Evaluation of developmental toxicity studies of glyphosate with attention to cardiovascular development.

The herbicide glyphosate has undergone multiple safety tests for developmental toxicity in rats and rabbits. The European Commission's 2002 review of available glyphosate data discusses specific heart defects observed in several individual rabbit developmental toxicity studies, but describes the evidence for a potential causal relationship as equivocal. The present assessment was undertaken to analyze the current body of information generated from seven unpublished rabbit studies in order to determine if glyphosate poses a risk for cardiovascular malformations. In addition, the results of six unpublished developmental toxicity studies in rats were considered. Five of the seven rabbit studies (dose range: 10-500 mg/kg/day) were GLP- and testing guideline-compliant for the era in which the studies were performed; a sixth study predated testing and GLP guidelines, but generally adhered to these principles. The seventh study was judged inadequate. In each of the adequate studies, offspring effects occurred only at doses that also caused maternal toxicity. An integrated evaluation of the six adequate studies, using conservative assumptions, demonstrated that neither the overall malformation rate nor the incidence of cardiovascular malformations increased with dose up to the point where severe maternal toxicity was observed (generally ≥150 mg/kg/day). Random occurrences of cardiovascular malformations were observed across all dose groups (including controls) and did not exhibit a dose-response relationship. In the six rat studies (dose range: 30-3500 mg/kg/day), a low incidence of sporadic cardiovascular malformations was reported that was clearly not related to treatment. In summary, assessment of the entire body of the developmental toxicity data reviewed fails to support a potential risk for increased cardiovascular defects as a result of glyphosate exposure during pregnancy.

Introduction

Glyphosate, the active ingredient in popular herbicide formulations such as Roundup, AquaMaster and Vision branded products, is the most commonly used herbicide in the US (Grube, 2011). Specific usage statistics are not readily available for Europe, but are assumed to mirror those of the US. Glyphosate acts by targeting the enzyme enolpyruvylshikamate phosphate synthase in plants (Williams et al., 2012). Although this enzyme is important in the synthesis of several essential amino acids in plants, it is not found in animals. For this reason, glyphosate is considered to be generally safe to people and other mammals when used according to the manufacturer’s instructions. Nevertheless, due to its widespread use and the large number of glyphosate manufacturers, glyphosate has been subjected to numerous safety tests to protect health. In a monograph developed to support the European Commission’s 2002 review of glyphosate (BBA, 1998–2000; European Commission, 2002), the authors discuss specific heart defects observed in individual rabbit developmental toxicity studies of glyphosate, however they describe the evidence for a potential causal relationship as equivocal. Based on data selected from these studies, others have alleged there is evidence of teratogenicity and have called for a new risk assessment of glyphosate (Antoniou et al., 2012).

The present critical analysis assesses the glyphosate developmental toxicity database available to European regulatory agencies in order to determine if there is, in fact, a cause for concern for cardiovascular defects or other malformations. Rabbit and rat developmental toxicity studies on glyphosate conducted by member companies of the European Union (EU) Glyphosate Task Force were made available to the authors of this paper for the purpose of this analysis. These included seven developmental toxicity studies conducted in rabbits as well as six developmental toxicity studies conducted in rats. A PubMed search of the peer-reviewed literature through May 2012 was also conducted in an attempt to identify other studies of developmental glyphosate exposure and heart/cardiovascular malformations. No studies were found to be focused on cardiovascular defects as a result of in utero glyphosate treatment. A few published studies examined the effects on the fetal development of in utero exposure to glyphosate-based herbicide formulations (Dallegrave et al., 2003, 2007; Daruich et al., 2001); none of these studies, however, addressed visceral malformations. Therefore, the focus of the present analysis is on developmental toxicity studies of glyphosate that were conducted to fulfill regulatory requirements, particularly those in the rabbit. Each of the seven rabbit developmental toxicity studies has been critically evaluated with attention to whether the database as a whole is of sufficient quality to determine glyphosate’s teratogenic potential in rabbits, particularly for the cardiovascular system. Details of these analyses are found in the Appendix. The findings from six rat developmental toxicity studies conducted with glyphosate for regulatory purposes are also addressed, paying particular attention to heart and cardiovascular defects. Finally, the rabbit and rat data are briefly discussed in the context of the available epidemiological data for glyphosate.

Rabbit developmental toxicity database

A total of seven developmental toxicity studies of glyphosate have been conducted in the rabbit, the designs of which are summarized in Table 1. These studies, which are critically evaluated in the Appendix, involved testing in three different rabbit strains (New Zealand white, Japanese white and Dutch belted) and covered a wide range of glyphosate doses, from 10 to 500 mg/kg/day. This range includes doses that caused overt maternal toxicity (150 mg/kg/day and above); in some cases, the maternal toxicity observed was substantial. Two of these studies (Suresh, 1993; Tasker, 1980a) had insufficient numbers of fetuses available for assessment at the high dose (500 and 350 mg/kg/day, respectively).

Table 1.

Maternal and developmental NOAELs from six sufficient rabbit developmental toxicity studies of glyphosate.

Study	No. of animals per group	Exposure period	Doses (mg/kg/day)	Maternal NOAEL (mg/kg/day)	Offspring NOAEL (mg/kg/day)
Moxon (1995)	20	GD 7–19†	0, 100, 175, 300	100	175
Coles & Doleman (1996)	18	GD 7–19	0, 50, 200, 400	200	≥400
Brooker et al. (1991a)	16–20	GD 7–19	0, 50, 150, 450	50	150
Hojo (1995)	18	GD 7–19†	0, 10, 100, 300	100	≥300
Tasker et al. (1980a)	16–17	GD 6–27	0, 75, 175, 350	75	≥175
Suresh (1993)	15–26	GD 6–18	0, 20, 100, 500	100	≥100
Bhide & Patil (1989)	15	GD 6–18	0, 125, 250, 500	–‡	–‡

†Moxon (1995) designated the day of insemination as GD 1 and Hojo (1995) designated the day after insemination as GD 0. The exposure periods here have been adjusted to be comparable to the other studies which used GD 0 as the day of insemination.

‡Due to significant limitations in study design and data reporting, this study was considered inadequate for determining NOAELs.

