de Sanjose, S.; Smedby, K.E.; Chiu, B.C.; Zhang, Y.; Mbulaiteye, S.M.; Monnereau, A.; Turner, J.J.; Clavel, J.; Adami, H.O.; Chang, E.T.; Glimelius, B.; Hjalgrim, H.; Melbye, M.; Crosignani, P.; di Lollo, S.; Miligi, L.; Nanni, O.; Ramazzotti, V.; Rodella, S.; Costantini, A.S.; Stagnaro, E.; Tumino, R.; Vindigni, C.; Vineis, P.; Becker, N.; Benavente, Y.; Boffetta, P.; Brennan, P.; Cocco, 426 E. T. CHANG AND E. DELZELL P.; Foretova, L.; Maynadie, M.; Nieters, A.; Staines, A.; Colt, J.S.; Cozen, W.; Davis, S.; de Roos, A.J.; Hartge, P.; Rothman, N.; Severson, R.K.; Holly, E.A.; Call, T.G.; Feldman, A.L.; Habermann, T.M.; Liebow, M.; Blair, A.; Cantor, K.P.; Kane, E.V.; Lightfoot, T.; Roman, E.; Smith, A.; Brooks-Wilson, A.; Connors, J.M.; Gascoyne, R.D.; Spinelli, J.J.; Arm- strong, B.K.; Kricker, A.; Holford, T.R.; Lan, Q.; Zheng, T.; Orsi, L.; Dal Maso, L.; Franceschi, S.; La Vecchia, C.; Negri, E.; Serraino, D.; Bernstein, L.; Levine, A.; Friedberg, J.W.; Kelly, J.L.; Berndt, S.I.; Bir- mann, B.M.; Clarke, C.A.; Flowers, C.R.; Foran, J.M.; Kadin, M.E.; Pal- tiel, O.; Weisenburger, D.D.; Linet, M.S.; Sampson, J.N. Etiologic heterogeneity among non-Hodgkin lymphoma subtypes: the Inter- Lymph Non-Hodgkin Lymphoma Subtypes Project. J. Natl. Cancer Inst. Monogr. 2014, 2014, 130-144. [75] Becker, N. Epidemiology of multiple myeloma. Recent Results Can- cer Res. 2011, 183, 25-35. [76] Glaser, S.L.; Chang, E.T.; Clarke, C.A.K., T.H. Chapter 1. Epidemiol- ogy. In Hodgkin Lymphoma. A Comprehensive Overview. 2nd ed.; Engert, A.; Younes, A. Eds.; Springer: New York, 2015; 3-26. [77] Linet, M.S.; Devesa, S.S.; Morgan, G.J. Chapter 44. The Leukemias. In Cancer Epidemiology and Prevention, 3rd Ed.; Schottenfeld, D.; Frau- meni, J.F., Jr. Eds.; Oxford University Press: New York, 2006; 841- 871. [78] Stiernstrom, E.L.; Holmberg, S.; Thelin, A.; Svardsudd, K. A prospec- tive study of morbidity and mortality rates among farmers and rural and urban nonfarmers. J. Clin. Epidemiol. 2001, 54, 121-126. [79] McCurdy, S.A.; Carroll, D.J. Agricultural injury. Am. J. Ind. Med. 2000, 38, 463-480. [80] Kavvoura, F.K.; Liberopoulos, G.; Ioannidis, J.P. Selection in reported epidemiological risks: an empirical assessment. PLoS Med. 2007, 4, e79. [81] Pahwa, M.; Harris, S.A.; Hohenadel, K.; McLaughlin, J.R.; Spinelli, J. J.; Pahwa, P.; Dosman, J.A.; Blair, A. Pesticide use, immunologic con- ditions, and risk of non-Hodgkin lymphoma in Canadian men in six provinces. Int. J. Cancer. 2012, 131, 2650-2659. [82] Acquavella, J.; Farmer, D.; Cullen, M.R. A case-control study of non- Hodgkin lymphoma and exposure to pesticides. Cancer. 1999, 86, 729-731. [83] Wester, R.C.; Melendres, J.; Sarason, R.; McMaster, J.; Maibach, H.I. Glyphosate skin binding, absorption, residual tissue distribution, and skin decontamination. Fundam. Appl. Toxicol. 1991, 16, 725-732. [84] Acquavella, J.F.; Alexander, B.H.; Mandel, J.S.; Gustin, C.; Baker, B.; Chapman, P.; Bleeke, M. Glyphosate biomonitoring for farmers and their families: results from the Farm Family Exposure Study. Envi- ron. Health Perspect. 2004, 112, 321-326. [85] Niemann, L.; Sieke, C.; Pfeil, R.; Solecki, R. A critical review of glypho- sate findings in human urine samples and comparison with the expo- sure of operators and consumers. J. Verbr. Lebensm. 2015, 10, 3-12. [86] U.S. EPA. Glyphosate (CASRN 1071-83-6). Integrated Risk Informa- tion System. Carcinogenicity Assessment for Lifetime Exposure, 1993. Available at http://www.epa.gov/iris/subst/0057.htm. (accessed Nov 2015). [87] Kier, L.D.; Kirkland, D.J. Review of genotoxicity studies of glyphosate and glyphosate-based formulations. Crit. Rev. Toxicol. 2013, 43, 283-315. [88] Greim, H.; Saltmiras, D.; Mostert, V.; Strupp, C. Evaluation of carci- nogenic potential of the herbicide glyphosate, drawing on tumor incidence data from fourteen chronic/carcinogenicity rodent studies. Crit. Rev. Toxicol. 2015, 45, 185-208. [89] Fasal, E.; Jackson, E.W.; Klauber, M.R. Leukemia and lymphoma mortality and farm residence. Am. J. Epidemiol. 1968, 87, 267-274. [90] Milham, S., Jr. Leukemia and multiple myeloma in farmers. Am. J. Epidemiol. 1971, 94, 507-510. [91] Sandin, S.; Hjalgrim, H.; Glimelius, B.; Rostgaard, K.; Pukkala, E.; Askling, J. Incidence of non-Hodgkin’s lymphoma in Sweden, Den- mark, and Finland from 1960 through 2003: an epidemic that was. Cancer Epidemiol. Biomarkers Prev. 2006, 15, 1295-1300. [92] Holford, T.R.; Zheng, T.; Mayne, S.T.; McKay, L.A. Time trends of non-Hodgkin’s lymphoma: are they real? What do they mean? Can- cer Res. 1992, 52, 5443s-5446s. [93] Thygesen, L.C.; Nielsen, O.J.; Johansen, C. Trends in adult leukemia incidence and survival in Denmark, 1943-2003. Cancer Causes Con- trol. 2009, 20, 1671-1680. [94] Hjalgrim, H. On the aetiology of Hodgkin lymphoma. Dan. Med. J. 2012, 59, B4485. [95] Hirabayashi, Y.; Katanoda, K. Comparison of time trends in multiple myeloma incidence (1973-1997) in East Asia, Europe and United States, from Cancer Incidence in Five Continents, Vols IV-VIII. Jpn. J. Clin. Oncol. 2008, 38, 720-721. [96] Morton, L.M.; Wang, S.S.; Devesa, S.S.; Hartge, P.; Weisenburger, D.D.; Linet, M.S. Lymphoma incidence patterns by WHO sub- type in the United States, 1992-2001. Blood. 2006, 107, 265-276. [97] Button, K.S.; Ioannidis, J.P.; Mokrysz, C.; Nosek, B.A.; Flint, J.; Robinson, E.S.; Munafo, M.R. Power failure: why small sample’- size undermines the reliability of neuroscience. Nat. Rev. Neurosci. 2013, 14, 365-376. Appendix Literature search methods The authors conducted a search of MEDLINE via PubMed using the following search string, which includes Chemical Abstracts Service (CAS) Registry Numbers for glyphosate and its salts: (glyphosat OR glifosat OR glyfosat OR gliphosat OR Roundup OR Round-up OR 1071-83-6 OR 38641-94-0 OR 70901-12-1 OR 39600-42-5 OR 69200-57-3 OR 34494-04-7 OR 114370-14-8 OR 40465-66-5 OR 69254-40-6 OR (aminomethyl w phosphonic) OR 1066-51-9 OR pesticid OR herbicid OR organophosphorus compounds [MeSH] OR pesticides [MeSH] OR herbicides [MeSH]) AND (leukemi OR leukaemi OR lymphoma OR NHL OR lymphopoietic OR hemato OR hema- topoie or hematolog OR lymphoid OR myeloid OR myeloma OR leukemia [MeSH] OR lymphoma [MeSH] OR multiple mye- loma [MeSH]) AND (cases OR controls OR case-control OR cohort). As of June 23, 2015, this search string identified a total of 11,755 articles in PubMed. We conducted additional targeted searches in PubMed, Web of Science, and Google Scholar using simpler keyword combinations such as (glyphosate AND lym- phoma), (pesticides AND lymphoma), and (herbicides AND lym- phoma). References also were identified from the bibliographies of recent review articles. Altogether, a total of 12,709 articles were identified from these combined sources (Fig. A1). Based on a review of titles and abstracts, 321 articles were identified as potentially containing estimates of the association between glyphosate exposure and LHC risk, and were obtained for further evaluation. Forty-seven of these articles con- tained the word “glyphosate” or “Roundup” (or alternative spellings of these terms) in the text; as specified earlier, articles that did not mention glyphosate were ineligible for inclusion. Following a review of the full text of each of the 47 articles mentioning glyphosate, 19 articles were ultimately deemed eligible for inclusion. JOURNAL OF ENVIRONMENTAL SCIENCE AND HEALTH, PART B 427 Figure A1. Flow chart of literature identification and selection process. 428 E. T. CHANG AND E. DELZELL 12,709 articles identified from PubMed, Web of Science, Google Scholar, and reference lists 12,388 articles excluded based on title and abstract -- w 321 full-text articles retrieved 27 4 articles excluded based on absence of uglyphosate" -- and "Roundup" in text ,~ 4 7 full-text articles reviewed for eligibility 28 articles excluded based on relevance and available --data w 19 articles included in review 12 non-Hodgkin lymphoma 2 Hodplymphoma 7 multiple m yelom a 3 leukemia EXHIBIT D This article was downloaded by: [John M. DeSesso] On: 04 January 2012, At: 07:29 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Journal of Toxicology and Environmental Health, Part B Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/uteb20 Developmental and Reproductive Outcomes in Humans and Animals After Glyphosate Exposure: A Critical Analysis Amy Lavin Williams a , Rebecca E. Watson b & John M. DeSesso a c a Exponent, Inc, Alexandria, Virginia, USA b SNBL USA, Everett, Washington, USA c Georgetown University School of Medicine, Washington, DC, USA Available online: 27 Dec 2011 To cite this article: Amy Lavin Williams, Rebecca E. Watson & John M. DeSesso (2012): Developmental and Reproductive Outcomes in Humans and Animals After Glyphosate Exposure: A Critical Analysis, Journal of Toxicology and Environmental Health, Part B, 15:1, 39-96 To link to this article: http://dx.doi.org/10.1080/10937404.2012.632361 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. Journal of Toxicology and Environmental Health, Part B, 15:39-96, 2012 Copyright © Taylor & Francis Group, LLC ISSN: 1093-7404 print / 1521-6950 online DOI: 10.1080/10937404.2012.632361 DEVELOPMENTAL AND REPRODUCTIVE OUTCOMES IN HUMANS AND ANIMALS AFTER GLYPHOSATE EXPOSURE: A CRITICAL ANALYSIS Amy Lavin Williams1, Rebecca E. Watson2, John M. DeSesso1,3 1Exponent, Inc, Alexandria, Virginia, USA 2SNBL USA, Everett, Washington, USA 3Georgetown University School of Medicine, Washington, DC, USA Glyphosate is the active ingredient of several widely used herbicide formulations. Glyphosate targets the shikimate metabolic pathway, which is found in plants but not in animals. Despite the relative safety of glyphosate, various adverse developmental and reproductive problems have been alleged as a result of exposure in humans and animals. To assess the developmental and reproductive safety of glyphosate, an analysis of the available literature was conducted. Epidemiological and animal reports, as well as studies on mechanisms of action related to possible developmental and reproductive effects of glyphosate, were reviewed. An evaluation of this database found no consistent effects of glyphosate exposure on reproductive health or the developing offspring. Furthermore, no plausible mechanisms of action for such effects were elucidated. Although toxicity was observed in studies that used glyphosate-based for- mulations, the data strongly suggest that such effects were due to surfactants present in the formulations and not the direct result of glyphosate exposure. To estimate potential human exposure concentrations to glyphosate as a result of working directly with the herbicide, available biomonitoring data were examined. These data demonstrated extremely low human exposures as a result of normal application practices. Furthermore, the estimated exposure concentrations in humans are >500-fold less than the oral reference dose for glyphosate of 2 mg/kg/d set by the U.S. Environmental Protection Agency (U.S. EPA 1993). In conclu- sion, the available literature shows no solid evidence linking glyphosate exposure to adverse developmental or reproductive effects at environmentally realistic exposure concentrations. Glyphosate, shown in Figure 1, is the active ingredient of several widely used herbicide formulations including Roundup, AquaMaster, and Vision branded products. First approved for the broad-spectrum control of weeds in 1974 (Franz et al. 1997), the applications for glyphosate have been expanded over the years, making it one of the most commonly used herbicides worldwide. Today, glyphosate- based herbicide formulations are used in over 100 countries, in almost all phases of agri- cultural, industrial, silvicultural, and residential weed control. In 2001 (the most recent year for which usage statistics are available from the The authors acknowledge the Monsanto Company for funding and for providing its unpublished glyphosate and surfactant toxicity study reports. Address correspondence to Dr. John M. DeSesso, Exponent, Inc., 1800 Diagonal Road, Suite 300, Alexandria, VA 22314, USA. E-mail: jdesesso@exponent.com U.S. Environmental Protection Agency [EPA]), an estimated 85-90 million lb of glyphosate was applied in the U.S. agricultural sector alone, making it the number one pesticide active ingredient used in the United States (Kiely et al. 2004). Glyphosate’s rapid rise in popularity since its first introduction is not only because of its effectiveness in controlling the growth of unwanted vegetation, but also in large part due to its relative safety for humans and animals. As previously noted by Williams et al. (2000), glyphosate has properties of both an acid and a base; consequently, the chemical 39 D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 (.?\ Taylor & Francis ~ Taylor&FrancisGroup 40 A. L. WILLIAMS ET AL. FIGURE 1. Chemical structures of glyphosate (A) and its major breakdown product, AMPA (B). has several pKa values (<2, 2.6, 5.6, 10.6; WHO 1994). It has a high water solubility of 12 mg/ml and reported octanol-water parti- tion coefficients (log Kow) of -2.8 and -1.7 (Nielsen et al. 2009; WHO 1994). Glyphosate acid is typically referred to as the technical- grade material, and commercial formulations contain glyphosate in the form of a salt (i.e., potassium, isopropylamine, or ammonium). In addition to glyphosate and water, com- mercial formulations also typically include a surfactant system, which enables the herbicide product to adhere to the surface of leaves, allowing penetration of the active ingredient (Franz et al. 1997). For instance, in the case of Roundup, there is no single formulation, but rather a range of Roundup-branded products that differ in their surfactants and other ingredi- ents depending on their intended use. One of the more common classes of surfactants used in glyphosate formulations is polyoxyethyle- neamines (POEA) (Bradberry et al. 2004). Surfactants may possess their own toxicological properties; these are discussed in more detail in the following and in Bradberry et al. (2004). In plants, glyphosate inhibits enolpyru- vylshikimate phosphate synthase, an enzyme required for the synthesis of several essential aromatic amino acids (Franz et al. 1997). This metabolic pathway is common to all plants, making glyphosate an effective, nonselective herbicide. Because the shikimate pathway is not shared by members of the animal king- dom, glyphosate is not expected to adversely affect humans and other mammals under nor- mal use conditions. While classified under the herbicide class phosphonomethyl amino acids, glyphosate is often mischaracterized as an organophosphate. This is likely due to the molecular structure being an organic molecule containing a phosphorus atom. However, clini- cal reports describing incidents of human inges- tion of glyphosate do not reflect the classic symptoms for organophosphate poisoning (sali- vation, lacrimation, urination, and defecation; Costa 2008). Glyphosate is not anticipated to persist in the environment for extended peri- ods of time following application. Glyphosate is nonvolatile and binds tightly to most soils, making it unlikely to migrate to groundwa- ter or reach nontarget plants. Over time, it is degraded by microbes in soil and natural waters into substances such as carbon dioxide and phosphate (Giesy et al. 2000). These fac- tors limit the degree of exposure of nontarget species to glyphosate, thus further increasing its relative safety. Based on these facts, in addition to results from animal toxicity studies, the U.S. Environmental Protection Agency (U.S. EPA), as well as other regulatory agencies worldwide, has deemed glyphosate to be of low to minimal toxicity to humans via reasonably anticipated exposure routes (U.S. EPA 1993). Potential routes for human exposure to glyphosate include inadvertent ocular expo- sure during herbicide mixing and application; dermal exposure due to mixing/application or contact with treated plants; and oral expo- sure through the ingestion of treated crops or contaminated water, although accidental inges- tion of larger amounts by children and adults and intentional ingestion for suicidal purposes have been reported. Inhalation of glyphosate is anticipated to be minimal because of the chemical’s nonvolatility. In fact, the U.S. EPA registration requirement for an acute inhala- tion study was waived for glyphosate due its nonvolatile nature (U.S. EPA 1993). Data related to dermal exposures indi- cate extremely low skin absorption rates. In non-Good Laboratory Practices (GLP) studies D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 0 HO"'--. //0 H 0 P N- LI / ~ '-...../"'-OH HO Glyphosate (CAS RN l 071-83-6) AMPA (aminomethylphosphosphonic acid; CAS RN l 066-51-9) GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 41 conducted with rhesus monkeys, the mea- sured absorption of glyphosate applied to the skin ranged from 0.4 to 2.2% (Wester et al. 1991; Maibach 1986), depending on the type of material applied to the skin (diluted or undiluted herbicidal formulation versus pure glyphosate), the duration of exposure, and the applied volume. In vivo and in vitro human dermal absorption studies showed that <2% of Roundup is absorbed as either a concen- trated or diluted spray (Franz 1983; Wester et al. 1991). Several recent in vitro human der- mal absorption studies have been conducted on glyphosate or glyphosate-based formula- tions. Nielsen et al. (2009) tested a variety of compounds including glyphosate using an OECD 428-like design with 0.4% glyphosate absorbed through the skin and 0.7% recov- ered in or on the skin; total glyphosate recov- ery was 92%. This type of in vitro study is designed to examine the penetration of a sub- stance across excised skin as a measure of potential dermal absorption (OECD 2004). The study by Nielsen et al. (2009) was conser- vative compared to established guidelines in that (i) the duration of exposure was 48 h without an interim wash; (ii) no tape strips were removed from the skin to differentiate between biologically available glyphosate and that contained in top stratum corneum lay- ers; and (iii) test cells were occluded with parafilm, resulting in hydration of the skin and potentially enhanced dermal permeabil- ity. In a later study that involved a similar experimental protocol but with an interim wash at 6 h (Nielsen 2010), glyphosate absorption was only 0.1% of dose and 0.1% was recov- ered in or on the skin; total glyphosate recov- ery was 100%. Ward (2009a; 2009b; 2009c) conducted OECD-compliant studies on three different glyphosate-based formulations, test- ing concentrate and two dilutions representing the range of field concentrations for each of the three formulations (dilutions ranging from 12.5× to 200×). These studies were conser- vative in that exposures were for a full 24 h with no interim washes. Total glyphosate recov- ery in these experiments ranged from 98.6 to 106%. All experiments exhibited extremely low glyphosate biological availability (total absorbed + remaining in skin after tape strip- ping), ranging from less than 0.05 to 0.123% for the concentrates and from less than 0.14 to 0.8% for the dilute formulations. Based on studies conducted in the rat, oral absorption also appears to be limited. Following administration of a single oral dose (5.6-10 mg/kg), approximately 30-36% of the glyphosate is absorbed in the rat, as deter- mined by measurements of the area under the curve for whole blood and urinary excretion data (Brewster et al. 1991; Chan and Mahler 1992; Williams et al. 2000). Oral absorption is further reduced to 19-23% by the appli- cation of extremely high doses of glyphosate (e.g., 1000 mg/kg) and in repeat-dosing regi- mens (Chan and Mahler 1992; Williams et al. 2000). Based on its ionic nature and high water solubility, glyphosate does not bioac- cumulate to any appreciable levels (Brewster et al. 199; WHO 1994). Following ingestion of a relatively large dose in rats, the blood glyphosate concentrations peaked at around 1-2 h post administration; urinary elimina- tion is complete at approximately 12 h post dosing (Chan and Mahler 1992). Additional data show that glyphosate is poorly metabo- lized and generally eliminated in the urine and feces unchanged (Brewster et al. 1991). Despite the well-established safety of glyphosate for humans by regulatory agencies, it was suggested that chronic, low-level expo- sure may lead to developmental and repro- ductive health problems, particularly for men and women residing in agricultural areas asso- ciated with heavy herbicide use (Solomon et al. 2009). This notion was developed based primarily on results from recently pub- lished in vitro and animal studies conducted using glyphosate-based herbicide formulations (Benachour et al. 2007; Dallegrave et al. 2003; 2007; Dariuch et al. 2001; Gasnier et al. 2009; Marc et al. 2005; Pagenelii et al. 2010; Richard et al. 2005; Romano et al. 2010). In order to determine whether appropriate data exist to support this claim, a thorough evaluation of the scientific literature was con- ducted. Experimental investigations conducted D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 42 A. L. WILLIAMS ET AL. by the Monsanto Company in support of reg- ulatory requirements were made available to the authors. These studies were compliant both with contemporary regulatory guidelines and GLP. In addition, research reports published in the open scientific literature were identi- fied through automated searches of PubMed, SciFinder, and ToxLine. Critiques of all the reviewed studies are included herein, along with rationales for using or discounting the reported results for the purpose of human health risk assessment. Finally, assessments published by international organizations and regulatory agencies were examined as sup- porting documentation. Emphasis was placed on identifying potentially adverse reproductive health and developmental effects; however, a review of available biomonitoring data was also conducted to understand anticipated exposure levels for humans. ASSESSMENT OF POTENTIAL REPRODUCTIVE HEALTH AND DEVELOPMENTAL EFFECTS Humans are the primary focus for this evaluation of possible reproductive and devel- opmental effects. Consequently, published epi- demiology studies addressing the potential for such outcomes upon exposure to glyphosate- containing herbicides were first assessed. Next, animal studies (both published reports as well as unpublished studies owned by Monsanto) addressing appropriate toxicity endpoints were reviewed. Finally, published mechanistic stud- ies using glyphosate and glyphosate-based herbicidal formulations were evaluated to determine whether a plausible mechanism of action (MOA) could be established to explain how glyphosate may contribute to reproductive and/or developmental problems in humans and other mammals. Epidemiological Evaluation Epidemiology is the study of disease pat- terns and the factors that play a role in their occurrence. The strongest epidemiological designs are cohort and case-control studies. Cohort studies are longitudinal prospective or retrospective investigations of persons exposed to an agent of interest. Exposed and unexposed populations are identified, and the prevalence of the disease or condition of interest within both populations over time is then assessed. Disease occurrences that are followed prospec- tively are more likely to have accurate expo- sure information, because confounding factors associated with inaccurate recall are reduced. Retrospective cohort studies rely on the accu- racy of information entered in various reg- istries (for example, birth and employment registries) and appropriate exposure recall of interviewees. Case-control studies identify persons who have developed a particular disease (cases) and then retrospectively investigate the histo- ries and habits of this population to determine whether any differences exist between the case population and another of disease-free individ- uals (controls) who are matched for age, sex, body weight, and exposure to environmental factors. When evaluating the results of case- control studies, recall bias among the cases and controls, the accuracy of the reporting physi- cian and/or hospital (if birth registries are used), and the possibility of interviewer bias must be considered. Other types of epidemiological investiga- tions, including cluster analyses, general obser- vational studies, and cross-sectional studies, are less rigorously designed than cohort and case-control studies. Cluster analyses report observations of isolated disease cases, often related to exposures to a specific agent. While considerable detail about each particular case is often available, the small number of cases results in insufficient statistical power to estab- lish an association between exposure and a specific disease. General observational stud- ies investigate whether exposure to an agent is related to the outcome of interest, but the numbers of cases and controls, the selection criteria, and/or the type(s) of controls are not as robust as those in the case-control and cohort studies. Often, general observational studies are based on birth records, employment records, death certificates, or surveys of cases D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 43 and controls. Cross-sectional studies are obser- vational studies that primarily ascertain the dis- ease incidence at a moment in time for a given population, as opposed to dealing with indi- viduals and their histories. Thus, cross-sectional studies report disease rates rather than cases, which precludes identifying many confounding factors, which are often based on data for each individual. Although a substantial body of data exists regarding the adverse reproductive and devel- opmental effects of pesticide exposures in general, few epidemiology studies have been conducted to specifically assess the poten- tially adverse effects associated with glyphosate exposure. Based on a review of the scientific lit- erature, only 11 epidemiology studies evaluat- ing pregnancy outcomes or reproductive health as they relate to glyphosate exposures were identified (Tables 1 and 2). With the excep- tion of Sanin et al. (2009), who only examined glyphosate but did not conduct detailed expo- sure assessments, none of these studies was designed to specifically assess exposures to glyphosate; rather, these studies address pesti- cide and/or herbicide exposures, categorized by use or chemical group. In the majority of these publications, glyphosate is only men- tioned in passing, or is grouped with other herbicides such as phosphonamino herbicides or organophosphate pesticides (an incorrect classification of glyphosate). Developmental Effects The Ontario Farm Family Health Study (OFFHS) was initi- ated in 1990/1991 to retrospectively examine the possible associations between various pesticide exposures and adverse develop- mental and reproductive outcomes. Based on the 1986 Canadian Census of Agriculture, 7379 farms in Ontario were identified as likely to be full-time family-run operations. Through telephone screening, a subset of 2946 couples was identified as eligible for study based on residence on or near the farm year-round and the female partner being ≤44 years of age. In total, 1898 couples provided completed TABLE 1. Epidemiological Studies Assessing Glyphosate Exposure and Potential Developmental Effects Study Agent Exposure Study population Endpoints Outcome Savitz et al. 1997 Glyphosate Male, self-reported (0-3 mo prior to conception) 3984 pregnancies Ontario Farm Family Health Study (OFFHS) Miscarriages, preterm deliveries, SGA births No effect Arbuckle et al. 2001 Glyphosate Male or female, self-reported (0-3 mo prior to 1st trimester) 395 spontaneous abortions OFFHS Spontaneous abortions No effect Bell et al. 2001a Phosphate pesticides Female, via maternal address (1-20 wk gestation) 74 fetal deaths due to congenital anomalies 20 wk gestation to 24 h after birth Fetal deaths Increased odds Bell et al. 2001b Phosphate pesticides Female, via maternal address (1-20 wk gestation) 413 fetal deaths due to causes other than above 20 wk gestation to 24 hr after birth Fetal deaths No effect Rull et al. 2006 Glyphosate Female, via maternal address (periconception) 731 NTD (anencephaly, spina bifida cystica, other sub-types) Neural tube defects No effect Garry et al. 2002 Herbicides, insecticides, fungicides, fumigants Male or female, self-reported 1532 live births Birth defects Increased risk with exposure to all 4 classes (in column 2) D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 44 A. L. WILLIAMS ET AL. TABLE 2. Epidemiological Studies Assessing Glyphosate Exposure and Potential Effects of Reproductive Health Study Agent Exposure Study population Endpoints Outcome Curtis et al. 1999 Glyphosate Male or female, self reported (0-2 mo prior to pregnancy attempts) 2,012 planned pregnancies OFFHS Fertility ♀: Decrease♂: Increase Use on farm: Increase Larsen et al. 1998 Spermatotoxic. pesticides Male, self-reported (1 yr prior to child’s birth) 904 pregnancies Time to pregnancy No effect Sanin et al. 2009; Solomon et al. 2009 Glyphosate Female, via residence in region of aerial glyphosate application 2751 first pregnancies Time to pregnancy No effect Greenlee et al. 2003 Herbicides Female, self-reported (2 yr prior to pregnancy attempts) 322 cases of female infertility Infertility Increased Farr et al. 2004 Glyphosate Female, self-reported (12 mo previous) 3103 premenopausal women Agricultural Health Study Menstrual cycle factors (cycle length, intermenstrual bleeding, missed periods) No effect questionnaires from the farm operator, hus- band, and wife (response rate of 64%). These couples reported a total of 5853 pregnancies. Pregnancy outcomes were determined through maternal self reports. Based on the OFFHS, three separate papers investigating the adverse effects of glyphosate exposure on prenatal development or reproductive health were published (Savitz et al. 1997; Curtis et al. 1999; Arbuckle et al. 2001). Other papers published from the OFFHS either did not address potential effects of glyphosate expo- sure or did not examine prenatal development and/or reproductive health. Pregnancy outcome. In Savitz et al. (1997), data from the OFFHS were used to examine the possible association of male pesticide exposure with adverse pregnancy outcomes. Out of a total of 5853 OFFHS preg- nancies, 3984 were included in this study. Approximately 40% of those included in the analysis occurred 10 yr or more before the study commenced, which likely introduces some recall bias. The majority of those preg- nancies not included in the study were omitted because the pregnancy did not occur while in residence on the farm. Pregnancies were clas- sified according to outcome (single live birth, miscarriage, stillbirth, preterm, small for ges- tational age [SGA], etc.), but were not con- firmed through medical records. Males were asked about their farm activities over the past 5 years. Activities involving the mixing and/or application of crop herbicides, crop insecti- cides or fungicides, livestock chemicals, yard herbicides, and/or building pesticides met the study requirements for direct pesticide expo- sure. All reported activities were assumed