Nov 10 2017

Glyphosate Not Associated with Cancer

IARC-Headquarters_ExteriorIn March of 2015 the International Agency for Research on Cancer (IARC), part of the World Health Organization (WHO), published their assessment on glyphosate, Monsanto’s popular weedkiller, classifying it as 2a – a probable carcinogen. This was like red meat to the anti-GMO crowd, and even sparked class action suits against Monsanto and may lead to banning use of the chemical in the EU.

There were significant problems with the IARC report, however. First – it is at odds with every other expert review of the scientific literature on glyphosate. I review the evidence here, citing many expert panel reviews, all conclude that the evidence does not support a link between glyphosate and risk of cancer. The IARC conclusion is a clear outlier, which reasonably prompts questions as to why their designation stands out.

We also need to put the IARC classification of 2a – probable carcinogen, into context. This is the same classification that the IARC gave to drinking hot beverages or eating red meat. Overall they tend to err on the side of caution when making their classification.

But there were problems that go beyond where the IARC sets their threshold for “probable.” Two main criticisms have emerged. The first is the lack of transparency. Reuters has published a series of articles on the issue, outlining, for example, that when the EPA reviewed the safety of glyphosate they also published a 1300 + page document that outlines the entire deliberative process. The IARC produced no such document.

Further, Reuters was able to obtain copies of the draft report and shows that the final report differs in significant ways. They found 10 major changes or omissions from the draft to the final copy, every one in the direction of emphasizing the risks of glyphosate. It is not known who made these edits, and the IARC responded by essentially instructing their scientists not to discuss the confidential deliberative process.

Far more important, however, is the accusation that the lead IARC scientist knew of unpublished data (because he was involved in the research) that showed no correlation between glyphosate and cancer, but this data was not considered in the review. So the lead scientist excluded his own data from the final analysis.

That data has now been published. 

The study comes from the Agricultural Health Study. Here are the results:

Among 54 251 applicators, 44 932 (82.8%) used glyphosate, including 5779 incident cancer cases (79.3% of all cases). In unlagged analyses, glyphosate was not statistically significantly associated with cancer at any site. However, among applicators in the highest exposure quartile, there was an increased risk of acute myeloid leukemia (AML) compared with never users (RR = 2.44, 95% CI = 0.94 to 6.32, Ptrend = .11), though this association was not statistically significant. Results for AML were similar with a five-year (RRQuartile 4 = 2.32, 95% CI = 0.98 to 5.51, Ptrend = .07) and 20-year exposure lag (RRTertile 3 = 2.04, 95% CI = 1.05 to 3.97, Ptrend = .04).

This is the best and largest set of data to date, and it was negative. The possible association with AML requires further discussion, as I am confident it will be seized on by those with an anti-glyphosate agenda. First and most importantly, this association was not statistically significant. This means it is almost certainly noise in the data. Given the number of possible correlations being examined, non-significant possible correlations are almost inevitable.

There are two other reasons to think this association is noise – there was no difference between the 5 year and 20 year exposure lag. If this were a true cause and effect, we would expect the lag time to matter. Even more significant, however, is the fact that previous possible correlations were between glyphosate and non-Hodgkins lymphoma (NHL). That is the association that led to the IARC classification. There was no association with NHL is this data, just a non-significant association with AML. This is exactly what we expect to find with random noise – different correlations in different sets of data. Such correlations don’t mean anything until they are replicated in an independent set of data.

So the bottom line is that this large data set is essentially negative regarding any association between glyphosate and cancer. If the IARC had taken this data into consideration it may have (and it seems should have) changed their conclusions. They knew about this data, but chose to ignore it.

The issue of glyphosate is controversial because it has become a focal point for an ideological struggle. For the anti-GMO movement it is the poster-child of corporate malfeasance. For corporations this is an example of activist government overreach.

I tend to think that both sides are correct, at least to an extent. We should not trust corporations, meaning that we should not just assume they will be good corporate citizens, never abuse their power, or that they will consider the public interest over their shareholders. There is overwhelming evidence that, generally speaking, this is not a good assumption. Corporations look after their own bottom line. That is why we need regulations, transparency, and oversight to protect consumer and public interests.

I also don’t think we can trust activist organizations, nor can we assume that government agencies will act without ideological bias. Again, history tells a very different story.

What we need, therefore, are professional disinterested reviewers. We need scientific experts to review objective evidence, and investigative journalists to make sure there is transparency. They don’t always do their job optimally either, but the whole system acts as a set of checks and balances.

The story of glyphosate and the IARC review is a microcosm of all this. We see multiple different interests, each with a different narrative interpretation of reality, fighting over what is, at the end of the day, a scientific question. What is the safety of glyphosate in the context of how it is used and compared to other alternatives? The best we can do is to have multiple independent experts review all the evidence and give us a transparent assessment. If a consensus emerges and that consensus includes the opinion that there is sufficient evidence to reach a conclusion, then that conclusion is probably the most reliable answer we can get.

In the case of glyphosate we actually have a large set of data with multiple independent reviews concluding it is relatively safe as used, and is superior to most other herbicides. The IARC review is an outlier, and the process used has come under significant criticism suggesting bias.

In any case, the recent published data from the AHS renders all previous reviews obsolete. This new data argues strongly against any link between glyphosate and cancer. In light of this, the IARC should update their classification, as their now obsolete classification is actively being used as a basis of lawsuits and regulations.


32 responses so far