The seven rabbit developmental toxicity studies vary considerably in their quality: the numbers of animals per dose group, the spacing of doses, the extent of documentation and detail provided and the specific types of data reported. Five of the studies stated that they followed good laboratory practices (GLP) specific to the time period in which they were conducted (Brooker et al., 1991a; Coles and Doleman, 1996; Hojo, 1995; Moxon, 1996; Suresh, 1993). Another study was conducted prior to the establishment of GLP requirements, but appears to have generally adhered to GLP principles (Tasker et al., 1980a). In the seventh study (Bhide & Patil, 1989), it is not clear to what extent GLP practices were followed, but it is unlikely that this study was fully GLP-compliant because the description of study results is extremely limited and inappropriate animals appear to have been included in the calculations for certain endpoints. All these studies were conducted according to developmental toxicity testing guideline requirements current at the time they were initiated and provided quality assurance audits.

As these studies were all done in different laboratories, there is considerable disparity across studies in the classification of various anomalies as major malformations, minor malformations or variations and in the terminology used to describe these findings. Further, three of the studies (Bhide & Patil, 1989; Hojo, 1995; Suresh, 1993) did not report anomalies by individual fetus. Therefore, for these studies, it is not possible to determine whether certain fetuses showed multiple anomalies or if anomalies occurred in combination. The study by Suresh (1993) also used some terminology that is not standard for heart defects in developmental toxicity studies (e.g. seal-shaped heart, dilated heart), which makes interpretation of the findings difficult. Certain cardiovascular changes reported in the Brooker et al. (1991a) study (e.g. retroesophageal right subclavian artery) are considered variations in other laboratories (Appendix), these are discussed in more detail below. Because of inappropriate methods and the poor reporting of data, the Bhide & Patil (1989) study was considered inadequate for assessing glyphosate’s potential for developmental toxicity in rabbits. The remaining six rabbit studies formed the basis for our analysis. While the individual studies may fall short of current guidelines (mainly because the desired number of rabbits per group has increased and the exposure period has been extended beyond GD18), these shortcomings are overcome when one considers the overall database. More specifically, the exposure period in each of these studies extends well before and after the period of organogenesis for the cardiovascular system. Additionally, the studies cover a broad and well-distributed range of 15 different glyphosate exposures ranging from 10 to 500 mg/kg/day. Finally, the combined database from these studies includes evaluation of 347 total litters (99 controls and 247 treated) and 2990 fetuses (834 controls and 2156 treated). Based on these elements, the overall database of six adequate rabbit developmental studies is considered to be robust for the purposes of risk assessment.

To address whether the six adequate studies exhibited evidence of selective offspring sensitivity to glyphosate treatment in utero, the no observed adverse effect levels (NOAELs) for maternal toxicity and developmental effects were determined (Table 1). Maternal toxicity was most commonly evidenced in the rabbit studies by diarrhea and reduced food intake, which generally occurred at doses of 150 mg/kg/day or higher. Additionally, maternal weight loss and deaths generally occurred at the highest doses. Table 1 also shows that offspring effects due to glyphosate, when observed in a particular rabbit developmental toxicity study, always occurred at the same dose or doses as those associated with maternal toxicity. This does not mean that injury to the fetus necessarily occurred as a direct result of maternal toxicity, but rather, when exposures to glyphosate were kept below the doses that cause maternal toxicity, the developing offspring did not exhibit any adverse effects. Therefore, selective offspring sensitivity to glyphosate is not apparent from these studies.

Post-implantation loss was quite variable across studies. Four of the six adequate studies (Hojo, 1995; Moxon, 1996; Suresh, 1993; Tasker, 1980a) reported no statistically significant increase in post-implantation loss in three different strains of rabbits at exposure levels as high as 500 mg/kg/day. In comparison, Coles & Doleman (1996) reported an increase in post-implantation loss at 200 mg/kg/day, but not at 400 mg/kg/day; consequently, a dose–response pattern was not established in this study. Brooker et al. (1991a) reported increased post-implantation loss at doses of 50 mg/kg/day and above (mean = 19.5 ± 19.8%, 15.3 ± 17.2% and 21.0 ± 11.8% for the 50, 150 and 450 mg/kg/day dose groups, respectively), but noted that post-implantation loss in the concurrent control group (5.7 ± 7.2%) was lower than in historical controls (mean: 12.9%; range: 6.5–17.5%), while post-implantation loss in treated litters was within or slightly higher than the historical control range. Post-implantation loss has a high degree of variability as demonstrated by the standard deviations around this endpoint in the six studies reviewed. This variability is common in the rabbit. Other historical control databases have reported mean percent post-implantation loss in the rabbit of 8.1% (range: 2.8–17.7%) and 9.1% (range: 0.6–23.4%) (Holson et al., 2006 and MARTA, 1997, respectively). Consequently, without a clear dose–response pattern established across the six studies reviewed, it is unlikely that these findings are biologically significant.

As previously noted, the rabbit developmental toxicity data for glyphosate have been previously described as equivocal with regard to cardiovascular defects (BBA, 1998–2000; European Commission, 2002). To address this issue, data were extracted from each study for malformations and variations (Appendix). Two of the studies (Brooker et al., 1991a; Suresh, 1993) suggested a possible association of cardiovascular anomalies with treatment, but the data were not clear-cut; these are discussed in more detail in the Appendix. In addition, two studies (Hojo, 1995; Moxon, 1996) reported an increase in skeletal defects at the high dose of 300 mg/kg/day. These anomalies appeared to be the result of reduced ossification, which is likely related to delayed development (evidenced by reduced fetal body weights observed at the high dose), or were not clearly dose-related. Based on this information and our evaluation of the combined data, we concluded that glyphosate treatment was not associated with an increase in malformations in rabbits. The remaining discussion focuses on cardiovascular defects only.

