EOS contradicts the EU review’s conclusion that glyphosate is not genotoxic, citing evidence that:
Roundup increases the frequency of gender-linked recessive lethal mutations in fruit flies (Kale et al, 1995), DNA adducts in the livers and kidneys of mice (Peluso et al, 1998) and sister chromatid exchanges in human lymphocytes (Vigfusson and Vyse, 1980);
Mice injected with glyphosate and Roundup show an increased frequency of chromosome damage and increased DNA damage in bone marrow, liver and kidney (Bolognesi et al, 1997);
GBHFs cause DNA damage in human cells (Gasnier et al, 2009);
In sea urchin embryos, GBHFs and AMPA (the environmental degradation product of glyphosate, aminomethylsulphonic acid) alter cell cycle checkpoints by interfering with DNA repair (Marc et al, 2002; 2004a,b; Belle et al, 2007) and cause inhibition of RNA transcription and delayed hatching (Marc et al, 2005); and
An epidemiology study in Ecuador found more extensive DNA damage in people living in an area that was aerially sprayed with glyphosate compared with those living 80 km away (Paz-y-Mino et al, 2007).
The genotoxicity of glyphosate, its metabolite AMPA and GBHFs (with and without surfactants including POEA) has been reviewed by Williams et al (2000), Kier and Kirkland (2013) and the Australian DoHA (1985, 1991, 1992 and 2005), US EPA (1993), WHO (1994) EU (1998) and JMPR (2004b). In addition to assays for gene mutation in bacteria and cultured mammalian cells, the investigated end-points included tests for DNA damage and repair in vitro and chromosomal aberrations (clastogenicity) in vitro and in vivo. All the reviews agreed that the vast majority of studies within the highly extensive database had clearly negative outcomes, and concluded that glyphosate, AMPA and GBHFs do not present a genotoxicity hazard. Furthermore, POEA is not mutagenic (Stegeman and Li, 1990; Williams et al, 2000).
The JMPR and/or EU reviews (both performed by the German BVL) covered four of the studies cited by EOS (2011) as demonstrating genotoxic activity. However, as outlined below, the BVL concluded that the findings were also consistent with cytotoxicity (cellular injury or death not caused by damage to genetic material), and commented that assessment of these data was complicated by a lack of information on product composition, reporting limitations, and by the use of some test systems which were of uncertain relevance for the assessment of risk to humans.
Kale et al (1995) obtained positive results in a test for lethal mutations in fruit flies (Drosophila melanogaster) after larvae were treated with a Roundup product (41% glyphosate IPA salt with POEA surfactant) or Pondmaster (41% glyphosate IPA salt with alkyl sulphate surfactant). Dosing conditions were not specified but the test insects were exposed to concentrations close to the LC50. The BVL considered that it would have been very difficult for the investigators to distinguish between deaths from lethal mutations and deaths from the anticipated high toxicity.
Using a rd32P-postlabelling assay, Peluso et al (1998) found a weak, dose-related increase in DNA adducts in the liver and kidney of mice injected IP with 400, 500 and 600 mg/kg of a Roundup product containing 30.4% glyphosate IPA salt with alkyl sulphate surfactant. No adducts were seen with glyphosate IPA alone at 130 or 270 mg/kg, or in a control group. While agreeing that the finding was an indication of possible DNA damage, the BVL regarded the biological significance as equivocal because DNA adducts can occur naturally or arise from increases in endogenous metabolite levels, as well as from direct interaction with chemicals. The BVL also questioned the relevance of IP administration to normal exposure conditions, and criticised the absence of any positive control group, individual animal data and information on the DNA adducts’ structure.
Vigfusson and Vyse (1980) observed a weak but statistically significant increase in the frequency of sister chromatid exchanges (SCEs) in human lymphocytes incubated with a Roundup product (composition unspecified) at 250 and 2500 µg/mL. The BVL observed inconsistencies in the results, in that a dose response occurred in cells from only one of the two donors, and the statistically increased values from one donor lay below the control values from the other.
