How to 'Prove' a Chemical is Dangerous: The Glyphosate Case Study

Related articles

The anti-biotech group GM Watch recently touted the results of a new study as evidence that the EPA has underestimated the risk posed by the weedkiller glyphosate. It's an illustration of what goes wrong when you force data to conform to a predetermined conclusion.

If you want to show that any chemical is dangerous, here's a three-step process that will consistently yield the desired result: 

  1. Visit Pubmed, type in your chemical of interest and the medical issue you want to link it to.
  2. Find the study or studies with the conclusion you need.
  3. Disregard any other research as irrelevant.

I searched for the weed killer “glyphosate” and “reproductive health” in preparation for this article. Within seconds, I had 59 peer-reviewed papers at my fingertips that I could have used to argue that glyphosate is a reproductive toxicant. If you see an activist group tout some study as clear evidence that a chemical is harmful, they've almost certainly followed this process.

The anti-biotech group GM Watch, for instance, claimed recently that “glyphosate damages blood-testis barrier and causes poor quality sperm,” based on this just-published paper (referred to as "Liu et al." below). The researchers fed rats chow containing glyphosate at doses that correspond to levels the Environmental Protection Agency (EPA) deems safe for humans, yet some of the animals experienced a decrease in sperm quality—suggesting that chronic glyphosate exposure induces reproductive toxicity. You can see why GM Watch was interested in the paper:

"The high dose tested, 50 [milligrams per kilogram of body weight per day] … is relevant to regulation of glyphosate because it is only 1/20th of the no-observed adverse effect level (NOAEL) of the 1000 mg/kg bw/d … Based on this finding, the obvious conclusion is (though the researchers do not state as much) that the EPA’s ADI [acceptable daily intake] is incorrect — it should be lowered."

Here's the problem: regulatory agencies usually don't draw conclusions based on a single study—and a deeply flawed one at that, as we'll see below. They look at all the available data, then make a determination about the risk a chemical poses to human health. When we look beyond this one paper, we can see why the EPA set the existing limits for glyphosate exposure.

Stuffing animals with weed killer

Sometimes I feel bad for lab animals. That's because scientists stuff them full of chemicals to see what happens, in an effort to determine if they pose a risk to humans. Experiments of this sort have been done many times over the years with glyphosate and two results show up repeatedly:

  • Animals sometimes experience negative outcomes if they consume massive quantities of glyphosate, far above the reference dose (RfD) the EPA sets for human exposure. The EU's Assessment Group on Glyphosate recently published a nearly 600-page analysis of the toxicological research on glyphosate full of examples.
  • Even at these unrealistic exposures, researchers can't consistently reproduce the harmful effect in question; and when they do, the results aren't always statistically significant, as the EU Food Safety Authority reported several years ago. Consider the following examples from a 2017 review of the data on glyphosate's reproductive toxicity:

Glyphosate was not a reproductive toxicant in the majority of rodent multigenerational reproductive studies up to the highest doses tested with HED [human equivalent] doses greater than 170 times higher than the short- term RfD and more than 800 times higher than the subchronic RfD. One multigenerational reproductive study reported decreased spermatid counts and delayed male puberty at an HED dose about 750 times higher than the subchronic RfD, but effects were not replicated in other studies

[A]nother study found increased testes weights but no effects on sperm motility, sperm counts, or estrous cycles at comparable doses. Increased testes and ovary weights were reported in a chronic mouse study at HEDs over 6,000 times higher than the chronic RfD [my emphases]

What about the new study?

The fact that GM Watch wants to elevate a single study over this massive body of peer-reviewed evidence, as well as the conclusions of 16 regulatory and scientific agencies, should set off alarm bells in your head. But what can we say about the new study itself? A lot. Here are just two critical issues with Liu et al. First, the authors claimed that “An accumulating body of evidence” suggests that glyphosate is an endocrine disruptor. [1] This is false, and we can see why by looking at a 2020 review the authors themselves cited. Here's that paper's conclusion:

"Based on an analysis of the comprehensive toxicology database for glyphosate and the literature, this review has concluded that glyphosate does not have endocrine-disrupting properties through estrogen, androgen, thyroid and steroidogenic modes of action." [my emphasis]

In other words, glyphosate does not interact with the pathways necessary to damage the endocrine system. The EPA also reached this conclusion after reviewing all the available data in 2015. What does this mean for the study GM Watch is so excited about? “Glyphosate doesn't bind the estrogen receptor, indicating that the initial step in the paper's model that leads to an adverse effect is flawed,” Bayer environmental toxicologist Steven Levine, lead author of the 2020 review, told ACSH by email.

Second, the researchers tried to show that glyphosate-induced oxidative stress led to decreased sperm quality, but they made a serious mistake along the way. “Measuring an increase in reactive oxygen with the kit they used is not a reliable indicator of oxidative stress,” Levine added. The US National Toxicology Program agrees. Using multiple reliable assays to measure oxidative stress, the NTP has shown that glyphosate does not induce oxidative stress in human cell lines.

Humans aren't lab rats

Replicating the results of animal models and cell-culture studies is important, but the results of human epidemiological research are no help to GM Watch either. For example, biomonitoring studies of farmers (who have the highest glyphosate exposure of any of us) indicate that the weed killer doesn't cause reproductive damage. According to a 2012 review:

"These data demonstrated extremely low human exposures as a result of normal [pesticide] 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[ay] set by the U.S. Environmental Protection Agency. In conclusion, the available literature shows no solid evidence linking glyphosate exposure to adverse developmental or reproductive effects at environmentally realistic exposure concentrations."

If farmers are exposed to glyphosate in quantities far below the EPA reference dose, what are the chances trace amounts of the weed killer in your food will damage your reproductive system? Slim to none is my guess. But you don't have to believe me. Let's return to the EU's Assessment Group on Glyphosate (AGG). They reached the same conclusion:

"… On basis of the available information … the AGG does not consider the criteria for classification with respect to reproductive toxicity ... to be fulfilled. The AGG proposes that classification of glyphosate as toxic for reproduction is not justified."

I'm not implying that you should believe someone just because they have advanced degrees, or that challenges to scientific consensus are always wrong. People with lots of letters after their names are often mistaken—even if many of them agree. But that doesn't negate the simple fact that we have to account for all the evidence we have before drawing conclusions. Bloggers at GM Watch still haven't grasped this important point. Fortunately, we don't have to make the same mistake.


[1] My colleague Dr. Josh Bloom adds that "'endocrine disruptor' is a catch-all phrase used by anti-chemical types to scare everyone."