to extend backward beyond the 5-yr period cov- ered by the questionnaire. Each pregnancy was classified as exposed or not exposed based on whether the male partner partook in an activity involving pesticide exposure for 1 mo or more during the 3 mo prior to the time of conception or during the month of conception itself. Based on information pro- vided by the farm operators, the analyses were further refined, taking into considera- tion the specific pesticides used on the farms during the preconception/conception period. These methods for exposure assessment likely introduced substantial exposure misclassifica- tion, especially for those pregnancies that occurred prior to the 5-yr period covered by the questionnaire. Substantial recall bias is also likely since fathers of pregnancies with adverse D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 45 outcomes are more likely to recall pesticide use during the preconception period than fathers of pregnancies with normal outcomes. Study authors controlled for potential confounders (including age of parents, level of education, jobs outside the farm, smoking habits, alcohol consumption, and caffeine use) in their anal- yses, as appropriate. The study results show that glyphosate exposure of males was not associated with an increased risk of miscar- riage, preterm delivery, or SGA delivery by their female partners. Spontaneous abortions. Using the OFFHS data, Arbuckle et al. (2001) further dissected out the influence of pre- and postconception pesticide exposures on spontaneous abortion, or miscarriage risks. Self-reported miscarriages of less than 20 wk of gestation were divided into 2 groups: those of less than 12 wk of gestation and those occurring between wk 12 and 19 of gestation. Out of a total of 3936 pregnancies, 395 spontaneous abortions were reported; all but 5 were reported to have been medically confirmed and 57% occurred before 12 wk of gestation. The approximate 10% rate for spontaneous abortions reported in this study is substantially lower than the approx- imate 12-25% rate reported for the general population (Everett 1997; Wilcox et al. 1999). Recall was greater than 5 yr for 64% of the spontaneous abortions and greater than 10 yr for 34% (Arbuckle et al. 1999), thus allowing for some recall bias, as previously discussed. For each pregnancy, a monthly agricultural and res- idential pesticide use history was constructed from questionnaire data provided by the farm operator, husband, and wife. Pesticide expo- sures were analyzed using two exposure win- dows: the preconception period (defined as the three calendar months prior to concep- tion and the month of conception, combined) and the postconception period (defined as the three calendar months of the first trimester). Although not specifically discussed in this arti- cle, it is likely that the exposure data were extrapolated in part, as was done in Savitz et al. (1997) and Curtis et al. (1999; discussed later); thus, substantial exposure misclassifi- cation may have occurred. Because strong confounding variables were not apparent from previous analyses of the data (Arbuckle et al. 1999), only crude odds ratios (OR) were cal- culated. Preconception exposure of fathers to glyphosate showed an elevated risk for spon- taneous abortions between wk 12 and 19 of gestation, although this risk was not statistically significant (OR = 1.7, 95% confidence interval [CI] = 1.0-2.6). No other pre- or postconcep- tion exposures to glyphosate demonstrated an increased risk of spontaneous abortions either before 12 wk or between wk 12 and 19 of gestation. A classification and regression tree analysis was used to explore possible statisti- cal interactions between exposures and various risk factors for spontaneous abortion. No sta- tistically significant interactions for glyphosate exposure were apparent. Overall, the results fail to demonstrate a significant association of glyphosate exposure with the risk of sponta- neous abortion and need to be considered cautiously in light of substantial exposure mis- classification that is likely to have taken place and the lack of adjustment made to the OR to account for the influence of confounding variables. Fetal deaths. Bell et al. (2001a; 2001b) conducted two case-control studies to exam- ine the possible associations between mater- nal pesticide exposure and fetal death due to congenital anomalies or other causes. In the congenital anomalies case-control study (Bell et al. 2001a), 73 cases were identified from the 1984 vital statistics records of 10 California counties. Cases were limited to fetal deaths identified as due to congenital anomalies and occurring after 20 wk of gestation; these included 43 neonatal deaths within the first 24 h after birth. All 611 controls were ran- domly selected from normal live births occur- ring in 1984 and frequency matched with cases by county and maternal age. Sites of pesticide application in the 10 counties for 1983-1984 were determined from the California Pesticide Use Report database and linked to maternal addresses according to their township, range, and section (TRS-a unique location identifier used by the Public Land Survey System and equal to 1 square D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 46 A. L. WILLIAMS ET AL. mile in area). In the narrow definition of exposure, a pregnancy was considered to be exposed if pesticides were applied to land within the same TRS as the maternal address. In the broader definition, a pregnancy was considered to be exposed if pesticides were applied within any of the eight surround- ing TRS or the same TRS as the maternal address. Exposures were assigned for each day of each woman’s pregnancy for 327 differ- ent pesticides, which were categorized into 5 separate classes (phosphates, carbamates, pyrethroids, halogenated hydrocarbons, and endocrine disruptors). Glyphosate (referred to as “glyphosphate” in the article) was classified as a phosphate/thiophosphate/phosphonate pesticide (by far, the broadest pesticide cate- gory evaluated in the study). Exposures were then analyzed according to 3 different peri- ods of gestational age: exposure during gesta- tional weeks 1-20; exposure during gestational weeks 1-13; and exposure during gestational weeks 3-8. A fourth exposure definition was also implemented, which further restricted the definition of nonexposure to be no exposure to any of the 5 classes of pesticides during ges- tational weeks 3-8. Although this method of assigning exposure is not subject to recall bias (a major strength), it still allows for some misclas- sification. In addition, pregnancies that were exposed only once during a particular expo- sure period would be analyzed together with those that were exposed multiple times during the same time period. Using the broad definition of exposure (i.e., pesticide exposure in the maternal TRS or one of the eight adjoining TRS), no sta- tistically increased OR of fetal death were associated with exposure to phosphates dur- ing any of the exposure periods examined. In contrast, a statistically significant risk of fetal death due to congenital anomalies was asso- ciated with exposure to pyrethroids (discussed in erratum Bell et al. 2001c) and halogenated hydrocarbons. These risks rose as the defini- tion of the exposure period was tapered down from anytime during wk 1-20 to only dur- ing wk 3-8 of gestation. Using the narrow definition of exposure (within the same TRS as the maternal address), findings for halo- genated hydrocarbons and pyrethoids lost sta- tistical significance. Exposure to phosphates, on the other hand, showed a statistically significant association with fetal death due to congeni- tal anomalies, which increased as the defini- tion of exposure became more limited. That is, phosphates exposure during wk 1-20 of gestation was associated with an OR of 2 (95% CI = 1-4), while exposure during wk 3-8 only was associated with an OR of 3 (95% CI = 1.4-6.5). It is not clear what conclu- sions, if any, can be drawn from these study results with regards to glyphosate exposure. Results specific to glyphosate are not avail- able, and, as previously mentioned, glyphosate is only one of the many pesticides included in the broad phosphates category. In fact, the vast majority of pesticides included in this cat- egory are organophosphate insecticides, which act by phosphorylating the acetylcholinesterase enzyme of insects and mammals-a mecha- nism of action entirely different from that of glyphosate, which targets an enzyme found only in plants and some microorganisms. This suggests that glyphosate was inappropriately included in the phosphates category, and fur- ther, that a risk for congenital anomalies cannot be inferred for glyphosate exposure. In the case-control study of fetal death due to other causes (Bell et al. 2001b), 314 cases were identified from the 1984 vital statistics records of the same 10 California counties as the previous study. These cases included 86 neonatal deaths within 24 h of birth, but excluded deaths for pregnancies of less than 20 wk, those due to congenital abnor- malities, and other causes not likely to be related to environmental exposures (i.e., mul- tiple births and umbilical cord compression). Controls were identified from normal live births in 1984 and frequency matched by maternal age and county. As in the previous study (Bell et al. 2001a), there were 611 controls; how- ever, it is not clear whether these controls are the same in both studies. Exposures were deter- mined according to the same methods used in the previous study (and thus, are subject to the same misclassification issues), and were D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 47 analyzed according to trimester and month of gestation. Because fetal deaths that occurred at a later gestational age had greater opportunity to be exposed than those that occurred ear- lier on, analyses were adjusted for gestational length. Adjusted hazard ratios and 95% CI were calculated. Overall, no pesticide class was sta- tistically associated with fetal death using either the broad or narrow definitions of exposure as analyzed according to trimester. Birth defects. Garry et al. (2002) con- ducted a cross-sectional study to examine the reproductive health outcomes of pesticide applicators and their families. Approximately 3000 residents of the Red River Valley of Minnesota were licensed to apply pesticides from 1991 to 1996. Half were randomly selected for study, and of these, 1070 volun- teered to participate. Enough detailed informa- tion regarding reproductive health outcomes and pesticide use was obtained for 695 fam- ilies (536 with children) and 1532 live births. Births occurred from 1968 to 1998, with over half of the births occurring before 1978 (almost 20 yr prior to the study’s initiation). Parent- reported health information (obtained through written questionnaire) was confirmed through birth certificates and medical records, when possible. Information regarding pesticide use was obtained initially via telephone survey, and approximately 6 mo later by written question- naire. Seventy children with congenital birth defects were identified. Parents of children with birth defects did not differ from parents of children without birth defects in terms of age at time of child’s birth, smoking status, or con- sumption of alcoholic beverages. Interestingly, the frequency of birth defects identified dur- ing the first year of life as reported in this study was significantly higher than that observed in an earlier cohort study (Garry et al. 1996) that only used medically confirmed cases of birth defects (26.1 vs. 18.9 per 1000 live births). Data suggest that the method for study sub- ject selection was biased in the cross-sectional study in such a way as to over-represent fam- ilies of children with birth defects. Pesticide exposure was assessed according to specific classes of use (herbicide only; herbicide and insecticide; herbicide, insecticide, and fungi- cide; herbicide, insecticide, and fumigant; and use of all four pesticide classes), with use of herbicides-only being the referent group for comparison purposes. Use of all four classes of compounds was associated with a quantitatively increased incidence of hav- ing children with birth defects compared to use of herbicide alone. Although the study authors clearly indicated that developmen- tal neurobehavioral disorders would not be considered in their detailed analyses due to the lack of uniformity in such diagnoses, a detailed analysis was preformed. Garry et al. (2002) reported that 43% (6/14) of children with parent-reported attention deficit disor- der (ADD)/attention deficit hyperactivity disor- der (ADHD) had parents that used phospho- namino herbicides, with an OR of 3.6 (95% CI = 1.35-9.65). Glyphosate and “Roundup” (specific formulation not specified) were the only herbicides in this class mentioned by name. No other data related to glyphosate and “Roundup” were reported. No conclusions can be drawn from this finding. Only 14 cases of ADD/ADHD were reported in a total of 1,532 live births-an incidence substantially lower than the 7% reported for the general population (Bloom and Dey 2006). Further, these are parent-reported cases that have not been confirmed through medical records and, as Bloom and Dey (2006) stated, such diag- noses can be highly unreliable. Without proper ascertainment of these neurobehavioral disor- ders, and in light of the fact that their inci- dence was lower than that normally observed in the general population, this finding pro- vides no reliable insight concerning the poten- tial adverse developmental health effects of glyphosate. Other papers published based on the Red River Valley cohort either did not address potential effects of glyphosate exposure or did not examine reproductive health and/or developmental issues. Neural tube defects. Rull et al. (2006) assessed pesticide exposure using two population-based case-control studies of neural tube defects (NTD) conducted by the California Birth Defects Monitoring D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 48 A. L. WILLIAMS ET AL. Program (Shaw et al. 1995; 1999). Cases were confirmed with diagnoses of anencephaly, spina bifida cystica, craniorhachischisis, and iniencephaly in Californian infants and fetuses delivered between 1987 and 1991. Controls included a random sample of normal singleton births and fetal deaths from the same time frame. Residential, medical, reproductive, occupational, nutritional, and family histories were taken by telephone (on average of 3.8 yr after date of delivery; Shaw et al. 1999) or in-person interviews (on average 5 mo after delivery date; Shaw et al. 1995). Maternal addresses during the calendar month of con- ception and the month following conception were geo-coded to latitude and longitude coordinates, and buffer zones of 500- and 1000-m radii were determined as potential exposure zones. This information was then used to determine possible pesticide expo- sures based on pesticide-use report data from the California Department of Pesticide Regulation. The risk of NTD was estimated for 59 specific pesticides using both single- and multiple-pesticide exposure models, taking into consideration the potentially confound- ing variables of ethnicity, education level, smoking status, and vitamin use. A hierar- chical logistic regression model was also run to minimize the number of false-positive results due to simultaneous analysis for mul- tiple pesticides. Risks for anencephaly and spina bifida, specifically, were also calculated according to pesticide class. As in Bell et al. (2001a; 2001b), glyphosate was inappro- priately categorized as an organophosphate pesticide; however, NTD risks were calculated for each pesticide separately rather than by pesticide class. NTD risks were increased for foreign-born Latina mothers, those who did not complete high school, and moth- ers who did not take vitamins. Glyphosate exposure during the peri-conception period was not associated with an increased risk of NTD (single pesticide model: OR = 1.5, 95% CI = 1-2.4; multiple-pesticide model: OR = 1.5, 95% CI = 0.8-2.9; hierarchical logistic regression model: OR = 1.4, 95% CI = 0.8-2.5). Reproductive Health Fertility. Curtis et al. (1999) conducted a retrospective cohort study to look at possible effects of pesticide exposure on fecundability. The study design was the same as that of Savitz et al. (1997), and based on the same 1898 couples and 5853 pregnancies identified in the OFFHS. Only planned pregnancies, for which women noted discontinuing a method of birth control in order to conceive, were eval- uated. These included 2012 pregnancies, 67% of which were conceived at least 5 yr prior to the study and 36% of which occurred at least 10 yr prior. The substantial amount of time that exists between the time of conception and the start of the study for the majority of these preg- nancies may have introduced some recall bias, although Curtis et al. (1999) indicated that an analysis limited to those pregnancies conceived within 5 years of the study gave results simi- lar to those obtained through an analysis based on all of the pregnancies combined. Detailed reproductive histories were not taken to con- firm time to pregnancy claims or to explore other factors that might influence fecundability, including frequency of intercourse, breastfeed- ing history, reproductive health issues, or men- strual cycle characteristics. Through completed questionnaires provided by farm operators, husbands and wives, monthly histories of pes- ticide usage were constructed for each farm and extrapolated back to years prior to 1991 (the year to which the questionnaire specifically referred). Because it assumes that practices do not change on the farm over time and that all pesticide-related activities occur at the same time each year, this method of determining exposures lends itself to substantial exposure misclassification. For each pregnancy, exposure to certain classes of pesticides, as well as spe- cific pesticide chemicals (including glyphosate), was determined on a yes/no basis for each month that the couple tried to conceive and the 2 mo prior. Exposures were also classi- fied by whether the husband, wife, or both participated in a pesticide activity. For each exposure type, Curtis et al. (1999) calculated a conditional fecundability ratio (CFR)-that is, the ratio of probabilities of conceiving in any D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 49 given month for the exposed group over that of the unexposed group. Results showed that 75% of the planned pregnancies were con- ceived in 3 mo or less. Reduced fecundability was associated with women’s exposure to glyphosate (regardless of men’s activity), but this association was not statistically significant (CFR = 0.61, 95% CI = 0.3-1.26). The find- ing may also be related to factors other than pesticide exposure, such as the influence of heavy manual labor on the menstrual cycle, or reduced frequency of intercourse. In con- trast to the preceding finding, glyphosate use on the farm (with no reported pesticide-related activities by women or men) was associated with a numerical increase, albeit not statisti- cally significant, in fecundability (CFR = 1.26, 95% CI = 0.94-1.69). Further, men’s expo- sure to glyphosate demonstrated a statistically significant rise in fecundability (CFR = 1.3, 95% CI = 1.07-1.56); however, this association was downplayed as a random chance finding. Overall, Curtis et al. (1999) found no significant adverse effects on time to pregnancy associ- ated with glyphosate exposure. Regardless, the study is severely limited by the likelihood of widespread exposure misclassification and the exclusion of information related to common factors that affect fecundability. Larsen et al. (1998) investigated whether time to pregnancy was associated with the use of pesticides during farming practices in Denmark. In total, 904 male farmers ranging in age from 18 to 50 yr and living in the Jutland were identified from the Danish Ministry of Agriculture lists of traditional and organic farm- ers. Information on the cohort was collected via telephone interviews between October 1995 and May 1996. Reproductive histories focused primarily on questions related to the youngest child. Information was also collected on potential confounders such as last con- traceptive method, smoking habits, age, and female parity. Participants were divided into four exposure groups according to their pesti- cide use in the year before the youngest child was born: traditional farmers spraying pesti- cides (n = 450), traditional farmers who did not spray the pesticides themselves (n = 72), organic farmers (n = 94), and those not involved in farming at the time the youngest child was conceived (n = 66); this last group was excluded from the final analyses. Those farmers never married (n = 36), without chil- dren (n = 97), or whose youngest child was conceived due to failure of the birth control method (n = 89) were also excluded from the final analyses. Farmers actively involved in spraying pesticides were asked about the num- ber of hectares treated, type of tractor and spraying equipment, use of protective equip- ment, and type of crops. These farmers were then assigned to three index groups based on this information. Cumulative potential expo- sures were determined based on the total number of years of pesticide use. From a list of possible pesticides, farmers were asked to identify those pesticides used in the year prior to the youngest child’s birth. Glyphosate was considered a potentially spermatotoxic pesti- cide by Larsen et al. (1998), although the basis for this assumption was not provided. Time to pregnancy data for the organic farmers and two groups of traditional farmers were ana- lyzed using a Cox regression model. For reasons that are not clear in from the study report, those showing time to pregnancy greater than 12 mo were excluded from the evaluations. The fecundability ratio (FR) between traditional farmers who actively participated in pesticide application and organic farmers was 1.03 (95% CI = 0.75-1.4). Interestingly, increased cumu- lative exposure to pesticides was associated with a significantly decreased time to preg- nancy, but only in the 11-15 years category (FR = 1.61, 95% CI = 1.07-2.4). In addition, the use of three or more “spermatotoxic” pes- ticides in the year prior to the youngest child’s birth exerted no significant effect on the time to pregnancy (FR = 0.88, 95% CI = 0.66-1.18). No information specific to glyphosate was pre- sented. Because recall was greater than 5 yr for 52% of the traditional farmers who sprayed pesticides, these data likely suffer from some exposure misclassifications. Further, because those without children and those with greater than 12 mo to pregnancy were excluded from the analyses, it is possible that an effect of D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 50 A. L. WILLIAMS ET AL. pesticide exposure may have been underesti- mated. Time to pregnancy was also evaluated in women from five Colombian regions that dif- fered in glyphosate use for the eradication of illicit crops (Sanin et al. 2009; Solomon et al. 2009). The regions for study included Boyacá and Sierra Nevada de Santa Marta, both of which did not conduct aerial spraying of glyphosate; Nariño and Putamayo, both of which conducted aerial spraying of glyphosate for the control of illicit crops; and Valle del Cauca, a more developed region that reported aerial spraying of glyphosate on sugar cane crops. In total, 2751 women aged 25 yr of older who reported their first pregnancy within the last 5 yr and no contraceptive use in the year prior to pregnancy were enrolled in the study between August 2004 and February 2005. Enrollments were conducted through house-to-house visits. Refusals to participate were only reported in Valle del Cauca and amounted to approximately 3% of the women queried in that region. Total participants from each region varied between 531 and 582. Information on the number of months of sexual intercourse prior to pregnancy was collected. Data on potential confounders were also col- lected, including age at first pregnancy, part- ner relationship, work and medical histories, lifestyle practices (smoking and use of drugs, coffee, and alcohol), and the work status and lifestyle practices of the father. Specific infor- mation regarding glyphosate exposures was not gathered; rather, exposures were assumed to relate to the degree of glyphosate aerial appli- cation registered for the region. Fecundability OR were calculated with 95% CI. In a restricted analysis, those women who reported seeking medical treatment for possible fertility issues (n = 159) were excluded. Substantial dif- ferences in the mean time to pregnancy in months were observed among the 5 Colombian regions of interest, with the lowest mean of only 3 mo reported in Boyacá and the great- est mean of 14 mo reported in Valle del Cauca. These differences in time to pregnancy did not correspond, however, with the use of glyphosate for aerial eradication of illicit crops. Thus, these data do not indicate an association between glyphosate use and increases in time to pregnancy. Greenlee et al. (2003) conducted a case- control study to determine risk factors for female infertility in an agricultural region. Cases were women, ages 18-35 yr, who sought treatment at an obstetrics/gynecology (OB/GYN) department in Wisconsin between June 1997 and February 2001 for either pri- mary or secondary infertility. (Infertility was defined as experiencing at least 12 mo of unprotected intercourse without conceiving a pregnancy ending in a live birth.) Cases of infertility due to surgical causes or male infer- tility were excluded from the study. Controls were women, ages 18-35 yr, seeking prena- tal care in their first trimester of pregnancy at the same OB/GYN department during the same period of enrollment and who conceived in less than 12 mo of trying; those reporting ever having trouble conceiving or maintain- ing a pregnancy were excluded. Controls were frequency-matched with cases based on age and clinic service date for a total of 322 cases and 322 controls. Data were gathered by tele- phone interview regarding activities during the 2 yr prior to the reported pregnancy attempt dates. Before the interview, subjects were pro- vided with a list of pesticides and possible exposure scenarios to review. Based on logistic regression models, cases and controls were sim- ilar in age; however, infertile women tended to work outside the home more, to be less educated, to smoke and consume more alco- holic beverages, to have gained weight steadily during adulthood, and to have older male part- ners than controls. Cases also spent significantly more time reviewing the pesticide exposures list than did controls (29.3 versus 18.5 min, respectively), indicating possible recall bias. An analysis of agricultural variables showed that a woman’s exposure to herbicides at any time during the 2-yr period prior to trying to con- ceive was statistically associated with an ele- vated risk of infertility. Interestingly, this variable only reached statistical significance after adjust- ing for confounding variables (crude OR = 2.3, 95% CI = 0.9-6.1; adjusted OR = 26.9, D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 51 95% CI = 1.9-385). The large change in the OR following adjustment, especially in light of the lack of substantial effect on the other agricultural variables evaluated, seems suspect. Furthermore, it was based on small numbers of cases (21/322) and controls (13/322), indicat- ing that the vast majority of both had no expo- sure to herbicides during the period of concern. While glyphosate was reported to be the most commonly used herbicide by both cases and controls (54 and 36 women, respectively), it is unclear why these numbers are greater than numbers of cases and controls reporting use of herbicides during the 2-yr period of interest. Based on questionable data and the fact that glyphosate use, specifically, was not evaluated, no conclusions can be drawn from this study regarding glyphosate’s association with female infertility or lack thereof. Menstrual cycle characteristics. The Agricultural Health Study is a prospective cohort study with more than 50,000 pesticide applicators and more than 32,000 of their spouses from North Carolina and Iowa. Data on the cohort were collected either by writ- ten questionnaire or by telephone between 1993 and 1997 in phase I of the study. Based on these data, Farr et al. (2004) examined the effects of pesticide use on the menstrual cycle characteristics of 3103 women. The cohort was limited to female private pesticide applicators or the spouses of private pesticide applicators who completed the Female and Family Health questionnaire and were identified as pre- menopausal and between the ages of 21 and 40 yr, had a body mass index between 15 and 40, and were not pregnant, breastfeeding, or taking oral contraceptives. Five menstrual cycle characteristics were assessed: short cycles (24 d or less), long cycles (36 d or more), irregular cycles, missed periods (no periods for an interval of more than 6 wk in the last 12 mo), and bleeding/spotting between periods within the last 12 mo. Women were asked about their use of 50 different pesticides, including the average number of days per year they mixed and applied these pesticides. Women were grouped into three exposure categories based upon their responses: 0 d, 1-9 d, and 10 d or more. For analysis purposes, the pesticides of interest were grouped into three categories: endocrine disruptors, those with ovarian effects, and estrous cycle disruptors. Glyphosate was listed as a pesticide with ovarian effects. Women who reported ever mixing and applying pesticides were less likely to have irregular periods (OR = 0.55, 95% CI = 0.41-0.75), but more likely to report missed periods (OR = 1.6, 95% CI = 1.3-2). Controlling for the average number of days worked in the fields quantitatively buffered these findings (OR changed from 0.55 and 1.6 to 0.61 and 1.3, respectively). Limiting the analyses to those women exposed to probable/possible hormonally active or ovo- toxic pesticides only slightly strengthened the associations. Findings for some specific pesticides were reported; however, no signif- icant associations were noted for glyphosate exposure. Other papers published from the Agricultural Health Study did not address potential effects of glyphosate exposure on reproductive health/development issues. Summary-Epidemiology Studies As previously mentioned, the body of epidemi- ological data regarding potentially adverse reproductive health or pregnancy outcomes associated with glyphosate use is scant. Only 11 such studies were identified, and none of these-with the exception of Sanin et al. (2002)-was designed to specifically assess the effects of glyphosate exposure. Furthermore, all of these studies suffer from likely exposure misclassifications, as previously discussed. Of the six studies examining potential devel- opmental effects, only Garry et al. (2002) claimed to observe a possible adverse outcome (increased risk of ADD/ADHD) associated with phosphonamino herbicide