On both sides of the Atlantic, a battle is raging between starkly opposed views of what science tells us about risks to our health emanating from our surroundings, including our food, water, and the wider environment.
This battle often pits advocates, nongovernmental organizations (NGOs), politicians, and partisan scientists, who have little ability or inclination to evaluate the evidence on its merits, against scientists and regulators who confine themselves to evaluating the evidence but are typically characterized as corrupt apologists for industry. Unfortunately, confining oneself to the scientific issues simply does not have the same impact in the public arena as holding up the specter of an imminent threat to our well-being, and the less sensationalist party in this lop-sided contest is often left to tear its hair out in frustration.
At the moment, a major focus in this battle is glyphosate, the most widely used herbicide (i.e., weed killer) in the United States. Glyphosate is the active ingredient in Monsanto’s Roundup, and a number of major crops have been genetically engineered to be resistant to Roundup.
In response to a new report alleging a cancer risk from trace residues of glyphosate, which has sparked renewed pressure from anti-pesticide NGOs, this past June the European Union voted against re-registering glyphosate for agricultural use in Europe but agreed to an 18-month extension of the license. In the U.S. glyphosate is also at the center of a tangled regulatory battle.
Glyphosate and compounds containing the chemical, like Roundup, have been reviewed for health effects by a number of health agencies, including the European Food Safety Authority (EFSA), the Food and Agricultural Organization (FAO), Germany’s Institute for Environment and Human Security (BfR), and the U.S. Environmental Protection Agency (EPA). These agencies have concluded that glyphosate is unlikely to pose a carcinogenic risk to humans.
However, in March 2015, the International Agency for Research on Cancer (IARC) classified glyphosate as a “probable carcinogen.” IARC stated that its determination was based on “clear evidence of cancer in experimental animals, limited evidence for cancer for humans from real-world exposures, of exposed farmers, and also strong evidence that it can damage the genes and other toxicological studies.”
A number of other recent determinations by IARC have perplexed both scientists and the public, most famously those pertaining to coffee, red meat, and cell phones. It has become clear that IARC takes a very different approach to assessing carcinogens than most other agencies. Rather than taking into account real-world exposure to an agent and what is known about its ability to cause cancer at various exposure levels – risk assessment – IARC’s assessment takes into account the potential to cause cancer under theoretical conditions – hazard assessment. This approach is considered outmoded and unscientific by many scientists and regulatory agencies (accepted manuscript Boobis A, Cohen, S, Dellarco V, et al. “Classification schemes for carcinogenicity based on hazard-identification have become outmoded and serve neither science nor society,” Regulatory Toxicology and Pharmacology, Oct. 2016).
Under IARC’s “hazard assessment,” weak and difficult-to-interpret findings can be given undue weight and can take precedence over other stronger types of evidence that do not point to a threat, such as that from epidemiologic studies in humans and whole animal studies.
Furthermore, for all IARC’s claims of impartiality and rigor, in the case of glyphosate, its assessment is slipshod when it comes to critically examining what the key human and animal studies actually show.
Animal carcinogen testing involves taking a given strain of mice or rats and dividing them into different groups – usually receiving different doses of the substance under study as well as a control group that is given a placebo. The experiment runs for the lifespan of the test animals – 1-2 years. As animals die, their tissues are examined by a pathologist. And at the end of the study, remaining animals are sacrificed, and their organs are examined. Benign and malignant changes in different organs are recorded.
What is crucial in evaluating the evidence from these studies is to look at all of the findings within each study for meaningful differences in the incidence of tumors between the exposed groups and the control group. A further expectation is that if tumors are caused by the agent, one should see a greater “tumor yield” at higher exposure levels, referred to as a “dose-response relationship.” It is also important to compare what is found in male test animals with what is found in female animals. Finally, one needs to compare the results of different studies conducted in the same animal species to see whether there is consistency. Above all, one is looking for consistency and for a strong signal indicating that there is unambiguous evidence of the effect one is looking for.
In an article in the European Journal of Cancer Prevention, this past August the biostatistician Robert E. Tarone critically examined IARC’s glyphosate assessment as well as the published articles cited by the Agency. Tarone spent most of his career at the National Cancer Institute (NCI) and then at the International Epidemiology Institute, which was founded by former NCI scientists. He was also a statistical editor for the Journal of the National Cancer Institute.