Examination of the data from the six rabbit studies showed a variety of malformations of the heart and great vessels. These included: dilated aorta/narrow pulmonary artery; narrow aorta/dilated pulmonary artery; hypoplasia of the pulmonary artery; interventricular (IV) septal defect; cardiomegaly; single ventricle, thickened ventricle walls; dilated ventricle; retro-esophageal right subclavian artery; interrupted aorta; right subclavian artery arising from aortic arch; “seal-shaped” heart. If glyphosate treatment was associated with congenital heart defects and malformation of the great vessels in rabbits, then the prevalence of these defects would be anticipated to increase with dose and the overall malformation rate would also be anticipated to increase. However, as can be seen from the malformation incidence tables in the Appendix, cardiovascular malformations generally occurred in the rabbit studies at a low incidence across all dose groups. Further, in most studies, they did not exhibit a positive dose–response, and oftentimes, clusters of malformations occurred in the same fetuses.

In order to further discern whether there might be an association between exposure of rabbits to glyphosate and cardiovascular malformations, the following conservative assumptions were made so that the malformation data from the six adequate studies could be combined. First, all three rabbit strains (Japanese white, New Zealand white and Dutch belted) were assumed to be equally sensitive to glyphosate. Second, small differences in treatment duration across studies were assumed not to affect the incidence of cardiovascular malformations because all treatment paradigms covered the critical period of heart and great vessel development (i.e. GD 8–17; DeSesso, 2012). Third, cardiovascular malformations were categorized depending on the type of cardiovascular defect and what is known about the underlying morphogenetic processes. For instance, several defects are related to development of the aorticopulmonary septum and are grouped together. As an example, Brooker et al. (1991a) reported that many fetuses with IV septal defects exhibited other cardiovascular defects that included enlarged aorta/stenotic pulmonary artery or the converse (stenotic aorta/enlarged pulmonary artery). During formation of the outflow tract from the ventricles, neural crest cells migrate from the hindbrain region into the truncus arteriosus where they contribute to and direct the growth of the aorticopulmonary septum (Hutson & Kirby, 2003; Kirby et al., 1983; Sadler, 2011). The aorticopulmonary (spiral) septum (Figure 1) grows as a pair of ridges that divide the truncus arteriosus into equally sized halves: the aorta and the pulmonary artery (DeSesso & Venkat, 2010). At its inferior end, the aorticopulmonary septum forms the upper portion (membranous portion) of the IV septum. Consequently, malformations relating to a disproportionately sized aorta and pulmonary septum, as well as IV septal defects of the upper region, are all related to displacement of the developing aorticopulmonary septum (DeSesso & Venkat, 2010).

Figure 1.

Division of the outflow tract by the aorticopulmonary (spiral) septum. In the top diagram, the aorticopulmonary septum is forming by the growth and merging of the conotruncal ridges in the walls of the outflow tract. This process divides the outflow tract into the atrioventricular canals (precursors of the aorta and pulmonary artery). In the lower diagram, the spiral septum has completed the separation of the outflow tract into the equally sized aorta (for systemic circulation) and pulmonary artery (for the pulmonary circulation). The most inferior part of the spiral septum will contribute to the upper membranous portion of the IV septum. (Modified from DeSesso & Venkat, 2010).

Based on this information, those cardiac defects that involved perturbations of aorticopulmonary septum development were combined based on the premise that glyphosate might cause all or any of these defects by acting on a single developmental process. Data from all numerically similar dose groups (e.g. data from all three studies that treated rabbits at 100 mg/kg/day) were combined into a single entry.

Evaluation of the resulting tabulation (Table 2) shows that there was no increase in cardiovascular malformations at doses that were not overtly toxic to the pregnant rabbits (i.e. generally at doses over 150 mg/kg/day). The two most commonly observed malformations involved the aorticopulmonary septum and dilated heart The incidence of aorticopulmonary septum-related defects in the combined control groups was 1/770 (0.1%); in the combined glyphosate-treated groups the incidence was 6/1939 (0.3%). More than half of these affected fetuses were found in litters exposed to one of the highest doses (450 mg/kg/day). Doses of 150 mg/kg/day and above were generally associated with maternal toxicity, including severe weight loss and death. If doses of 300 mg/kg/day and above are not considered because of the confounding maternal toxicity issues, then the incidence of the defects in glyphosate-treated animals is 2/1388 (0.1%). Thus, these data show that the overall incidence of aorticopulmonary septum-related defects in offspring from mothers exposed to glyphosate at doses below those that cause severe maternal toxicity is similar to that seen in non-exposed rabbits.

Table 2.

Combined and grouped (number and percentage) cardiovascular malformations from six rabbit developmental toxicity studies.

Dose (mg/kg/day)	0	10	20	50	75	100	150	175	200	300	350	400	450	500
Total number of fetuses evaluated at each dose	770	130	78	261	114	374	112	200	119	256	38	134	95	28
Defects related to displaced aorticopulmonary (spiral) septum including ventricular septal defects	1B (0.1%)					1H (0.3%)	1B (0.9%)						4B (5.0%)
Dilated heart			4S (5.1%)			4S (1.1%)								2S (7.1%)
Dilated ventricles	1S (0.1%)					1S (0.2%)								1S (3.6%)
Cardiomegaly						1S (0.2%)
Single heart ventricle, thickened ventricle walls						1M (0.2%)				1M (0.4%)
Retroesophageal right subclavian artery							3B (2.7%)						2B (2.1%)
“Seal-shaped” heart	1S (0.1%)					1S (0.2%)
Acephalic animal with heart defects				1B (0.4%)					1C (0.8%)
Cebocephalic animal with heart defects	1M (0.1%)

B = Brooker et al. (1991a); C = Coles & Doleman (1996); H = Hojo (1995); M = Moxon (1996); S = Suresh (1993).

The other prevalent cardiovascular malformation reported was dilated heart. All observations of this finding occurred in a single study (Suresh, 1993). There was also one case of cardiomegaly at 100 mg/kg/day in the same study. None of the other five adequate studies reported dilated hearts or cardiomegaly. Furthermore, neither the criteria used to diagnose dilated heart nor measurements of the hearts were provided in the study report, so it is not possible to directly compare the dilated heart findings to the hearts of the more than 2500 fetuses in the other studies.

Finally, an examination of the overall rate of cardiac malformations across the six studies did not support a dose–response correlation with glyphosate exposure. Based on this analysis, it appears that prenatal glyphosate exposure is not associated with increased cardiovascular defects in rabbits.

