I'm sure many toxicologists around the world were a little puzzled by the recent carcinogenicity category assigned to glyphosate (i.e. the active ingredient in "Roundup" and found in over 750 products worldwide) by the International Association for Research on Cancer, the specialized cancer agency of the World Health Organization. (By the way, it was "Group 2A," meaning "probably carcinogenic to humans.")
Some of the confusion and speculation in the subsequent news stories (now over 16,000 strong) is likely rooted in past carcinogenicity reviews and what may be an incomplete consideration of the evidence. Settling speculation on whether the latest chemical threat can actually lead to the biological outcome of concern is best done through a M³ODE ANALYSIS (i.e. Mode of Action) approach.
Past Carcinogenicity Reviews
First and foremost, the EPA has reviewed this as the primary regulator in the US. The IRIS program in 1993 determined that glyphosate was not classifiable due to lack of animal data, but that there was no evidence it was genotoxic. EPA is also currently performing "registration review" on glyphosate and additional details are expected this year. [WHO, through its International Program on Chemical Safety, agreed with this conclusion in its 1994 review, and again in 2004, through the Joint Meeting of the FAO Panel of Experts on Pesticide Residues in Food and the Environment. ]
Most recently, the federal agency for risk assessment in Germany performed a comprehensive review in 2014 and concluded the following:
This is not an exhaustive review of the cancer reviews performed by major regulatory programs around the world. It must be noted that IARC reviews are "qualitative scientific reviews and are not meant to be formal human risk assessments" (such as the one performed most recently by Germany and quoted above) and the conclusions drawn by IARC - which shall appear on more detailed form as a Monograph at some later point in time - appeared appear draw most heavily from three cohort studies (one from the United States, one from Sweden and one from Canada) reporting increased risk of non-Hodgkin lymphoma, a group of related blood cancers. While the Agricultural Health Study, a very large cohort of ~ 80,000 of farmers in North Carolina and Iowa studied by the National Cancer Institute, "did not show a significantly increased risk of non-Hodgkin's lymphoma," this was clearly not strong enough evidence for NCI scientist and IARC Working Group Chair, Aaron Blair.
Incomplete Consideration of the Evidence?
Again, the full Monograph will be released, but the sneak peek of the analysis provided in the Lancet over the weekend gives the impression that more recent evidence specific to glyphosate (and not part of a larger study examining groups of pesticides) was not considered:
- Private consultant Larry Kier published a review of the available human biomonitoring study data for glyphosate-based formulations (i.e. jugs of actual Roundup, including the other formulants, like surfactants) in the March 2015 Critical Reviews in Toxicology. His conclusion was these formulations "do not appear to present significant genotoxic risk under normal conditions of human or environmental exposures," which was found to be in agreement with his previous assessment of the available experimental genotoxicity data. These papers strongly suggest a non-genotoxic mode of action, if anything.
- The most recent and comprehensive analysis of both public and private (i.e. data submitted by companies to regulatory authorities) experimental glyphosate data in rat and mouse was published by the industry Glyphosate Task Force also in the March 2015 Critical Reviews in Toxicology. They review nine (9) rat studies and (5) mouse studies performed from 1981 through 2009, along with other data. The publication is very thorough and worth reviewing, but beyond my attention span here. The final conclusion is worth noting, references left in for your pleasure:
- "The absence of a glyphosate-related mechanism for
carcinogenesis, the huge volume of genotoxicity data studies indicating no likely mutagenic or DNA-reactive potential (Kier and Kirkland 2013), combined with the lack of epidemiological evidence for glyphosate-induced cancer (Mink et al. 2012), and the lack of carcinogenicity in multiple rodent carcinogenicity assays, are depicted in a causal inference grid in Figure 2, as put forth by Adami et al. (Adami et al. 2011). The overwhelming weight of the available evidence, demonstrating a lack of both biological plausibility and epidemiological eff ects, draws a compelling conclusion that glyphosate ’ s carcino- genic potential is extremely low or non-existent." [causal inference grid found below]
- "The absence of a glyphosate-related mechanism for
- The science & engineering consulting firm Exponent published two very comprehensive reviews of available epidemiologic data for both cancer and non-cancer health outcomes in 2012 and 2011, respectively. Their conclusions for both were that no "consistent pattern of positive associations indicating a causal relationship" could be found; however, they noted much uncertainty surrounding exposure measurement in general for any individual pesticide and recommended greater use of biomarkers in future epidemiologic research.
- Two scientists from IARC performed a systematic review and meta-analysis of over forty (40) studies across 30 years examining pesticide use and non-Hodgkin lymphoma published in 2014. They found "associations between pesticides and NHL subtypes were reported; B cell lymphoma was positively associated with phenoxy herbicides and the organophosphorus [sic] herbicide glyphosate" in a "handful of papers." They concluded: "Despite compelling evidence that NHL is associated with certain chemicals, this review indicates the need for investigations of a larger variety of pesticides in more geographic areas, especially in low- and middle-income countries, which, despite producing a large portion of the world’s agriculture, were missing in the literature that were reviewed." However, the finding specific to glyphosate is that "[t]here was a positive association between exposure to organophosphorus [sic] herbicide, glyphosate, and B cell lymphoma;" however, the 95th percentile surrounding the relative ratio for the meta-analysis was 1.1 to 3.6 -- when this statistic includes 1.0 on the low end, it is likely not a statistically significant finding. [Ed. note: glyphosate is a "glycine phosphonate" and not an organophosphorus.]
Looking at the causal inference grid of Adami et al. (2011), both biological plausibility of glyphosate being a carcinogen is low as is the level of epidemiologic evidence, making the overall likelihood of a causal relationship between glyphosate and carcinogenicity "unlikely."
In the case of chemicals where there is a lack of biological data and perhaps a compelling case made by the epidemiology data, there are simple tools and tiered approaches to understanding the biological pathway through which effects might operate (i.e. mode of action, toxicity pathway, adverse outcome pathway ... however you decide to describe M³ODE). In the specific case of glyphosate, this is probably overkill, but use of toxicogenomics approaches to screen the entire transcriptome for gene expression activity can be a "quick and dirty" approach to understanding if a chemical operates through some selective mode of action (Or many. Or none.) There have been several such attempts in the literature as well as supporting data submitted to EPA to help understand glyphosate's potential role as an endocrine disruptor (i.e. Endocrine Disruptor Screening Program).
In any event, the critics of any chemical may not be silenced with additional studies demonstrating or refuting the biological plausibility of health effects; however, collecting data in a tiered fashion that is fit for purpose (for the regulatory need, the social need, etc.) can be done using these modern methods that only use animals as a last resort. Following the core principles of M³ode, M³argin and M³anagement can get you there.