Glyphosate (often known by its original brand name, Roundup) is the most widely used agricultural chemical in the world. What, if anything, is it doing to people?
I bring this up because of some recent (and seemingly contradictory) news items. A group of farmers is suing Monsanto, the compound’s original developers, because they claim that the company knew of (and deliberately minimized) risk of non-Hodgkin’s lymphoma from exposure to it. At the same time, two branches of the UN, the World Health Organization and the Food and Agriculture Organization) have come out with a statement that glyphosate is “unlikely to pose a carcinogenic risk in humans”. And this only a year after another UN agency, the International Agency for Research on Cancer, stated what looks like the exact opposite, that it could “probably” be a cause of cancer in humans. Later on last year, the European Food Safety Authority said that glyphosate is “unlikely to pose a carcinogenic hazard”.
So the situation is confused, and that confusion is compounded by the amazingly low signal/noise on this subject if you start searching for information. The list of web sites that have “glyphosate = death” articles on them is beyond numbering, because a lot of natural food advocates hate the compound and hate Monsanto. What you have to try to do, though, is figure out how good the reasons are for that hate. The problem is, most of these sites seem to be repeating the same allegations, and thus have little to add to the debate. The sources that a lot of them are using for their arguments are not helping much, either, such as the papers of Stephanie Seneff and co-authors.
I went into one of those in detail here, and after looking into the case that it makes, I am willing to dismiss out of hand anything else Seneff has to say on the subject. It’s that bad. You will hear that “MIT researchers” have “proven” that glyphosate does X and Y and Z, and that this work is “published in peer-reviewed journals”, but nothing like that is true. Seneff has done no actual studies on glyphosate; she doesn’t work in a lab. Those papers are rehashes of stuff from the literature, piles of speculation and dot-connecting, and they’re invariably published in low-quality pay-to-play journals that do little or no actual refereeing of their contents. And their content is yet another problem – as shown in that link above, the paper that goes on and on about glyphosate’s effect on gut bacteria does not manage to cite any of the papers that have studied. . .glyphosate effects on bacteria. It not only doesn’t cite them, it seems to pretend that this research does not even exist, probably because all these papers contradict the fundamental ideas that Seneff’s tower of speculation is built on. She’s going around now saying that half of all children are going to be autistic (because of glyphosate), and that it’s also a root cause of not only cancer, but Alzheimer’s and a whole list of other diseases. If your knowledge of glyphosate’s toxicology comes only from reading the Seneff papers, I feel pity for you, because you have a lot of work ahead of you if you want to actually understand anything about it.
All right, with that out of the way, what are the facts, if any? First off, let’s take a look at those two UN-based statements. This article at Wired does a good job explaining what’s going on: the difference is that the IARC is looking at the question from a “Is there any possible way, under any conditions at all, that glyphosate could be a carcinogen?”, while the FAO and WHO are giving an answer to the questions “Is glyphosate actually causing cancer in people?”. If you’re not used to thinking like a toxicologist, then these two questions might still seem to be pretty similar. In fact, if you went out and took a public opinion poll and asked everyone if we should make sure to ban anything in food that has been linked to causing cancer, you’d get a large majority of vigorous agreement. But the problem is, as long as you’re willing to allow pretty much any conditions, pretty much anything can conceivably cause cancer.
A toxicologist interviewed in the Wired article makes a great analogy to explain hazard versus risk, which is what we’re talking about here. Sharks are a hazard. They are fierce predators with sharp teeth, and they most definitely have attacked humans in the past. But for most people, sharks are not much of a risk. “Risk”, technically speaking, refers to your chances of being harmed under real-world conditions, while “hazard” refers to the potential for harm. A large shark has plenty of potential to do damage to you, but should you be worried about that happening? It depends. If you’re sitting in your car, probably not (you have other, more immediate risks to worry about). If you’re swimming off a tropical beach, though, that may be another matter. And if you’re doing so after cutting your foot and leaving a trail of blood in the water, I would seriously consider the shark possibility and act accordingly: your risk has increased to what most people would find unacceptable levels.