exposure of the parents of affected children. However, the diagnostic criteria for these neurobehavioral disorders are notoriously inconsistent, none of the study cases was confirmed through a review of medical records, and the study’s parent-reported incidence of ADD/ADHD is lower than that of the general population. Thus, this study should be considered unin- formative. In Bell et al. (2001a), parental D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 52 A. L. WILLIAMS ET AL. exposure to phosphate pesticides was statisti- cally associated with an increased risk of fetal death due to congenital anomalies; however, glyphosate was inappropriately included in this class of chemicals. Therefore, no conclusions regarding glyphosate exposure can be drawn from this analysis. None of the other develop- mental outcomes studies reported a statistically increased risk of adverse pregnancy outcomes associated with glyphosate exposure (Arbuckle et al. 2001; Bell et al. 2001b; Rull et al. 2006; Savitz et al. 1997). Of five studies that assessed the potential reproductive health effects of glyphosate (Curtis et al. 1999; Farr et al. 2004; Greenlee et al. 2003; Larsen et al. 1998; Sanin et al. 2009), none reported significant adverse outcomes associated with exposure. Greenlee et al. (2003) suggested that female infertility may be influenced by herbicide exposure, but data are questionable and too few exposed women were included in the study to address glyphosate exposure. In conclusion, although the database addressing these health issues is extremely limited, the totality of availabil- ity of epidemiological evidence fails to link glyphosate exposure with significant adverse reproductive health or pregnancy outcomes. Animal Studies Twelve animal studies assessing the poten- tial developmental or reproductive health outcomes associated with glyphosate expo- sure were identified. Some of these stud- ies tested pure glyphosate, while others tested commercial herbicide formulations, the formulation surfactant alone, or the major envi- ronmental breakdown product of glyphosate (aminomethylphosphonic acid: AMPA). These studies also varied greatly in their qual- ity. Some studies were conducted under Good Laboratory Practices (GLP compliant) and/or according to the health effects test- ing guidelines set by the U.S. EPA or Organization for Economic Cooperation and Development (OECD); others used limited numbers of animals per group, inadequate study designs, inappropriate controls, and inadequately-identified test materials. For the purposes of review, these studies are grouped according to the specific test agent used. Developmental Studies Glyphosate. Two developmental toxicity studies were conducted using glyphosate acid (Table 3). Although these studies were con- ducted prior to the establishment of GLP, they both received quality assurance audits by the testing facility and were essentially guideline- compliant. In the first study by the International Research and Development Corporation (IRDC 1980a), pregnant female Charles River COBS CD rats were dosed once daily by gavage with 0, 300, 1000, or 3500 mg/kg/d glyphosate on gestational days (GD) 6-19 (25 animals per group). The high dose in this study is 1750-fold higher than the oral reference dose of 2 mg/kg/d set for glyphosate by the U.S. EPA (1993) and more than 10,000-fold higher than the highest estimated dose of glyphosate measured in biomonitoring studies (discussed in a subsequent section of this arti- cle). Individual doses were determined based on GD 6 body weights. Animals were exam- ined daily for clinical signs of toxicity and body weights were recorded at appropriate intervals. Food consumption was not reported. On GD 20, all surviving animals were sacrificed and developmental effects of treatment assessed. The numbers of corpora lutea, implantations, resorptions, and live and dead fetuses were recorded. Dams were examined for gross mor- phological changes of internal organs. Fetuses were weighed, sexed, and examined for mal- formations and variations. Approximately half of the fetuses were fixed in Bouin’s solution for visceral examination via sectioning; the other half were prepared for evaluation of skeletal morphology. Gross malformations were not reported separately from soft tissue and skeletal malformations. Clinical signs including soft stools, diar- rhea, breathing rattles, and inactivity were noted in rats dosed with 3500 mg/kg/d of glyphosate. Six of 25 rats in this group died before the end of study. Red matter around the nose, mouth, forelimbs, and dorsal head were noted in animals prior to death, and stomach hemorrhages were noted in two of D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 TA BL E 3. D ev el op m en ta lA ni m al St ud ie s C on du ct ed us in g G av ag e Ad m in ist ra tio n of G ly ph os at e, PO EA ,A M PA ,o rR ou nd up N um be r( % )f et us es w ith m al fo rm at io ns A ni m al m od el (n um be rp er gr ou p) A ge nt (e xp os ur e du ra tio n) D os e (m g/ kg / d) N um be r of gr av id fe m al es N um be r (m ea n) of co rp or a lu te a N um be r (m ea n) of im pl an ta - tio ns N um be r (% ) re so rp - tio ns N um be r (% )l iv e fe tu se s M ea n fe ta l w ei gh t( g) N um be ro f fe tu se s (li tte rs )w ith m al fo rm a- tio ns G ro ss Vi sc er al Sk el et al M at er na le ffe ct s Re fe re nc e C ha rle s Ri ve r C O BS C D ra ts (2 5) G ly ph os at e (G D 6- 19 ) 0 22 34 9 (1 5. 9) 33 0 (1 5. 0) 14 31 6 3. 5 3 (3 ) nr 2 (2 ) 1 (1 ) IR D C 19 80 a 30 0 20 30 3 (1 5. 2) 24 1 (1 2. 1) ∗∗ 4 23 7∗ 3. 7 0 (0 ) nr 0 (0 ) 0 (0 ) 10 00 21 33 8 (1 6. 1) 31 0 (1 4. 8) 10 30 0 3. 6 0 (0 ) nr 0 (0 ) 0 (0 ) 35 00 23 23 7 (1 4. 8) 21 7 (1 2. 8) ∗ 21 ∗ 19 6∗ 3. 2∗ ∗ 10 (3 ) nr 7 (2 ) 9 (2 ) 6/ 25 de at hs ;v ar io us sig ns of cl in ic al to xi ci ty ;d ec re as ed w ei gh tg ai n du e to w ei gh tl os s on G D 6- 9 D ut ch be lte d ra bb its (1 6) G ly ph os at e (G D 6- 27 ) 0 14 10 8 (9 .0 ) 71 (5 .9 ) 8 63 33 .4 0 (0 ) nr 0 (0 ) 0 (0 ) 2/ 16 da m s ab or te d on G D 22 ;s of t st oo ls/ di ar rh ea IR D C 19 80 b 75 16 15 2 (1 0. 1) 12 0 (8 .0 ) 6 11 4∗ 30 .9 3 (3 ) nr 0 (0 ) 3 (3 ) 1/ 16 de at hs on G D 26 ; so ft st oo ls/ di ar rh ea 17 5 14 11 6 (1 0. 5) 67 (6 .1 ) 2 65 29 .9 2 (2 ) nr 0 (0 ) 2 (2 ) 1/ 16 da m s ab or te d on G D 27 ;2 / 16 de at hs on G D 22 an d 25 ; in cr ea se d so ft st oo ls/ di ar rh ea 35 0 16 51 (8 .5 ) 43 (7 .2 ) 5 38 29 .3 2 (1 ) nr 2 (1 ) 0 (0 ) 1/ 17 da m s ab or te d on G D 23 ;1 0/ 17 de at hs by G D 21 ;i nc re as ed so ft st oo ls/ di ar rh ea , na sa ld isc ha rg e SD C rl: C D BR ra ts (2 5) M O N 08 18 (P O EA ) (G D 6- 15 ) 0 24 37 6 34 1 16 (6 .0 ) 32 5 (9 4. 0) 3. 5 2 (2 ) 1 (0 .3 ) 1 (0 .3 ) 0 (0 ) H ol so n 19 90 15 23 37 7 31 6 21 (6 .7 ) 29 5 (9 3. 3) 3. 6 2 (2 ) 2 (0 .7 ) 0 (0 ) 0 (0 ) 10 0 22 35 6 32 0 16 (5 .3 ) 30 4 (9 4. 7) 3. 6 0 (0 ) 0 (0 ) 0 (0 ) 0 (0 ) In fre qu en tc lin ic al sig ns ; de cr ea se d fo od co ns um pt io n 30 0 15 34 4 21 6 11 (5 .1 ) 20 5 (9 4. 9) 3. 4 4 (4 ) 1 (0 .5 ) 3 (1 .5 ) 1 (0 .5 ) 6/ 25 de at hs ;c lin ic al sig ns ;d ec re as ed w ei gh tg ai n; de cr ea se d fo od co ns um pt io n (C on tin ue d) 53 D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 TA BL E 3. (C on tin ue d) N um be r( % )f et us es w ith m al fo rm at io ns A ni m al m od el (n um be rp er gr ou p) A ge nt (e xp os ur e du ra tio n) D os e (m g/ kg / d) N um be r of gr av id fe m al es N um be r (m ea n) of co rp or a lu te a N um be r (m ea n) of im pl an ta - tio ns N um be r (% ) re so rp - tio ns N um be r (% )l iv e fe tu se s M ea n fe ta l w ei gh t( g) N um be ro f fe tu se s (li tte rs )w ith m al fo rm a- tio ns G ro ss Vi sc er al Sk el et al M at er na le ffe ct s Re fe re nc e SD C rl: C D BR ra ts (2 5) A M PA (G D 6- 15 ) 0 24 38 7 36 1 16 (4 .4 ) 34 5 (9 5. 6) 3. 5 2 (2 ) 1 (0 .3 ) 1 (0 .3 ) 0 (0 ) H ol so n 19 91 15 0 24 40 7 36 6 19 (5 .3 ) 34 7 (9 4. 7) 3. 5 4 (4 ) 1 (0 .3 ) 2 (0 .6 ) 3 (0 .9 ) 40 0 24 39 4 36 0 14 (3 .9 ) 34 6 (9 6. 1) 3. 4 0 (0 ) 0 (0 ) 0 (0 ) 0 (0 ) In cr ea se d m uc oi d fe ce s/ so ft st oo ls 10 00 24 39 7 36 3 14 (3 .8 ) 34 9 (9 6. 2) 3. 3∗ 2 (2 ) 2 (0 .6 ) 0 (0 ) 0 (0 ) In cr ea se d m uc oi d fe ce s/ so ft st oo ls W ist ar ra ts (1 4- 16 ) Ro un du p (G D 6- 15 ) 0 15 17 1ˆ (1 1. 4) 15 7ˆ (1 0. 5) 3ˆ (2 .4 ) 15 4ˆ 5. 1 nr 0 (0 ) nr 24 ˆ@ (1 5. 4) D al le gr av e et al . 20 03 50 0 15 17 4ˆ (1 1. 6) 14 7ˆ (9 .8 ) # ˆ( 3. 3) 14 8ˆ 5 nr 3ˆ (2 .0 ) nr 49 ˆ( 33 .1 ) 75 0 16 16 2ˆ (1 0. 1) 18 6ˆ (1 1. 6) 24 ˆ( 2. 6) 16 2ˆ 5. 1 nr 1ˆ (0 .6 ) nr 68 ˆ( 42 .0 ) D ec re as ed fo od co ns um pt io n 10 00 14 90 ˆ( 12 .9 ) 80 ˆ( 11 .4 ) 5ˆ (3 .8 ) 75 ˆ 5. 1 nr 0 (0 ) nr 43 ˆ( 57 .3 ) 7/ 14 da m s di ed G D 7- 14 ;d ec re as ed w ei gh tg ai n; de cr ea se d fo od co ns um pt io n N ot e. ∗ S ta tis tic al ly di ffe re nt fro m co nt ro l, p < .0 5; ∗∗ st at ist ic al ly di ffe re nt fro m co nt ro l, p < .0 1. ˆ, Au th or s di d no td ist in gu ish be tw ee n liv e an d de ad fe tu se s; m ea ns re po rte d in pa pe r fo rh ig h- do se gr ou p ba se d on on ly se ve n su rv iv in g da m s; nu m be ro fc or po ra lu te a an d im pl an ta tio ns ha ve be en ca lc ul at ed ba se d on th e m ea ns an d nu m be ro fd am sr ep or te d in th e pa pe r; nu m be r of fe tu se s w ith gr os s an d sk el et al m al fo rm at io ns ha ve be en ca lc ul at ed ba se d on th e pe rc en ta ge s an d nu m be r of fe tu se s re po rte d in th e pa pe r; nu m be r of re so rp tio ns ha ve be en ca lc ul at ed as th e di ffe re nc e be tw ee n th e nu m be ro ff et us es re po rte d an d th e ca lc ul at ed nu m be ro fi m pl an ta tio ns .# ,N um be ro fr es or pt io ns ca nn ot be ca lc ul at ed fro m th e pr ov id ed da ta . @ ,N um be ro fa ni m al s w ith sk el et al m al fo rm at io ns ca nn ot be ca lc ul at ed fro m th e pe rc en ta ge pr ov id ed by th e au th or s. 54 D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 55 six rats at necropsy. Maternal body weights in the glyphosate-treated groups were not significantly different from control, although the body weight gain in the high-dose group was numerically decreased compared to con- trol. Statistically significant decreases in the mean numbers of implantations and viable fetuses per dam in the 300-mg/kg/d treat- ment group were noted; however, because no marked effects of treatment were noted in the next higher dose group of 1000 mg/kg/d (and implantations occurred before treatment), these changes were attributed to random chance. In the 3500-mg/kg/d glyphosate treat- ment group, a statistically significant increase in the number of resorptions, significant decreases in the mean numbers of implantations and viable fetuses per dam, and diminished mean fetal body weights were observed compared to controls (possibly related to reduced maternal weight gains at the high dose). No significant change in the mean number of corpora lutea per dam was noted in this treatment group. Because ovulation and implantation occurred prior to dosing, the differences in corpora lutea and implantations were not considered to be related to glyphosate treatment. No apparent malformations were observed in fetuses from the 300- and 1000-mg/kg/d treatment groups. Two control fetuses had soft tissue malformations and one had a skeletal malformation; these were found in a total of three different litters. In the 3500-mg/kg/d glyphosate group, 10 fetuses had malforma- tions; these included 7 soft-tissue malforma- tions and 9 skeletal malformations. Upon closer examination, however, it was noted that these malformations were primarily minor, and included dwarfish and bent tails. Further, the malformations occurred in only three litters, with the same anomaly often presenting itself multiple times in a single litter (data not shown). Based on the types of malformations and their occurrence in limited litters, IRDC (1980a) con- cluded that these findings likely represented a litter effect of genetic origin. An increased incidence of unossified sternebrae (a variation, possibly related to the reduced fetal weights and a developmental delay) was also reported in fetuses at 3500 mg/kg/d. Overall, this study indicates a no-observable-adverse-effect level (NOAEL) of 1000 mg/kg/d glyphosate for both maternal and developmental toxicity. In IRDC (1980b), female Dutch belted rab- bits were inseminated on GD 0 using semen from 4 proven male rabbits. Impregnated does were administered 0, 75, 175, or 350 mg/kg/d glyphosate by gavage on GD 6-27 (16 rab- bits per group). Animals were examined daily for mortality and clinical signs of toxicity. Body weights were recorded at appropriate intervals. Food consumption rates were not recorded. Dams that did not survive to the end of study were necropsied to determine cause of death. All surviving animals were sacrificed on GD 28. The numbers of corpora lutea, implanta- tions, resorptions, and live and dead fetuses were recorded. All fetuses were weighed, sexed internally, examined for external and visceral malformations (via dissection), and prepared for skeletal examination using alizarin red. Gross malformations were not reported sepa- rately from soft tissue and skeletal malforma- tions in this study. Soft stools and diarrhea were noted in all treatment groups, but showed a dose-dependent rise in incidence in dams treated with 175 and 350 mg/kg/d glyphosate compared to controls. Animals at 350 mg/kg/d also demonstrated an increase in nasal dis- charge. Maternal body weight changes were highly variable across groups throughout the study and no significant differences compared to controls were noted. Abortions occurred in 2 rabbits from the control group, and in 1 rabbit in each of the 175- and 350-mg/kg/d treatment groups. The reason for the rela- tively high abortion rate among control animals (2/16) is not known and was not discussed in the original study report. One, 2, and 10 rab- bits died before the end of study in the 75-, 175-, and 350-mg/kg/d glyphosate treatment groups, respectively. The causes of maternal death were determined for 5 of 13 animals, but were not consistent across the group. A statisti- cally significant elevation in number of viable fetuses per dam treated with 75 mg/kg/d was noted, but this result was considered to be a random occurrence because it was not D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 56 A. L. WILLIAMS ET AL. observed in the 2 higher treatment groups. Compared to controls, glyphosate treatment exerted no marked effect on the numbers of corpora lutea, implantations, resorptions, live and dead fetuses, fetal sex ratios, or fetal weights. Glyphosate treatment also had no significant effect on the incidence of fetal malformations and variations. Based on mor- tality and clinical signs at 350 mg/kg/d, the NOAEL for maternal toxicity is considered to be 175 mg/kg/d. Although no apparent devel- opmental toxicity was observed at any dose, 175 mg/kg/d is considered the NOAEL for developmental toxicity as well because too few fetuses were available at the high dose of 350 mg/kg/d for adequate toxicological assessment. POEA-Surfactants used in a number of commercial herbicide formulations. One GLP- and guideline-compliant developmental toxic- ity study was conducted with POEA (Holson 1990; Table 3). Pregnant female Sprague- Dawley Clr:CD BR rats were dosed once daily by gavage with 0, 15, 100, or 300 mg/kg/d POEA on GD 6-15 (25 animals per group). Animals were examined daily for clinical signs of toxicity, and body weights and food con- sumption were recorded at appropriate inter- vals. On GD 20, animals were sacrificed and any developmental effects of treatment on the resulting offspring were determined by vis- ceral dissection and alizarin staining for osseous effects. In this study, 6/25 animals treated with 300 mg/kg/d POEA died before the end of study and clinical signs of maternal toxicity were observed in the remaining animals at this dose level. In addition, significant decreases in weight gain and food consumption during the treatment period were observed. Infrequent signs of clinical toxicity and significantly lower food consumption on GD 6-9 were recorded for animals in the 100-mg/kg/d treatment group. However, despite these obvious signs of maternal toxicity, no significant compound- related developmental effects on the offspring were observed. The rate of malformations in the treated groups was within the historical control range for the lab. From these results, Holson (1990) concluded that 15 mg/kg/d POEA was the NOAEL for maternal toxicity and 300 mg/kg/d POEA was the NOAEL for developmental toxicity. Aminomethylphosphonic acid (AMPA)- Major environmental breakdown product of glyphosate. One GLP- and guideline- compliant developmental toxicity study tested AMPA (Holson 1991; Table 3). Pregnant female Sprague-Dawley Crl:CD BR rats were dosed once daily by gavage with 0, 150, 400, and 1,000 mg/kg/d AMPA on GD 6-15 (25 ani- mals per group). Animals were examined once daily for clinical signs of toxicity, and body weights and food consumption were recorded at appropriate intervals. On GD 20, animals were sacrificed and the effects of treatment on development of the resulting offspring were recorded. Increased incidences of mucoid feces, soft stools, and hair loss were observed in animals treated with 400 and 1000 mg/kg/d AMPA and these effects appeared to rise in a treatment-related manner. Offspring were examined by visceral dissection and alizarin staining for osseous effects. No marked developmental effects of treatment on the offspring were observed except for a statis- tically significant decrease in fetal weights at 1000 mg/kg/d. This finding was within the range of historical control data from the laboratory at which the study was conducted and was influenced by the results of two litters. All malformations in the treated groups were considered to be of spontaneous origin and unrelated to treatment. Based on these results, Holson (1991) concluded that 400 mg/kg/d AMPA was the NOAEL for both maternal and developmental toxicity. Commercial herbicide formulation. One non-guideline developmental study was con- ducted using an unspecified commercial for- mulation of “Roundup,” which was reported to consist of 360 g/L glyphosate and 18% (w/v) POEA (Dallegrave et al. 2003; Table 3). Sixty pregnant Wistar rats were divided into 4 treatment groups and dosed once daily by gavage with 0, 500, 750, or 1000 mg/kg/d “glyphosate-Roundup” on GD 6-15. The treatment description provided was ambigu- ous as to whether the dosages stated were D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 57 for “Roundup” or the active ingredient, glyphosate. Body weights were recorded daily, and food and water consumption was recorded at appropriate intervals. On GD 21, animals were sacrificed and the effects of treatment on fetal development were recorded. It is difficult to draw any conclusions regarding the devel- opmental effects of Roundup from this study. Not only are the treatment doses unclear, but the number of animals per group is significantly lower than the recommended 25 rats/group, and the highest dose group was further reduced by half due to animal deaths. Furthermore, few data are actually presented in the article. Rather, maternal body weight is reported as relative weight, with weights on GD 0 consid- ered 100%. Likewise, food and water intake are presented graphically as relative intakes, although it is not clear from the article to what values these reported intakes are nor- malized. Similarly, fetal findings are presented as percentages or unsubstantiated mean val- ues throughout the article, which complicates interpretation. When one converts these mean reproductive indices to actual totals, one dis- covers that data presented are flawed (see Table 3). For example, an animal should have at least as many implantation sites as fetuses and the number of corpora lutea should be greater than (or at least never less than) the number of implantation sites; however, in the 500-mg/kg/d treatment group, more fetuses were reported than implantation sites. This brings into question the reported resorption rate for this treatment group. Further, in the 750-mg/kg/d treatment group, more implan- tation sites than corpora lutea were reported. Dallegrave et al. (2003) reported a dose- related increased incidence of skeletal alter- ations of 15.4, 33.1, 42, and 57.3% in con- trol, 500-, 750-, and 1000-mg/kg/d treatment groups, respectively. The most frequently observed alterations included incomplete ossi- fication of the skull and enlarged fontanels. Interestingly, examination of the list of skeletal alterations observed in this study showed an extremely high prevalence of incomplete ossi- fication of various bone structures. These find- ings are signs of a developmental delay that correct themselves within a brief period. It is important to note that the methods described to fix and stain the fetal skeletons for evalu- ation are unusual and it is possible that the method led to artifacts that were falsely cat- egorized as alterations. The standard method calls for fetuses to be fixed in alcohol and macerated with potassium hydroxide before staining (Dawson 1926; Wilson 1965). In this study, however, fetuses were fixed in forma- lin, and later immersed in a solution of trypsin before staining with alizarin red. Since trypsin is a proteolytic enzyme, it could have digested some of the peptide bonds of the bone matrix, resulting in areas that would appear as if they were incompletely ossified. Based on the use of these questionable methods, and the obvi- ously flawed reporting of data, it is not possible to draw any conclusions regarding the devel- opmental effects of “Roundup” treatment from this article. Furthermore, because a commer- cial formulation was used, it is not possible to attribute any observed effects to glyphosate specifically. Reproductive Studies Glyphosate. Three multigenerational reproductive studies tested glyphosate (Table 4), two in Sprague-Dawley rats (Schroeder 1981; 1982; Reyna 1990) and one in Wistar-derived rats (Moxon 2000). Although conducted prior to the establishment of GLP, the Schroeder (1981; 1982) study received a quality assurance audit by the testing facility and generally adhered to cur- rent testing guidelines. In this study, diets initially contained 0, 30, 100, and 300 ppm glyphosate and were adjusted weekly to maintain approximate glyphosate dose levels of 0, 3, 10, and 30 mg/kg/d throughout. The test substance was administered starting 63 d prior to mating of the first generation and continuously thereafter. Three generations of parents (F0, F1, and F2) were raised to maturity and mated. Each generation was mated twice (24 females and 12 males per treatment group per mating) with a rest period of at least 14 d between weaning and mating, resulting in 2 sets of litters per generation. The first litter of each generation was sacrificed at weaning and D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 TA BL E 4. Re pr od uc tiv e A ni m al St ud ie s C on du ct ed U sin g G ly ph os at e or PO EA A dm in ist ra tio n in th e D ie t M ea n of fs pr in g w ei gh t( g) N um be rs ur vi vi ng of fs pr in g (% ) A ni m al m od el (n um be r/ gr ou p) Ag en t( ex po su re du ra tio n) M at in g (re su lti ng pr og en y) D os e N um be r (% ) gr av id fe m al es M ea n ge st at io na l le ng th (d ay s) N um be r de ad fe tu se s (m ea n) N um be r liv e fe tu se s (m ea n) Pe rc en t PN D 0 m al e pu ps M ea n nu m be r pu ps w ea ne d/ lit te r PN D 0 PN D 4 PN D 21 PN D 0- 4 PN D 4- 21 Re fe re nc e SD C rl: C D ra ts ˆ (2 4 F; 12 M ) G ly ph os at e (6 3 d pr io rt o m at in g th ro ug h w ea ni ng of lit te rF 3b ) F0 (F 1a ) 0 m g/ kg / d 19 (9 5) 22 .1 2 (0 .1 ) 21 8 (1 1. 5) 50 .3 10 .7 6 9. 9 41 .1 21 0 (9 6. 3) 19 2 (9 8. 5) Sc ho ed er 19 81 ; 19 82 3 m g/ kg / d 21 (9 5. 5) 21 .8 3 (0 .1 ) 26 8 (1 2. 8) 48 .5 12 .4 5. 8 9. 3 37 .7 25 1 (9 3. 7) 24 7 (9 8. 4) 10 m g/ kg / d 16 (8 4. 2) 21 .8 4 (0 .3 ) 19 6 (1 2. 3) 51 .5 11 .9 5. 9 9. 4 39 .7 19 4 (9 9. 0) 19 2 (9 9. 0) 30 m g/ kg / d 19 (9 0. 5) 21 .8 1 (0 .1 ) 22 1 (1 1. 6) 50 .7 11 .3 6 9. 6 39 .2 21 7 (9 8. 2) 21 5 (9 9. 1) F0 (F 1b ) 0 m g/ kg / d 19 (9 5) 22 3 (0 .2 ) 22 3 (1 1. 7) 47 .1 11 .3 6. 1 9. 9 40 .9 21 8 (9 7. 8) 21 5 (9 8. 6) 3 m g/ kg / d 19 (8 2. 6) 21 .8 11 (0 .6 ) 23 2 (1 2. 2) 50 11 .4 6. 1 9. 7 43 .2 22 3 (9 6. 1) 20 6 (9 2. 4) ∗∗ 10 m g/ kg / d 12 (7 0. 6) 22 4 (0 .3 ) 15 3 (1 2. 8) 47 .1 10 .9 5. 8 9 37 .9 14 5 (9 4. 8) 12 0 (8 2. 8) ∗∗ 30 m g/ kg / d 18 (8 1. 8) 21 .9 5 (0 .3 ) 22 6 (1 2. 6) 50 11 .4 6. 2 9. 9 36 .9 22 5 (9 9. 6) 19 4 (9 0. 7) ∗∗ F1 (F 2a ) 0 m g/ kg / d 18 (1 00 ) 21 .9 3 (0 .2 ) 21 6 (1 2. 0) 43 11 .7 5. 8 9. 4 41 20 1 (9 3. 1) 19 9 (9 9. 0) 3 m g/ kg / d 20 (8 7) 21 .8 0 (0 .0 ) 23 6 (1 1. 8) 55 .9 11 .6 6 9. 7 43 .4 23 1 (9 7. 9) ∗ 23 1 (1 00 ) 10 m g/ kg / d 17 (9 4. 4) 21 .9 0 (0 .0 ) 21 6 (1 2. 7) 48 .6 12 .4 6 9. 1 39 .7 21 4 (9 9. 1) ∗∗ 21 1 (9 8. 6) 30 m g/ kg / d 18 (9 4. 7) 22 7 (0 .4 ) 20 7 (1 1. 5) 49 .1 11 .1 6. 2 9. 4 40 .3 20 6 (9 9. 5) ∗∗ 20 0 (9 7. 1) F1 (F 2b ) 0 m g/ kg / d 15 (8 8. 2) 21 .9 6 (0 .4 ) 18 6 (1 2. 4) 48 .4 11 .9 5. 9 9. 4 41 .1 17 8 (9 5. 7) 17 8 (1 00 ) 3 m g/ kg / d 15 (7 8. 9) 21 .9 6 (0 .4 ) 18 7 (1 2. 5) 48 .7 12 .7 5. 7 9. 2 41 .1 16 6 (8 8. 8) ∗ 16 5 (9 9. 4) 10 m g/ kg / d 14 (8 2. 4) 22 .1 3 (0 .2 ) 18 4 (1 3. 1) 49 .5 12 .7 5. 8 9. 5 41 .3 18 1 (9 8. 4) 17 8 (9 8. 3) 30 m g/ kg / d 14 (7 3. 7) 22 .1 4 (0 .3 ) 14 7 (1 1. 3) 51 11 .1 6. 4 10 .3 41 .3 14 4 (9 8. 0) 14 4 (1 00 ) F2 (F 3a ) 0 m g/ kg / d 23 (9 5. 8) 21 .9 5 (0 .2 ) 26 8 (1 1. 7) 51 .4 11 6 9. 5 37 .1 26 6 (9 9. 3) 25 4 (9 5. 5) 3 m g/ kg / d 20 (1 00 ) 22 18 (0 .9 ) 22 2 (1 1. 1) 49 .8 10 .6 6. 1 9. 4 36 .8 21 9 (9 8. 6) 19 0 (8 6. 8) ∗ 10 m g/ kg / d 16 (8 0) 21 .9 2 (0 .1 ) 20 2 (1 2. 6) 48 .7 11 .8 6 9. 5 37 .3 20 2 (1 00 ) 18 8 (9 3. 1) 30 m g/ kg / d 17 (9 4. 4) 21 .9 5 (0 .3 ) 20 0 (1 1. 8) 54 .4 11 6. 3 9. 8 36 .7 19 8 (9 9. 0) 18 7 (9 4. 4) F2 (F 3b ) 0 m g/ kg / d 22 (9 5. 7) 21 .9 5 (0 .2 ) 24 7 (1 1. 2) 45 .7 11 .3 5. 9 8. 8 38 .1 24 1 (9 7. 6) 23 7 (9 8. 3) 3 m g/ kg / d 16 (8 4. 2) 21 .9 6 (0 .4 ) 19 7 (1 2. 3) 46 .2 12 .7 5. 8 9 39 .6 19 2 (9 7. 5) 19 1 (9 9. 5) 10 m g/ kg / d 16 (8 0) 22 .1 2 (0 .1 ) 20 8 (1 3. 0) 45 .7 12 .4 6. 1 9. 4 39 .9 20 2 (9 7. 1) 19 8 (9 8. 0) 30 m g/ kg / d 19 (9 0. 5) 21 .9 10 (0 .5 ) 18 7 (9 .8 ) 45 .5 9. 9 6. 3 9. 1 38 .5 18 3 (9 7. 9) 17 8 (9 7. 3) Sp ra gu e D aw le y Ra ts # ◦ (3 0 F; 30 M ) G ly ph os at e (1 1 w k pr em at in g th ro ug h w ea ni ng of lit te rF 2b ) F0 (F 1) 0 pp m 24 (8 2. 8) 22 .2 9 1 (0 .0 ) 31 8 (1 3. 3) 49 .7 na 6. 1 9. 4# 52 30 7 (9 5. 4) (9 9. 0) Re yn a 19 90 20 00 pp m 29 (9 6. 7) 22 .2 2 5 (0 .2 ) 36 2 (1 2. 5) 53 .5 na 6. 1 9. 3# 50 .4 35 2 (9 7. 5) (9 9. 6) 10 ,0 00 pp m 28 (9 6. 6) 22 .5 3 (0 .1 ) 35 5 (1 2. 7) 51 na 6. 3 9. 5# 49 .8 35 2 (9 9. 4) ∗ (9 9. 6) 30 ,0 00 pp m 28 (9 3. 3) 22 .2 6 4 (0 .1 ) 32 3 (1 1. 5) 50 .5 na 6. 3 9. 6# 46 .9 ∗∗ 32 3 (1 00 )∗ ∗ (9 8. 4) F1 (F 2a ) 0 pp m 28 (9 3. 3) 22 .4 4 2 (0 .1 ) 33 7 (1 2. 0) 45 .1 na 6. 1 9. 7 53 .3 32 5 (9 6. 7) (1 00 .0 ) 20 00 pp m 24 (8 5. 7) 22 .5 8 6 (0 .3 ) 30 8 (1 2. 3) 52 .3 na 6 9. 4 51 .2 28 4 (9 0. 9) (9 6. 7) 10 ,0 00 pp m 24 (8 2. 8) 22 .6 1 4 (0 .2 ) 27 5 (1 1. 5) 48 na 6. 1 9. 5 50 .1 ∗ 26 3 (9 6. 2) (9 9. 5) 30 ,0 00 pp m 26 (8 9. 7) 22 .5 8 0 (0 .0 ) 28 1 (1 0. 8) 49 .1 na 6. 3 9. 8 45 .9 ∗∗ 27 9 (9 9. 4) (1 00 .0 ) F1 (F 2b ) 0 pp m 16 (6 4. 0) 22 .4 4 2 (0 .1 ) 19 1 (1 1. 9) 49 .7 na 6. 2 9. 8 50 .8 18 8 (9 8. 8) (1 00 .0 ) 20 00 pp m 21 (8 4. 0) 22 .5 5 4 (0 .2 ) 21 7 (1 0. 9) 55 .8 na 6 9. 7 51 .8 19 3 (9 1. 1) (9 9. 3) 10 ,0 00 pp m 19 (7 9. 2) 22 .3 7 3 (0 .2 ) 25 0 (1 3. 2) 54 .4 na 6. 2 9. 3 51 23 9 (9 6. 5) (9 4. 1) 30 ,0 00 pp m 25 (9 6. 2) 22 .5 4 (0 .2 ) 26 8 (1 0. 7) 48 .1 na 6. 3 9. 7 43 .4 ∗∗ 25 8 (9 6. 9) (1 00 .0 ) 58 D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 A lp k: A P f SD (W ist ar -d er iv ed ) ra ts † (2 6 F; 26 M ) G ly ph os at e (1 0 w k pr em at in g th ro ug h w ea ni ng fl itt er F2 a) F0 (F 1a ) 0 pp m 23 (8 8. 5) 22 .3 18 27 7 (1 2. 1) 53 .8 10 .6 5. 6 9. 1 42 .7 22 4 (8 9. 3) 22 3 M ox on 20 00 10 00 pp m 22 (8 4. 6) 22 .3 8 25 3 (1 1. 5) 49 .8 10 .0 6. 0 8. 8 41 .4 22 5 (8 8. 4) 21 9 30 00 pp m 22 (8 4. 6) 22 .2 3 26 1 (1 1. 8) ∗∗ 53 .3 10 .9 5. 8 8. 7 40 .4 22 8 (9 2. 7) 22 8 10 ,0 00 pp m 24 (9 2. 3) 22 .2 10 29 1 (1 2. 0) 50 .5 11 .2 5. 9 8. 3 38 .5 ∗ 26 2 (9 4. 8) ∗∗ 25 7 F1 (F 2a ) 0 pp m 22 (8 4. 6) 22 .3 13 22 6 (1 0. 2) 50 .9 9. 8 6. 2 9. 5 43 .6 20 7 (9 6. 2) 20 6 10 00 pp m 23 (8 8. 5) 22 .2 9 25 4 (1 0. 9) 52 .2 9. 9 6. 1 9. 8 45 .5 21 8 (9 0. 2) ∗ 21 8 30 00 pp m 21 (8 0. 8) 22 .1 3 25 1 (1 3. 0) ∗∗ 53 .8 11 .9 ∗ 6. 1 9. 2 42 .5 22 7 (9 3. 0) ∗ 22 6 10 ,0 00 pp m 25 (9 6. 2) 22 .0 ∗∗ 8 29 6 (1 1. 9) 54 .4 11 .1 6. 0 9. 3 43 .9 27 8 (9 3. 5) 27 8 Sp ra gu e- D aw le y ra ts ¶( 20 F; 20 M fo rF 0; 31 / 32 fo r co nt ro ls, 24 / 26 fo r 10 00 pp m ) M O N 08 18 (P O EA )(1 0 w k pr em at in g th ro ug h PN D 4 of lit te rF 2a ) F0 (F 1a ) 0 pp m 16 (8 0. 0) 22 .3 nr N R (1 3. 9) nr nr 7. 3 9. 9 54 .2 N R (9 7. 9) N R (9 8. 4) Kn ap p 20 07 10 0 pp m 15 (7 8. 9) 22 .3 nr N R (1 3. 1) nr nr 7. 4 10 .5 55 .4 N R (9 3. 8) N R (9 8. 3) 30 0 pp m 14 (7 3. 7) 22 .2 nr N R (1 3. 9) nr nr 7. 2 10 .0 54 .6 N R (9 7. 7) N R (1 00 ) 10 00 pp m 15 (8 8. 2) 22 .2 nr N R (1 1. 8) nr nr 7. 3 9. 8 54 .8 N R (8 2. 2) N R (9 8. 1) F1 (F 2a ) 0 pp m 28 (9 3. 3) 21 .9 nr N R (1 4. 4) nr nr 7. 0 9. 8 N / A nr N / A 10 0 pp m N / A N / A N / A N / A N / A N / A N / A N / A N / A N / A N / A 30 0 pp m N / A N / A N / A N / A N / A N / A N / A N / A N / A N / A N / A 10 00 pp m 21 (9 5. 5) 21 .9 nr N R (1 4. 8) nr nr 7. 2 9. 8 N / A nr N / A Sp ra gu e D aw le y ra ts + (1 2 F; 12 M ) M O N 81 09 (P O EA ) 0 pp m 11 (9 1. 7) 21 .5 nr N R (1 4. 4) nr nr 6. 6 8. 5 N / A N R (9 4. 2) N / A Kn ap p 20 08 30 pp m 11 (9 1. 7) 21 .9 nr N R (1 3. 9) nr nr 6. 9 9. 1 N / A N R (9 7. 0) N / A 10 0 pp m 10 (8 3. 3) 21 .9 nr N R (1 3. 9) nr nr 6. 7 8. 7 N / A N R (8 8. 7) N / A 30 0 pp m 12 (1 00 ) 21 .8 nr N R (1 4. 8) nr nr 6. 9 9. 3 N / A N R (9 5. 9) N / A 20 00 pp m 12 (1 00 ) 21 .9 nr N R (8 .0 )∗ ∗ nr nr 6. 2 8. 6 N / A N R (3 6. 3) ∗∗ N / A M O N 08 18 (P O EA ) 10 00 pp m 12 (1 00 ) 21 .7 nr N R (1 3. 6) nr nr 6. 7 8. 7 N / A N R (9 8. 7) N / A N ot e. ∗ S ta tis tic al ly di ffe re nt fro m co nt ro l, p < .0 5; ∗∗ st at ist ic al ly di ffe re nt fro m co nt ro l, p < .0 1. ˆ, PN D 4- 21 su rv iv al da ta fo r F1 a co nt ro lg ro up ex cl ud es lit te r of sin gl e fe m al e w hi ch ga in ed a pu p du rin g th is pe rio d; PN D 4- 21 su rv iv al da ta fo r F1 b hi gh do se gr ou p ex cl ud e lit te r of sin gl e fe m al e th at es ca pe d fro m ca ge on PN D 20 .