Bolognesi et al (1997) examined the effects of glyphosate and a Roundup product (30.4% glyphosate IPA salt with alkyl sulphate surfactant) on several end-points:
i) A SCE assay in cultured human lymphocytes from two female donors was positive with glyphosate at 1–6 mg/mL and Roundup at 100 and 330 µg/mL. The formulation was cytotoxic at higher concentrations. The BVL criticised the statistical analysis, as data from the donors were pooled and individual values were not provided.
ii) A weakly positive alkaline elution assay for single-strand DNA breaks and formation of alkali-labile sites in DNA suggested possible transient DNA damage in the liver and kidney of mice, four hours after IP injection with glyphosate or Roundup at 300 and 900 mg/kg respectively. The BVL noted that IP injection was an inappropriate route because the test chemicals could be directly cytotoxic to the tissues within the peritoneal cavity. Furthermore, the outcome was inconsistent with three other studies in which glyphosate did not cause cytogenetic damage, mutation or DNA adduction in mice treated IP at up to 1000 mg/kg bw.
iii) One day after treatment as described in (ii), measurement of 8 hydroxydesoxyguanosine (OHdG) adducts revealed evidence of increased oxidative metabolism / injury in the liver (with glyphosate only) and kidney (with Roundup only). The BVL suggested that the finding may elucidate a mechanism of toxicity but is not evidence of genotoxicity.
iv) In a bone marrow micronucleus assay, groups of three male mice received two IP doses of glyphosate (150 mg/kg) or Roundup (225 mg/kg) at 24—hour intervals, and were killed for assessment six and 24 hours after the final dose. A weakly positive response was obtained with Roundup at both time points, and glyphosate at 24 hours. With respect to glyphosate, the BVL highlighted the inconsistency between the positive outcome and other micronucleus assays, which were negative in rats treated at up to 1000 mg/kg IP and in mice receiving up to 5000 mg/kg PO. Furthermore, Bolognesi’s assay did not comply with the relevant OECD Test Guideline, as the treated groups contained fewer than the recommended five animals and only one dose was tested, precluding the assessment of dose-response. It was unclear when the control mice were killed, weakening the validity of the statistical comparison. The BVL also commented that the formulation (although not the active) may have caused cytotoxicity in the bone marrow, as evidenced by a decrease in the ratio between polychromatic and normochromatic erythrocytes. Cytotoxicity may therefore have affected the frequency of chromosomal aberrations. There was apparently no data on the mutagenicity of the alkyl sulphate surfactant present in the tested Roundup product.
Using the Comet assay, Gasnier et al (2009) measured single- and double-stranded DNA breakage and alkali-labile DNA damage in HepG2 liver cancer cells in vitro after 24 hours of incubation with Roundup Grands Travaux, a product containing glyphosate at 400 g/L together with unidentified adjuvants (see assessment in Appendix 3). The test cells were exposed at 1, 2.5, 5, 7.5 and 10 ppm. The pro-mutagen benz[a]pyrene (50 µM) was used as positive control. The test product had no effect at the two lowest concentrations but caused a dose-dependent increase in DNA strand breaks at 5, 7.5 and 10 ppm (50, 60 and 75% breakage compared with 35% in negative controls and 95% in positive controls). However, Gasnier et al also reported that the test product was cytotoxic against HepG2 cells at concentrations of 5 ppm upwards, with an LC50 of 12 ppm. It is therefore possible that the increased DNA strand breakage seen at 5–10 ppm was secondary to cellular injury or death, rather than arising directly from damage to DNA by the test product. Furthermore, it is unclear which component(s) of Roundup Grands Travaux was biologically active, as the effects of glyphosate or adjuvant(s) alone were not tested.
The Australian DoHA (2005) assessment found that Marc et al (2005) had demonstrated that Roundup (diluted to glyphosate concentrations of up to 4 mM) delayed RNA synthesis, transcription of the hatching enzyme and hatching of sea urchin embryos by ca two hours. There was only a marginal effect on cell division indicating the delay was not due to any cell-cycle effect. Pure glyphosate at up to 8 mM had only a weak effect on hatching (a delay of 30 min). Marc et al also reported that POEA was “highly toxic to the embryos leading to irreversible damage” but provided no supporting data. The DoHA considered the sea urchin model as being of “dubious” value for human health risk assessment, given that glyphosate had already been tested by validated methods.
In an investigation of associations between genotoxic risk and aerial application of glyphosate-based herbicides for control of illicit crops, Bolognesi et al (2009) performed a cytogenic biomonitoring study on agricultural workers in Colombia. In areas where glyphosate was sprayed, blood samples were taken prior to application and then at five days and four months post-application. Chromosomal damage and cytotoxicity in lymphocytes were evaluated by cytokinesis-block micronucleus assay. Compared with Santa Marta, where organic coffee is grown without pesticides, the baseline frequency of binucleated cells with micronuclei (BNMN) was significantly greater in subjects from four other regions. However, only gender, region and older age were associated with baseline BNMN frequencies, and glyphosate was not used in one of the two regions where the highest frequencies of BNMN were found. In three regions, a significant increase in BNMN frequency occurred five days after glyphosate was applied, which reversed in one of these regions within four months post-application. The study authors concluded that genotoxic damage associated with glyphosate application was small and transient, and the genotoxic risk was low.