Rat developmental toxicity database

The six developmental toxicity studies of glyphosate conducted in the rat are discussed in the Appendix and summarized in Table 3. These studies involved testing in two different rat strains (Wistar and Sprague–Dawley) and covered a wide range of glyphosate doses up to 3500 mg/kg/day, which is well above the current limit dose for toxicity studies of 1000 mg/kg/day. With the exception of Tasker et al. (1980b), all studies conformed to internationally accepted general principles of GLPs and were conducted according to OECD 414 (1981) and US EPA 83-3 guideline requirements. The study by Tasker et al. (1980b) predated the establishment of US EPA and OECD guidelines, but it received quality assurance audits by the testing facility and appeared to be well-conducted and essentially guideline-compliant. As with the rabbit studies, the rat developmental toxicity studies of glyphosate varied in the numbers of animals per dose group, the spacing of doses, the extent of documentation and detail provided, and the specific types of data reported. Nevertheless, for the purposes of this evaluation, all six rat studies were considered adequate for assessing the developmental toxicity potential of glyphosate.

Table 3.

Maternal and developmental NOAELs from six sufficient rat developmental toxicity studies of glyphosate.

Study	No. of animals per group	Exposure period	Doses (mg/kg/day)	Maternal NOAEL (mg/kg/day)	Offspring NOAEL(mg/kg/day)
Moxon (2002)	22–24	GD 6–15†	0, 250, 500, 1000	≥1000	≥1000
Wood (1996)	22–25	GD 6–15	0, 100, 500, 1000	≥1000	≥1000
Hatakenaka (1995)	22–24	GD 6–15	0, 30, 300, 1000	300	≥1000
Brooker et al. (1991a)	23–25	GD 6–15	0, 300, 1000, 3500	1000	1000
Suresh (1991)	20–30	GD 6–15	0, 1000	≥1000	≥1000
Tasker et al. (1980b)	20–23	GD 6–19	0, 300, 1000, 3500	1000	1000

†Moxon (1995) designated the day of finding sperm as GD 1. The exposure period here has been adjusted to be comparable to the other studies which used GD 0 as the day of insemination.

The NOAELs for maternal toxicity and developmental effects as assessed for the six rat developmental toxicity studies are shown in Table 3. Maternal body weight was not affected in any of the studies at exposure levels lower than 3500 mg/kg/day. Further, there were no dose-related effects on intrauterine parameters at doses of 1000 mg/kg/day and below. Maternal NOAELs were determined to be ≥1000 mg/kg/day for all studies except Hatakenaka (1991) (Table 3), which reported loose stools in a few dams at that exposure. No treatment-related effects were observed in the offspring at doses of 1000 mg/kg/day and below. Consequently, the offspring NOAELs for these studies were ≥1000 mg/kg/day and equal to or greater than the maternal NOAELs in each study (Table 3). Further, no treatment-related effects of glyphosate on structural development of the offspring were observed (Table A10). Generally, malformations (including cardiovascular malformations) were limited to 1–3 fetuses in 1–2 litters in the exposed groups and occurred at incidences as low as or lower than those in the control group. Overall, the rat developmental toxicity studies do not show any evidence of cardiovascular or other types of malformations as a result of glyphosate exposure at doses of up to 3500 mg/kg/day.

Table A10.

Maternal and fetal outcome data from the developmental toxicity studies of glyphosate in rats.

Strain (#/group)	Duration of Treatment	Dose (mg/kg/day)	No. gravid females	No. maternal deaths	No. litters examined	Mean % post- implantation loss	Mean No. live fetuses	Mean fetal body wts (gms)	No. malformed fetuses (litters)	Cardiovascular malformations	Maternal toxicity	Ref.
Wistar (24)	GD 6–15†	0	22	0	22	9.9 ± 15.5	12.9 ± 2.4	4.86 ± 0.29	1 (1)	None	None	Moxon (2002)
		250	24	0	24	4.0 ± 5.1	12.4 ± 3.4	5.02 ± 0.33	1 (1)	None	None
		500	23	0	23	7.8 ± 10.8	13.1 ± 2.7	4.95 ± 0.29	1 (1)	None	None
		1000	24	0	23	5.8 ± 8.3	12.9 ± 2.9	4.96 ± 0.27	2 (2)	None	None
Sprague–Dawley (25)	GD 6–15	0	23	0	23	4.9 ± 5.6	14.1 ± 3.3	3.81 ± 0.32	3 (3)	One IV septal defect and persistent truncus arteriosis, 1 retro- esophageal right-sided aortic arch	None	Wood (1996)
		100	24	0	24	4.4 ± 4.7	13.8 ± 2.2	3.99 ± 0.47	1 (1)	One IV septal defect	None
		500	22	0	22	6.1 ± 7.0	14.0 ± 1.8	3.76 ± 0.29	0	None	None
		1000	25	0	25	5.2 ± 6.8	14.0 ± 3.1	3.79 ± 0.40	0	None	None
Sprague–Dawley Crj:CD (24)	GD 6–15	0	23	0	23	7.0 + 6.1	13.7 ± 4.1	M: 3.6 ± 0.4 F: 3.3 ± 0.3	2 (1)	None	None	Hatakenaka (1995)‡
		30	24	0	24	6.8 ± 7.8	15.0 ± 2.1	M: 3.6 ± 0.2 F: 3.4 ± 0.3	1 (1)	None	None
		300	24	0	24	7.4 ± 8.0	14.9 ± 2.8	M: 3.5 ± 0.4 F: 3.4 ± 0.4	3 (2)	One right aortic arch, 1 IV septal defect	None
		1000	22	0	22	8.4 ± 9.1	15.4 ± 2.1	M: 3.6 ± 0.2 F: 3.4 ± 0.2	5 (2)	One IV septal defect	Loose stool
Sprague–Dawley (25)	GD 6–15	0	23	0	23	6.1	13.7	3.96	1 (1)	None¶	None	Brooker et al. (1991b)§
		300	23	0	23	7.3	12.7	3.90	2 (2)	None	None
		1000	25	0	25	5.7	13.2	3.89	1 (1)	One IV septal defect	None
		3500	25	3	22	3.6	13.1	3.71++	3 (2)	One IV septal defect	Salivation, loose feces, noisy respiration, wet coats, gasping
Wistar	GD 6–15	0	30	0	30	8	8.7	3.6 ± 0.4	External/visceral 5 (5)|| Skeletal 17 (8) #	None	None	Suresh (1991)
		1000	20	0	20	11	7.9	3.7 ± 0.3	External/VIsceral 0 Skeletal 10 (6)#	None	None
Sprague–Dawley COBS CD rats (25)	GD 6–19	0	22	0	22	4.2 ± 5.7	14.4 ± 1.3	3.5 ± 0.2	3 (3)	None	None	Tasker et al. (1980b)$
		300	20	0	20	1.4 ± 3.5	11.9 ± 4.4*	3.7 ± 0.7	0	None	None
		1000	21	0	21	3.1 ± 5.6	14.3 ± 2.1	3.6 ± 0.2	0	None	None
		3500	23	6	16	14.3 ± 24.0⊥	11.5 ± 4.1*	3.2 ± 0.3**	10 (3)$	None	6/25 deaths; various signs of clinical toxicity; decreased weight gain due to weight loss on GD 6–9