# ,A ve ra ge gl yp ho sa te in ta ke in th e 20 00 -, 10 ,0 00 -, an d 30 ,0 00 -p pm gr ou ps du rin g th e pr em at in g pe rio d re po rte d to be 13 2, 66 6, an d 19 83 m g/ kg / d, re sp ec tiv el y, fo rF 0 m al es ;1 60 ,7 77 ,a nd 23 22 m g/ kg / d, re sp ec tiv el y, fo r F 0 fe m al es ; 14 0, 71 1, an d 22 30 m g/ kg / d, re sp ec tiv el y, fo r F 1 m al es ; an d 16 3, 80 4, an d 25 36 m g/ kg / d, re sp ec tiv el y fo r F 1 fe m al es ; PN D 4 fe ta lw ei gh ts re po rte d pr ec ul lin g of lit te rs . †, Av er ag e gl yp ho sa te in ta ke in th e 10 00 -, 30 00 -, an d 10 ,0 00 -p pm gr ou ps du rin g th e pr em at in g pe rio d re po rte d to be 99 ,2 93 ,a nd 98 5 m g/ kg / d, re sp ec tiv el y, fo r F 0 m al es ; 10 4, 32 3, an d 10 54 m g/ kg / d, re sp ec tiv el y, fo r F 0 fe m al es ; 11 7, 35 2, an d 11 61 m g/ kg / d, re sp ec tiv el y, fo r F 1 m al es ; an d 12 3, 37 1, an d 12 18 m g/ kg / d, re sp ec tiv el y, fo r F 1 fe m al es ; da y of bi rth de sig na te d PN D 1 in st ud y; fo rp ur po se s of da ta pr es en ta tio n in ta bl e, da y of bi rth is re de sig na te d as PN D 0; pu ps w ea ne d on PN D 29 (P N D 28 ); m ea n of fs pr in g w ei gh ts in ta bl e av er ag e of m ea n w ei gh ts fo rm al es an d fe m al es se pa ra te ly ;P N D 4- 21 pe rc en ts ur vi va ln ot ca lc ul at ed .¶ ,A ve ra ge PO EA in ta ke in th e 10 0- ,3 00 -, an d 10 00 -p pm gr ou ps du rin g th e pr em at in g pe rio d re po rte d to be 6, 18 an d 60 .5 m g/ kg / d, re sp ec tiv el y, fo r F 0 m al es an d 7, 20 .7 ,a nd 70 .4 m g/ kg / d, re sp ec tiv el y, fo r F 0 fe m al es ; m ea n of fs pr in g w ei gh tp re se nt ed in ta bl e is th e av er ag e of m ea n m al e an d fe m al e pu p w ei gh ts co m bi ne d; fo r th e F 1 ge ne ra tio n, on ly th e co nt ro la nd hi gh -d os e gr ou ps w er e m at ed an d pu ps w er e sa cr ifi ce d on PN D 4; N R = no t re po rte d; N / A = no ta pp lic ab le .+ ,A ve ra ge M O N 81 09 in ta ke in th e 3- 0, 10 0- ,3 00 -, an d 20 00 -p pm gr ou ps du rin g th e pr em at in g pe rio d re po rte d to be 2, 8, 23 ,a nd 13 4 m g/ kg / d, re sp ec tiv el y, in m al es an d 3, 9, 26 ,a nd 14 8 m g/ kg / d, re sp ec tiv el y, in fe m al es ;a ve ra ge M O N 08 18 in ta ke re po rte d to be 76 an d 86 m g/ kg / d in m al es an d fe m al es ,r ep sp ec tiv el y, in th e 10 00 -p pm gr ou p; ex po su re s w er e fo r1 4 d pr em at in g th ro ug h la ct at io n; pu ps sa cr ifi ce d on PN D 4; m ea n of fs pr in g w ei gh tp re se nt ed in ta bl e is th e av er ag e of m ea n m al e an d fe m al e pu p w ei gh ts co m bi ne d; N R = no tr ep or te d; N / A = no ta pp lic ab le . 59 D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 60 A. L. WILLIAMS ET AL. necropsied. Offspring of the second litter were randomly selected for mating to produce the next generation for study. Histopathology was conducted on 10 males and 10 females from each control and high dose treatment group for each parental generation (F0, F1, and F2) and offspring of generation F3b. Clinical obser- vation data, mean body weights, and food consumption were comparable across control and treatment groups for all generations (data not shown). Mating, fertility, and pregnancy indices showed considerable variability across the study, but no consistent dose-related trends were evident. Mean gestation length was comparable among control and treatment groups for each mating and in all generations, as were mean numbers of total, live, and dead pups per dam and the male/female sex ratios of pups. Pup weights throughout weaning and the mean number of pups weaned per litter were similar among control and treatment groups for each mating and all generations. Statistical differences in postnatal pup survival indices were noted between control and some treated groups in each generation; however, no dose-related trends could be discerned. It should be noted that in the second mating of the F0 generation, reduced pup survival in the treated groups for postnatal days (PND) 4-21 was mainly attributed to high pup mor- tality in one or more litters at each treatment level; as such, differences in pup survival indices between control and treated groups were concluded to not indicate an adverse effect of treatment. Terminal body, organ, and organ/body relative weights were comparable across all control and treatment offspring of the F0 and F1 generations, and for males of the F2 generation (data not shown). F2 female offspring from the treatment groups, however, exhibited significantly lower liver/body weight ratios compared to controls, although no dose-related trend was apparent. In addition, mean spleen weights were higher in the F2 mid-dose females compared to controls, but the low- and high-dose weights were comparable to control values. Because a clear dose response was not observed across the generations, these data were not considered indicative of a treatment-related adverse effect. An equivocal increase in tubular dilation of the kidney observed in the high-dose male F3b pups was not considered to be related to treatment (Schroeder 1982); further, it was not observed in a second study (Reyna 1990; see later discussion) conducted at much higher doses. Gross postmortem observations and histological evaluations of offspring from all generations also failed to demonstrate any treatment-related adverse effects. No NOAEL values were reported by the study authors. The study by Reyna (1990) was conducted according to established U.S. EPA guidelines of the time. In this study, male and female Sprague-Dawley rats (30/gender/group) were fed diets containing 0, 2000, 10,000, or 30,000 ppm glyphosate starting approximately 11 wk prior to mating. It is of note that the high dose exceeds by 50% the current limit dose for dietary studies (20,000 ppm). Also, the lowest dietary concentration (2,000 ppm) is almost sevenfold higher than the highest dietary concentration (300 ppm) used in the study of Schroeder (1981; 1982). These diets continued to the end of study and through- out all generations. Glyphosate intakes dur- ing the premating period averaged 132, 666, and 1983 mg/kg/d for F0 males and 160, 777, and 2322 mg/kg/d for F0 females in the 2000-, 10,000-, and 30,000-ppm groups, respectively; average compound intakes dur- ing the premating period were slightly higher for the F1 generation. Litters of the first gen- eration were culled to 8 pups each on PND 4 and weaned on PND 21, at which time 30 rats/gender/group were randomly selected for creation of the F2 generation. The selected animals were allowed a 14-wk growth period before being mated twice, resulting in creation of the F2a and F2b generations. All animals were observed twice daily for mortality/morbidity, and body weights were recorded weekly. Food consumption was also recorded weekly up until mating, after which determinations were continued through gestation and lactation for females only. Weekly assessments for clinical signs of toxicity were also made on adults. Pup weights and signs of clinical toxicity were D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 61 recorded on PND 0, 4, 14, and 21. All litters were culled to 8 pups each on PND 4. F0 and F1 adults were examined by gross necropsy. All culled pups, those that died postnatally, those not selected for mating, and all F2a and F2b weanlings were also examined by gross necropsy. The F0 and F1 adult ovaries and testes (including epididymides) were weighed. Histopathology using H&E stain was conducted on all tissues retained from control and high- dose treatment groups of the F0 and F1 gen- erations and on one weanling per gender per litter from these treatment groups of the F2b generation. Clinical signs included soft stools, reduced food intake, and decreased body weights in male and female rats of both the F0 and F1 generations fed 30,000 ppm glyphosate. Mean body weights of high dose animals were main- tained at 8-11% below control throughout the study. Body weight gains during gestation, however, were comparable among females from the control and high-dose groups. Body weight effects were not observed in the middle- and low-dose treatment groups. Glyphosate treatment exerted no marked effect on mat- ing, pregnancy, or fertility indices. Gestational lengths were also unaffected. The mean num- ber of pups per dam of the F0 generation’s high- dose group was numerically reduced com- pared to control. A similar, although less substantial, difference between control and high-dose animals was noted as a result of the first, but not the second, mating of the F1 gen- eration. Because the differences in litter size between high-dose and control groups were not statistically significant and not observed as a result of all matings, it is unlikely the effect was a result of treatment. The percentages of live and dead pups and the male/female sex ratios were similar across treatment groups for all generations. Mean pup weights at birth and initial weight gains were comparable across all treatment groups and generations. As ani- mals reached the age of weaning (PND 21), however, weight gains for the high-dose pups had significantly waned compared to controls for all generations. It was postulated that as pups began supplementing their milk intake with consumption of the prepared diets toward the end of the lactation period, food intake of pups in the high dose groups likely lagged behind that of control animals. No gross or microscopic pathological changes related to glyphosate treatment were noted for adult ani- mals or their offspring. The kidney effects noted in the Schroeder (1981; 1982) study (which used lower doses than those in this study) were not confirmed. Based on these results, 10,000 ppm (approximately 694 mg/kg/d, per Williams et al., 2000) is considered the NOAEL for systemic toxicity and 30,000 ppm (approx- imately 2132 mg/kg/d, per Williams et al., 2000) is considered the NOAEL for reproduc- tive and developmental toxicity. Moxon (2000) conducted an investigation according to more current U.S. EPA guidelines. In this study, male and female Wistar-derived rats (26/gender/group) were fed diets contain- ing 0, 1000, 3000 and 10,000 ppm glyphosate starting 10 wk before mating. These diets continued to the end of study and through- out two generations. Glyphosate intakes during the premating period averaged 99, 293, and 985 mg/kg/d for F0 males and 104, 323, and 1054 mg/kg/d for F0 females in the 1000-, 3000-, and 10,000-ppm groups, respectively; average compound intakes during the pre- mating period were slightly higher for the F1 generation. The day of birth was designated PND1. The study did not report whether lit- ters were culled. F1A litters were weaned on PND 29 and 26 rats/gender/group selected to become the F1 parental generation. Mating for production of the F2A litter commenced after another 10-wk premating period. Males were terminated following littering; females were ter- minated on or soon after PND 29, the day of weaning. Treatment-related reduced body weights were noted for F1 males at 10,000 ppm. However, glyphosate did not adversely affect reproductive performance and exerted no adverse effect on pup survival, litter size, or the pup sex ratio for either the F1A or F2A lit- ter. Although pup birth weight was not affected by treatment in both the F1A and F2A litters, F1A pup body weights in the high-dose group D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 were lower than for controls throughout the lactation period; this finding was not observed for the F2A litters. F1 males selected for mat- ing had a subsequent reduction in body weight during the premating period. Preputial sepa- ration and day of vaginal opening were both unaffected by glyphosate treatment in the F1 animals (Table 5). Further, no marked effects of treatment were noted on sperm (sperm num- ber, motility or morphology; see Table 5) in F0 and F1 males or on the number of primordial and small growing ovarian follicles in high-dose F1 females (F0 animals and other dose groups were not evaluated; data not shown). Estrous cycle length was significantly reduced at the high dose in F1 females; however, because the change was marginal, it was not con- sidered to be treatment related. Glyphosate treatment did not significantly affect F0 or F1 organ weights and was not associated with any macroscopic or microscopic findings in either parental animals or pups. Due to the reduced body weights at the high dose of F1A pups during lactation and F1 males dur- ing the premating period, the systemic and offspring/developmental NOAEL were both 3000 ppm (approximately 335 mg/kg/d dur- ing the premating period). Due to the absence of effects on fertility and reproductive perfor- mance, the reproductive NOAEL is considered to be the highest dose tested, 10,000 ppm (approximately 1105 mg/kg/d during the pre- mating period). POEA. Two reproductive/developmental screening studies were conducted with POEA surfactants covering a range of carbon chain lengths and degrees of polyalkoxylation (Tables 4 and 5; Knapp 2007; 2008). In the first study (Knapp 2007), groups of male and female Sprague-Dawley rats (20/gender/group) were administered a POEA incorporated into the diet at concentrations of 0, 100, 300, and 1000 ppm for 70 d premating through mat- ing (males) or through mating, gestation, and lactation (females). POEA intake during the premating period averaged 6, 18, and 60.5 mg/kg/d for F0 males and 7, 20.7, and 70.4 mg/kg/d for F0 females in the 100-, 300-, and 1000-ppm groups, respectively. F1 pups TA BL E 5. Se xu al M at ur at io n, Sp er m Pa ra m et er s, an d Es tro us C yc lic ity D at a fro m G ly ph os at e an d PO EA Re pr od uc tiv e St ud ie s in Ra ts A ni m al m od el (n um be r pe rg ro up ) A ge nt G en er at io n D os e (p pm ) A no ge ni ta l di st an ce (m m ), M / F D ay of pr ep ut ia l se pa ra tio n D ay of va gi na l op en in g Ri gh tc au da w ei gh t( g) N or m al sp er m (% ) Sp er m m ot ili ty (% ) St ra ig ht lin e ve lo ci ty (µ m / s) N um be ro f sp er m (1 07 / g ca ud a) N um be ro f sp er m (1 07 / g te st is) Es tro us cy cl e le ng th (d ) Re fe re nc e A lp k: A P f SD (W ist ar - de riv ed ) ra ts # (2 6 F; 26 M ) G ly ph os at e F0 0 N A N A N A 0. 25 8 ± 0. 03 1 97 .9 85 .9 ± 8. 8 55 .7 ± 7. 9 51 3 ± 16 0 55 ± 7 3. 97 ± 0. 87 M ox on 20 00 10 00 N A N A N A 0. 23 8 ± 0. 02 5∗ 98 .1 81 .5 ± 11 .5 55 .4 ± 11 .3 46 9 ± 21 3 N A 4. 38 ± 1. 25 30 00 N A N A N A 0. 24 7 ± 0. 03 3 95 .9 83 .5 ± 16 .1 56 .4 ± 10 .8 47 7 ± 18 2 N A 3. 88 ± 0. 68 10 ,0 00 N A N A N A 0. 25 6 ± 0. 02 4 97 .9 85 ± 8. 5 56 .3 ± 10 .3 55 0 ± 31 0 53 ± 10 3. 67 ± 0. 68 F1 0 N A 47 .1 ± 2. 3 35 .3 ± 1. 3 0. 25 5 ± 0. 03 4 99 .0 78 .1 ± 15 .9 48 .9 ± 12 .4 44 4 ± 18 3 56 ± 7 4. 26 ± 0. 58 10 00 N A 46 .6 ± 1. 7 35 .3 ± 1. 5 0. 26 3 ± 0. 03 7 99 .0 82 .7 ± 18 .8 48 .1 ± 15 .7 50 3 ± 30 8 N A 4. 12 ± 0. 24 30 00 N A 47 .2 ± 1. 6 35 .2 ± 1. 6 0. 25 9 ± 0. 03 0 98 .9 79 .7 ± 14 .2 47 .4 ± 15 .6 44 7± 17 7 N A 4. 05 ± 0. 35 10 ,0 00 N A 48 .0 ± 2. 1 35 .9 ± 1. 6 0. 25 5 ± 0. 03 1 99 .2 78 .8 ± 11 .9 46 .6 ± 12 .5 41 9 ± 25 9 55 ± 12 3. 94 ± 0. 35 ∗∗ Sp ra gu e- D aw le y ra ts † M O N 08 18 (P O EA ) F1 0 4. 99 / 3. 09 43 .0 ± 1. 93 33 .1 ± 1. 85 N R 99 .4 83 ± 8. 5 N R N R 78 .9 ± 12 .1 2 4. 2 ± 0. 38 Kn ap p 20 07 10 0 5. 00 / 3. 11 42 .9 ± 2. 10 32 .8 ± 1. 47 N R 99 .2 79 ± 10 .3 N R N R 79 .4 ± 8. 95 4. 4 ± 0. 55 30 0 5. 03 / 2. 90 ∗ 43 .7 ± 1. 61 33 .5 ± 1. 13 N R 99 .3 84 ± 7. 4 N R N R 78 .0 ± 8. 75 4. 2 ± 0. 54 10 00 5. 09 / 3. 03 42 .8 ± 1. 53 33 .0 ±0 .7 1 N R 98 .9 77 ± 12 .5 N R N R 71 .2 ± 13 .2 8 4. 2 ± 0. 50 N ot e. ∗ S ta tis tic al ly di ffe re nt fro m co nt ro l, p < .0 5; ∗∗ st at ist ic al ly di ffe re nt fro m co nt ro l, p < .0 1. # ,A ve ra ge gl yp ho sa te in ta ke in th e 10 00 -, 30 00 -, an d 10 ,0 00 -p pm gr ou ps du rin g th e pr em at in g pe rio d re po rte d to be 99 ,2 93 ,a nd 98 5 m g/ kg / d, re sp ec tiv el y, fo rF 0 m al es ;1 04 ,3 23 ,a nd 10 54 m g/ kg / d, re sp ec tiv el y, fo rF 0 fe m al es ;1 17 ,3 52 ,a nd 11 61 m g/ kg / d, re sp ec tiv el y, fo rF 1 m al es ;a nd 12 3, 37 1, an d 12 18 m g/ kg / d, re sp ec tiv el y, fo rF 1 fe m al es .† Av er ag e PO EA in ta ke in th e 10 0, 30 0 an d 10 00 pp m gr ou ps du rin g th e pr em at in g pe rio d re po rte d to be 6, 18 ,a nd 60 .5 m g/ kg / d, re sp ec tiv el y, fo rF 0 m al es an d 7, 20 .7 ,a nd 70 .4 m g/ kg / d, re sp ec tiv el y, fo rF 0 fe m al es ;N R = no tr ep or te d. 62 D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 63 (3/gender/litter) were weaned on PND 21/22, and then were administered POEA in the diet at mg/kg/d target doses equal to the mean POEA intake of the F0 generation animals until PND70. At PND70, F1 animals selected for breeding in the control and high-dose groups (2/gender/litter) were administered 0 or 1000 ppm POEA, respectively, in the diet either through mating (males) or mat- ing, gestation, and lactation (females). The majority of experimental parameters evaluated in this study were unaffected by treatment, including survival and clinical condition, body weight and food consumption, reproductive performance, organ weights, macroscopic and microscopic morphology of the F0 and F1 parental generations; clinical condition and body weights, anogenital distance, preputial separation and vaginal opening, estrous cyclicity, spermatogenic endpoints, testosterone and thyroid hormone levels of the F1 generation; and clinical condition, body weights, litter viability and postnatal survival of the F2 litters. In the F0 high-dose group, a significant increase was observed in the mean number of implantation sites that could not be accounted for by resorptions or pups born (live or dead). This finding was accompanied by a reduced mean number of pups born and a decreased live litter size in the high dose group. Three dams in the high dose group had litters of only two to four pups each, two of which showed total litter loss by PND4; this finding contributed to lower PND4 postnatal survival at the high dose. Upon breeding of the F1 gen- eration, however, none of these findings was reproducible. Because the observed changes were not reproducible between generations, and in some cases were not considered statisti- cally significant, these findings were considered equivocal. Knapp (2007) considered 300 ppm (approximately 20 mg/kg/d) to be the NOAEL for reproductive and developmental toxicity in this screening study. Knapp (2008) conducted a second reproductive/developmental screening study according to the OECD 422 test guideline and using two different POEA surfactants. Groups of Sprague-Dawley rats (12/gender/group) were administered either the same POEA from the 2007 study (POEA 1) at a concentration of 1000 ppm in the diet (to assess whether the equivocal litter effects seen at the high dose in the 2007 study could be repeated) or another POEA surfactant (POEA 2) at concentrations of 0, 30, 100, 300, or 2000 ppm in the diet. Treatment was administered from 14 d prior to mating for up to 72 d, including during gestation and lactation in the females. POEA 1 intake during the premating period averaged 76 mg/kg/d in males and 86 mg/kg/d in females administered 1000 ppm in the diet. POEA 2 intake during the premating period averaged 2, 8, 23, and 134 mg/kg/d for males and 3, 9, 26, and 148 mg/kg/d for females in the 30-, 100-, 300-, and 2000-ppm groups, respectively. In the group administered 1000 ppm POEA 1, all animals survived to scheduled necropsy with the exception of 1 female with dystocia that died on PND1 and a second female euthanized with a ruptured uterus on GD30. Because dystocia was noted in one of the F1 control-group females in the 2007 study, this finding was not considered treatment related. Parental systemic toxicity (mean body weight losses, lower mean body weight gains and reduced food consumption for both genders) was observed with POEA 2 at 2000 ppm, but not in any other treatment group. No treatment-related effects were observed on any reproductive parameters assessed, organ weights, or macroscopic and microscopic histology. The mean number of implantation sites was reduced and the number of unaccounted-for sites increased at 2000 ppm of POEA 2. Further, the mean live litter size on PND 0, mean number of pups born, and postnatal survival to PND4 were reduced in this treatment group. Compared to the control, mean pup weight on PND1 was also quantitatively reduced. Based on these findings, Knapp (2008) considered the NOAEL for POEA 2 to be 300 ppm (approximately 23 mg/kg/d). With regarding to the repeat study with POEA 1, no test substance-related signs of systemic toxicity, reproductive effects, or effects on pup survival or morphology were noted at any dose; thus, Knapp (2008) D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 considered the NOAEL for this POEA to be 1000 ppm (approximately 81 mg/kg/d). Reproductive/Developmental Data From Other Animal Studies Glyphosate. The National Toxicology Program (NTP) conducted a 13-wk glyphosate feeding study in F344/N rats and B6C3F1 mice (Chan and Mahler 1992). Groups of 10 male and 10 female rats and mice were administered feed containing 0, 3125, 6250, 12,500, 25,000, and 50,000 ppm glyphosate for 13 wk. Average glyphosate consumption in the 3125-, 6250-, 12,500-, 25,000-, and 50,000-ppm groups was 205, 410, 811, 1678, and 3393 mg/kg/d, respectively, for males and 213, 421, 844, 1690, and 3393 mg/kg/d, respectively, for females. An additional 10 ani- mals per gender and species were included at each dose level for evaluation of hematological and clinical pathology parameters. At the end of study, necropsies were conducted on all animals. For the screening of potential repro- ductive toxicity, epididymal tail, epididymal body, and testicular weights, sperm motility, sperm counts, and testicular spermatid head counts were evaluated for male rats and mice in control and three highest dose groups (12,500, 25,000, and 50,000 ppm glyphosate). Similarly, vaginal cytology and estrous cycle lengths were evaluated for female rats and mice from the same dose groups. Glyphosate treatment did not significantly affect survival of either rats or mice, but did reduce terminal body weights of male rats and mice in the 25,000- and 50,000-ppm treatment groups (Table 6). Mean body weight of female mice in the highest glyphosate treatment group (50,000 ppm) was also numerically affected (data not shown). The weights of the left testis as well as the cauda and corpus of the left epididymis were not affected by glyphosate treatment of rats or mice. Similarly, no marked effects of treatment on sperm motility, spermatid counts, and spermatid head counts were observed in either species. A sta- tistically significant decrease in concentration of spermatozoa in fluid withdrawn from the caudal epididymis compared to controls was TA BL E 6. Re pr od uc tiv e En dp oi nt s A ss es se d in a 13 -w k G ly ph os at e Fe ed in g St ud y in M ic e an d Ra ts (C ha n an d M ah le r1 99 2) A ni m al m od el (n um be rp er gr ou p) D os e (p pm ) Ex po su re Te rm in al bo dy w ei gh ts (g )ˆ Le ft ep id id ym al ta il w ei gh t( g) Le ft te st es w ei gh t( g) Le ft ep id id ym is w ei gh t( g) Sp er m co nc en tra tio n (1 06 ) Sp er m m ot ili ty (% ) Sp er m at id co un t (m ea n/ 10 −4 m l) Sp er m at id he ad s (1 07 / te st is) Sp er m at id he ad s (1 07 / g te st is) Es tro us cy cl e le ng th (d ) F3 44 / N ra ts (1 0 M , 10 F) 0 D ai ly, 13 w k 38 5 ± 5 0. 17 0 ± 0. 00 4 1. 54 ± 0. 03 0. 44 8 ± 0. 00 7 61 0 ± 36 91 ± 1 70 .1 5 ± 3. 00 14 .0 3 ± 0. 60 9. 10 ± 0. 35 4. 90 ± 0. 10 12 50 0 D ai ly, 13 w k 35 0 ± 5 0. 16 8 ± 0. 00 6 1. 52 ± 0. 05 0. 43 7 ± 0. 01 6 56 1 ± 23 92 ± 1 65 .3 3 ± 5. 49 13 .0 7 ± 1. 10 8. 48 ± 0. 64 5. 00 ± 0. 07 25 00 0 D ai ly, 13 w k 34 0 ± 5∗ 0. 16 7 ± 0. 00 4 1. 56 ± 0. 03 0. 44 0 ± 0. 00 4 48 5 ± 23 ∗∗ 92 ± 2 67 .2 3 ± 2. 05 13 .4 5 ± 0. 41 8. 63 ± 0. 30 4. 90 ± 0. 10 50 00 0 D ai ly, 13 w k 30 5 ± 7∗ ∗ 0. 17 9 ± 0. 00 6 1. 56 ± 0. 02 0. 45 2 ± 0. 00 7 48 6 ± 23 ∗∗ 91 ± 1 69 .0 0 ± 1. 71 14 .0 6 ± 0. 35 9. 04 ± 0. 20 5. 40 ± 0. 21 ∗ B6 C 3F 1 m ic e 10 M , 10 F) 0 D ai ly, 13 w k 32 .0 ± 1. 0 0. 01 5 ± 0. 00 1 0. 11 0 ± 0. 00 2 0. 04 4 ± 0. 00 1 11 62 ± 44 91 ± 1 67 .2 0 ± 2. 30 2. 15 ± 0. 07 19 .6 1 ± 0. 92 4. 06 ± 0. 05 # 12 50 0 D ai ly, 13 w k 31 .9 ± 0. 9 0. 01 4 ± 0. 00 1 0. 11 1 ± 0. 00 3 0. 04 3 ± 0. 00 2 13 70 ± 13 0 91 ± 1 63 .1 8 ± 3. 06 2. 02 ± 0. 10 18 .1 7 ± 0. 71 4. 00 ± 0. 00 25 00 0 D ai ly, 13 w k 29 .4 ± 0. 7∗ 0. 01 4 ± 0. 00 1 0. 11 1 ± 0. 00 2 0. 04 4 ± 0. 00 1 11 89 ± 60 92 ± 1 61 .9 3 ± 1. 92 1. 98 ± 0. 06 17 .8 7 ± 0. 60 4. 00 ± 0. 00 50 00 0 D ai ly, 13 w k 27 .2 ± 0. 4∗ ∗ 0. 01 4 ± 0. 00 1 0. 11 0 ± 0. 00 3 0. 04 2 ± 0. 00 1 13 08 ± 97 89 ± 1 65 .4 0 ± 2. 89 2. 09 ± 0. 09 18 .9 9 ± 0. 73 4. 00 ± 0. 00 # N ot e. Av er ag e gl yp ho sa te co ns um pt io n in th e 31 25 -, 62 50 -, 12 ,5 00 -, 25 ,0 00 -, an d 50 ,0 00 -p pm gr ou ps w as 20 5, 41 0, 81 1, 16 78 ,a nd 33 93 m g/ kg / d, re sp ec tiv el y, fo r m al es an d 21 3, 42 1, 84 4, 16 90 , an d 33 93 m g/ kg / d, re sp ec tiv el y, fo r fe m al es . ˆ, Te rm in al bo dy w ei gh ts ar e m ea n va lu es fo r m al es on ly. ∗ S ta tis tic al ly di ffe re nt fro m co nt ro l, p ≤ .0 5; ∗∗ st at ist ic al ly di ffe re nt fro m co nt ro l, p ≤ .0 1. # ,n = 9. 64 D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 65 observed in male rats treated at the two highest glyphosate concentrations. In addition, rat estrous cycle length among animals exposed to 50,000 ppm glyphosate was longer than that of controls, which is consistent with the reported weight loss. Similar effects on spermatozoa concentrations and estrous cycle lengths were not observed in treated mice. The biological significance of these findings is not clear; however, in the opinion of Chan and Mahler (1992), these findings were not considered evidence of adverse effects on the reproductive system. Yousef et al. (1995) investigated the effects of subchronic glyphosate treatment on semen characteristics in New Zealand white rabbits (Table 7). The study consisted of a 6-wk preliminary period in which no treatment was provided, followed by a 6-wk treatment period, and finally a 6-wk recovery period in which treatment was discontinued. Groups of 4 glyphosate-treated rabbits received either 1/100 LD50 (low dose) or 1/10 LD50 (high dose) administered orally in a gelatin cap- sule. The exact doses of glyphosate adminis- tered cannot be determined from the study report because neither the LD50 value from which the doses were determined nor the study from which the LD50 value was obtained was reported. Further, it is not clear if animals were dosed daily or weekly, or whether an herbicide formulation or pure glyphosate was used. On a weekly basis, body weights were recorded and ejaculates were obtained using a teaser doe. The following parameters were measured for each ejaculate sample: volume, sperm concentration, percent dead and abnor- mal sperm, methylene blue reduction times (MBRT; an indicator of sperm quality), initial fructolytic activity (an indicator of sperm vital- ity), and sperm osmolality. In general, body weights and all sperm parameters appeared to be adversely affected by treatment and showed some improvement during the recov- ery period. Based on these results, one cannot conclude that glyphosate treatment induced a harmful effect on semen quality. The dosages and frequency of exposure are unknown. Furthermore, the methods do not indicate whether control animals were sham-handled during the treatment period. If animals were not sham-handled, then the effects of treat- ment may be stress related. Based on statistical analyses, Yousef et al. (1995) were not able to detect a dose-response relationship for any of the measured parameters, with the excep- tion of dead sperm per ejaculate during the recovery period. The lack of an overall dose response further suggests that the observed results are random, rather than a direct effect TABLE 7. Overall Mean Semen Characteristics of New Zealand White Rabbits Treated Orally With A Glyphosate-Based Formulation (Yousef et al. 1995) Dose Body weight (kg) Semen volume (ml) Sperm conc. (× 104/cc) Abnormal sperm (%) Dead sperm (%) Methylene blue reduction time (min) Initial fructose (mg/100 ml) Semen osmolality Pretreatment period 0 2.944 ± 0.030 0.88 264 9.4 6.6 5.07 337 248 1/100 LD50 2.979 ± 0.060 0.83 265 9.7 6.4 5.22 324 255 1/10 LD50 3.173 ± 0.050∗ 0.88 262 10.3 6.5 5.07 336 253 Treatment period 0 3.008 ± 0.020 0.83 413 12.5 8.9 3.53 359 283 1/100 LD50 2.811 ± 0.060∗ 0.6∗ 242∗ 21.9∗ 19.5∗ 6.54∗ 281∗ 252∗ 1/10 LD50 3.125 ± 0.030 0.62∗ 262∗ 22.6∗ 21.4∗ 7.26∗ 267∗ 261∗ Recovery periodˆ 0 3.108 ± 0.010 0.82 596 20.4 4.1 3.48 312 278 1/100 LD50 2.816 ± 0.070∗ 0.68∗ 473∗ 25.7∗ 6.2 5.0∗ 298 284 1/10 LD50 3.368 ± 0.020∗ 0.73∗ 467∗ 24.1∗ 7.5∗ 5.29∗ 297 278 Note. ˆ, Data from only three animals per group were recorded for the recovery period. ∗Significantly different from control, p < .05. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 66 A. L. WILLIAMS ET AL. of glyphosate treatment. In addition, the rabbits used in the study were quite small (approxi- mately 3 kg); buck rabbits used for mating pur- poses are usually larger, often weighing 4-5 kg. This suggests the animals may not have been fully mature. Based on the aforementioned shortcomings (failure to report numerical data, lack of detail in methods description, group sizes that were too small, and absence of a dose response), it is not possible to draw conclusions regarding the effects of glyphosate treatment on male rabbit fertility. Commercial herbicide formulations. Dallegrave et al. (2007) conducted a non- guideline developmental-reproductive study using a commercial formulation of Roundup (exact formulation unspecified) reported to contain 360 g/L glyphosate and 18% (w/v) POEA (Tables 8 and 9). Sixty pregnant female Wistar rats (15/group) were gavaged daily with 0, 50, 150, or 450 mg/kg/d “glyphosate-Roundup” throughout pregnancy and lactation. As previously discussed in ref- erence to an earlier study by Dallegrave et al. (2003), it is not clear whether the stated doses are of glyphosate or the Roundup formulation. Maternal body weights were recorded daily during pregnancy. Litter size, the numbers of living, dead, and viable pups, and the sex ratio of the offspring were recorded at birth. From the end of lactation until puberty, pup weights were recorded weekly and both general and sexual development of the offspring followed. One male and one female per litter were sacrificed at puberty (65 d of age) as well as at the age of adulthood (140 d of age) to investigate systemic and reproductive effects of treatment. For the females, sacrifices occurred on the first estrus after 65 or 140 d of age. Organ weights were recorded relative to total body weights. Additional investigations were also conducted on males. The numbers of TABLE 8. Reproductive Outcome and Sexual Development Data for Rats Treated With Roundup Through Pregnancy and Lactation (Dallegrave et al. 2007) Males Females Animal model (number per group) Exposure Dose (mg/kg/d) Maternal relative weight gain (%) Number of live fetuses (number per litter) Number of PD0 males/ females Mean offspring weight at birth (g) Day of testis descent Day of preputial separation Day of vaginal opening Wistar rats (15) Through pregnancy and lactation 0 36.8 160 (10.7) 75/85 5.77 15.0 31.7 34.9 50 38.7 161 (10.8) 80/81 5.97 15.0 31.7 37.6∗ 150 38.8 165 (10.3) 83/82 5.90 15.0 31.5 36.9∗ 450 40.9 162 (10.8) 93/69 6.05 14.8 30.7∗ 36.7∗ Note. ∗Significantly different from control, p < .05. TABLE 9. Effects of In Utero and Lactational Roundup Exposure on Sperm Production-Related Parameters (Dallegrave et al. 2007) Dose (mg/kg/d) Daily sperm production (× 106) Number of sperm (× 106) Sperm transit rate (d) Abnormal sperm (%) Tubules with spermatogenesis (%) Tubule diameter (µm) Testosterone level (ng/ml) Puberty (65 d of age) 0 11.1 44.2 4.1 8.6 84 166.7 5.2 50 12.2 53.9 4.6 16.7∗ 77 159.6 4 150 12.1 67.2 5.3 9.2 78.7 159.6 3.2 450 11.7 57.4 5.1 11.6 75 160.4 1.5∗ Adulthood (140 d of age) 0 20.5 344.7 17.7 5.4 92 180.8 3.9 50 15.3∗ 251.0∗ 17.5 8.3 73.5 187.3 3.4 150 19.7 368.7 20.2 8.4 74.5 173.4 6.3 450 14.7∗ 257.1∗ 18.5 7.7 65 185.1 3.3 Note. ∗Significantly different from control, p < .05. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 67 homogenization-resistant spermatid per testis and spermatozoa per epididymis tail were counted using a hemocytometer. Daily sperm production was determined by dividing the number of spermatids per animal by 6.1 d, and the epididymal sperm transit rate was cal- culated by dividing the number of epididymal sperm by the daily sperm production rate. Sperm morphology was assessed via micro- scopic examination of 200 sperm rinsed from the deferens ducts. Histological examination of the testes was conducted on five testes per treatment group to assess mean tubule diam- eter, the number of tubules with elongated spermatids, and the general condition of the testicular tissues. Finally, blood testosterone levels were determined by radioimmunoassay. At the doses administered in this study, maternal toxicity was not observed and repro- ductive outcome data (number of pups, sex ratio, etc.) and pup weights were unaffected by treatment. A non-dose-related delay in vaginal opening in females and early preputial sep- aration in the high dose males were noted; however, these findings were all within the nor- mal physiological range for the species and in line with historical control data. No other sig- nificant effects on the female offspring were observed. Male offspring at puberty exhibited a statistically increased percentage of abnormal sperm at the low but not medium or high dose, suggesting a random finding. A dose-related decrease in blood testosterone levels was also observed at puberty, with the finding at the high dose being significantly different from control. Interestingly, this result is contrary to what would be expected if the early preputial separation were a true finding. Further, the effect on testosterone levels was no longer evi- dent at the age of adulthood. In adulthood, daily sperm production and sperm number were also significantly reduced in the low- and high-dose groups, but not the medium-dose group, compared to controls; thus, no dose- related findings were observed. Finally, upon histological examination of the testis, a reduc- tion in elongated spermatids and the presence of vacuolization at puberty and degeneration of the tubular lumen at adulthood in some of the animals in the treated groups were noted. Unfortunately, the micrographs provided in the study report are at too low of a magnification and too small to draw conclusions; however, other findings are evident in the micrographs- specifically, enlarged interstitial cells-that the investigators fail to mention in their report. This obvious omission suggests that Dallegrave et al. (2007) have limited experience doing these types of histological examinations and that the findings may be an artifact of processing rather than a true effect of exposure. In addition, none of the guideline-compliant reproductive studies nor the NTP 13-wk subchronic study discussed previously-all of which involved much greater glyphosate exposures-reported such testicular anomalies. Two non-guideline studies were conducted using Herbicygon, a commercial herbicide for- mulation containing glyphosate; no other for- mulation information was provided (Daruich et al. 2001; Beuret et al. 2004). In both stud- ies, the activity levels for certain enzymes in the liver (as well as in the heart and brain in Daruich et al., 2001) were measured both in dams exposed to Herbicygon via drink- ing water and in their fetuses. Female Wistar rats were mated and then divided into con- trol and treatment groups, with eight rats per group. Exposures began on GD 1 and were continued throughout gestation to GD 21. Control animals received tap water. In Daruich et al. (2001), treated animals received drink- ing water with 0.5% or 1% “glyphosate solution (w/v)”; in Beuret et al. (2004), treated ani- mals received 1% “glyphosate solution (w/v).” Despite the assertion that animals were dosed with glyphosate, the test solutions of drink- ing water were most likely prepared with Herbicygon. Because the glyphosate concen- tration of Herbicygon is not provided, it is not possible to know exactly how much glyphosate the treated rats received in these experiments. Further, Herbicygon is a commercial herbicide formulation, and as such most likely contains a surfactant. Thus, any effects noted in these studies cannot be definitively attributed to glyphosate, the surfactant, or a combination of the two ingredients. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 68 A. L. WILLIAMS ET AL. In Dariuch et al. (2001), body weights, food intake, and water consumption were mea- sured daily. After 2 wk of treatment, it was noted that treated rats had reduced their food and water intake compared to controls. In an attempt to account for the possible effects of restricted diet on the study results, a fourth treatment group of six rats was added to the study. This low diet group did not receive Herbicygon treatment, but was provided with minimal food and water (10 g of rat feed and 10 ml drinking water daily, administered sometime after a period of regular food and water consumption). On GD 21, maternal and fetal livers, hearts, and brains were removed, washed, and stored at -20◦C for subsequent analysis. Fetal organs were pooled. Tissues were homogenized, centrifuged to obtain the cytosolic fraction, and analyzed to measure the enzymatic activities of isocitrate dehydro- genase, glucose-6-phosphate dehydrogenase, and malic dehydrogenase. Herbicygon-treated dams consumed significantly less food and water and gained significantly less weight than controls (mean weight gains of 80.7 and 52.79 g in the 0.5% and 1% treatment groups, respectively, versus 92 g in the control group). Maternal liver (but not heart and brain) weights were also significantly decreased. Animals in the restricted diet group also gained signifi- cantly less weight during gestation than controls (49.51 g versus 92 g in the controls) and dis- played statistically smaller livers than control animals. These results suggest that the effect of treatment on body and organ weights may be due to reduced food and water intakes rather than a direct effect of Herbicygon treat- ment. It is difficult to draw any further conclu- sions from this study. Although various statistical increases and decreases in enzymatic activ- ity of maternal and fetal organs were noted (data were provided graphically in original study and are not reproduced here), a con- sistent effect of treatment was not observed and dose-response relationships were gener- ally lacking. In addition, only the cytosolic fractions from organ homogenates were eval- uated. Thus, the information gathered may be misleading because the enzymes monitored are found in both the cytosol and mitochon- dria. For example, in rat brain, only 45% of malic enzyme activity is in the cytosolic iso- form (Vogel et al. 1998). Furthermore, despite the inclusion of a low diet control group, a possible effect of diet restriction on enzy- matic activity cannot be completely ruled out for the Herbicygon-treated animals. Several investigators found that food restriction (in the absence of toxicants) affects the activity of many enzymes, including those examined in this study (Boll et al. 1996; Goodridge et al. 1996; 1998; Martin et al. 1990; Martins et al. 1985; 1986; Nagy et al. 1978; Sachan and Das 1982; Sassoon et al. 1968; Xie et al. 1995). Daruich et al. (2001) did not state exactly when the low diet treatment group began receiv- ing its restricted diet, and the average food and water consumption data presented in the study report suggest that this may not have occurred until as late as halfway through gesta- tion; thus, although data show that the period of restricted diet was sufficient to affect total maternal weight gain and liver weights, it may not have been sufficient to affect changes in organ enzyme activities. In order to appro- priately control for reductions in food and water consumption in the Herbicygon treat- ment groups, pair-fed control animals should have been included. Thus, based on the use of inappropriate controls, treatment with a com- mercial herbicide rather than pure glyphosate, unknown exposure levels, and a lack of consis- tent dose-response data, conclusions regarding the results of Daruich et al. (2001) cannot be made. Dosing in the second study using Herbicygon (Beuret et al. 2004) was as discussed earlier. It should be noted that Beuret et al. (2004) incorrectly referred to glyphosate as an organophosphate pesticide. In this study, body weights, food intake, and water consumption were measured daily. On GD21, serum samples were taken from dams for measurement of lipid peroxidation using a thiobarbituric acid-reactive substances (TBARS) assay. Maternal and fetal livers were homogenized and assays were con- ducted to measure levels of lipid peroxidation D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 69 products, and glutathione peroxidase, catalase, and superoxide dismutase (SOD) activities. Although not specifically stated in the methods section, it appears that fetal livers from each litter were pooled for analyses. Treated dams consumed significantly less food and water and gained significantly less weight during gestation than controls (mean weight gain of 53 g versus 92 g for controls). Liver weights were also reduced in treated rats compared to controls. Despite these maternal differences, average fetal body and liver weights did not appear to be affected by treatment. Because the number of fetuses per litter is not provided in the study report, it is not known whether the reduced body weight gain in the dams affected the number of fetuses surviving to term. Serum lipid peroxidation levels of the dams were not affected by treatment. Lipid peroxidation levels in the livers of treated dams and their fetuses, however, were increased over those of controls (dams: 1.6 ± 0.05 µg TMP/g tissue versus 0.9 ± 0.02 in the controls; fetuses: 9.8 ± 2.7 versus 2.4 ± 0.9). Glutathione peroxidase activity levels were also elevated with treatment in the fetuses (13.28 ± 0.58 µmol NADPH/min/g tissue versus 9.03 ± 1.01 in controls), but not in dams. Liver catalase and SOD activity levels were not affected by treatment. It is not possible to draw any conclusions regarding the effects of Herbicygon treatment from these data because no restricted diet controls were included in the study. Other studies showed that dietary restriction may affect lipid peroxidation and glutathione peroxidase activity levels (Kim et al. 1996; Mura et al. 1996; Rao et al. 1990); therefore, it is not known whether the effects observed resulted from treatment or reduced food and water intake. Furthermore, even if the effects observed in this study were directly related to treatment, whether they were due to glyphosate or other agents included in the Herbicygon formulation (including a surfactant) cannot be determined. In summary, because of inadequate information regarding dosing, limited sample numbers, and the lack of appropriate controls, no conclusions can be made regarding the effects of glyphosate on liver enzyme activity in treated dams and their offspring. Romano et al. (2010) conducted an experimental investigation regarding preputial separation in male Wistar rats using Roundup Transorb containing 480 g/L glyphosate. Groups of newly weaned male rats (n = 16-18) were gavaged daily from PND23 to PND53 with 0, 5, 50, or 250 mg/kg/d of Roundup Transorb. It is not known whether the rats in each group were taken from the same or different litters to control for potential litter effects. Rats were evaluated daily for balanopreputial separation beginning on PND 33. On PND 53, the rats were sacrificed, the testes and adrenal glands were weighed and examined both histologically and morphometrically, and serum hormone assessments were conducted for testosterone, estradiol and corticosterone. Although data were not shown, no marked effects of treatment on body weights were reported. Daily exposure to Roundup Transorb was reported to be associated with a significant delay in the time of preputial separation in the mid- and high-dose groups; body weight at preputial separation, however, was unaffected. It should be noted, however, that the age of preputial separation reported for the control rats (37 d) is extremely early compared to that reported in various other studies (40-43 d of age). Further, Romano et al. (2010) did not report whether the assessment was conducted blinded, whether the observations were made at the same time each day, or whether signs of incomplete separation or persistent threads were observed at any time. The extremely early age of preputial separation reported in this study suggests that the investigators may not have distinguished incomplete from complete separation. Thus, these data are likely unreliable. Relative testicular weights were increased at the high dose and relative adrenal weights were elevated in both the mid-dose and high- dose groups. It should be noted, however, that these data were highly variable. Although no abnormal pathology was reportedly observed in these tissues, the seminiferous epithelial D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 70 A. L. WILLIAMS ET AL. height was reduced in a dose-related manner and the luminal diameter of the seminiferous tubules increased in all treatment groups com- pared to controls. The micrographs presented in support of these findings are of poor qual- ity, exhibiting numerous fixation artifacts (e.g., shrinkage) that prevent an accurate estima- tion of tubule lumen diameter or seminiferous epithelial cell height. Furthermore, the sper- matogenic cycle does not stabilize and become synchronous for some time after puberty; thus, the findings observed may have simply been due to maturational variation among individ- ual tubules in the peripubertal rat. Thus, the morphometric analyses conducted are unre- liable and confounded by tubules being at different stages of maturation. Finally, Romano et al. (2010) report a dose-related reduc- tion in serum testosterone levels compared to controls, with testosterone levels at the high dose being approximately 50% lower than control; serum estradiol and corticosterone concentrations were unaffected. The reduced testosterone concentrations are in contradic- tion to the reported rise in testicular weights. In conclusion, the results of this study lack the scientific rigor necessary to support a defini- tive scientific conclusion and do not offset the findings of previous large, definitive, and GLP- compliant studies concluding that Roundup and glyphosate do not adversely affect repro- ductive development. Conclusions-Developmental and Reproduction Studies Based on a review of the available developmental and reproduc- tive studies, no data exist from studies that have been conducted using Good Laboratory Practices (GLP) protocols and/or according to established testing guidelines to indicate that glyphosate, POEA surfactants, or commercial glyphosate herbicides are developmental or reproductive toxicants. While a few studies claimed adverse reproductive or develop- mental effects (Dallegrave et al. 2003; Yousef et al. 1995; Dariuch et al. 2001; Beuret et al. 2004; Romano et al. 2010), these studies suffer from numerous inadequacies in design and reporting. Many of these studies appear to (1) have used commercial herbicide formulations rather than pure glyphosate or surfactant and have not followed up with additional studies to determine if findings were due to the pesticide active ingredient or another formulation component, (2) have failed to include appropriate controls, (3) have used inadequate numbers of animals per treatment group, and (4) have not clearly stated doses or dose rates. Furthermore, no consistent dose-related trends in effects were observed in these studies. The studies with glyphosate that have been conducted using GLP protocols and/or according to established testing guidelines found no marked effects of treatment on reproduction or in offspring, despite significant toxicity in treated dams (Holson 1990; 1991; IRDC 1980b; Moxon 2000; Reyna 1990; Schroeder 1981). The exception is a single rat teratology study that found an increase in resorptions, a decrease in the number of fetuses per dam, and reduced fetal weights associated with gavage administration of 3500 mg/kg/d glyphosate on GD 6-19 (IRDC 1980a). These effects, however, were associated with significant maternal toxicity, including death in 6/25 rats treated at this dose. It is important to note that the current limit dose for oral gavage studies is 1000 mg/kg/d for studies required by regulatory agencies. In conclusion, animal data, as a whole, indicate that glyphosate is not a selective developmental or reproductive toxicant. Mechanistic Studies A number of studies have been performed to investigate the potential impact of glyphosate and glyphosate-based herbicides on a variety of biological processes. The vast majority of these studies used in vitro models or non- mammalian in vivo models, such as the sea urchin. When possible, toxicity data derived using the same model system, but different types of test substances (pure glyphosate ver- sus glyphosate-based formulations), are high- lighted. These data provide an indication of the relative impacts of glyphosate versus other for- mulation additives on the observed effect(s). D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 71 The studies presented here are categorized according to the biological processes exam- ined. Emphasis is placed on those processes that could contribute to developmental and reproductive perturbations, although some of the information provided also relates to general mechanisms of toxicity. Cell Cycle/Transcriptional Inhibition Studies Several studies were performed to ascertain whether glyphosate is likely to inhibit cell cycle progression and transcription (Table 10). These studies were conducted using the sea urchin (Lytechinus variegatus) model, which is often used to assess aquatic toxicity but is of questionable value in gauging human health risk. In addition, the majority of these studies use herbicide formulations containing glyphosate as the test article, rather than glyphosate alone. Medina et al. (1994) first examined the impact of a formulation identified as “Roundup” on the sea urchin. In this study, the advantages of the use of the sea urchin as a biomarker of toxicity were discussed; particu- larly, its sensitivity to a variety of compounds and the ease of handling the model. Sea urchin embryos were exposed to 480 g/L “Roundup” (with a final concentration of 1.4 × 10−4 M glyphosate) 3 min after appearance of the fertilization membrane. It should be noted that this concentration of glyphosate is 34-fold greater than the allowable maximum contami- nant level of glyphosate in drinking water (U.S. EPA 2009a). The investigators observed that the Roundup-treated eggs exhibited deformed or destroyed nuclear elements, as well as a per- forated nuclear membrane. It is noted, how- ever, that (1) Roundup formulations contain surfactants; (2) the observations are consistent with the effects of a surfactant; and (3) the impact of glyphosate alone was never assessed. In Marc et al. (2002), the impacts of glyphosate alone and a formulation identified as “Roundup” (containing 170 g/L of isopropy- line glyphosate salt) on the cell cycle was examined. Concentrations of Roundup <1.0% were not lethal to urchin embryos; however, treatment with concentrations ≥0.8% (a con- centration much higher than what would be used for herbicidal purposes) led to a delay in the time to M-phase entry in the first cell divi- sion following fertilization. It was also reported that although pure glyphosate ≤25 mM exerted no marked effect on cell division, adding progressively larger amounts of glyphosate to 0.2% Roundup (which already contains 2 mM glyphosate) induced the delay in cell division. The data to support these claims, however, are weak. In particular, when glyphosate was added at increasing concentrations, no dose- response relationship was evident. Further, no statistics were shown, suggesting that the sig- nificance of the glyphosate-induced effects was not tested. Thus, the claim that glyphosate potentiates the action of Roundup on cell divi- sion is not supported by the data. Because CDK1/cyclin B is an important modulator of progression into the M phase of mitosis, Marc et al. (2002) next assessed the kinetics of CDK/cyclin B activation using histone H1 as a substrate for phosphorylation. Roundup was found to inhibit CDK/cyclin B activation and to reduce protein production, as indicated by a methionine incorporation radioassay. The amounts and phosphorylation status of cyclin B were also determined by Western blot analysis of whole embryo extracts; however, no clear differences between untreated and Roundup- treated cells were observed. Based on these results, no conclusions can be made regard- ing the effects of glyphosate on cell division. In addition, the study had several design flaws. For instance, appropriate controls were not included in these experiments. Given that cell division is highly affected by pH, tempera- ture and ionic concentration, a relatively non- toxic solution with these same characteristics should have been used as the negative con- trol. Furthermore, glyphosate alone was not examined. Rather, evaluations involved the herbicidal formulation Roundup, which con- tains surfactants having the potential to affect cell division. The impact of a formulation identified as Roundup on cell division and activation of CDK1/cyclin B was further investigated by Marc et al. (2003). Data demonstrated that 0.8% Roundup (glyphosate concentration D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 72 A. L. WILLIAMS ET AL. TABLE 10. Sea Urchin Embryo Assays Assessing the Ability of Glyphosate and Glyphosate-Based Formulations to Inhibit Cell Cycle Progression and Transcription Study Basic experimental design Findings Medina et al. 1994 20 µl suspension of fertilized sea urchin eggs exposed to 480 g/L Roundup (containing 1.4 × 10−4 M glyphosate) 3 min after appearance of the fertilization membrane; eggs observed for ∼24 h until the pluteus (free-swimming larvae) stage. Roundup-treated eggs exhibited deformed or destroyed nuclear elements, as well as a perforated nuclear membrane. Because a herbicidal formulation was tested, findings cannot be specifically attributed to glyphosate. Also, concentrations of Roundup used in this study are not environmentally relevant. Marc et al. 2002 Effects of 0.8% Roundup (containing 8 M glyphosate), 8mM pure glyphosate, and 0.2% Roundup supplemented with concentrations of glyphosate up to 10 mM, on the 1st cell division in sea urchin embryos were investigated; ∼100 embryos were scored per treatment group. The kinetics of CDK/cyclin B activation were also measured using H1 protein as a substrate. Roundup exposure was associated with an increase in first cell division delay, and inhibits CDK/cyclin B activation; 8 mM glyphosate had no impact on urchin cell division; supplemental glyphosate added to 0.2% Roundup induced cell division delay, but no dose-response relationship was observed. Concentrations of Roundup and glyphosate used in this study are not environmentally relevant. Marc et al. 2003 Impact of 0.8% Roundup (containing 8 M glyphosate) exposure on CDK/cyclin B activation at selected times following fertilization (≤120 min) investigated using H1 histone as a phosphorylation substrate. Authors report that Roundup blocked CDK/cyclin B activation, but urchins underwent cell division, albeit delayed. Because a herbicidal formulation was tested, findings cannot be specifically attributed to glyphosate. Also, concentrations of Roundup used in this study are not environmentally relevant. Marc et al. 2004a Effect of various concentrations of glyphosate-containing herbicides (Roundup3Plus, Amega, Cargly, Cosmic, and Roundup Biovert) on time of 1st cell division postfertilization assessed. Glyphosate concentrations of herbicide preparations tested ranged from 0.1-30 mM. All herbicides tested delayed the 1st cell division in a dose-dependent manner, but response across herbicides was independent of glyphosate concentration. Because herbicidal formulations were tested, findings cannot be specifically attributed to glyphosate. Also, herbicide concentrations used in this study are not environmentally relevant. Marc et al. 2004b Whether 10 mM Roundup inhibits CDK/cyclin B activation by preventing dephosphorylation of CDK1/cyclin B tyrosine 15 complex was examined using affinity purification and Western blot analysis; cells were examined at the time of 1st cell division postfertilization. The effects of 10 mM pure glyphosate and Roundup 3plus (at a concentration equivalent to 10 mM glyphosate) on phosphatase activity of the cdc25C recombinant protein and embryo extracts were also assessed. Roundup 3Plus exposure caused a 30 min delay in CDK1 tyrosine phosphorylation and was associated with a 70% inhibition of DNA synthesis; neither 10 mM pure glyphosate nor the concentration of Roundup 3Plus (containing 10 mM glyphosate) was associated with a change in cdc25C protein. Because only the herbicidal formulation was tested for effects on CDK1 phosphorylation, the findings cannot be specifically attributed to glyphosate. Also, concentrations of glyphosate and Roundup used in this study are not environmentally relevant. Marc et al. 2005 Impact of 0.2%, 0.4 %, and 0.6% Roundup (containing 2, 4, and 6 mM glyphosate), 30-900 mg/L POEA, and 0.2% Roundup with 8 mM supplemental glyphosate on the percentage of embryos hatching, and the delay in hatching time was observed with phase contrast microscopy. Transcriptional activity of embryo suspensions exposed to 0.2%, 0.4%, and 0.6% Roundup quantified by incorporation of 5-[3H]-uridine. Expression of sea urchin hatching enzyme mRNA (SgHE) in urchins exposed to 1% Roundup (10 mM glyphosate) measured by RT-PCR. 0.2-0.6% Roundup was associated with a dose-dependent decrease in percentage of embryos hatching, and an increase in hatching delay; addition of 8 mM glyphosate to 0.2% Roundup increased the hatching delay (but no statistics are provided to show that this is significant) and 8 mM glyphosate alone had no effect; a dose-dependent decrease in urchin embryo transcription seen with 0.1-0.8% Roundup; 30-900 mg/L POEA led to irreversible embryonic damage or lethality. Concentrations of Roundup and glyphosate used in this study are not environmentally relevant. Also, data suggest POEA, not glyphosate, is responsible for adverse effects of Roundup to sea urchin embryos. of 8 mM) slows first cell divisions in the sea urchin when applied after fer- tilization. The impact of CDK activity in treated and untreated sea urchin embryos was measured by affinity-purifying CDK1/cyclin B at selected times following fertilization (≤120 min), and determining the kinase activity of the enzyme complex using H1 as D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 73 a substrate for phosphorylation. The results from this experiment are difficult to inter- pret. Although Marc et al. (2003) claimed that Roundup treatment reduced CDK1/cyclin B complex activity, the figure presenting these findings shows only one-sided upper value standard error (SE) values for the control group, and no SE bars for the point of max- imum CDK1/cyclin B activation in the con- trols. Further, although the sea urchin embryos treated with Roundup were stated not to undergo CDK1/cyclin B activation, enough activation was apparently present to induce cell division, albeit delayed. In addition, using an assay to examine protein synthesis via incor- poration of radiolabeled methionine, Roundup appeared to decrease protein production dur- ing the first 2 h, which could potentially inhibit or delay various reproduction processes. In the last set of data presented, Marc et al. (2003) examined cyclin B abundance and phosphory- lation status at 60 and 75 min after fertilization using an antibody detection method. As shown previously (Marc et al. 2002), Roundup did not significantly affect CDK1/cyclin B activa- tion. Overall, data presented in this study did not clarify whether the delayed phosphoryla- tion of cyclin B observed following Roundup treatment is due to the delay in cell division or vice versa. Furthermore, because an herbicidal formulation was used in these experiments, no conclusions can be made regarding the poten- tial actions of glyphosate alone on cell cycle division. In Marc et al. (2004a), the effects of a variety of glyphosate-based herbicides on cell cycle progression in the sea urchin embryo were investigated. Herbicides assayed included Roundup 3plus, Amega, Cargly, Cosmic, and Roundup Biovert. The glyphosate concentra- tions of the herbicide preparations tested ranged from 0.1 to 30 mM. The percent- age of embryos undergoing the first postfer- tilization cell division was assessed by phase microscopy at 60-min intervals up to 300 min postfertilization. All herbicides tested inhib- ited cell cycle progression; however, the effects observed were not proportional to the glyphosate content of the herbicides. When tested at equivalent glyphosate concentrations, Amega, Cosmic, and Cargly were all more effective than Roundup 3plus and Roundup Biovert in delaying the first cell division. These results suggest that a formulation ingredient other than glyphosate may be mediating this effect. Cytological observations revealed no aberrant chromosome morphology in relation to the delay in cell cycle progression for any of the compounds tested. Marc et al. (2004b) then went on to examine whether Roundup 3plus inhibits CDK1/cyclin B activation by preventing dephosphorylation of the complex at tyrosine 15. Sea urchin cells were treated with Roundup 3plus at a concentration equivalent to 10 mM glyphosate, after which CDK1/cyclin B com- plex was affinity-purified from embryo extracts at 10-min intervals postfertilization. Following extraction from the beads, the affinity-purified proteins were resolved by gel electrophoresis. Using Western blot analysis, cyclin B and CDK1 protein expression were assessed, as well as the tyrosine phosphorylation of CDK1. For these experiments, CDK1 abundance was deemed to be not affected by treatment, and was therefore used as a gel-loading con- trol; however, CDK1 expression against that of another protein that is more commonly used for such purposes was not evaluated. Considering that these experiments were designed to examine effects on CDK1/cyclin B activation, the use of CDK1 expression as a loading control seems highly inappropriate. From this experiment, Marc et al. (2004b) reported that herbicide treatment delayed the cyclin B pattern changes associated with