About that GMO feeding study in rats... yes, THAT one

Ever since it was published, I'd been wanting to do a takedown of Gilles-Eric Seralini's much-discussed paper finding that rats fed Monsanto's Roundup corn developed tumors at a greater rate than a control group of rats fed non-GMO (genetically modified organism) corn. It turns out I don't really need to write my own article, since plenty of people have done a great job of writing it for me. I really liked this one from the LA Times by Michael Hiltzik.

I'm going to do some choice copy and pasting here:

The research in question is a paper published a few weeks ago by a team led by French biologist Gilles-Eric Seralini. Its findings were explosive: Laboratory rats fed for up to two years on genetically modified corn of a type widely used in the U.S. developed huge, grotesque tumors.

The paper claimed to be "the first detailed documentation of long-term deleterious effects arising from the consumption" of the corn. Seralini found very similar effects in rats fed high dosages of Roundup, a widely used pesticide that the corn had been engineered to tolerate, and in rats fed a combination of the corn and Roundup.

Holy crap, one might rightly think. That's terrifying. But remember that thing I wrote about a few weeks ago about how something being published doesn't necessarily mean it's correct, true, or backed by good science, and that's why we need more double-checking of our colleagues' work? Unfortunately, Seralini's paper is so obviously flawed that even a grad student like myself could pick out mistakes, and I'm astonished it was published at all. For one thing, even the way that it is being billed as some kind of vanguard study, for the first time observing the effects of GMOs in the diet, is entirely false (emphasis mine):

By the way, Seralini's paper isn't the first long-term study of genetically modified foods in the American diet, by a long shot. The same journal that published Seralini's paper (Food and Chemical Toxicology) published a survey of 12 studies of genetically modified corn, soybeans and rice tested on rats, cows, salmon or monkeys for up to two years, and in general found no evidence of any health hazards.

So in stark contrast to the fantastical findings in Seralini's paper, twelve other studies in other animals over the same duration found nothing, and that's just in one journal. This is an anti-GMO argument I see a lot -- for some reason there is this idea that there isn't research being done on the effects of GMOs. I don't know where people are getting this information, because there are tons of studies, both past and present, that mostly show nothing. Seralini's is an astonishing exception.

But okay. Let's not just dismiss it based in it being a lone voice of opposition against a stack of contrasting evidence; after all, sometimes you only need one iconoclast to drastically alter knowledge. Hiltzik nicely broke down the issues with the paper itself:

Among the most common critiques of the experiment is that Seralini used an insufficient number of control rats — 180 test rats were fed genetically modified corn, Roundup or both, but only 20 control rats were fed a purportedly normal diet. Critics say that's too small a control group to be statistically valid.

Of course you'll see more tumor growth in a group that's nine times as large as another group. The proportions are so off that the scientific community dismisses even percentage-based differences between the two groups, because those differences will be more likely to be due to chance. In fact, I could find no evidence in the paper that the authors even tested for statistical significance in the mortality and instances of tumor development between groups. This is a huge -- HUGE! -- red flag. They use misleading statements like "In the female cohorts, there were 2–3 times more deaths in all treated groups compared to controls by the end of the experiment and earlier in general." Announcing from which group rats died earlier is a completely meaningless statement when one group has 180 possible rats that can die and another group only has 20; furthermore, the phrase "2-3 times more deaths" is a tricky phrase because it sounds bad, but it actually only indicates quantity. Saying "2-3 times as likely to die" would indicate tested  -- and (hopefully) verified by statistical significance -- increased odds of death, but "2-3 times more deaths" only means 2-3 times as many rats died in the treatment groups than the control groups, which is again a big DUH since 180 rats vs. 20 rats, well -- you get the picture.

Moreover, the researchers identified no dose-related response: The rats fed higher doses of pesticide or GM corn didn't consistently get sicker than those fed lower doses. In fact, some rats fed higher doses did better than the others.

Seralini offered no explanation why rats fed a pesticide should show the same pathology as rats fed genetically modified corn but not the pesticide, although Roundup and genetically modified corn are totally different things with, one would presume, different effects on the organism. That points to another shortcoming of the paper, which is that there's no explanation or even hypothesis of why either impurity should produce the tumors Seralini found.

"They don't show a plausible biochemical or molecular mechanism for the effect," observes Kevin Folta, a plant biologist at the University of Florida who has written critically about the Seralini paper. "It happens with two completely independent treatments, the herbicide and the [genetically modified] product, and to get the same unusual response from both is beyond suspicious."

The ultimate complicating factor is that the strain of lab rat Seralini used is predisposed to tumors, especially mammary tumors. By about 2 years of age, 80% of these rats will have them, on average. Therefore, the longer the experiment proceeds, the cloudier the data become, because most of the rats would eventually be tumor-ridden anyway. In other words, the length of the study isn't a virtue, as Seralini contends, but may be a flaw.

The dose-response and lack-of-mechanism stuff is bad, but not ultimately completely damning. Papers are published all the time that show some kind of association, but don't yet propose a causal mechanism in the body that explains the observation. Usually, what that means is that the scientific community accepts such a paper as an interesting observation worthy of further study, not as a dogmatic assertion of biological truth, and that's an important distinction against the reception of this paper: this is suddenly being treated by the anti-GMO crowd like absolute truth, even without those biological explanations, and that's bad.

And why isn't there a really viable biological explanation for Seralini's observations? Hello! Did you see that last paragraph? Allow me to repeat: "The ultimate complicating factor is that the strain of lab rat Seralini used is predisposed to tumors, especially mammary tumors. By about 2 years of age, 80% of these rats will have them, on average." Consider that with the fact that, as the article mentions, Seralini's control group was tiny, and you've got a pretty obvious foregone conclusion. Of course rats that grow tumors anyway will grow tumors when you feed them, well, anything.

Regarding Prop 37 itself, which Hiltzik argues against in his article: the scientist in me thinks it's really silly, because in my view the arguments against GMOs hedge into woo-based misinformation about how it has to be bad for you because it's not natural. It's that, or else people don't like Monsanto so they lash back at their GMO products. The pragmatist in me, though, thinks people can just have their labeled food. They can choose to eat what they want, even if it's based on exaggerated misinformation, because why not. In most other instances, I support people's decisions to control matters regarding their own bodies (being pro-vaccination is a notable exception.) I'm not entirely convinced, as the No on 37 lobbyists claim, that it will lead to increased food taxes/costs, but that's something I'll need to look into a little more. In short, I can't really endorse this proposition, because I feel it's based on quackery and fluff, but if labeled food increases the precious comfort and perceived bodily autonomy of the people of California, then all right hippies -- have at it.