How Do You Assess if a Chemical Causes Cancer?

Years of testing glyphosate, part of the most widely used herbicide in the world, has shown us that the chemical is not carcinogenic. Why can’t we believe it?

Activists pull down a giant bottle of weedkiller as they demonstrate in favor of a Glyphosate ban by the European Union in front of the European Union Commission headquarter in Brussels on July 19, 2017.
Activists pull down a giant bottle of weed killer as they demonstrate in favor of a Glyphosate ban by the European Union in Brussels on July 19. Thierry Charlier/AFP/Getty Images

Can anyone make sense of the debate over glyphosate, the active molecule in the most widely used herbicide in the world? According to a study by University of California–San Diego researchers published last year, urine samples from a group of 100 southern California residents showed that levels of glyphosate and its metabolite, aminomethylphosphonic acid have increased fifteenfold between 1993 and 2016.

Unsurprisingly, the publication of this research added considerable fuel to the regulatory fire sweeping Brussels and Washington alike over the use of the chemical, even though the study is limited in scope and has no bearing on the weed killer’s health effects. On Nov. 27, the European Commission agreed to renew the molecule’s market reauthorization for five years, after serious opposition from member states. EU members still retain the right to ban the chemical domestically if they so wish. Meanwhile, on Capitol Hill, lawmakers from the House Committee on Science, Space, and Technology have opened an investigation into the International Agency for Research on Cancer, an arm of the World Health Organization that receives U.S. funding, because the agency stands accused of editing its findings to support a 2015 decision that glyphosate is probably carcinogenic.

Several things are going on here that are worth untangling. At the heart of the problem is the messy fact that when scientists and policymakers carelessly substitute risk for hazard, flawed conclusions are drawn.

We are generally bad at understanding risk. Determining the carcinogenicity of any foreign bodies that ends up in our ecosystems involves assessing degrees of uncertainty. Unlike with pharmaceutical trials, it is impossible to carry out a randomized clinical trial with glyphosate or any other weed killer.

It goes without saying that weed killers should be toxic to their intended target and benign to humans. Animal studies can be a tool for gauging a biocide’s carcinogenic potential, but they can only provide hazard signals, as multiple studies have shown that simply extrapolating research conducted on rodents to humans rarely produces accurate results. Animal testing therefore has limited value in determining carcinogenicity, because it does not necessarily allow researchers to draw conclusions of causation between a given substance and the occurrence of cancer in test animals, let alone in humans.

This is an important factor to consider with regards to the glyphosate debate. When IARC announced in June 2015 that glyphosate was “probably carcinogenic” to humans, Kate Guyton, a toxicologist and lead author of the IARC monograph, stated that “because the evidence in laboratory animals was sufficient and the evidence in humans was limited, this places [glyphosate] in Group 2A [of probable carcinogens].” It was later revealed that IARC scientists had removed findings from studies that concluded glyphosate to be noncarcinogenic before publishing the final version. The edits were made in the monograph’s chapter on animal studies, which crucially informed IARC’s assessment that glyphosate causes cancer.

The debate sparked by IARC’s evaluation highlights why human studies are so essential. Indeed, one key study—whose initial findings were not included in IARC’s literature review due to their internal prohibition on considering unpublished data—is the Agricultural Health Study, a long-term observational analysis of the health effects of herbicides on 89,000 farmers and their families in Iowa and North Carolina. Running since 1993, the AHS has consistently failed to find that glyphosate use is linked with increased risk of cancer. Parts of the study, whose failure to find any evidence of glyphosate’s carcinogenicity was already well-known among IARC staff, were finally published earlier in November.

These findings have been backed up by other studies as well.
Last year, the U.N.’s Food and Agriculture Organization and the World Health Organization presented a joint review report on pesticide residue in indirectly exposed persons, including farming and production workers’ families, as well as consumers. The report did not find any evidence of increased risk of cancer from glyphosate exposure. Instead, after having examined the epidemiological evidence of occupational exposures, the report concluded, “Glyphosate is unlikely to pose a carcinogenic risk to humans by food intake.”

The European Food Safety Administration also performed human-exposure studies as part of its assessment of glyphosate. Published in October 2015, their assessment found that the observed risk of cancer in humans induced by glyphosate was very low and the causal links virtually nonexistent.

It is important to note that unlike EFSA and other agencies, IARC is responsible for identifying a potential cancer hazard and “does not measure the likelihood that cancer will occur (technically called ‘risk’) as a result of exposure to the agent,” while U.S. and European agencies are concerned with defining the real risk for the general public. This means that differentiating between absolute risk in the lab and true risk for human health is essential. In glyphosate’s case, this true risk assessment concerns whether the substance is linked with non-Hodgkin’s lymphoma—a hypothesis that so far has been backed by IARC, but have not been proven in any human studies.

So what should regulators do? Although it has been subject to outsized scrutiny, glyphosate’s benefit-risk ratio is one of the highest of all weed killers derived from organic chemistry. It is less toxic than common chemicals like aspirin. It is therefore surprising that media coverage has been irrational and even hysterical. One of the main arguments brought forward by the opponents of biocides is the precautionary principle, which simply states that precautionary measures should be taken to avoid harm even if said harm is uncertain. The obvious problem with this argument is the fact that it is not equally valid or universally applicable.

In the case of extremely high doses of radiation (over 1,000 rad), for instance, the causality of its use and effect are well-established, with unambiguous and predictable outcomes. The same is not true for glyphosate, where there is no established causality between higher cancer risk and exposure. Denying this stretches the precautionary principle to the point of rendering it nearly meaningless, precluding any uncertainty that might arise from human activity—and thereby any new scientific developments.

To expand our understanding of the mechanics of pathogenesis in relation to uncertainty, more studies and innovation in applicable criteria are required. We need organizational innovation that collects comprehensive data as well as fundamental research tracing the cellular action of biocides during human exposure.

But in any case, as the facts stand today, there is no indication that glyphosate raises the risk of cancer. IARC is not assessing the risk, they are simply meant to be alerted in case of hazards.

In this case because of the absence of evidence against glyphosate, we should be aware of the potential for hazard, but the chemical should be considered noncarcinogenic. Otherwise, the purpose of science itself, which will always entail some degree of uncertainty, is utterly undermined.