Tarone’s exacting reexamination of the studies evaluated by IARC is a “lesson from the master” on how to review evidence rigorously and not be fooled by what one wants to find.
According to Tarone, probably the worst thing IARC did was to ignore mouse data that contradicted the story line supporting carcinogenicity.
In addition, IARC highlighted certain results from studies of Sprague-Dawley rats, as Tarone describes:
The highlighting of selective marginally significant tumor increases in a single study without noting the complete absence of supporting evidence of tumor increases in two other studies using the same rat strain is a highly questionable scientific practice. Once again, a synthesis of the data from all three rat studies does not provide evidence in support of the hypothesis that glyphosate is associated with increased liver or thyroid C-cell tumor rates in Sprague–Dawley rats.
This is only one of the many examples in which the results highlighted by IARC do not withstand careful scrutiny.
Tarone concludes his discussion of animal studies as follows:
Glyphosate would not have been classified by IARC as a probable human carcinogen except for the Working Group’s conclusion that there was sufficient evidence of carcinogenicity in animals. When all relevant data from the rodent carcinogenicity studies of glyphosate relied upon by the Working Group are evaluated together, it is clear that the conclusion that there is sufficient evidence that glyphosate is an animal carcinogen is not supported empirically. Even a conclusion that there is limited evidence of animal carcinogenicity would be difficult to support on the basis of the rodent carcinogenicity assays of glyphosate reviewed by the IARC Working Group.
As far as the epidemiology is concerned, as mentioned earlier, there is a general consensus that the evidence that glyphosate is a human carcinogen is weak. IARC points to an association of glyphosate exposure with an increased risk of non-Hodgkin’s lymphoma (NHL). But Tarone comments:
…the case made by IARC for a possible role of glyphosate in the etiology of NHL is quite weak. For example, the only significant finding reported for NHL and glyphosate in a US study (De Roos et al., 2003) is of questionable evidentiary weight. Glyphosate was one of 47 different pesticides evaluated for associations with NHL in this pooled analysis of case–control studies, and each pesticide was assessed using two different statistical methods. A significant association was reported between glyphosate and NHL for only one of the statistical methods applied.
In other words, IARC selected one result from among the many comparisons arising from applying two methods to the analysis of so many different pesticides. This selecting of a favored result exaggerates the significance of any risk due to glyphosate.
IARC’s questionable finding that glyphosate is a “possible carcinogen” has been taken up by advocates and NGO’s concerned about pesticides, including the National Resources Defense Council. A bizarre episode showcases the clash of different interest groups and interpretations regarding glyphosate. In October 2015, the EPA briefly released a report, labeled “final report,” stating that glyphosate is not likely a carcinogen. But the agency quickly took the report down, saying it was a draft that was not meant to be published. The EPA’s unexplained actions have provoked a firestorm of suspicion and charges that the agency is caving into anti-pesticide activists, as well as counter-charges that the agency is unduly influenced by commercial giants like Monsanto. A new meeting of the EPA Science Advisory Panel is scheduled for December. But changes to the make-up of the panel have provoked concern as to whether the Agency will give an unbiased assessment of the issue. A Congressional Committee is now investigating the EPA’s process regarding glyphosate.
All of this points up just how politicized questions like the safety/carcinogenicity of glyphosate have become. The subtle and difficult-to-interpret results of animal experiments and studies of agricultural workers easily lend themselves to what different specialists with different points-of-view may wish to find in them. Meanwhile, the overlay of strong beliefs and ideological commitments threatens to obscure what the science has to say on a question of enormous economic importance.
What is at stake in this latest iteration of the clash over environmental threats is enormous. First, there is the possibility that a product that is cheap, safe, and effective will be restricted or banned, reducing crop yields and requiring the substitution of products about which less is known and which may pose a greater danger. Second, controversies like this cause unnecessary confusion and alarm in the public and divert attention from issues that really do matter. Third, by doing so, they add to the already considerable distrust of science.
This article originally appeared in Forbes as “While Unlikely to Be Carcinogenic, The Herbicide Glyphosate Is A Symptom Of A Deep Social Pathology” and has been republished with permission from the author.
Geoffery Kabat is an epidemiologist at the Albert Einstein College of Medicine and the author of Getting Risk Right: Understanding the Science of Elusive Health Risks. Follow him on Twitter @GeoKabat.