Archives of Toxicology

, Volume 86, Issue 11, pp 1791–1793

Comment on “Glyphosate impairs male offspring reproductive development by disrupting gonadotropin expression” by Romano et al. 2012


    • Exponent
  • Amy L. Williams
    • Exponent
Letter to the Editor

DOI: 10.1007/s00204-012-0894-3

Cite this article as:
DeSesso, J.M. & Williams, A.L. Arch Toxicol (2012) 86: 1791. doi:10.1007/s00204-012-0894-3

Romano et al. (2012) recently published the results of a study in which pregnant rats were treated with a glyphosate-based herbicide formulation from gestational day 18 to postnatal day 5. Various parameters related to sexual behavior and reproductive development were assessed in the resulting male offspring. Based on their results, the study authors claim that glyphosate exposure in the perinatal period causes changes in males that are associated with hypersecretion of androgens. There are a number of issues with this study, however, that preclude such conclusions.

First, as a premise for this study, the authors incorrectly assert that glyphosate is a “potential endocrine chemical disruptor.” As evidence, they cite three in vitro studies of altered aromatase activity conducted in a single laboratory (Benachour et al. 2007; Gasnier et al. 2009; Richard et al. 2005), and their own earlier work suggesting reduced serum testosterone concentrations in response to treatment (Romano et al. 2010). These studies, however, were conducted using glyphosate-based herbicide formulations rather than the active ingredient alone. Exposures were at high concentrations;1 for instance, in the study by Benachour et al. (2007), the concentration of glyphosate in the tissue culture medium was reported to be 1.27 mM (approximately 200,000 μg/L), which is not relevant to concentrations in target tissue concentrations or environmental media. Further, although sporadic positive responses were reported, aromatase activity was generally unaffected in the few experiments that tested glyphosate by itself. It should be additionally noted that the EPA OPPTS guideline for conducting the aromatase assay (U.S. Environmental Protection Agency 2009a) clearly warns that all glassware and apparatus used need to be free of detergent residue; this is because microsomes are extremely sensitive to the surfactants found in detergents. Surfactants are a major component of the various herbicide formulations tested in these studies and are the likely cause for the observed aromatase inhibition. Therefore, the claim that glyphosate is an endocrine disruptor is not substantiated.

Second, it is not clear to what agent(s) the animals were exposed. The study authors indicate that Roundup Transorb (480 g/L glyphosate, 648 g/L isopropylamine salt, and 594 g/L inert ingredients) was the product administered. However, they report no additional details regarding the composition of the herbicide formulation used such as the identity and amount of surfactant(s) in the formulation. This is important because in other studies, Levine et al. (2007) have shown that steroidogenesis in Leydig cell cultures is inhibited by the presence of surfactants, which perturb membrane function of mitochondria. An appropriate control group that was exposed to the surfactant(s) only was not included in the study nor was there a group exposed to glyphosate only. As such, the study results are confounded.

Third, the authors did not describe any methods to control for potential litter effects, and it is not clear whether this was done or not. Because the dams were the unit of exposure in these experiments, the exposed litter (the dam) is the appropriate unit of analysis. This is particularly important when assessing these experiments because dosing of the dams occurred from gestational day 18 through postnatal day 5; however, glyphosate does not cross well into milk (WHO 1994). This means that most of the offspring exposure occurred during the last 4 days of gestation only; exposure during the postnatal period primarily affected the dams. Furthermore, because littermates are known to be more similar to each other than offspring derived from separate litters (Holson and Pearce 1992), the observed differences between groups may be due to animals being derived from the same limited number of litters rather than a true effect of treatment during gestation.

Fourth, we note that Romano et al. present data for, and statistically analyze, at least 40 diverse end points that include body and organ weights, behavioral observations, sperm parameters, histological analyses, serum hormone levels, and mRNA expression. It is not possible to discuss all of the end points in this letter. Rather, our comments will be restricted to a few examples that illustrate why there are concerns about this study.

The authors evaluated the effects of treatment using a non-standardized protocol (the sexual partner preference test). This testing protocol that is not used by regulatory agencies has minimal historical data and is susceptible to confounding if precautions to avoid environmental cues (including odors and pheromones from previously tested animals, as well as visual or auditory stimuli) are not utilized. The methods provide no evidence that the study investigators recognized or took actions to minimize such confounders. Furthermore, the rats from treated litters gravitated more toward females than did controls, which appears to be in the “normal” direction. Thus, the interpretation of these findings is problematic.