Fine – but what about your risk when, say, you visit an aquarium? Remember, the hazard a shark poses has not changed during all this – he’s still hungry and he still has a mouthful of teeth. The shark is the shark. Your risk of being bitten by him has, in fact, increased a great deal when you visit an aquarium – you’ve gone from a place (your home, your car) where there are (one assumes) no sharks whatsoever, and now you’re in the same room with one. True, you’re separated by a thick pane of glass, and true, it’s hard to come up with a plausible chain of events that would lead to said shark chomping on your leg, but it’s undeniable that this is much more possible than it was in the parking lot outside the aquarium or back in your bed. The odds are still vanishingly low, and not many people worry on their visit the aquarium about the chances of being bitten by a shark (nor should they) – but if you write things up from the right angle, that visit can look wildly dangerous. “RISK OF SHARK ATTACK NOW INCREASED BY FACTOR OF ONE MILLION!”
This may sound like a silly example, but that’s basically what happened recently with the IARC and its announcement on bacon being a cause of cancer. Under real-world conditions, eating a normal amount of bacon raise your risk of colorectal cancer by an amount too small to consider. But it does appear to be raising it by a reproducible, measurable amount, and therefore bacon (and other processed meats) are in the IARC’s category 1. (Here’s an article on those categories). It’s important to note that some hypothetical substance that reproducibly, in human studies, gives anyone cancer every single time they touch it would also be in category 1, the same as a hypothetical substance that reproducibly, in human studies, raises a person’s risk of cancer by one millionth of a per cent. Same category. These categories are not arranged by relative risk – they’re arranged by how good the evidence is. Glyphosate is in category 2A, which means that there is evidence from animal studies, but limited/insufficient evidence from humans as of yet. Here’s the list of everything in category 2A, and you’ll note that it’s a mixed bag, with entries that range from ethylene dibromide to shift work that disrupts circadian rhythms. So yes, by the standards of the available evidence, glyphosate is in the same cancer hazard category as working the night shift, or working as a hairdresser. (Working nights as a hairdresser is presumably worse). You will not see these comparisons made on the web sites that proclaim Roundup as a known human carcinogen, but they’re completely appropriate by IARC standards, and they’re the only agency by whose standards glyphosate is a cancer hazard.
Now to the case of those farmers. They’re suing, as mentioned, because of a putative link between glyphosate and non-Hodgkin’s lymphoma, NHL. When you look into the medical literature, the best you can say is that no relationship between these forms of cancer and glyphosate exposure can be established, based on the available evidence. Some associations from meta-analysis of the literature barely reach statistical significance, but at a level that could easily be due to bias or confounding variables. If you want everyone to take numbers like this at face value, then all of medical science is going to be thrown into confusion, because we normally need a lot stronger evidence than this. Here’s the IARC’s meta-analysis for comparison. They rank the evidence for an association between glyphosate and B-cell lymphoma higher, but that 2016 analysis linked first says that “Meta-analysis is constrained by few studies and a crude exposure metric, while the overall body of literature is methodologically limited and findings are not strong or consistent“.
As mentioned earlier, the IARC and the EFSA have disagreed about whether there’s any connection at all (here’s an EFSA viewpoint, and here’s the IARC response). My own take is that if it’s possible for toxicologists at this level of expertise to be in such disagreement on the subject, then the weight of the evidence must not be very strong. There’s been a recent call for glyphosate to be prioritized in current toxicology evaluations so that new data can be obtained, and it’s hard to disagree with that. It’s such a widely used compound that we need to try to settle this question – on the other hand, it’s such a widely used compound that if there were easily quantifiable risks, you’d figure that we wouldn’t still be arguing about it at this point.
For what it’s worth, new cases of non-Hodgkin’s lymphoma went up in the US from 1975 to about 1995, and have been more or less flat since then, while the use of glyphosate has increased steadily during this entire time period. There are a lot of confounding variables in there, but if there were the sort of clear association that the more alarmed web sites like to tout, these numbers would not look the way that they actually do. What I would wonder about, though, are agricultural workers themselves – if there is any connection at all, that’s where you’d expect to see it. A study from Saskatchewan found an association between farm work and NHL, as have many others, but this seems to have been possibly linked to several pesticides, not exposure to glyphosate (a herbicide). In general, the NHL/farming link seems to be real, although the evidence for it has been described as “markedly inconsistent” and has not been pinned down to any particular variable. The IARC itself, in their meta-analysis which led them to classify glyphosate in category 2A, noted that “the wide variety of chemical and microbial exposures that occur simultaneously in agricultural production makes disentangling the effects of these factors challenging“.
So that’s the state of the art: there is, from what I can see, nothing very clearly linking glyphosate to human cancer. There’s certainly room for more evidence to come in, though, and it looks like we’re going to need it, because this is a topic that’s never going to go away until we have more data.