†Moxon (2002) designated the day of finding sperm as GD1. The exposure period listed here was adjusted using GD 0 as the day of finding sperm.

‡Hatakenaka (1995) did not report a combined mean fetal weight, but rather reported the mean fetal weight for males (M) and females (F) separately. Individual animal data were not available to calculate the combined mean fetal weight. Mean post-implantation loss also was not reported but was calculated by the present authors based on data provided in the study report.

¶One small IV septal defect was considered a variation by the authors.

§Brooker et al. (1991b) did not provide standard deviation values for mean post-implantation loss, mean number of live fetuses, or mean fetal body weights.

||Undescended testis and unascended kidneys were considered minor malformations by the authors but are included here.

#Several bilobed vertebral centra and delayed ossification of various bones were reported as major malformations, but none fit the author’s definition of a major malformation. Individual fetal data were incompletely reported, so it is difficult to determine which type of defects which fetus and litter. The number of fetuses (litters) given here is taken from Table A9 in Suresh (1991).

⊥Post-implantation loss percentages and standard deviations calculated from individual animal data in Tasker et al. (1980b); statistical significance was not calculated.

$Includes six fetuses in one litter with a syndrome of bent tail, open eyelids, missing kidneys and ureters, and various skeletal defects and three fetuses in another litter with dwarfism. All malformations were seen in the historical controls.

+ + p < 0.01, Kruskal–Wallis followed by distribution-free Williams’ test; litter was the statistical unit.

*p < 0.05, ANOVA followed by Dunnett’s test; statistical unit was not specified.

**p < 0.01, ANOVA followed by Dunnett’s test; statistical unit was not specified.

Discussion and conclusions

The 13 developmental toxicity studies summarized above and discussed in detail in the Appendix have been submitted to regulatory agencies in support of the registration of glyphosate. Analyses by the regulatory agencies have not supported the claim that glyphosate causes cardiovascular defects or other developmental effects (BBA, 1998–2000; EPA, 1993; European Commission, 2002). At the time of the US EPA’s assessment, only the studies by Tasker et al. (1980a,b) were available for evaluation. The European Commission’s review (European Commission, 2002), however, included the examination of four of the rabbit studies (Bhide & Patil, 1989; Brooker et al., 1991a; Suresh, 1993; Tasker et al., 1980a) and three of the rat studies (Brooker et al., 1991b; Suresh 1991; Tasker et al., 1980b) discussed herein. In a related monograph (BBA, 1998–2000), the results from two of the rabbit studies reviewed by the European Commission were characterized as equivocal for cardiovascular developmental effects. None of the three rabbit developmental toxicity studies that were not evaluated by the European Commission (Coles & Doleman, 1996; Hojo, 1995) showed a potential for cardiovascular defects.

Based on our assumptions underlying the integrated assessment of data across studies (equal strain sensitivity, insignificant differences in timing of exposure and shared morphogenetic processes of certain defects), the overall conclusion of our analysis of the potential for glyphosate to cause malformations, and cardiovascular defects in particular, is that there is no increased risk at the levels of exposure below those that caused maternal toxicity. This conclusion is in agreement with that of regulatory agency reviews as well as the limited data available from epidemiology studies showing no increased risk of congenital defects with exposure (Bell et al., 2001a,b,c; Garry et al., 2002; Rull et al., 2006; reviewed in Williams et al., 2012). It should be noted, however, that these studies investigated exposures to several pesticides and were not specific to glyphosate. More recently, a detailed review of epidemiology studies of glyphosate and non-cancer endpoints found no evidence of a causal relationship between glyphosate exposures and malformations (Mink et al., 2011). Finally, a review of the available biomonitoring data demonstrates that human exposure as a result of normal glyphosate application practices is extremely low, often below the limits of analytical detection (Williams et al., 2012). In conclusion, this analysis of the developmental toxicity data available for glyphosate exposure confirms that there is no evidence of an increased risk of cardiovascular defects as a result of glyphosate exposure.

Table A1.

Maternal and fetal outcome data for New Zealand white rabbits treated with glyphosate on gestational days 7–19† (Moxon 1996).

	0 mg/kg/day	100 mg/kg/day	175 mg/kg/day	300 mg/kg/day
Maternal data
No. animals on study	20	20	20	20
No. non-gravid	2	0	1	1
No. gravid does dead or sacrificed in extremis	1	2	2	2
No. that aborted	1	2	1	2
Embryo/fetal data
Total No. litters examined‡	17	18	17	17
Mean No. corpora lutea	10.8 ± 2.2	11.0 ± 1.6	11.1 ± 1.3	11.2 ± 1.4
Mean No. implantations	9.65 ± 2.06	9.00 ± 1.78	9.12 ± 2.50	9.82 ± 1.88
Mean % pre-implantation loss	10.7 ± 11.0	18.2 ± 11.1	18.1 ± 20.8	12.8 ± 11.9
Mean No. embryo/fetal death	NR	NR	NR	NR
Mean No. viable fetuses	8.41 ± 1.80	8.17 ± 2.20	7.94 ± 2.19	8.47 ± 2.32
Mean % post-implantation loss	11.7 ± 12.0	9.5 ± 16.7	12.1 ± 9.7	13.6 ± 16.6
Mean fetal body weight (g)	44.4 ± 4.3	43.3 ± 3.9	43.2 ± 5.7	40.7 ± 7.8*
Total fetuses (litters) with malformations
Major external/visceral	2 (2)	1 (1)	0 (0)	2 (2)
Minor external/visceral	12 (8)	7 (5)	9 (8)	11 (7)
Major skeletal	3 (2)	0 (0)	0 (0)	1 (1)
Minor skeletal	58 (16)	82 (18)**	59 (16)	79 (17)**
Total fetuses (litters) with variations
External/visceral	0	0	0	0
Cardiovascular	1 (1)	1 (1)	0	1 (1)
Skeletal	119 (17)	129 (18)	116 (17)	132 (17)**