When discussing changes in serum hormone concentrations and other reproductive parameters, it is important to insure that the assays provide consistent results with control values that are within the expected range. Several parameters evaluated in this paper were also examined in a 2010 publication by the same group of investigators (Romano et al. 2010). These include serum testosterone and estradiol concentrations; measures of seminiferous tubule morphology; and age and weight at preputial separation (PPS). Both studies involved oral exposure of Wistar rats to the commercial herbicide formulation Roundup Transorb and both included a dose of 50 mg/kg/day. However, the durations and timing of exposure differed; specifically, the 2012 study involved treatment of maternal animals from gestational day 18 to postnatal day 5, while the 2010 study treated male offspring directly from postnatal day 23 to postnatal day 53. Because of the different exposure regimens, one would not expect the findings after glyphosate exposure to be the same in both studies. However, the control animals were treated with deionized water, and the shared parameters were evaluated at similar times (on PND53 in 2010, on PND60 in 2012). Therefore, one would expect that the findings for the control animals would be similar across studies; however, as shown in Table 1, this is not the case. In fact, the differences in control values across the two studies are greater than the magnitude of change seen in the individual studies with glyphosate treatment.
Table 1

Control values reported in Romano et al. (2010, 2012)


Romano et al. (2010)

Romano et al. (2012)

Age at preputial separation (days)



Body weight at preputial separation (g)




Evaluated on PND53

Evaluated on PND60

Serum testosterone concentrations (ng/dL)

154 ± 12.9


Serum estradiol concentrations (pg/mL)

31.5 ± 1.2


Tubular diameter (μm)

265.7 ± 4.8

466.9 ± 14.3

Epithelial height (μm)

85.8 ± 2.8

91.7 ± 2.2

Luminal diameter (μm)

94.0 ± 5.7

256.9 ± 13.6

Romano et al. reported PPS occurred at approximately 37 days of age in their 2010 report, but at 47 days of age in their 2012 report. In contrast, studies by other groups of investigators have reported that PPS in Wistar rats typically occurs at PND 40-43 (Schneider et al. 2005; Botelho et al. 2009; Zambrano et al. 2005; Engelbregt et al. 2000; Martini et al. 2006; Goetz et al. 2007; Bell et al. 2007; Zorrilla et al. 2009; Stoker et al. 2000, 2002, 2004). In the EPA OPPTS guideline for conducting the male pubertal assay (U.S. Environmental Protection Agency 2009b), the EPA has established acceptable ranges of age and body weight values for PPS in control Wistar rats of 40.1–46.1 days of age and 177.3–241.0 g, respectively. The values for control animals in both of the studies by Romano et al. (2010) fall outside this acceptable range, missing either below or above the accepted range of control values. Thus, not only do the values of the two studies fail to agree with each other, they do not fall within the range of generally accepted values.

Based on experience gained in validating the male pubertal assay, the EPA guideline also provides an acceptable range of serum testosterone concentrations for control male Wistar rats on PND53 of 0–4.7 ng/mL (0–46.6 ng/dL) (U.S. Environmental Protection Agency 2009b). In published male pubertal assays conducted by Stoker et al. (2002, 2004), control male Wistar rat serum testosterone concentrations were reported to be 2.2 ng/mL (22.1 ng/dL) and 1.6 ng/mL (16.2 ng/dL) on PND53. The serum testosterone values reported by Romano et al. (2010) are not in line with these values or the range of acceptable values outlined by the EPA based on validation efforts.

Taken together, the shortcomings in this paper erode any confidence that these experiments are able to demonstrate disruption in the development or function of the male reproductive system in offspring whose dams were treated with glyphosate.


Concentrations of glyphosate in surface water have been reported in the literature, with maximum concentration values of 12, 40, and 75–90 μg/L (Byer et al. 2008; Struger et al. 2008; Botta et al. 2009).


Conflict of interest

This Letter to the Editor was not funded by outside sources. The authors have previously published a report on the potential developmental toxicity of glyphosate that was funded, in part, by Monsanto.

Copyright information

© Springer-Verlag 2012