activation of CDK1 during the first postfertil- ization cell division. Furthermore, treatment delayed CDK1 tyrosine phosphorylation by 30 min compared to control and this delay corresponded with the delay in CDK1/cyclin B activation. To evaluate whether the delay in phosphorylation was due to an effect of treatment on phosphatase activity, the effects of both 10 mM glyphosate and Roundup 3plus (at a concentration equivalent to 10 mM glyphosate) on the phosphatase activity of recombinant cdc25C protein and embryo D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 74 A. L. WILLIAMS ET AL. extracts were assessed. Neither of the treat- ments induced changes in the phosphatase activity of recombinant cdc25C or embryo extracts. Next, DNA synthesis, as measured by the incorporation of radiolabeled thymidine, was assessed at various times postfertiliza- tion. During the first cell division, herbicide treatment inhibited DNA synthesis by approxi- mately 70% compared to control. From these results, it is not clear how herbicide treatment may mediate an inhibition of DNA synthesis or how such an effect may translate to a delay in CDK1 tyrosine phosphorylation. Furthermore, the fact that Marc et al. (2004b) did not present data using pure glyphosate in the DNA synthesis experiment is interesting, especially since glyphosate was purportedly used alone in the phosphatase assays. It appears likely that the observed effects on DNA synthesis are not mediated by glyphosate, but rather by another component of the Roundup 3plus formulation. In the final study, Marc et al. (2005) exam- ined the influence of various glyphosate formu- lations on transcription and sea urchin hatch- ing kinetics. For most experiments, Roundup 3plus was used; however, other formulations including Cargly, Cosmic, and Roundup Biovert were also tested in some assays. Hatching was observed with phase-contrast microscopy and expression of sea urchin hatching enzyme mRNA (SgHE) was measured by reverse- transcription polymerase chain reaction (RT- PCR). Transcriptional activity was quantified by incorporation of 5-[3H]-uridine in a sea urchin embryo suspension. Actinomycin D, a known transcription inhibitor, was used as a positive control. In the first experiment measuring the effect of Roundup 3plus on hatching kinetics, sea urchin embryos at the morula stage (after 4-6 cycles of cell division) were exposed to 2, 4, or 6 mM of Roundup (30 replicates per concentration). The morula stage was chosen because previous studies showed that Roundup delayed the first cell divisions (Marc et al. 2002; 2003), and thus, experiments focused on the impact of Roundup on later cell divi- sions and transcription. Interestingly, the posi- tive control agent was applied 10 min following fertilization rather than at the morula stage. Why the test agent and positive control were not applied at the same developmental stage or what the effects may be of application at different stages is not known. A concentration- related decrease in percent embryos hatching after Roundup 3plus treatment was observed. The delay in hatching time due to administra- tion of 8 mM pure glyphosate, 0.2% Roundup, and 0.2% Roundup supplemented with 8 mM glyphosate was also measured. Test agents were again administered during the morula stage (four trials per treatment group). Pure glyphosate delayed hatching by 33 ± 6 min, 0.2% Roundup resulted in a 128 ± 30 min inhibition, and the 0.2% Roundup plus 8 mM glyphosate supplementation resulted in a delay of 205 ± 30 min. Thus, although glyphosate alone exerted little effect, co-administration of additional glyphosate with Roundup increased hatching delay time. Marc et al. (2005) inter- preted these results to mean that the sur- factant included in the Roundup formulation was not solely responsible for Roundup’s effect on hatching time; however, a statistical anal- ysis of the results was not conducted, and only four replicates were run per treatment group. Interestingly, in another experiment, pure glyphosate tested at concentrations of ≤8 mM exerted no marked effect on hatch- ing time. The four glyphosate-containing for- mulations (Roundup Biovert, Roundup 3plus, Cargly, and Cosmic), however, all produced delays in hatching, although some formula- tions were more potent than others. Despite the lack of effect with neat glyphosate, Marc et al. (2005) concluded that glyphosate might be detrimental because all four of the formula- tion products led to hatching time delays. In a final experiment, it was reported that Roundup 3plus applied at the morula stage of sea urchin development decreased transcription, as indi- cated by a decrease in 5-[3H]-uridine incor- poration. Further, SgHE mRNA expression was reduced for 2 h postfertilization; however, the results of only a single experiment were shown to support this claim and no statistical analyses were conducted. Overall, these experiments demonstrated that glyphosate-based herbicidal formulations impact cell divisions in the sea D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 75 urchin embryo; however, data do not pro- vide evidence that glyphosate is the cause of these effects. In fact, these studies indicate that glyphosate itself is significantly less toxic to sea urchin embryos than the commercial her- bicidal products, suggesting that the observed effects are due to another component of the formulations. A study by Amouroux et al. (1999) was con- ducted that did not address glyphosate, but rather examined the toxicity of three com- monly used mild surfactants in sea urchins. The results of this study provide support that the effects observed in sea urchin embryos following herbicide application are due to a component of the formulations rather than to glyphosate itself. In this study, the effects of cocamido propyl hydroxyl sultaine (CAS), mag- nesium laureth sulfate (Mg LES), and decyl glucoside (APG) on inhibition of egg cleav- age, calcium homeostasis, intracellular pH, sodium and potassium contents, protein and DNA synthesis, and protein phosphorylation were measured. All surfactants tested pro- duced inhibition of cleavage at concentrations lower than those commonly used in consumer products. In addition, both CAS and Mg LES induced changes in membrane permeability and ionic disequilibrium. APG was found to alter intracellular pH and decrease DNA syn- thesis. Although POEA, the surfactant used in Roundup and many of the other commercially available glyphosate-based herbicides, was not specifically examined in this study, these find- ings suggest that toxicity to sea urchin eggs appears to be a common feature of surfactants. Thus, the findings of similar toxicity upon appli- cation of herbicidal formulations containing similar surfactants should be considered unre- markable. Summary-Cell Cycle/Transcriptional Inhibition Studies Overall, results using the sea urchin model showed that exposure to high concentrations of glyphosate-based herbicide formulations and substances used on consumer products lead to cell cycle delay. Despite these findings, the relevance of such studies for the human health risk assessment of glyphosate is questionable. The relationship between cell division in sea urchin eggs directly exposed to high concentrations of pesticides versus the effects in humans exposed dermally or orally to much lower concentrations of glyphosate-based herbicides is tenuous at best. Generally, concentrations ≥8 mM glyphosate were used in these studies (Medina et al. 1994; Marc et al. 2002; 2003; 2004a; 2004b; 2005). This concentration equates to an average body burden of 1.8 g isopropylamine glyphosate/kg body weight. For a 55-kg person, this would be equal to 100 g glyphosate, or the amount that would be found in 0.6 L of Roundup, if it were to be directly ingested (Kutzman and DeSesso 2003). Further, the majority of experiments addressing the impact of glyphosate in the sea urchin model were conducted using Roundup- branded or other herbicide formulations, rather than neat glyphosate. Evidence that glyphosate, and not the surfactants present in these formulations, was involved in the observed effects is lacking. Finally, there is not sufficient evidence to support the notion of Marc et al. (2003) that glyphosate potentiates the toxicity of Roundup-branded herbicides. Endocrine Disruption In recent years, many environmental pol- lutants have been suspected to contribute to endocrine disruption; however, only a few have been scientifically proven to disrupt the endocrine system at environmentally rele- vant concentrations (WHO 2002). Mechanistic studies to ascertain whether glyphosate might produce adverse developmental or repro- ductive effects by interfering with the func- tioning of the endocrine system have been conducted (Table 11). These studies are var- ied in their approach and examine poten- tial effects on steroid hormone production and placental enzyme activity. In a number of cases, glyphosate-based formulations con- taining surfactant systems were evaluated for aromatase activity using microsomes. These studies are flawed from the outset because microsomes are denatured by low concentra- tions of surfactants and detergents. This is noted in the U.S. EPA Endocrine Disruptor Screening D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 76 A. L. WILLIAMS ET AL. TABLE 11. Mechanistic Studies Assessing the Potential Endocrine-Disrupting Effects of Exposure to Glyphoste and Glyphosate-Based Formulations Study Basic experimental design Findings Petit et al. 1997 Recombinant yeast system expressing the estrogen receptor (ER): Estrogenic potential of various chemicals, including 10−8 to 10−4 M glyphosate, tested in yeast cells expressing the rainbow trout ER linked to a lacZ reporter gene; cells treated to test agents for 4 h. Glyphosate did not demonstrate estrogenic activity. Lin and Garry 2000 Estrogen-responsive MCF-7 cells: Response of MCF-7 cells to Roundup or glyphosate exposure assessed; cell proliferation after a 7-d exposure period in presence and absence of steroid growth factor-deficient FBS examined by flow cytometry; cell viability and apoptosis examined after 72 h of incubation by flow cytometry and propidium iodide. Cell proliferation increased with exposure to both Roundup and glyphosate, but response was similar with and without FBS, suggesting it was mediated through a nonestrogenic pathway; no cytotoxicity or apoptosis observed due to glyphosate exposure. Meulenberg 2002 Displacement of estradiol (E2) from human sex hormone binding globulin (SHBG): Displacement of tritiated E2 from SHBG by different concentrations of various test agents (including glyphosate) measured in vitro. Glyphosate reported to have shown ambiguous results for E2 displacement from SHBG. Xie et al. 2005 Rainbow trout vitellogenin assay: Ability of 0.11 mg/L glyphosate and other herbicides to induce vitellogenin expression in trout assessed. Glyphosate was not found to have estrogenic activity in this assay. Kojima et al. 2004 Human ERα, ERβ, and androgen receptor (AR) binding: More than 200 pesticides were tested for agonist or antagonist activity at human ERα, ERβ, and AR transfected into Chinese hamster ovary cells; ≤10−5 M glyphosate tested. Glyphosate was not noted to affect hormone binding in any of the receptor subtypes tested. Walsh et al. 2000 Steroidogenic acute regulatory (StAR) protein synthesis: Impact of Roundup (with 180 g/L glyphosate) and other herbicides on steroidogenesis in MA-10 Leydig tumor cells was assessed by measuring progesterone production by radioimmunoassay; levels of StAR mRNA assessed using Northern blots. 20-100 µg/ml Roundup, but not pure glyphosate, caused a significant dose-dependent decrease in progesterone production; 25 µg/ml Roundup did not influence overall protein levels, but decreased levels of StAR mRNA. Levine et al. 2007 Inhibition of progesterone production in MA-10 mouse Leydig cells: MA-10 cells were exposed for 2 h to Roundup with and without glyphosate, as well as to various surfactants; the hCG-stimulated increase in progesterone production was measured following incubation; impact of surfactants on StAR protein levels was assessed by Western blot on hCG-stimulated and nonstimulated MA-10 cells; impact of treatment on mitochondrial membrane function was determined by JC-1 cationic dye. Exposure to surfactants, as well as to Roundup with and without glyphosate, was associated with a decrease in hCG-progesterone production, decreased expression of the StAR protein, and a decrease in mitochondrial membrane function. Richard et al. 2005 Aromatase activity and mRNA levels in JEG3 cells and placental and equine testicular microsomes: Aromatase activity in JEG3 cells treated 1 and 18 h with 0.2-2% Roundup (or corresponding concentrations of glyphosate) measured by radioimmunoassay; aromatase mRNA expression measured by RT-PCR. Aromatase activity in microsomes from full-term placentas and equine testes also assessed upon 15 min exposure to Roundup or glyphosate. JEG3 cells: 0.2-2% Roundup has significantly greater impact on cell viability than glyphosate of corresponding concentrations; aromatase activity significantly increased at 1 h and significantly decreased at 18 h after exposure to 0.01% Roundup; aromatase mRNA also decreased at 18 h following Roundup exposure; ≤0.8% glyphosate for 1 or 18 h had no effect on aromatase activity. Microsomes: Aromatase activity decreased at >0.05% Roundup and >0.5% glyphosate. Concentrations of Roundup and glyphosate used in this study are not environmentally relevant. (Continued) D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 77 TABLE 11. (Continued) Study Basic experimental design Findings Benachour et al. 2007 Aromatase activity in JEG3 and human embryonic kidney 293 cells and placental and equine testicular microsomes: Cell viability and aromatase activity following 1, 24, or 48 h of treatment with 1-2% Roundup or equivalent concentrations of glyphosate assessed as above; cultures treated in either serum-containing or serum-free media. 293 Cells were more sensitive than JEG3 cells; cells in serum-free media were more sensitive than those in serum-containing media; Roundup was substantially more cytotoxic than glyphosate; Roundup decreased aromatase activity in microsomes in temperature-responsive manner. Concentrations of Roundup and glyphosate used in this study are not environmentally relevant. Also, the pH values of the test agents were not adjusted appropriately. Gasnier et al. 2009 Aromatase activity and anti-estrogenicity in HepG2 cells and anti-androgenicity in MDA-MB-453-kb2 cells: Aromatase activity following 24 h of treatment with glyphosate or 1 or 4 herbicide formulations; anti-estrogenicity and anti-androgenicity assessed following 24 h of treatment with same test compounds; incubations done in serum-free media. Herbicide formulations inhibited aromatase activity and exhibited dose-dependent antiestrogenic and antiandrogenic activity; results were not proportional to glyphosate concentration of formulations; ≤0.3% glyphosate has no effect on aromatase or estrogenic activity; androgenic activity altered by glyphosate, but not in dose-dependent manner. Results with glyphosate alone suggest no endocrine modulating activity. Results with formulations confounded by presence of surfactants and other ingredients. Hokanson et al. 2007 Gene expression in MCF-7 cells: Gene expression following 18 h of exposure to 0.001-0.1% of a glyphosate-containing herbicide was assessed by DNA microarray and RT-PCR. Treatment altered gene expression, but of seven genes selected for further study, disregulation was confirmed by RT-PCR for only three. Because a herbicidal formulation was tested, findings cannot be specifically attributed to glyphosate. Also, no evidence indicates that these changes were mediated through endocrine-disruption. Paganelli et al. 2010 Neural crest cell marker expression in Xenopus laevis embryos: Expression of various neural crest cell markers following exposure of 2-cell stage embryos to 1/5000-1/3000 dilutions of Roundup Classic or injection with 500 pg glyphosate. Treatments reduced neural crest cell marker expression and appeared to be associated with cranial malformations; possible involvement of retinoic acid pathway hypothesized. Glyphosate solution was not pH-adjusted and was injected into embryos, making relevance to environmental exposures questionable. Program Test Guideline OPPTS 890.1200: Aromatase (Human Recombinant) (U.S. EPA 2009b), which clearly warns that all glassware and apparatus used in the microsome prepa- rations need to be free of detergent residue. Furthermore, if detergent residues compromise study viability testing, measurable concentra- tions of detergent-like substances would cer- tainly overload such in vitro systems, and thus do not represent a viable approach to investi- gating endocrine disruption. Petit et al. (1997) screened various herbicides, fungicides, insecticides, xeno- biotics, and phytoestrogens for estrogenic potency using two in vitro systems: a recombinant yeast system expressing the rainbow trout estrogen receptor, and rain- bow trout hepatocyte cultures. Yeast cells containing a lacZ reporter gene linked to 2 estrogen-responsive elements were treated in culture at 10−8 to 10−4 M of each test agent for 4 h. 17 β-Estradiol was used as the positive control. β-Galactosidase activity, dependent on expression of the lacZ gene, was measured in Miller units using a colorimetric substrate. To ensure that the absence of a response was not due to toxicity, cell density measure- ments were made before and after treatment, although the data for agents that were not estrogenic were not shown. Glyphosate treat- ment exerted no marked effect on the basal level of β-galactosidase activity. Only those test D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 78 A. L. WILLIAMS ET AL. agents shown to be positive for estrogenicity in the yeast system, plus 11 other randomly selected test compounds, were evaluated in the trout hepatocyte cultures for expression of the vitellogenin gene, as determined by slot blot analysis; glyphosate was not among those tested. One weakness of this study is that the description of methods is not clear as to whether pure glyphosate or a glyphosate-based herbicide was tested. Nevertheless, these data provide no evidence of estrogenic activity. Lin and Garry (2000) investigated whether certain herbicides and fungicides commonly used in the Red River Valley of Minnesota might induce proliferation of the estrogen- responsive MCF-7 cell line. MCF-7 cells were seeded in media containing either regular fetal bovine serum (FBS) or steroid growth-factor- deficient FBS (produced through prior treat- ment with 10% charcoal dextran). Following a 48-h incubation, the cells were then treated with different dilutions of test chemicals, 10−9 M estradiol (positive control), or solvent vehi- cle (negative control). After 7 d in culture, cell numbers and viability of harvested cells were assessed using a fluorescence-activated cell sorter. In separate experiments, cytotoxic- ity (following 72-h incubation of MCF-7 cells in various concentrations of test agents) and apoptosis (using propidium iodide staining) were evaluated by flow cytometry. Both the “Roundup”-branded formulation (identified as containing 0.99% glyphosate) and its active ingredient, glyphosate, were found to induce proliferation of MCF-7 cells. This occurred in media containing either regular or steroid growth-factor-deficient FBS, suggesting that the proliferative effect was mediated through a nonestrogenic pathway. Maximal induction lev- els ranged from 121 ± 10.3% for 10 µg/ml “Roundup” in regular FBS and 135 ± 3.5% for 2.28 × 10−4 M glyphosate in steroid growth-factor-deficient FBS. None of the test agents used in these experiments was shown to be cytotoxic at the concentrations used in the 7-d proliferation studies. In addition, neither glyphosate nor “Roundup” was found to induce apoptosis. While these results sug- gest that glyphosate may be able to induce cell proliferation, this response is not mediated through an estrogenic pathway. Using an in vitro system, Meulenberg (2002) tested the ability of various endoge- nous steroids, pharmaceutical agents, pesti- cides, and pollutants to displace estradiol (E2) from human sex hormone-binding glob- ulin (SHBG), a high-affinity, but low-capacity, hormone-binding protein found in the blood that functions in the transport of sex hormones and protects against their degradation. Changes in the binding capacity of SHBG affect the free concentrations of various sex hormones. Because it is assumed that only the free frac- tion of such hormones exerts biological activity, such changes may result in hormonally medi- ated changes in the organism. Microtiter plates were coated with rabbit anti-SHBG antibody, and using these plates, SHBG was isolated overnight from the serum of pregnant women. Following several washes, tritiated E2, along with the test compound, was added to the microtiter plates. Following 48 h of incubation, supernatant was removed from the plates and the amount of radioactivity in the media was measured using a scintillation counter. Because testosterone is known to possess a threefold greater affinity for SHBG than E2, testosterone was used as a positive control. The binding of varying concentrations of test agents was refer- enced to the standard curve for testosterone. Affinity of these compounds for SHBG was defined as an ability to displace tritiated E2 to an extent comparable to that of testosterone. Meulenberg (2002) indicated that glyphosate demonstrated ambiguous results for displace- ment of E2 from SHBG, although actual exper- imental data were not shown. Because no data were presented for independent review, conclusions on whether glyphosate affects the ability of SHBG to bind sex hormones in the blood cannot be made. In Xie et al. (2005), the estrogenic potency of glyphosate, three non-glyphosate-based herbicides, and two types of ethyoxylate- containing surfactants (R-11 and Target Prospreader Activator [TPA]) was determined using the in vivo rainbow trout vitellogenin (VTG) assay. In fish, adult female production D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 79 of VTG is mediated by estrogenic activity; thus, VTG expression is thought to serve as a biomarker for chemicals likely to alter estrogenic activity in fish and other animals. In this study, exposure of fish for 7 d to 0.11 mg/L glyphosate exerted no significant effect on VTG levels, suggesting that glyphosate is unlikely to alter estrogenic activity. Kojima et al. (2004) tested more than 200 pesticides for their ability to act as ago- nists and antagonists to 2 human estrogen receptor (hER) subtypes, hERα and hERβ, and a human androgen receptor (hAR). For each hormone receptor of interest, Chinese ham- ster ovary cells were transfected with the appropriate cDNA expression vector, along with a reporter plasmid containing either an estrogen-responsive element or an androgen- responsive element, and a Renilla luciferase expression vector (used as an internal con- trol for determining transfection efficiency). After 3 h of transfection, cells were incubated for 24 h with varying concentrations of test agent. To assess antagonistic activity to hERα, hERβ, and hAR, the appropriate transfected cells were co-incubated with test agent and either 10−11 M E2, 10−10 M E2, or 10−10 M 5α-dihydroxytestosterone (DHT), respec- tively. Following incubation, expression of the response element-linked luciferase reporter was measured and normalized against that of the Renilla transfection control vector. Agonist activity was measured as the concentration showing 20% relative effective activity (REC20) as 10−10 M E2, 10−9 M E2, and 10−9 M DHT at the hERα, hERβ, and hAR, respectively. Antagonist activity was expressed as the 20% relative inhibitory concentration (RIC20)-that is, the concentration of test agent producing 20% inhibition of activity of 10−11 M E2, 10−10 M E2, or 10−10 M DHT at the hERα, hERβ, and hAR, respectively. Although not completely clear from the methods section of the paper, it appears that Kojima et al. (2004) deemed a test agent positive for agonist or antagonist activity when, at the range of concentrations tested (10−5 to 10−8 M), the test agent demon- strated greater activity than the REC20 or RIC20, respectively. The values presented in the study are the mean and standard deviations derived from at least three independent experiments. Although glyphosate was tested, it was not identified as a chemical possessing agonist or antagonist activity at any of the three recep- tor sites evaluated. It is noteworthy that spe- cific tests for cell toxicity were not conducted, although assays were conducted at concentra- tions ≤10−5 M to minimize cytotoxicity. Based on these results, glyphosate did not appear to affect hormone binding at the hERα, hERβ, or hAR. In Walsh et al. (2000), investigators assessed whether glyphosate or Roundup might affect the synthesis of the steroidogenic acute regu- latory (StAR) protein. The StAR protein, located on the outer mitochondrial membrane, trans- ports cholesterol to the inner mitochondrial membranes (Granot et al. 2002). It was pos- tulated that this protein might be particularly sensitive to environmental toxicants in general because its active precursor form is both highly labile and critically dependent on trophic hor- mone stimulation. Translocation of cholesterol across the mitochondrial membranes is a rate- limiting step in steroidogenesis, so slight dis- ruptions of StAR function and/or synthesis may potentially produce adverse effects. In this study, Roundup (180 g/L glyphosate) signif- icantly inhibited steroidogenesis (as seen by decreased progesterone production in MA- 10 cells) by inhibiting StAR protein expression at concentrations of 20-100 µg/ml. It is note- worthy that glyphosate alone, however, did not exert an effect on steroidogenesis or pro- tein production at any concentration tested (0-100 µg/ml), indicating that the effect on StAR was dependent on other components of the herbicide formulation. Levine et al. (2007) investigated the potential role of the surfactant in a Roundup- branded formulation in the inhibition of progesterone production upon treatment of MA-10 mouse Leydig cells. In this study, MA-10 cells were exposed for 2 h to var- ious surfactants (LAS D-40 [a linear alkyl- benzene sulfonate], alcohol ethoxylate, lau- ryl sulfate [SDS], and benzalkonium chloride), as well as a concentrated Roundup-branded D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 80 A. L. WILLIAMS ET AL. lawn and garden herbicide (with 180 g/L glyphosate isopropylamine, and 6.53 g/L sur- factant [primarily POEA]), and Roundup blank (formulation without glyphosate). Both the Roundup-branded formulation and Roundup blank decreased the hCG-stimulated increase in progesterone production. In both cases, the median inhibition concentration (IC50) was approximately 5 mg/ml. IC50 values for the four other surfactants were similar to that of the Roundup branded formulation and Roundup blank, indicating that (1) the effect on pro- gesterone is largely attributable to the surfac- tant, and not glyphosate, and (2) surfactants, in general, decrease hCG-stimulated proges- terone production. The impact of the various surfactants on StAR protein levels was also assessed by Western blot analysis on hCG- stimulated and nonstimulated MA-10 cells. Exposure to the surfactants, Roundup-branded formulation, and Roundup blank resulted in reduced levels of the 30-kD form of StAR protein, but not the 37-kD precursor form. Because formation of the 30-kD form requires mitochondrial import and processing of the 37-kD precursor, the effect of treatment on mitochondrial potential, an indicator of proper mitochondrial membrane function, was mea- sured using the JC-1 cationic dye. Treated MA-10 cells demonstrated a loss of normal mitochondrial membrane potential, implying that proper import and processing of the 37- kD form of the StAR protein was disrupted upon treatment. This finding explains the pre- viously observed decrease in the 30-kD form of the StAR protein. In addition, this effect on mitochondrial membrane potential was seen for benzalkonium chloride and the alcohol ethoxylate surfactants, the Roundup branded formulation, and Roundup blank at concen- trations below those that affect steroidogen- esis. Overall, these results support the con- cept that the adverse effects of Roundup branded herbicidal formulations on steroidoge- nesis are not mediated by glyphosate exposure, but rather are due to a nonendocrine mech- anism of compromised mitochondrial mem- brane potential and altered permeability of cell membranes. Richard et al. (2005) examined aromatase activity and mRNA levels in JEG3 cells (derived from a human placental choriocarcinoma cell line) exposed to pure glyphosate or unspeci- fied Roundup. Because glyphosate affects the cytochrome P-450 activity of plants (Lamb et al. 1998), it was postulated that mammalian aromatase (also a cytochrome P-450 enzyme) might be adversely affected. It was also of inter- est to further investigate claims made in other studies that glyphosate and/or an unspeci- fied Roundup branded formulation induced reproductive/developmental disturbances. The “Roundup” formulation was diluted in water to concentrations of ≤2% based on the rec- ommended concentration for agricultural use of 1-2% in water. Concentrations of pure glyphosate equivalent to those present in the range of Roundup dilutions tested were also used. Aromatase activity was measured at 1 and 18 h post treatment by determin- ing the amount of tritiated water released from the radiolabeled aromatase substrate, [1β−3H]-androstenedione. RT-PCR to amplify aromatase and GAPDH (as an endogenous control) mRNA was performed. General cell viability was also measured. Roundup exerted a more pronounced effect on cell viabil- ity than equivalent concentrations of pure glyphosate, indicating that the formulation ingredients played an important role in cytotox- icity, as discussed previously for in vitro systems where surfactants were added. Pure glyphosate did not significantly affect aromatase activ- ity at 1 or 18 h at any concentration tested (≤0.8%, or the highest concentration at which marked cytotoxicity was not observed). Similarly, aromatase mRNA levels were not affected by 18 h of treatment with ≤0.1% glyphosate. Incubation of the cells in Roundup for 1 h, however, increased aromatase activ- ity at all concentrations examined (0.02-0.2%). In contrast, incubation in Roundup for 18 h induced a concentration-dependent decrease in aromatase activity at all concentrations tested (≤0.8%). Levels of aromatase mRNA were also significantly decreased upon 18 h of incubation with 0.02 and 0.06% con- centrations of Roundup. It was noted that D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 81 if glyphosate was combined with 0.02% Roundup, a greater fall in aromatase activity after 18 h of incubation was observed than was seen with 0.02% Roundup alone; how- ever, the concentration of pure glyphosate used in this experiment was not indicated. Richard et al. (2005) also measured aromatase activ- ity in microsomes prepared from human