NR = Not reported.

†Moxon (1995) designated the day of insemination as GD 1. The exposure period here has been adjusted to be comparable to the other studies which used GD 0 as the day of insemination. See text for details.

‡Includes litters that were aborted in the analysis.

*p < 0.05, ANOVA; litter is statistical unit.

**p < 0.05, Fisher’s exact test.

Table A2.

Maternal and fetal outcome data for New Zealand white rabbits treated with glyphosate on gestational days 7–19 (Coles & Doleman, 1996).

	0 mg/kg/day	50 mg/kg/day	200 mg/kg/day	400 mg/kg/day
Maternal data
No. animals on study	18	18	18	18
No. non-gravid	3	0	2	1
No. gravid does dead or sacrificed in extremis	1	0	1	2†
No. that aborted	0	0	0	0
Embryo/fetal data
Total No. litters examined	14	18	15	15
Mean No. corpora lutea	10.9 ± 2.2	10.5 ± 2.4	10.7 ± 2.1	11.5 ± 1.8
Mean No. implantations	9.5 ± 2.5	9.1 ± 2.3	8.9 ± 2.5	10.3 ± 2.3
Mean % pre-implantation loss	12.5 ± 18.2	13.6 ± 9.4	16.4 ± 15.5	9.3 ± 12.5
Mean No. embryo/fetal death	0.36 ± 0.63	0.33 ± 0.77	1.00 ± 1.00*	1.40 ± 2.35
Mean No. viable fetuses	9.1 ± 2.5	8.7 ± 2.4	7.9 ± 2.5	8.9 ± 2.6
Mean % post-implantation loss	3.7 ± 6.5	3.6 ± 8.5	11.5 ± 11.4*	12.1 ± 18.6
Mean fetal body weight (g)	41.5 ± 5.5	39.4 ± 5.6	41.7 ± 4.5	38.2 ± 5.2
Total fetuses (litters) with malformations	1 (1)	3 (2)	2 (2)	1 (1)
Total fetuses (litters) with cardiovascular malformations	0	0	1 (1)	0
Total fetuses (litters) with variations‡	41 (13)	50 (17)	39 (15)	51 (14)

†At least one of these deaths/sacrifices at 400 mg/kg/day was likely treatment-related.

‡Fetuses with both malformations and variations are included in the malformations tally; fetuses with only variations are captured here.

*p < 0.05, Kruskal–Wallis followed by the Mann–Whitney U test; litter was the statistical unit.

Table A3.

Maternal and fetal outcome data for New Zealand white rabbits treated with glyphosate on gestational days 7–19 (Brooker et al., 1991a).

	0 mg/kg/day	50 mg/kg/day	150 mg/kg/day	450 mg/kg/day
Maternal data
No. animals on study	19	19	16	20
No. excluded from study	1	0	0	1
No. non-gravid	0	6	1	5
No. gravid does dead or sacrificed in extremis	0	0	0	1
No. that aborted	0	1	0	0
Embryo/fetal data
Total No. litters examined	18	12†	15‡	13
Mean No. corpora lutea¶	11.5	12.4	11.7	11.3
Mean No. implantations¶	9.7	10.5	9.0	9.2
Mean % pre-implantation loss¶	14.6	15.4	23.4	18.8
Mean No. embryo/fetal death¶	0.6	1.8*	1.5*	1.8**
Mean No. viable fetuses¶	9.1	8.7	7.5	7.3
Mean % post-implantation loss§	5.7 ± 7.2	19.5 ± 19.8*	15.3 ± 17.2*	21.0 ± 11.8**
Mean fetal body weight (gms)¶	43.9	43.3	44.0	44.5
Total fetuses (litters) with malformations	3 (3)	3 (3)	5 (3)	6 (5)
Total fetuses (litters) with cardiovascular malformations||	1 (1)	1 (1)	4 (3)	5 (4)
Total fetuses (litters) with variations “anomalies”	29 (13)	26 (9)	26 (11)	16 (10)

†Analysis does not includes the one litter that was aborted at this dose.

‡Includes one female which aborted one embryonic death – referred to as “partial abortion”.

¶Standard deviation was not provided.

§Standard deviation values calculated from individual animal data in Brooker et al. (1991a).

||Exclusion of retroesophageal right subclavian artery reduces the numbers to 1 (1), 1 (1), 1 (1) and 4 (4) for 0, 50, 150 and 450 mg/kg/day, respectively.

*p < 0.05; **p < 0.01., Kruskal–Wallis test followed by non-parametric equivalent of Williams’ test; litter was the statistical unit.

Table A4.

Types and incidence of malformations by individual fetus (Brooker et al., 1991a).

