I spent the last week reviewing a number of documents provided to me by the Pest Management Regulatory Agency (PMRA) of Health Canada related to the federal government’s risk assessment of glyphosate, the active ingredient in the popular herbicde product RoundUp. I had requested these from PMRA to explore the extent to which these risks assessment by Canada and the US incorporate health endpoints other than cancer in their evaluation of the product’s safety.
The short version of my assessment is that the systematic reviews and risk assessments performed by PMRA and by the US Environmental Protection Agency (USEPA) exhibit systemic confirmation bias and exclude numerous important studies for spurious reasons. In general, they appear to evaluate new studies with an eye towards why they should be excluded rather than if they increase uncertainty regarding safety. Most often, this involves eliminating studies that use commercial formulations, i.e., RoundUp. These studies are not irrelevant, however, and any reasonable risk assessment would see these studies as increasing uncertainty regarding glyphosate’s safety.
I discuss both agencies here because PMRA’s email noted specifically that the US and Canada coordinated their processes and reviews. For the purposes of transparency, all of the documents that PMRA provided are linked at the bottom of this page, including their email to me in service to my request. The PMRA, it is worth noting, was prompt and transparent when responding to my inquiries.
My concern about the comprehensiveness of the PMRA risk assessment was prompted by the apparent emphasis in online discussions on cancers as the only health endpoints of concern. While carcinogenicity is an important piece of the puzzle here, it is not the only possible health issue of concern. Neurological toxicity, reproductive toxicitity, and impacts on dietary health and the microbiome are also areas of significant concern.
My working assumption going into this were twofold. First, I was concerned that the risk assessments do not incorporate a sufficiently comprehensive list of possible negative health endpoints, and that instead, findings relating to a lack of carcinogenicity were being extrapolated to statements about the herbicide’s safety in general. Specifically, I had a hunch that research related to the impacts of glyphosate on the gut microbiome and dietary health are not being sufficiently considered. Second, I was concerned that the risk assessments do not take a fully precautionary approach, and instead focus on narrow lines of evidence and a high threshold for showing possible harm. In other words, that evidence that is indirect or not sufficiently specific about dose rates of glyphosate is being excluded rather than considerd as a source of uncertainty and cause for precaution.
What I found was a combination of the two concerns. The sytematic reviews and risk assessments by the USEPA and PMRA, in my assessment, do look at multiple health endpoints but DO NOT take a sufficiently precautionary approach. They dismiss emerging areas of concern like impacts on the gut microbiome, and they exclude numerous studies outright for reasons that undermine the entire assessment.
First, let’s talk about why studies are excluded, and why they should have instead been included through the lenses of uncertainty. One reasons for exclusion recurs frequently in the USEPA’s documents: “no glyphosate measurement” — meaning that they found the details about glyphosate dosing in the studies inadequate or insufficient to evaluate the importance of their findings.
I have a number of objections to this rationale. The first is that I very easily found studies that were excluded on this basis that do indeed specify the doses of glyphosate used in the experiments. One set of examples are the publications from the “Ramazzini Institute 13-week pilot study,” which looked at the impact of glyphosate and RoundUp formulations on development, endocrine system, and gut microbiome in rats. Here are two publications, and both actually do provide clear dosing information in their abstracts and methods:
This one, for example, showed that glyphosate and RoundUp, at a dose level considered as safe, induce endocrine effects and altered reproductive developmental parameters in male and female rats.
This study on microbiome effects shows that exposures at doses considered safe are capable of modifying the gut microbiota in early development, particularly before the onset of puberty.
Why are both excluded? USEPA offers this general rationale:
There are numerous glyphosate formulations and providing a general product name, such as Roundup, does not provide the agency with information to ascertain the exact formulation used and determine all of its chemical components. Additionally, several studies were conducted in other countries and utilized formulations that are not registered in the United States. As a result, the active ingredients and other components of the formulation are unknown, and any potential effects cannot be attributed to glyphosate and/or defined glyphosate concentrations. (p. 2)
This rationale for exclusion is indefensible and could be interpreted as systematic cherry-picking. The PMRA and USEPA are only interested in evidence that specifically draws a line between the one active ingredient—glyphosate—and potential health outcomes, and they seem to be actively excluding any evidence of health impacts when established in experiments using commercial preparations. But unless there’s something I don’t undertand about the fate and transport of the herbicide, people are not, generally, just exposed to glyphosate. They are exposed to the commercially available products, which often include adjuvants—chemicals that are specifically designed to amplify the effectiveness of herbicides.
It makes little sense to explore the impact of a herbicide without also exploring its effect when paired with a chemical intended to increase its effectiveness. Regarding adjuvants, Clair et al. (2012) write, “the present regulatory in vivo tests with [glyphosate] alone to study chronic toxicity are hardly relevant.” Incidentally, their study, which shows that glyphosate-based herbicide induces necrosis in mature rat testicular cells, was excluded for not being sufficiently specific to glyphosate dosing.
Here is a specific example of the line being drwan: the study noted above on rat development and endocrine systems shows much stronger negative effects from "Roundup Bioflow” than glyphosate alone. The USEPA’s document construes this as “generally no adverse effects observed” (p. 8). It is hard to not interpret this as a sleight of hand.
We see a similar concern in how the PMRA’s decision in 2017 addresses newly emerging evidence for concern regarding the effects of the herbicide on the gut microbiome. They effectively dismiss studies that show correlation and construe the emerging evidence as being limited or irrelevant. (p. 30)
PMRA and USEPA are effectively taking an all or nothing approach here. A more appropriate framing for this risk assessment would entail a discussion of uncertainty and a precautionary approach that mandates a high level of certainty in regards to evidence of safety, not of harm. But this risk assessment is reversing that approach. Studies that explore newly identified concerns like impacts on the gut microbiome, which should not be dismissed because of how rapidly gut microbiome work is transforming how we understand all manner of diseases, should not be so easily dismissed.
Similarly, a precautionary approach ought be representative of the actual environment in which exposure occurs: which in this case is via commercially available products, not individual ingredients. What use is our set of laws for environmental and health protection if they only operate one chemical at a time and cannot account for synergistic and cumulative effects?
I do not think that the indivuduals performing these reviews and assessements are actively trying to suppress evidence. I do think, however, that they’re working with far too narrow a framing for the risk assessment and the kinds of evidence that can be considered.
The entire thing needs to be started from the beginning.
Here are the documents in question:
EPA - Dec 2017 Systematic Review
EPA - Jan 2019 Response to Comments on Interim Decision
PMRA - 2015 Proposed Re-evaluation Decision