full- term placental tissues incubated for 15 min with higher concentrations of Roundup and glyphosate (≤10 and 1.1%, respectively). In this case, Roundup and glyphosate significantly decreased aromatase activity at concentra- tions of >0.05% and ≥0.5%, respectively. Because significant cytotoxicity would not be expected at 15 min post treatment, the fall in aromatase activity likely is not due to cell death. Based on additional experiments using microsomes derived from equine testis, it was concluded that the rapid decrease in micro- somal aromatase activity is due to competitive inhibition; however, only data using Roundup are presented in the study. Based on these results, Richard et al. (2005) concluded that the additives in Roundup play a key role in its effect on aromatase, but that glyphosate itself might also elicit adverse effects. Although it was shown that pure glyphosate added to Roundup further reduced aromatase activity, the concentration of glyphosate required to elicit this effect was not indicated. Finally, in interpreting such findings for human health risk assessment, one needs to consider that the internal glyphosate concentration antici- pated to reach sensitive tissues is several orders of magnitude lower than those used in this study. Because these experiments were all con- ducted in an nonvalidated in vitro system using physiologically irrelevant concentrations and Richard et al. (2005) were thought to have greatly overinterpreted the results of their stud- ies, the French Ministry of Agriculture and Fish concluded that the study provided no use- ful information that was of value for human health risk assessment (Committee for Study of Toxicity 2005). As discussed previously, it is now recognized that testing surfactant- like substances in such a test system is not valid. A similar study using both JEG3 cells and the human embryonic kidney 293 cell line was conducted in the same laboratory to assess the effects of 1-2% concentrations of Roundup Bioforce (360 g/L acid glyphosate) and equivalent glyphosate concentrations on cell viability and aromatase activity (Benachour et al. 2007). The glyphosate solution used in many of these experiments was reported to have been pH adjusted to 5.8 (equivalent to the pH of 2% Roundup Bioforce solution). Following 1, 24, or 48 h of incubation, 293 cells were found to be more sensitive to the cyto- toxic effects of treatment than JEG3 cells; cells in serum-free media were more sensi- tive than those incubated in serum-containing media; and Roundup Bioforce was shown to be substantially more cytotoxic than glyphosate itself. In additional experiments, both Roundup Bioforce and glyphosate reduced aromatase activity in 293 cells cultured for 24 h in serum- free medium and human placental microsomes treated for 15 min. Roundup Bioforce was also demonstrated to affect aromatase activ- ity in equine testicular microsomes, and this effect appeared to be temperature respon- sive. The sensitivity of the cells incubated in serum-free media is not surprising. Serum supplementation of culture media provides cells with necessary nutrients and other pro- tective elements. Along these lines, it was reported that cells grown in the absence of serum were not viable after 60 h, regardless of treatment. Benachour et al. (2007) inter- preted these results to suggest that glyphosate is cytotoxic and possesses endocrine-disrupting properties. Because many of these experiments were conducted using serum-free media and the pH of the glyphosate solution was only adjusted to be equivalent to that of Roundup and not physiological pH, however, it is likely that many of the effects observed following treatment are due to changes in pH rather than a direct effect of glyphosate on cells. Ideally, the pH of the glyphosate solution should have been adjusted to physiological pH for these experiments. Alternatively, a negative control treatment using media that was pH adjusted to 5.8 should have been included. Interestingly, D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 82 A. L. WILLIAMS ET AL. in at least one of the experiments measuring the effects of Roundup treatment on aromatase activity in microsomal preparations, the pH of the Roundup was adjusted to physiological pH (7.4). Why the pH of the glyphosate solution was not similarly adjusted in these experiments is not clear. Given the confounding surfactant effects of damaging cell membranes, the value of these data is questionable. Follow-up investigations were conducted by Benachour and Séralini (2009) using endothelial cells from human umbilical cord vein (HUVEC), 293 embryonic kidney cells, and JEG3 placental cells. These studies assessed the cytotoxic potential and apoptosis associated with glyphosate, four glyphosate- based formulations, AMPA, and POEA. Thus, endocrine activity was not specifically evalu- ated and these studies are not discussed in detail herein. Gasnier et al. (2009) assessed the poten- tial for endocrine disruption in HepG2 cells. Additional experiments reported in this study evaluated cytotoxicity and genotoxicity, but these results are beyond the scope of the present analysis and are not discussed herein. In these experiments, both glyphosate and var- ious formulations used were pH 5.8 prior to cell treatment. To evaluate effects on aromatase activity, HepG2 cells in serum-free media were exposed for 24 h to non-cytotoxic concentra- tions of either glyphosate or one of four differ- ent commercial glyphosate-based formulations (Roundup Express with 7.2 g/L glyphosate; Bioforce/Extra 360 with 360 g/L glyphosate; Grand Travaux with 400 g/L glyphosate; or Grand Travaux plus with 450 g/L glyphosate). After treatment, cells were washed, and then treated with 200 nM of radiolabeled androstenedione. Aromatase mRNA levels were also measured by semiquantitative RT- PCR. Experiments were repeated thrice in trip- licate. Glyphosate alone at concentrations of up to 0.3% exerted no marked effect on aromatase activity or mRNA levels. In contrast, the four herbicide formulations-all of which con- tain various surfactants-inhibited aromatase activity and altered aromatase mRNA lev- els. Gasnier et al. (2009) noted that these effects were not proportional to the amount of glyphosate in the formulation, which sug- gests that the findings were due to other formulation components. In further experi- ments, HepG2 cells that had been transiently transfected with human ERα, ERβ, and a luciferase-linked estrogen-responsive reporter gene were treated in serum-free media with either glyphosate or 1 of the 4 formulations for 24 h. These incubations were done in the pres- ence of 10 nM 17β-estradiol to assess antie- strogenic potential. To assess antiandrogenic potential, MDA-MB-453-kb2 cells, which con- tain endogenous androgen receptors and a sta- bly transfected androgen-responsive reporter gene, were treated in serum-free media with either glyphosate or 1 of the 4 formulations for 24 h in the presence of 0.4 nM dihydrotestos- terone. Glyphosate alone exerted no significant antiestrogenic activity. At low concentrations, glyphosate treatment exhibited some antian- drogenic activity; however, as the glyphosate concentration increased, androgenic activity returned. Because the antiandrogenic activity did not increase with increasing glyphosate concentrations, it is unlikely to be related to treatment. In contrast, the four herbicide for- mulations inhibited both estrogenic and andro- genic activity. Again, these findings were not proportional to the amount of glyphosate in the formulations, suggesting the effects were due to the presence of other components in the formulations. As a whole, the results of this study suggest that glyphosate did not markedly affect endocrine activity. Further, as with other studies from this same group of investigators (Benachour et al. 2007; Benachour and Séralini 2009; Richard et al. 2005), this study is con- founded by the use of commercial formulations containing surfactants and other components that affect the integrity of cellular membranes and consequently produce false findings. Hokanson et al. (2007) examined gene expression in MCF-7 cells in response to treat- ment with 0.0001-0.1% dilutions of a herbi- cidal formulation containing 15% glyphosate (exact formulation not specified). Following 18 h of exposure, the expression of 1550 genes in treated and control cultures was evaluated D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 83 using a DNA microarray platform. Data showed that 680 genes were either upregulated or downregulated in response to glyphosate treat- ment; however, it is not clear whether the variability in gene expression of control cells was taken into account. The expression of seven of the genes was then examined in more detail using quantitative PCR. In this analy- sis, only three of the seven genes evaluated continued to display up- or downregulation; the other four failed to show disregulation in response to treatment. The 3 genes that continued to demonstrate a treatment-related effect (hypoxia inducible factor 1 [HIF1], early growth response 1 [EGR1], and chemokine lig- and 12 [CXCL12]) were said to also be affected by treatment with 3 × 10−10 M estrogen, which induced a response that was intermedi- ate between that of control and treatment with estrogen plus herbicide. Hokanson et al. (2007) interpreted these results to mean that glyphosate treatment altered estrogen regulation of gene expression; however, it cannot be determined whether the gene response may be due to formu- lation ingredients besides glyphosate or an effect of treatment on pH of the cell culture media. Furthermore, no evidence exists in the study to suggest that the effect of herbicide treatment was mediated through an estrogen- related pathway. Paganelli et al. (2010) studied the poten- tial for glyphosate or a glyphosate-based herbicide formulation (Roundup Classic, with 48% w/v glyphosate salt) to induce malforma- tions in developing Xenopus laevis embryos. Embryos at the 2-cell stage were exposed to 1/3000, 1/4000, or 1/5000 dilutions of Roundup Classic for an undisclosed period of time. Compared to untreated controls, treated embryos exhibited downregulation of the neu- ral crest markers slug, krox-20 in rhombomere 3, and N-tubulin along the three longitudi- nal domains of the posterior neural plate. In addition, the expression of shh (a mor- phogen), pax6 (essential for eye formation), otx2 (gene expressed in various parts of the developing eye), and sox9 (a transcription fac- tor expressed in cranial neural crest cells) was reduced in the embryos with herbicide formulation treatment at a 1/5000 dilution. In further experiments, embryos were directly injected with 500 pg glyphosate, again for an undisclosed period of time. Similar, albeit milder, downregulation of these neural crest markers was observed on the injected side of the embryos. To assess the functional changes associated with these changes in neural crest marker expression, embryos treated at the 2-cell stage were allowed to develop to stage 47, and then stained for skeletal analysis. Those treated with the herbicide formulation exhibited reduced cranial structures and eyes. Glyphosate injection resulted in similar malfor- mations. Additional experiments using embryos transiently transfected with a retinoic acid- responsive reporter gene suggested that treat- ment with the herbicide formulation enhanced endogenous retinoid activity and that this increased activity played a role in the induction of cranial malformations. Finally, chick embryos were treated in culture with the herbicide for- mulation and showed a similar downregulation of neural crest markers as was observed in the Xenopus embryos. The significance of these findings is unclear for several reasons. One drawback relates to the fact that glyphosate was not reported to have been pH adjusted; thus, the reported changes may have been due to the acidic nature of the test com- pound. Further, injection is an inappropriate route of exposure for assessing risk and it is not clear why the glyphosate was injected into the embryos rather than administered in the culture media like Roundup Classic. Overall, these findings require further substantiation in other labs using appropriate methods before the observations can be considered for risk assessment. Summary-Endocrine Disruption Over- all, these studies do not suggest that glyphosate is an endocrine disruptor. When tested alone, glyphosate was shown to be not estrogenic in a number of assay systems. Glyphosate did not activate the estrogen receptor or affect its ability to bind its normal endogenous lig- and in either in vitro or in vivo test systems (Gasnier et al. 2009; Kojima et al. 2004; D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 84 A. L. WILLIAMS ET AL. Petit et al. 1997; Xie et al. 2005); glyphosate also failed to displace estradiol from human sex hormone-binding globulin (Meulenberg 2002). Although a Roundup-branded formu- lation was able to alter StAR protein function (Walsh et al. 2000) and aromatase activity (Richard et al. 2005; Benachour et al. 2007), and inhibit progesterone production (Levine et al. 2007), these same effects generally were not observed when glyphosate was tested alone, suggesting that the responses might be due to another component of the pesticide formulation-likely a surfactant, as shown in the study by Levine et al. (2007), and likely via a non-endocrine-mediated mechanism. Finally, while both Roundup and its active ingredient, glyphosate, were able to induce the prolif- eration of estrogen-responsive MCF-7 cells in culture (Lin and Garry 2000), the use of steroid growth factor-deficient serum suggested that this response was not mediated through an estrogenic pathway. Reproductive Function Yousef et al. (1996) investigated the impact of glyphosate, as well as that of other pes- ticides, on the motility of human and rabbit sperm in vitro. This study was done, in part, to evaluate the utility of the motile rabbit sper- matozoa assay as a test system for predicting human responses to male reproductive toxi- cants. The concentration of glyphosate used cannot be determined because the study indi- cates that a glyphosate-based herbicide, and not pure glyphosate, was used in these experi- ments, and neither the commercial name nor the glyphosate concentration of this formu- lation was provided. Following incubation of sperm with varying concentrations of pesticides in either protein-free medium or medium con- taining bovine serum albumin (BSA), a sperm motility index (SMI) was calculated. This index was based on the percentage of sperm that were motile and the motility grade of the sperm (with values ranging from 0 in cases of no motility to 4 for cases of fast for- ward progressive movement). Fifteen minutes of incubation in BSA medium containing what was reported as 250, 500, or 1000 µM of the glyphosate-based test solution resulted in rabbit SMI values of 2.4, 2, and 1.8, respec- tively, versus a control SMI of 3.5. In contrast, the glyphosate-based test solution administered in protein-free medium for 15 min resulted in a rabbit SMI value of 0, regardless of the concentration, versus a control SMI value of 2.7. Following 60-min incubations with vary- ing concentrations of the glyphosate-based test solution, the IC50 values for rabbit sperm were 23.3 µM and 500 µM in protein-free medium and BSA medium, respectively. Similarly, the IC50 values for human sperm motility were 48.2 µM and 740 µM in protein-free medium and protein-containing (BSA) medium, respec- tively. Although these results suggested that the protein present in BSA-containing medium par- tially protected sperm from the harmful effects of treatment, little else can be concluded from this study. Because a herbicidal formu- lation was used rather than pure glyphosate, it is consistent with the aforementioned and reviewed studies that the observed results were due to the presence of surfactant rather than glyphosate. Furthermore, Yousef et al. (1996) did not mention whether they corrected the pH of the media following the addition of the pesticides. Certainly, a pH outside the normal range would adversely impact sperm motil- ity, regardless of treatment agent. Thus, the observed effects may have little to do with the actual agent administered in the study. Overall, this study provides no useful informa- tion regarding the potential adverse reproduc- tive effects of glyphosate for men. Conclusions-Mechanistic Studies Overall, the aggregate of available mecha- nistic data did not provide a plausible MOA by which glyphosate may produce adverse developmental or reproductive effects in humans. Many of these studies provide inadequate description of the test agent(s)- particularly, whether test systems were treated with pure glyphosate or a glyphosate-based commercial herbicide-and the final concen- trations of glyphosate to which test models were exposed. These deficiencies make it impossible to determine whether the observed D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 85 results may be attributed to glyphosate or another formulation ingredient, such as the surfactant. Furthermore, in the only study to test for this possibility (Levine et al. 2007), the results demonstrated that the observed effects were mediated through the surfactants present in the herbicidal formulations and consumer products. Finally, for the purposes of a human health risk assessment, these data provide little relevant information. For one, the concentra- tions administered in these in vitro studies are substantially higher than those anticipated to be experienced as a result of dermal contact or oral ingestion of glyphosate. In addition, these studies, by their very nature, do not take into account such factors as absorption, distribution, metabolism, and elimination, all of which play important roles in shaping human exposure responses. In conclusion, these data do not show a plausible and consistent mechanism by which glyphosate might produce developmental or reproductive disturbances in humans or animals. EVALUATION OF BIOMONITORING DATA Although the preceding hazard assessment for glyphosate failed to demonstrate any con- sistent evidence to indicate that glyphosate exposure may produce adverse developmental or reproductive health effects in humans, a review of the available biomonitoring data was considered pertinent to this evaluation in order to better understand the reasonably anticipated exposure levels for humans. Biomonitoring Studies To date, only a small body of biomonitoring data exists for assessing exposure levels associated with glyphosate field application (Table 12). These data are derived from studies looking at occupational pesticide levels in tree nursery workers (Lavy et al. 1992; 1993), those involved in the spray-clearing of brush (Cowell and Steinmetz 1990a; Jauhiainen et al. 1991), and members of farm and nonfarm families (Acquavella et al. 2004; 2005; Baker et al. 2005; Curwin et al. 2007a; 2007b; Mandel et al. 2005). Two other biomonitoring studies of glyphosate have been published (Abdelghani 1995; Centre de Toxicology du Quebec 1988), but neither study provides measures of indi- vidual systemic glyphosate concentrations, and thus they are not discussed in this review. Studies that measured glyphosate exposures via passive dosimetry only (for example, on cloth- ing, in air samples, or through hand washes alone) were also excluded from analysis, as these types of exposure measures do not pro- vide a predictive indicator of internal dose. TABLE 12. Estimated Glyphosate Doses Associated With Herbicide Application Study Sample size Dosimetry method Estimated glyphosate dose LLOMVa Spray-clearing of brush Cowell and Steinmetz 1990 16 Urinalysis (5/16 participants) 18.8 µg 0.01 µg/ml Passive (patch) 274 µg 0.1 µg/patch Jauhiainen et al. 1991 5 Urinalysis NDb 0.1 µg/mlc Passive (air) ≤15.7 µg/m3 0.3 µg/m3c Tree nursery work Lavy et al. 1992; 1993 14 Urinalysis ND 0.01 µg/ml Farm and nonfarm families Acquavella et al. 2004 48 farmers Urinalysis 4 µg/kgd 0.001 µg/mlc 48 spouses 0.04 µg/kgd 78 children 0.8 µg/kgd Curwin et al. 2007a; 2007b 65 farm children Urinalysis 0.11 µg/kge 0.0009 µg/mlc 51 nonfarm children 0.13 µg/kge Note. 1 ppb = 1 µg/L = 1 µg/1000 ml = 0.001 µg/ml. aLLOMV = Lower limit of method validation. bND = Not detectable. cAssay detection limit. dBased on highest reading registered. eBased on maximum likelihood model. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 86 A. L. WILLIAMS ET AL. Cowell and Steinmetz (1990a) measured glyphosate concentrations in the urine of forestry workers involved in the mixing and backpack spray application of a Roundup herbicide at three different locations. Although all 16 workers were involved in spray appli- cation of the herbicide, only 1 worker at each site prepared and mixed the Roundup herbicide prior to application. Air samples from the breathing zone of each worker were col- lected using an air filter and portable pump. Passive monitoring was conducted using hand washes and gauze patches placed at various predetermined locations on the workers’ cloth- ing. To determine the percent clothing pen- etration, patches were also worn underneath the clothes at sites adjacent to those where outside patches were attached. Urine sam- ples were collected on the day before, the day of, and 3 d following herbicide appli- cation. Twelve-hour composite samples from each worker were analyzed. Following sam- ple processing, glyphosate was quantified using high-pressure liquid chromatography and fluo- rescence detection. The lower limit of method validation (LLOMV) was reported to be 0.01 µg/ml for the urine samples, 0.5 µg for each air filter, and 0.1 µg per patch. For the pur- poses of exposure assessment, data less than the LLOMV were assumed to be equal to one- half the LLOMV. Applicator body doses were calculated based on the first 72 h following application. Only 5 of 16 workers had mea- surable glyphosate concentrations in their urine on the day of application; all other urine sam- ples were below the limits of detection. Based on analysis of the collected urine samples, the estimated average total body dose following spray application was 18.8 µg. In compari- son, the estimated average total body dose based on passive dosimetry measures was 274 µg and the average inhalation dose based on air sampling was 55.3 µg. Total body dose did not appear to correlate with specific occupation (mixing versus spray application). These data show that passive dosimetry esti- mates are approximately one order of mag- nitude higher than those based on biological measures. Jauhiainen et al. (1991) measured glyphosate concentrations in air and urine samples from five workers employed in the spray-clearing of forest brush. Workers were involved in the daily mixing of their own herbicide sprays, wore limited protective equipment (primarily helmets and gloves), and did not have access to wash facilities during their workday. A control group of five forest workers involved in the planting of trees was also evaluated. Air samples from the breathing zone of the workers were taken daily for 1 wk using a portable pump. Sampling times varied from 1 to 6 h. Urine samples were collected over the test week at the end of each workday, as well as after a 3-wk follow-up period. Following sample processing, glyphosate concentrations were measured by gas chromatography, with a detection limit of 0.1 ng/µl (0.3 µg/m3). Mid-week air samples contained less than 1.25 µg glyphosate/m3 air. The highest recorded air sample readings were 2.8 and 15.7 µg/m3. All urine glyphosate concentrations were below the limits of detection. Lavy et al. (1992; 1993) measured glyphosate exposure levels among conifer seedling nursery workers. Fourteen workers- including applicators, weeders, and scouts-were employed at two tree nurseries that used a Roundup herbicide. In this study, three different types of measurements were taken to assess potential and real exposures: dislodgeable residues, passive monitoring, and biological monitoring. To assess the amount of residual glyphosate that could be dislodged from conifer seedlings during contact with the plants, 100-g samples of fresh seedlings were shaken and rinsed under water for 45 s each. These measurements were made twice weekly over four spring/summer months. Passive monitoring of exposures was conducted using gauze patches attached to the clothing of work- ers at nine potential exposure points and via hand rinses of the workers taken at the end of the same workday. These measurements were taken 1 d/wk over the entire course of study and composited for each day of measurement to provide total passive exposures for each D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 87 worker. Biological monitoring involved collec- tion of total daily urine for each worker over 12 consecutive weeks. Twenty-four-hour sam- ples were also collected once weekly for 5 mo following the study period for each worker. Glyphosate concentrations were determined using the analytical procedures of Cowell and Steinmetz (1990b). The limit of detection for urine samples was 0.002 ppm and the lower limit of method validation was defined as 0.01 ppm. Of the 78 dislodgeable residue samples taken at 21 different sampling times, only 1 sample was positive for glyphosate residue, measuring 138.5 µg glyphosate. This finding indicates that dislodgeable residues are not a significant source of glyphosate exposure for nursery workers. Passive exposure measurements indicated that ankles and thighs received the greatest exposure, with 98% of exposures occurring at or below the thigh. Applicators received greater exposures than weeders. Scouts showed minimal exposure, with only 1 of 23 hand washes and 1 of 34 composited patch samples being positive for glyphosate. Normalizing the composite exposure values for body weight and exposure period resulted in average exposure levels of 7.2 × 10−4, 2.0 × 10−4, and 1.6 × 10−6 mg/kg/h for applicators, weeders, and scouts, respectively. In total, 355 urine samples were analyzed from the 14 workers over the course of study; however, all samples were below the limits of detection for glyphosate. These results suggested that, despite the level of passive exposures measured, actual internal doses of glyphosate received by the workers were minimal to nonexistent. The Farm Family Exposure Study was ini- tiated in 1999 and ultimately involved the biomonitoring of 95 families for glyphosate, 2,4-D, and chlorpyrifos exposure during years 2000 and 2001 (Acquavella et al. 2004; 2005; Baker et al. 2005; Mandel et al. 2005). Only the results related to glyphosate application are discussed herein. Families were randomly selected from listings of licensed pesticide applicators in South Carolina and Minnesota. Eligibility requirements were as follows: The family had to consist of the farmer, spouse, and at least 1 child between the ages of 4 and 18 yr; the family had to live on the farm and to farm at least 10 acres within 1 mile of the home, onto which it planned to apply 1 or more of the study pesticides within the study period as a part of normal operations; and the family members had to be willing to collect 24-h urine samples over 5 d, starting 1 d prior to the pesticide applica- tion through 3 d following application. Parents filled out pre- and postapplication question- naires detailing family activities and application practices. In addition, trained field staff were on hand to observe the pesticide application. Forty-eight of the 95 families provided speci- mens related to glyphosate application; these included specimens for 79 children. Urine sam- ples were analyzed for glyphosate using chela- tion ion exchange to concentrate and isolate the pesticide, followed by high-pressure liquid chromatography and fluorescence detection. Glyphosate findings were adjusted for recov- ery of the analyte using values obtained from spiked field- and travel-samples. Recovery was 69% for a 10-ppb sample and 78% for 100-ppb samples. The detection limit was 1 µg/l for a 100-ml urine sample. Twenty-nine percent of farmers applied glyphosate within 1 wk prior to their partic- ipation in the Farm Family Exposure Study. Glyphosate was applied using a tractor and boom sprayer in all cases. Twenty-nine per- cent of these farmers did not wear rubber gloves during the application process, 15% spilled pesticide during the mixing and/or load- ing stages of application, and 27% worked on their equipment during the application process. Only 60% of farmers had detectable glyphosate levels in their urine on application day, the day of highest glyphosate readings. By 3 d postap- plication, this number had declined to 27%. Urine concentrations of glyphosate ranged from below the limit of detection to 233 ppb. The geometric mean value for farmers was 3.2 ppb on application day, and declined to 1 ppb by postapplication day 3. Use of rub- ber gloves exerted the greatest influence on urinary concentrations. Other factors associ- ated with urine concentrations of glyphosate in D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 88 A. L. WILLIAMS ET AL. the farmers included the number of times the farmers mixed and loaded the glyphosate, use of an open-cab tractor, observed skin contact with the pesticide, and repair of the applica- tion equipment. The number of acres treated exerted no significant influence on urinary glyphosate concentrations. Only 2 of 48 spouses had detectable glyphosate concentrations in their urine on application day. The highest urine concentra- tion of glyphosate in a spouse was 3 ppb. No spouses participated in the pesticide appli- cation process. Nine of 78 children had detectable glyphosate concentrations in their urine on the day of application; all but 1 of these were reported either to have been present during or to have helped with the pesticide application. The highest glyphosate urinary value in a child was 29 ppb. Systemic doses of glyphosate were calcu- lated for all participants with detectable urine glyphosate concentrations. For each individual, the total amount of glyphosate excreted dur- ing the study period was determined, adjusting for incomplete excretion and pharmacokinetic recovery; this value was then divided by each individual’s body weight for determination of an individual’s systemic dose. Using these cal- culations, the maximum systemic dose for farmers was estimated to be 0.004 mg/kg and the geometric mean value was estimated to be 0.0001 mg/kg. Maximal systemic doses for spouses and children were estimated to be 0.00004 mg/kg and 0.0008 mg/kg, respec- tively. These values are all well below the oral reference dose for glyphosate of 2 mg/kg/d set by the U.S. Environmental Protection Agency (U.S. EPA 1993). Curwin et al. (2007a; 2007b) conducted a similar study of both farm and nonfarm fam- ilies residing in Iowa during the spring and summer of 2001. Exposure to seven target pesticides (atrazine, acetochlor, metolachlor, alachlor, chlorpyrifos, glyphosate, and 2,4-D) was examined; however, only the results for glyphosate are discussed herein. Study recruit- ment was done by convenience sampling. Study eligibility requirements were as follows: Households had to reside in 