	0 mg/kg/day	50 mg/kg/day	150 mg/kg/day	450 mg/kg/day
No. fetuses examined	163	104	112	95
Narrow ascending aorta, dorsally displaced pulmonary trunk, IV septal defect	1		1	1
Dilated ascending aorta/aortic arch, narrow pulmonary trunk; IV septal defect with enlarged left, reduced right ventricle				2
Retroesophageal right subclavian artery† (1 fetus at 150 mg/kg/day also had forelimb flexure; 1 fetus at 450 mg/kg/day with IV septal defect)			3	2
Acephaly; single dilated arterial trunk and carotid artery; right-sided descending aorta; IV septal defect, forelimb flexure and hindlimb brachydactyly		1
Sacral meningocoele occulta with slightly flattened cranium and minimal protrusion in occipital region	1
Bilateral small eye (areas of retinal folding and dysplasia)		1
Hydrocephaly and cebocephaly with fused and reduced nasals and premaxillae, fused nares, absent upper incisors			1
Cleft palate; forelimb flexure and brachydactyly				1
Reduced and fused thoracic vertebral arches with absent centrum; connected, branched and absent ribs	1
Spina bifida with lumbar kyphosis and flattened cranium; malrotated hindlimb		1

†Retroesophageal right subclavian artery is considered a variation by other laboratories. Removing this endpoint as a malformation would reduce the number of fetuses in this group to one fetus with forelimb flexure at 150 mg/kg/day and one fetus with IV septal defect at 450 mg/kg/day.

Table A5.

Maternal and fetal outcome data for Japanese white rabbits treated with glyphosate on gestational days 7–19† (Hojo, 1995).

	0 mg/kg/day	10 mg/kg/day	100 mg/kg/day	300 mg/kg/day
Maternal data
No. animals on study	18	18	18	18
No. non-gravid	0	0	0	0
No. gravid does dead or sacrificed in extremis	0	0	0	1
No. that aborted	0	2	0	2
No. with only resorptions	0	1	2	1
Embryo/fetal data
Total No. litters examined‡	18	15	16	14
Mean No. corpora lutea	10.2 ± 2.0	11.7 ± 2.2	12.1 ± 2.0	10.1 ± 2.3
Mean No. implantations	8.5 ± 2.8	9.8 ± 2.9	10.4 ± 2.9	8.6 ± 3.3
Mean % pre-implantation loss¶	17.8 ± 22.4	16.6 ± 17.0	15.2 ± 18.0	14.6 ± 25.2
Mean No. embryo/fetal death¶	0.7	1.1	1.0	0.6
Mean No. viable fetuses	7.8 ± 2.4	8.7 ± 3.2	9.4 ± 2.7	8.0 ± 3.2
Mean % post-implantation loss§	7.1 ± 8.8	13.8 ± 14.1	8.7 ± 10.5	6.5 ± 9.8
Mean fetal body weight (g) MALES	35.8 ± 8.1	37.3 ± 5.4	36.7 ± 3.3	36.2 ± 5.4
Mean fetal body weight (g) females	35.7 ± 6.7	36.1 ± 5.1	36.0 ± 3.9	34.9 ± 4.4
Malformations and variations
Total # litters (%) with malformations	1 (5.6)	3 (20.0)	3 (18.8)	5* (35.7)
Total # litters (%) with variations	16 (88.9)	14 (93.3)	16 (100.0)	8* (57.1)
Total # Fetuses (%) with malformations
External	0 (0.0)	0 (0.0)	2 (1.3)	0 (0.0)
Visceral	0 (0.0)	1 (0.8)	3 (2.0)	0 (0.0)
Cardiovascular	0	0	1 (1)	0
Skeletal	1 (0.7)	4 (3.1)	6 (4.0)	5 (5.4)
Total # Fetuses (%) with variations
Visceral	4 (2.9)	5 (3.8)	5 (3.3)	1 (0.9)
Skeletal	40 (28.6)	32 (24.6)	61* (40.7)	31 (27.7)

†Day of insemination adjusted to GD0 for comparison with other studies. See text for details.

‡Analysis does not include the litters that were aborted.

¶Mean and standard deviations not reported. Calculated from individual animal data in Hojo (1995).

§Standard deviations calculated from individual animal data in Hojo (1995).

*p < 0.05, Fisher’s exact test; litter is the statistical unit.

Table A6.

Maternal and fetal outcome data for Dutch belted rabbits treated with glyphosate on gestational days 6–27 (Tasker et al., 1980a).

	0 mg/kg/day	75 mg/kg/day	175 mg/kg/day	350 mg/kg/day
Maternal data
No. animals on study	16	16	16	17
No. non-gravid	2	0	2	0
No. gravid does dead or sacrificed in extremis	0	1	2	10
No. that aborted	2	0	1	1
Embryo/fetal data
Total No. litters examined†	12	15	11	6
Mean No. corpora lutea	9.0 ± 2.13	10.1 ± 1.64	10.5 ± 3.45	8.5 ± 1.87
Mean No. implantations	5.9 ± 2.39	8.0 ± 1.81	6.1 ± 2.84	7.2 ± 2.93
Mean % pre-implantation loss	NR	NR	NR	NR
Mean No. embryo/fetal deaths	NR	NR	NR	NR
Mean No. viable fetuses/litter	5.3 ± 2.73	7.6 ± 1.84*	5.9 ± 2.77	6.3 ± 2.25
Mean % post-implantation loss‡	16.7 ± 23.0	4.9 ± 8.0	2.5 ± 5.8	18.7 ± 13.5
Mean fetal body weight (g)	33.4 ± 7.27	30.9 ± 4.43	29.9 ± 7.21	29.3 ± 4.82
Total fetuses (litters) with malformations¶
External and visceral	0	0	0	2 (1)
Cardiovascular	0	0	0	0
Skeletal	0	3 (3)	2 (2)	0

NR = Not reported.

†Analysis does not include the litters that were aborted.

‡Calculated from individual animal data in Tasker et al. (1980a).

¶The incidences of variations were not reported in this study.

*p < 0.05, ANOVA followed by t-test for multiple comparisons; litter is the statistical unit.

Table A7.

Maternal and fetal outcome data for New Zealand white rabbits treated with glyphosate on gestational days 6–18 (Suresh, 1993).