1 of 10 counties in central or eastern Iowa and have at least 1 child under the age of 16 yr; nonfarm fam- ilies had to reside on land that was not used for farming and no one in the household could be employed in agriculture or the commer- cial application of pesticides; farm families had to use at least 1 of the 7 target pesticides. Twenty-five farm families (66 farm children) and 25 nonfarm families (52 nonfarm children) were enrolled in the study. Each household was visited twice during the study period and two urine samples were collected from partic- ipants at each visit (one from the evening and one from the following morning). Dust samples were collected during each visit according to standard practices established by the American Society for Testing Material (ASTM). Urine sam- ples were kept cool, then shipped frozen to the laboratory, where the samples were ana- lyzed for parent pesticides and metabolites by immunoassay. The limit of detection (LOD) for glyphosate was 0.9 µg/L. Urinary concentra- tions data were recorded as positive values at or above the LOD, positive values below the LOD, or nondetects. These data were then analyzed using two different approaches. In the maxi- mum likelihood estimation, urinary concentra- tions reported as either nondetects or at levels below the LOD were set at the LOD for the assay. In the mixed-effects modeling approach, positive urinary concentrations below the LOD were used as reported and nondetects were set at one-half the lowest positive concentration measured. Urinary creatinine levels were also measured and used to normalize for total daily urinary voids when estimating daily pesticide exposures. Only 30% of absorbed glyphosate was assumed to be excreted in the urine, and this information was used to correct for total glyphosate exposure. In the case of glyphosate, urinary concen- trations were above the limits of detection for 65-75% of the parent samples and for 81-88% of children’s samples. Furthermore, farm and nonfarm families did not significantly differ in their mean urinary concentrations of glyphosate. Curwin et al. (2007a; 2007b) sur- mised that this may be because glyphosate use is not restricted to agricultural practices, D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 89 but rather may be commonly seen in resi- dential settings as well. Geometric mean uri- nary concentrations of glyphosate (using the maximum likelihood model) were 1.4 µg/L (range: 0.13-5.4 µg/L) and 1.9 µg/L (range: 0.020-18 µg/L) in nonfarm and farm fathers, respectively; 1.2 µg/L (range: 0.062-5 µg/L) and 1.5 µg/L (range: 0.1-11 µg/L) in non- farm and farm mothers, respectively; and 2.7 µg/L (range: 0.1-9.4 µg/L) and 2 µg/L (range: 0.022-18 µg/L) in nonfarm and farm children, respectively. Mean urinary concentrations cal- culated using the mixed-effect model were similar. These estimated urinary concentrations of glyphosate from this study are all within the same approximate order of magnitude as those found in the Farm Family Health Study, discussed earlier. Based on these data, the geometric mean doses of glyphosate were estimated for both farm and nonfarm children. Again using the maximum likelihood model, the daily absorbed dose of glyphosate for farm children was esti- mated to be 0.11 µg/kg/d (range: 0.013-0.34 µg/kg/d). This was similar to the dose esti- mated for nonfarm children: 0.13 µg/kg/d (range: 0.037-0.33 µg/kg/d). However, these values are approximately eightfold lower than the 0.8-µg/kg/d glyphosate exposure esti- mated for farm children in the Farm Family Exposure Study and certainly lower than the oral reference dose for glyphosate of 2 mg/kg/d set by the U.S. EPA (U.S. EPA 1993). The reason for the discrepancy in values between the two studies is not clear, but likely relates to differences in adjustments made to account for total urinary void and incomplete excretion of glyphosate. Summary-Biomonitoring Data The body of biomonitoring data avail- able for glyphosate is limited at this time. Nevertheless, the data reviewed herein clearly show that the degree of systemic glyphosate exposure that occurs as a result of nor- mal application practices is exceedingly small, often below the limits of detection (espe- cially for those not intimately involved in the application process). In fact, the highest sys- temic dose estimated from these studies was 0.004 mg/kg (Acquavella et al. 2004), a value 500-fold below the daily oral reference dose for glyphosate of 2 mg/kg/d (U.S. EPA 1993). These findings indicate that the risk of substan- tial exposure as a result of glyphosate applica- tion practices is minimal at best. CONCLUSIONS An extensive, in-depth analysis of the avail- able scientific literature provides no appar- ent evidence to indicate that exposure to glyphosate is associated with the poten- tial to produce adverse developmental and reproductive effects in humans. While the body of epidemiological data for glyphosate is fairly limited, and none of the available studies (with the exception of Sanin et al. 2009) were designed specifically to assess the potential effects of glyphosate exposure, data as a whole reveal no developmental or reproductive health disturbances associated with exposure. In contrast to epidemiologi- cal data, the database of animal studies for glyphosate is relatively robust, including stud- ies of mice, rats, and rabbits exposed to glyphosate, various glyphosate-based herbici- dal formulations, the major glyphosate envi- ronmental breakdown product AMPA, and POEA surfactants included in some Roundup- branded herbicides. All guideline-compliant studies reviewed found no marked effects of glyphosate treatment on reproductive health or the developing offspring at non-maternally toxic doses (Holson 1990; 1991; IRDC 1980a; 1980b; Knapp 2007; 2008; Reyna 1990;Schroeder 1981). It should be noted that while a number of non-guideline-compliant studies claimed adverse developmental effects associated with glyphosate exposure (Beuret et al. 2004; Dallegrave et al. 2003; Dariuch et al. 2001; Yousef et al. 1995), these inves- tigations suffer from numerous inadequacies in design, which makes substantiation of their conclusions problematical. Furthermore, these studies all used commercially formulated glyphosate-based herbicides rather than pure D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 90 A. L. WILLIAMS ET AL. glyphosate. Thus, findings reported in these studies cannot be definitively assigned to glyphosate exposure. Similarly, review of the available mech- anistic data related to glyphosate fails to find a plausible MOA by which glyphosate may be able to induce adverse develop- mental or reproductive outcomes. It should be noted, however, that the body of avail- able studies suffers from numerous design inadequacies, particularly with regard to the type of test agents used (commercially avail- able glyphosate-based herbicides versus pure glyphosate). Furthermore, other than hypothe- sizing possible MOA, these data provide little relevant information that can be used in a human health risk assessment. Finally, a review of the limited body of available biomonitoring studies shows that, via reasonably anticipated exposure routes, human exposure to glyphosate is likely to be well below the daily oral reference dose for glyphosate of 2 mg/kg/d, as set by the U.S. EPA (1993). These data show that, regardless of any potential developmental and reproductive hazards that may be alleged based on misin- terpretation of results from animal and mecha- nistic studies, the levels of glyphosate to which humans are likely to be exposed are far below the range of doses considered to be safe by the U.S. and other regulatory agencies world- wide. In conclusion, a thorough evaluation of the available data demonstrates that exposure to environmentally relevant glyphosate concen- trations is not anticipated to produce adverse developmental and reproductive effects in humans. REFERENCES Abdelghani, A. A. 1995. Assessment of the exposure of workers applying herbicide mixtures (2,4-D+Roundup, Garlon- 3A+Roundup). Toxicity and fate of these mixtures in the environment. Summary report. Baton Rouge, LA: Louisiana Transportation Research Center, State project no. 736-14-0067. Acquavella, J. F., Alexander, B. H., Mandel, J. S., Gustin, C., Baker, B., Chapman, P., and Bleeke, M. 2004. Glyphosate biomonitoring for farmers and their families: Results from the Farm Family Exposure Study. Environ. Health Perspect. 112: 321-26. Acquavella, J. F., Gustin, C., Alexander, B. H., and Mandel, J. S. 2005. Implications for epidemiologic research on variation by pes- ticide in studies of farmers and their families. Scand. J. Work Environ. Health 31(Suppl. 1): 105-9. Amouroux, I., Pesando, D., Noël, H., and Girard, J.-P. 1999. Mechanisms of cytotox- icity by cosmetic ingredients in sea urchin eggs. Arch. Environ. Contam. Toxicol. 36: 28-37. Arbuckle, T.E., Lin, Z., and Mery, L. S. 2001. An exploratory analysis of the effect of pesticide exposure on the risk of sponta- neous abortion in an Ontario farm pop- ulation. Environ. Health Perspect. 109: 851-57. Arbuckle, T. E., Savitz, D. A., Mery, L. S., and Curtis, K. M. 1999. Exposure to phe- noxy herbicides and the risk of spontaneous abortion. Epidemiology 10: 752-60. Baker, B. A., Alexander, B. H., Mandel, J. S., Acquavella, J. F., Honeycutt, R., and Chapman, P. 2005. Farm Family Exposure Study: Methods and recruitment practices for a biomonitoring study of pesticide expo- sure. J. Expos. Anal. Environ. Epidemiol. 15: 491-99. Bell, E. M., Hertz-Picciotto, I., and Beaumont, J. J. 2001a. A case-control study of pesticides and fetal death due to congenital anomalies. Epidemiology 12: 148-56. Bell, E. M., Hertz-Picciotto, I., and Beaumont, J. J. 2001b. Case-cohort analysis of agricul- tural pesticide applications near maternal residence and selected causes of fetal death. Am. J. Epidemiol. 154: 702-10. Bell, E. M., Hertz-Picciotto, I., and Beaumont, J. J. 2001c. Pesticides and fetal death due to congenital anomalies: Implication of an erratum (letter). Epidemiology 12: 595-96. Benachour, N., Sipahutar, H., Moslemi, S., Gasnier, C., Travert, C., and Séralini, G. E. 2007. Time- and dose-dependent effects of Roundup on human embryonic and D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 91 placental cells. Arch. Environ. Contam. Toxicol. 53: 126-33. Benachour, N., and Séralini, G. E. 2009. Glyphosate formulations induce apopto- sis and necrosis in human umbilical, embryonic, and placental cells, Chem. Res. Toxicol. 22: 97-105. Beuret, C. J., Zirulnik, F., and Giménez, M. S. 2004. Effect of the herbicide glyphosate on liver lipoperoxidation in pregnant rats and their fetuses. Reprod. Toxicol. 19: 501-4. Bloom, B., and Dey, A. N. 2006. Summary health statistics for U.S. children: National health interview survey, 2004. Vital Health Stat. 10: 1-85. Boll, M., Weber, L. W., and Stampfl, A. 1996. Nutritional regulation of the activi- ties of lipogenic enzymes of rat liver and brown adipose tissue. Z. Naturforsch 51: 859-69. Bradberry, S. M., Proudfoot, A. T., and Vale, J. A. 2004. Glyphosate poisoning. Toxicol. Rev. 23: 159-67. Brewster, D. W., Warren, J., and Hopkins, W. E. 1991. Metabolism of glyphosate in Sprague- Dawley rat: Tissue distribution, identifica- tion, and quantitation of glyphosate-derived materials following a single oral dose. Fundam. Appl. Toxicol. 17: 43-51. Centre de Toxicology du Quebec. (1988). Étude de l’Exposition Professionnelle des Travailleurs Exposés au Glyphosate. Report no. ER89-1110. Unpublished report. Quebec, Canada: Ministère de Energie et des Ressources, Centre Hospitalier de Université Laval. Chan, P. C., and Mahler, J. F. 1992. NTP tech- nical report on toxicity studies of glyphosate (CAS no. 1071-83-6) administered in dosed feed to F344/N rats and B6C3F1 mice. NIH publication 92-3135. NTP Toxicity Report Series no. 16. Raleigh, NC: National Toxicology Program. Committee for the Study of Toxicity. 2005. Enquiry into the referral of the Committee for the Study of Toxicity by the DGAL regarding the article “Differential effects of glyphosate and Roundup on human placental cells and aromatase.” Richard S., Moslemi, S., Sipahutar H., Benachour N., Seralani G.E., Environ. Health Perspect. 2005 (in the press; online 24 February 2005). French Ministry of Agriculture and Fish. Minutes of the meeting of 14 December 2005, pp. 90-98. Costa, L. G. 2008. Toxic effects of pesticides. In Casarett & Doull’s toxicology, The basic sci- ence of poisons, 7th ed., ed. C. D. Klaassen, 883-930. New York, NY: McGraw-Hill. Cowell, J. E., and Steinmetz, J. R. 1990a. Assessment of forest worker exposures to glyphosate during backpack foliar applica- tions of Roundup herbicide. Report no. MSL- 9656. Unpublished report. St. Loius, MO: Monsanto Company. Cowell, J. E., and Steinmetz, J. R. 1990b. Assessment of forestry nursery workers expo- sure to glyphosate during normal operations. Report no. MSL-9655. Unpublished report. St. Loius, MO: Monsanto Company. Curtis, K. M., Savitz, D. A., Weinberg, C. R., and Arbuckle, T. E. 1999. The effect of pesticide exposure on time to pregnancy. Epidemiology 10: 112-117. Curwin, B. D., Hein, M. J., Sanderson, W. T., Striley, C., Heederik, D., Kromhout, H., Reynolds, S. J., Alavanja, M. C. 2007a. Urinary pesticide concentrations among children, mothers and fathers living in farm and non-farm households in Iowa. Ann. Occup. Hyg. 51: 53-65. Curwin, B. D., Hein, M. J., Sanderson, W. T., Striley, C., Heederik, D., Kromhout, H., Reynolds, S. J., Alavanja, M. C. 2007b. Pesticide dose estimates for children of Iowa farmers and non-farmers. Environ. Res. 105: 307-15. Dallegrave, E., Mantese, F. D., Coelho, R. S., Pereira, J. D., Dalsenter, P. R., and Langeloh, A. 2003. The teratogenic potential of the herbicide glyphosate- Roundup in Wistar rats. Toxicol. Lett. 142: 45-52. Dallegrave, E., Mantese, F. D., Oliveira, R. T., Andrade, A. J. M., Dalsenter, P. R., and Langeloh, A. 2007. Pre- and postnatal toxic- ity of the commercial glyphosate formulation in Wistar rats. Arch. Toxicol. 81: 665-73. Daruich, J., Zirulnik, F., and Gimenez, M. S. 2001. Effect of the herbicide glyphosate on D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 92 A. L. WILLIAMS ET AL. enzymatic activity in pregnant rats and their fetuses. Environ. Res. 85: 226-31. Dawson, A. B. 1926. A note on the stain- ing of the skeleton of cleared specimens with alizarin red S. Biotechnic Histochem. 1: 123-24. Everett, C. 1997. Incidence and outcome of bleeding before the 20th week of preg- nancy: prospective study from general prac- tice. Br. Med. J. 315: 32-34. Farr, S. L., Cooper, G. S., Cai, J., Savitz, D. A., and Sandler, D. P. 2004. Pesticide use and menstrual cycle characteristics among premenopausal women in the Agricultural Health Study. Am. J. Epidemiol. 160: 1194-204. Franz, T. F. 1983. Kinetics of cutaneous drug penetration. Int. J. Dermatol. 22: 449-505. Franz, J., Mao, M., and Sikorski, J. 1997. Glyphosate: A unique global herbicide. ACS Monograph 189. Washington, DC: American Chemical Society. Garry, V. F., Harkins, M. E., Erickson, L. L., Long-Simpson, L. K., Holland, S. E., and Burroughs, B. L. 2002. Birth defects, season of conception, and sex of children born to pesticide applicators living in the Red River Valley of Minnesota, USA. Environ. Health Perspect. 110(suppl. 3): 441-49. Garry V. F., Schreinemachers, D., Harkins, M. E., and Griffith, J. 1996. Pesticide appli- ers, biocides, and birth defects in rural Minnesota. Environ. Health Perspect. 104: 394-99. Gasnier, C., Dumont, C., Benachour, N., Clair, E., Chagnon, C.-M., and Séralini, G. E. 2009. Glyphosate-based herbicides are toxic and endocrine disuptors in human cell lines. Toxicology 262: 184-91. Giesy J. P., Dobson S., and Solomon K. R. 2000. Ecotoxicological risk assessment for Roundup herbicide. Rev. Environ. Contam. Toxicol. 167: 35-120. Goodridge, A. G., Klautky, S. A., Fantozzi, D. A., Baillie, R. A., Hodnett, D. W., Chen, W., Thurmond, D. C., Xu, G., and Roncero, C. 1996. Nutritional and hormonal regu- lation of expression of the gene for malic enzyme. Prog. Nucleic Acid Res. Mol. Biol. 52: 89-122. Goodridge, A. G., Thurmond, D. C., Baillie, R. A., Hodnett, D. W., and Xu, G. 1998. Nutritional and hormonal regulation of the gene for malic enzyme. Z. Ernahrungswiss. 37(suppl. 1): 8-13. Granot, Z., Silverman, E., Friedlander, R., Melamed-Book, N., Eimerl, S., Timberg, R., Hales, K., Hales, D., Stocco, D., and Orly, J. 2002. The life cycle of the Steroidogenic Acute Regulatory (StAR) protein: From tran- scription through proteolysis. Endocr. Res. 28: 375-86. Greenlee, A. R., Arbuckle, T. E., and Chyou, P.-H. 2003. Risk factors for female infertility in an agricultural region. Epidemiology 14: 429-36. Hokanson, R., Fudge, R., Chowdhary, R., and Busbee, D. 2007. Alteration of estrogen- regulated gene expression in human cells induced by the agricultural and horticultural herbicide glyphosate. Human Exp. Toxicol. 26: 747-52. Holson, J. F. 1990. A developmental toxicity study of MON 0818 in rats. Unpublished study report. Ashland, OH: WIL Research Laboratories, Inc. Holson, J. F. 1991. A developmental toxic- ity study of AMPA in rats. Unpublished study report. Ashland, OH: WIL Research Laboratories, Inc. International Research and Devlopment Corporation. 1980a. Teratology study in rats. Unpublished study report. Mattawan, MI: International Research and Development Corporation. International Research and Devlopment Corporation. 1980b. Teratology study in rabbits. Unpublished study report. Mattawan, MI: International Research and Development Corporation. Jauhiainen, A., Räsänen, K., Sarantila, R., Nuutinen, J., and Kangas, J. 1991. Occupational exposure of forest workers to glyphosate during brush saw spraying work. Am. Ind. Hyg. Assoc. J. 52: 61-164. Kiely, T., Donaldson, D., and Grube, A. 2004. Pesticide industry sales and usage: 2000 and D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 93 2001 market estimates. Report. Washington, DC: Office of Pesticide Programs, U.S. Environmental Protection Agency. Kim, J. D., McCarter, R. J., and Yu, B. P. 1996. Influence of age, exercise, and dietary restriction on oxidative stress in rats. Aging (Milano) 8: 123-29. Kojima, H., Katsura, E., Takeuchi, S., Niiyami, K., and Kobayashi, K. 2004. Screening for estrogen and androgen receptor activi- ties in 200 pesticides by in vitro reporter gene assays using Chinese hamster ovary cells. Environ. Health Perspect. 112: 524-31. Knapp, J. F. 2007. A reproductive/ developmental toxicity screening study of MON 0818 in rats. Unpublished study report. Report no. WIL-50282. Ashland, OH: WIL Research Laboratories, Inc. Knapp, J. F. 2008. A combined 28-day repeated dose oral (dietary) toxicity study with the reproduction/developmental toxicity screen- ing test of MON 8109 and MON 0818 in rats. Unpublished study report. Report no. WIL-50337. Ashland, OH: WIL Research Laboratories, Inc. Kutzman, R. S., and DeSesso, J. M. 2003. A critique of ‘Embryonic Cell Cycle for Risk Assessment of Pesticides at the Molecular Level’ by Marc et al. (Environ. Chem. Lett. 1: 8-12 [2003]) Monsanto Agricultural Company, St. Louis, MO. Lamb, D. C., Kelly, D. E., Hanley, S. Z., Mehmood, Z., and Kelly, S. L. 1998. Glyphosate is an inhibitor of plant cytochrome P450: Functional expres- sion of Thlaspi arvensae cytochrome P45071B1/reductase fusion protein in Eschericia coli. Biochem. Biophys. Res. Commun. 244: 110-14. Larsen, S. B., Joffe, M., Bonde, J. P., and the ASCLEPIOS Study Group. 1998. Time to pregnancy and exposure to pesticides in Danish farmers. Occup. Environ. Med. 55: 278-83. Lavy, T. L., Cowell, J. E., Steinmetz, R., and Massey, J. H. 1992. Conifer seedling nurs- ery worker exposure to glyphosate. Arch. Environ. Contam. Toxicol. 22: 6-13. Lavy, T. L., Mattice, J. D., Massey, J. H., and Skulman, B. W. 1993. Measurements of year-long exposure to tree nursery work- ers using multiple pesticides. Arch. Environ. Contam. Toxicol. 24: 123-144. Levine, S. L., Han, Z., Liu, J., Farmer, D. R. and Papadopoulos V. 2007. Disrupting mito- chondrial function with surfactants inhibits MA-10 Leydig cell steroidogenesis. Cell. Biol. Toxicol. 23: 385-400. Lin, N., and Garry, V. F. 2000. In vitro stud- ies of cellular and molecular developmental toxicity of adjuvants, herbicides, and fungi- cides commonly used in Red River Valley, Minnesota. J. Toxicol. Environ. Health A 60: 423-39. Maibach, H. I. 1986. Irritation, sensitiza- tion, photoirritation and photosensitization assays with a glyphosate herbicide. Contact Dermatitis 15: 152-56. Mandel, S., Alexander, B. H., Baker, B. A., Acquavella, F., Chapman, P., and Honeycutt, R. 2005. Biomonitoring for farm families in the Farm Family Exposure Study. Scand. J. Work Environ. Health. 31(suppl. 1): 98-104. Marc, J., Bellé, R., Morales, J., Cormier, P., and Mulner-Lorillon, O. 2004b. Formulated glyphosate activates the DNA-response checkpoint of the cell cycle leading to the prevention of G2/M transition. Toxicol. Sci.82: 436-42. Marc, J., Le Breton, M., Cormier, P., Morales, J., Bellé, R., and Mulner-Lorillon, O. 2005. A glyphosate-based pesticide impinges on transcription. Toxicol. Appl. Pharmacol. 203: 1-8. Marc, J., Mulner-Lorillon, O., and Bellé, R. 2004a. Glyphosate-based pesticides affect cell cycle regulation. Biol. Cell 96: 245-49. Marc, J., Mulner-Lorillon, O., Boulben, S., Hureau, D., Durand, G., and Bellé, R. 2002. Pesticide Roundup provokes cell division dysfunction at the level of CDK1/Cyclin B activation. Chem. Res. Toxicol. 15: 326-31. Marc, J., Mulner-Lorillon, O., Durand, G., and Bellé, R. 2003. Embryonic cell cycle for risk assessment of pesticides at the molecular level. Environ. Chem. Lett. 1: 8-12. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 94 A. L. WILLIAMS ET AL. Martin, R. J., Beverly, J. L., Hausman, D. B., and Bellinger, L. L. 1990. Effect of liver denervation on compensatory changes in food intake, body composition and hepatic enzyme induction after food restriction in rats. J. Nutr. 120: 893-99. Martins, R. N., Stokes, G. B., and Masters, C. L. 1985. Regulation of the multiple molecu- lar forms of rat liver glucose 6-phosphate dehydrogenase by insulin and dietary restric- tion. Biochem. Biophys. Res. Commun. 127: 136-42. Martins, R. N., Stokes, G. B., and Masters, C. L. 1986. Regulation of liver and brain hexose monophosphate dehydrogenases by insulin and dietary intake in the female rat. Mol. Cell. Biochem. 70: 169-75. Medina, H. S. G., Lopata, M. E., and Bacila, M. 1994. The response of sea urchin egg embryogenesis towards the effect of some pesticides. Arq. Biol. Technol. 37: 895-906. Meulenberg, E. P. 2002. A new test to identify endocrine disruptors using sex hormone- binding globulins from human serum. Eur. J. Lipid Sci. Technol. 104: 131-36. Moxon, M. E. 2000. Glyphosate acid: Multigeneration reproduction toxicity study in rats. Alderley Park, Macclesfield, UK: Central Toxicology Laboratory. Report no. CTL/P/6332. Mura, C. V., Gong, X., Taylor, A., Villalobos- Molina, R., and Scrofano, M. M. 1996. Effects of calorie restriction and aging on the expression of antioxidant enzymes and ubiquitin in the liver of Emory mice. Mech. Ageing Dev. 91: 115-29. Nagy, I. Kurcz, M., Baranyai, P., and Meites, J. 1978. Activity alterations of metabolic enzymes in the anterior pituitary of female rats during acute and chronic starvation, as well as after refeeding. Experientia 34: 545-47. Nielsen, J. B., Sørensen, J. A., and Nielsen F. 2009. The usual suspects-Influence of physicochemical properties on lag time, skin deposition, and percutaneous penetration of nine model compounds. J. Toxicol. Environ. Health A 72: 315-23. Nielsen, J. B. 2010. Efficacy of skin wash on dermal absorption: An in vitro study on four model compounds of varying solubil- ity. Int. Arch Occup. Environ. Health 83: 683-90. Organisation for Economic and Cooperative Development. 2004. OECD guideline for the testing of chemicals. Skin absoprtion: In vitro method. No. 428. 13 April. Available at http://ec.europa.eu/food/plant/protection/ evaluation/guidance/wrkdoc20_rev_en.pdf Paganelli, A., Gnazzo, V., Acosta, H., López, S. L., and Carrasco, A. E. 2010. Glyphosate- based herbicides produce teratogenic effects on vertebrates by impairing retinoic acid signaling. Chem. Res. Toxicol. 23: 1586-95. Petit, F., Le Goff, P., Cravédi, J.-P., Valotaire, Y., and Pakdel, F. 1997. Two complemen- tary bioassays for screening the estrogenic potency of xenobiotics: Recombinant yeast for trout estrogen receptor and trout hep- atocyte cultures. J. Mol. Endocrinol. 19: 321-35. Rao, G., Xia, E., Nadakavukaren, M. J., and Richardson, A. 1990. Effect of dietary restric- tion on the age-dependent changes in the expression of antioxidant enzymes in rat liver. J. Nutr. 120: 602-09. Reyna, M. S. 1990. Two generation repro- duction feeding study with glyphosate in Sprague-Dawley rats. Study no. MSL-10387. Unpublished study report. St. Louis, MO: Monsanto Agricultural Company. Richard, S., Moslemi, S., Sipahutar, H., Benachour, N., and Seralini, G.-E. 2005. Differential effects of glyphosate and Roundup on human placental cells and aromatase. Environ. Health Perspect. 113: 716-20. Romano, R. M., Romano, M. A., Bernardi, M. M., Furtado, P. V., and Oliveira, C. A. 2010. Prepubertal exposure to commer- cial formulation of the herbicide glyphosate alters testosterone levels and testicular mor- phology. Arch. Toxicol. 84: 309-17. Rull, R. P., Ritz, B., and Shaw, G. M. 2006. Neural tube defects and maternal residential proximity to agricultural pesticide applica- tions. Am. J. Epidemiol. 163: 743-53. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 GLYPHOSATE-DEVELOPMENTAL AND REPRODUCTIVE REVIEW 95 Sachan, D. S., and Das, S. K. 1982. Alterations of NADPH-generating and drug-metabolizing enzymes by feed restriction in male rats. J. Nutr. 112: 2301-6. Sanin, L.-H., Carrasquilla, G, Solomon, K. R., Cole, D. C., and Marshall, K. R. 2009. Regional differences in time to pregnancy among fertile women from five Colombian regions with different use of glyphosate. J. Toxicol. Environ. Health A 72: 949-60. Sassoon, H. F., Watson, J., and Johnson, B. C. 1968. Diet-dependence of rat liver glucose 6-phosphate dehydrogenase levels. J. Nutr. 94: 52-56. Savitz, D. A., Arbuckle, T., Kaczor, D., and Curtis, K. M. 1997. Male pesticide exposure and pregnancy outcome. Am. J. Epidemiol. 146: 1025-36. Schroeder, R. E. 1981. A three generation reproduction study in rats with glyphosate. Unpublished study report. East Millstone, NJ: Biodynamics, Inc. Schroeder, R. E. 1982. A three-generation reproduction study with glyphosate in rats. Addendum to pathology report. Unpublished study report. East Millstone, NJ: Biodynamics, Inc. Shaw, G. M., Schaffer, D., Velie, E. M., Morland, K., and Harris, J. A. 1995. Periconceptional vitamin use, dietary folate, and the occurrence of neural tube defects. Epidemiology 6: 219-26. Shaw, G. M., Wasserman, C. R., O’Malley, C. D., Nelson, V., and Jackson, R. J. 1999. Maternal pesticide exposure from multiple sources and selected congenital anomalies. Epidemiology 10: 60-66. Solomon, K. R., Marshall, E. J. P., and Carrasquilla, G. 2009. Human health and environmental risks from the use of glyphosate formulations to control the production of coca in Colombia: Overview and conclusions. J. Toxicol. Environ. Health A 72: 914-20. U.S. Environmental Protection Agency. 1993. Re-registration eligibility decision (RED), Glyphosate. Report no. EPA-738-R-93- 014. Washington, DC: U.S. Environmental Protection Agency, Office of Pesticide Programs and Toxic Substances. U.S. Environmental Protection Agency. 2009a. Drinking water contaminants. Available at: www.epa.gov/safewater/contaminants/ index.html (updated 21 May 2009). U.S. Environmental Protection Agency. 2009b. Endocrine Disruptor Screening Program test guideline OPPTS 890.1200: Aromatase (human recombinant). EPA 740-C-09-004. October. Available at www.epa.gov/safewater/contaminants/index. html (updated 21 May 2009) Vogel, R., Hamprecht, B., and Wiesinger, H. 1998. Malic enzyme isoforms in astrocytes: Comparative study on activities in rat brain tissue and astroglia-rich primary cultures. Neurosci. Lett. 247: 123-26. Walsh, L., McCormick, C., Martin, C., and Stocco, D. 2000. Roundup inhibits steroidogenesis by disrupting steroidogenic acute regulatory (StAR) protein expres- sion. Environ. Health Perspect. 108: 769-76. Ward, R. J. 2009a. 450 g/L Glyphosate SL for- mulation (MON 79545) in vitro absorption of glyphosate through human epidermis. Draft report. 19 February. Report no. JV2083-REG. St. Loius, MO: Monsanto Company. Ward, R. J. 2009b. 360 g/L Glyphosate SL for- mulation (MON 52276) in vitro absorption of glyphosate through human epidermis. Draft report. Report no. JV2084-REG. St. Loius, MO: Monsanto Company. Ward, R. J. 2009c. 480 g/L Glyphosate SL for- mulation (MON 79351) in vitro absorption of glyphosate through human epidermis. Draft report. Report no. JV2085-REG. St. Loius, MO: Monsanto Company. Wester, R. C., Melendres, J., Sarason, R., McMaster, J., and Maibach, H. I. 1991. Glyphosate skin binding, absorption, resid- ual tissue distribution, and skin decon- tamination. Fundam. Appl. Toxicol. 16: 725-32. Wilcox, A. J., Baird, D. D., and Weinberg, C. R. 1999. Time of implantation of the conceptus and loss of pregnancy. N. Engl. J. Med. 340: 1796-99. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12 96 A. L. WILLIAMS ET AL. Williams, G. M., Kroes, R., and Munro, I. C. 2000. Safety evaluation and risk assessment of the herbicide Roundup and its active ingredient, glyphosate, for humans. Reg. Toxicol. Pharmacol. 31: 117-65. Wilson, J. G. 1965. Methods for admin- istering agents and detecting malforma- tions in experimental animals. In Teratology: Principles and techniques, eds. J. G. Wilson and J. Warkany, 262-277. Chicago, IL: University of Chicago Press. World Health Organization. 1994. Glyphosate. Environmental health criteria 159. Available at: http://www.inchem.org/documents/ehc/ ehc/ehc159.htm World Health Organization. 2002. Global assessment of the state-of-the-science of endocrine disruptors, WHO/PCS/EDC.02.2, eds. T. Damstra, S. Barlow, A. Bergman, R. Kavlock, and G. Van Der Kraak. Available at: http://www.who.int/ipcs/publications/ new_issues/endocrine_disruptors/en Xie, E., Rao, G., Van Remmen, H., Heydari, A. R., and Richardson, A. 1995. Activities of antioxidant enzymes in various tissues of male Fischer 344 rats are altered by food restriction. J. Nutr. 125: 195-201. Xie, L., Thrippleton, K., Irwin, M. S., Siemering, G. S., Mekebri, A., Crane, D., Berry, K., and Schlenk, D. 2005. Evaluation of estrogenic activities of aquatic herbicides and surfactants using a rainbow trout vitel- logenin assay. Toxicol. Sci. 87: 391-98. Yousef, M. I., Bertheussen, K., Ibrahim, H. Z., Helmi, S., Seehy, M. A., and Salem, M. H. 1996. A sensitive sperm-motility test for the assessment of cytotoxic effect of pesticides. J. Environ. Sci. Health B 31: 99-115. Yousef, M. I., Salem, M. H., Ibrahim, H. Z., Helmi, S., Seehy, M. A., and Bertheussen, K. 1995. Toxic effects of carbofuran and glyphosate on semen characteristics in rabbits. J. Environ. Sci. Health B 30: 513-34. D ow nl oa de d by [ Jo hn M . D eS es so ] at 0 7: 29 0 4 Ja nu ar y 20 12