	0 mg/kg/day	20 mg/kg/day	100 mg/kg/day	500 mg/kg/day
Maternal data
No. animals on study	26	17	16	15
No. non-gravid	4	4	0	1
No. gravid does dead or sacrificed in extremis	2	0	4	8
No. that aborted	NR	NR	NR	NR
No. with only resorptions	0	0	0	1
Embryo/fetal data
Total No. litters examined	20	13	12	6†
Mean No. corpora lutea	11 ± 2.8	10 ± 2.4	10 ± 1.9	9 ± 2.0
Mean No. implantations	8 ± 2.0	8 ± 1.5	9 ± 1.8	6 ± 2.4
Mean % pre-implantation loss‡	48	29	20	37
Mean No. embryo/fetal death‡	0.90	1.38	2.00	1.67
Mean No. viable fetuses	6.7	6.1	6.4	5.6
Mean % post-implantation loss¶	13.5 ± 14.3	18.6 ± 13.1	23.4 ± 23.8	23.2 ± 39.0
Mean fetal body weight (g)	32 ± 5.3	35 ± 3.7#	35 ± 2.4#	33 ± 4.9
“Abnormal fetuses” (n; %)	1 (1)	2 (3)	0	0
Total fetuses (litters) with malformations
External	2 (2)	2 (1)	1 (1)	0 (0)
Visceral	4 (3)	6 (3)	6 (4)	8 (2)*
Cardiovascular	2 (2)	4 (3)	6 (4)	6 (2)
Skeletal	11 (4)	5 (3)	0 (0)	1 (1)
Total fetuses (litters) with minor malformations and variations§
External	0	0	1 (1)	0
Visceral	NR (9)	NR (5)	NR (7)	NR (2)
Skeletal	NR (20)	NR (13)	NR (11)	NR (5)

NR = Not reported.

†Only five litters were evaluated for developmental toxicity at 500 mg/kg/day; includes single litter that was aborted at this dose in the analysis.

‡Standard deviation not reported.

¶Calculated from data provided in Suresh, 1993; values do not exactly match those presented in the study report.

§Incidence was not reported by individual fetus; rather, the incidence of each type of defect was reported, but more than one may have been seen in the same fetus.

#Significantly higher than control by ANOVA followed by Dunnett’s test; litter is the statistical unit.

*Significantly different from control by chi-square test.

Table A8.

Types and incidence of individual malformations† (Suresh, 1993).

	0 mg/kg/day	20 mg/kg/day	100 mg/kg/day	500 mg/kg/day
No. fetuses examined	133‡	78	77	28
Acephaly, abdominal hernia, external nares absent, shortened upper jaw, tail short & kinky, dorsal displacement of genital tubercle; multiple associated skeletal malformations	1 (1)
Acrania, open eyelids, kinky tail, arthrogryposis and adactyly (one with microglossia, short upper jaw, thoracic and abdominal hernia, hemimelia, malformed skull, missing cervical centrum and arch; one with cleft palate and oligodactyly)		2 (1)
Seal-shaped heart	1 (1)
Cardiomegaly and seal-shaped-heart			1 (1)
Dilated heart		4 (3)*	4 (2)*	5 (2)*
Dilated ventricle¶	1 (1)		1 (1)	1 (1)
Cleft palate	1 (1)
Forelimb arthyrogryposis			1 (1)
Liver hematoma				1 (1)
Gall bladder absent				1 (1)
Hydronephrosis	1 (1)	1 (1)
Dilated ureter		1 (1)
Fused sternebrae		1 (1)
Malformed sternebrae	1 (1)	1 (1)
Displaced sternebrae	1 (1)
Missing ribs	4 (3)
Bifurcated ribs	2 (2)
Missing thoracic arch and centrum	3 (3)
Extra lumbar arch and centrum				1 (1)

†Single fetuses may be represented more than once.

‡One fetus was not examined for skeletal malformations.

¶It is unclear from the study report if the dilated ventricles are of the heart or brain. For the purposes of this review, it is assumed that this description relates to the ventricles of the heart.

*Significantly different from control by chi-square test.

Table A9.

Maternal and fetal outcome data for New Zealand white rabbits treated with glyphosate on gestational days 6–18† (Bhide & Patil, 1989).

	0 mg/kg/day	125 mg/kg/day	250 mg/kg/day	500 mg/kg/day
Maternal data
No. animals on study	15	15	15	15
No. non-gravid	2	1	1	3
No. gravid does dead or sacrificed in extremis	0	0	0	0
No. that aborted	0	0	0	2
Embryo/fetal data
Total No. litters examined	13	14	14	12‡
No. litters with no live fetuses	0	0	0	2
Mean No. corpora lutea	10.0 ± 1.69	10.1 ± 1.60	10.3 ± 1.44	9.8 ± 1.57
Mean No. implantations	9.0 ± 1.20	9.3 ± 1.33	9.4 ± 1.12	8.5 ± 1.05
Mean No. early resorptions	1.7 ± 3.22	1.1 ± 2.53	1.0 ± 2.56	1.9 ± 2.43
Mean % pre-implantation loss	21.3 ± 32.4	14.9 ± 24.09	14.7 ± 24.38	13.1 ± 6.34
Mean No. embryo/fetal death	0.07 ± 0.26	0.13 ± 0.35	0.27 ± 0.59	1.4 ± 2.20
Mean No. viable fetuses	7.3 ± 3.10	8.0 ± 2.59	8.0 ± 2.48	5.2 ± 3.03
Mean % post-implantation loss	NR	NR	NR	NR
Mean fetal body weight (gms) ¶	40.6 ± 16.6	47.1 ± 0.95	47.5 ± 1.38	48.7 ± 1.87
Total fetuses (litters) with malformations§
Total fetuses (litters) with malformations||	3 (3)	6 (6)	10 (10)	20 (14)
External	1 (1)	2 (2)	3 (3)	3 (3)
Visceral	1 (1)	4 (4)	5 (5)	12 (9)
Cardiovascular	0	1 (1)	1 (1)	2 (2)
Skeletal	1 (1)	0	2 (2)	5 (2)

NR = Not reported.

†Body weights, maternal endpoints and some developmental endpoints for all 15 animals in each group appear to be included in the data. It appears that only the gravid animals were included for data on sex ratio and fetal body weight.

‡The two litters that were aborted at this dose were included in the analysis.

¶Fetal body weight data are as reported in the study report; it is unclear if all 15 does were included in this calculation.

§The incidences of variations were not reported in this study.

||The total number appears to be the sum of the fetuses and litters with external, visceral and skeletal malformations. Thus, the number of litters in the 500 mg/kg/day dose group is reported as 14 when there were only 12 litters in the group.