This study has been debunked. Instead of posting a bunch of articles no one will read I'll just post this round up of the missteps this study has taken. It's better than any one of the articles I've read on the subject anyway.
Alright, I'm going to run through some of the problems from the study, so people don't have to dig through links themselves.
First, issues with the study author and initial reporting. The study was done by Gilles-Eric Séralini, who is a well known Green Peace funded anti-GMO activist (
Greenpeace Financed Scientist Wins Defamation Judgement in France - Hit & Run : Reason.com). While being associated with a group that has prior interest isn't a guarantee that his work is bad, it does raise an immediate red flag that says you should look at it a bit more closely. Additionally, while he does say that there is no conflict of interest in the paper, you should look at a amazon search in his name (
Amazon.com: Gilles-Eric Séralini: Books, Biography, Blog, Audiobooks, Kindle). He does make money off of the anti-GMO food position. Again, see the previous disclaimer.
With the initial reporting, the journalists who had access to the study in advance had to sign a contract saying that they would not talk to any scientists who were not directly involved in the study. This means that they had to rely on their own knowledge of GMO crops and study design for whatever they wrote. This generally means all early reporting will be biased in favor of the study's author's positions.
Second, problems with the study. The first one that comes to mind is that they had a control group of only 20 rats, which was then subdivided into two groups of 10 male, and 10 female rats. This is an incredibly small control. He also used 10 groups of 20 rats, similarly divided to test each of his 10 different variables (Which is rather complex). A major problem with this is that Sprague Dawley rats have a very high tumor incidence rate naturally. By very high, we are talking approximately 72% of female Sprague Dawley rats develop tumors within 2 years (
Spontaneous endocrine tumors in Sprague-Dawley rats., Suzuki et al, 1979; (
Spontaneous endocrine tumors in Spra... [J Cancer Res Clin Oncol. 1979] - PubMed - NCBI). If he was using a group of 1000 rats for each one, you could expect numbers very different from 72% to indicate something meaningful. Because he is looking at effectively 10 rats, it is not unreasonable for only 50% to get tumors, and its not unreasonable for 100% to get tumors, because probability.
They don't appear to have done any blinding on the tests, which can allow researcher bias to show through when analysing the rats. Blinding is a basic thing used in noncontroversial research. It certainly should be used in something like this.
Going back to the complexity of the experiment, it looks like the measured upwards of 47 different parameters in their testing. However, their statistical analysis is bad. Really bad. They didn't do basic things like checking for the statistical power of the design. They also leave massive gaps in their statistical methodology, like not giving any info on how they looked at mortality or tumor data. The appear to have done statistical fishing for interesting results, while ignoring a lot of standard analyses to see if their results mean anything. (
in which I blow a gasket and get very uppity about this GM-food study which appears to have everyone going nuts at the moment - michaelgrayer's posterous)
Another problem, is that they quote ”All data cannot be shown in one report and the most relevant are described here”. This is not done. Ever. This means that the authors almost certainly cherry picked only the best results to show (
https://embargowatch.wordpress.com/...onsor-engineered-embargo-to-prevent-scrutiny/).
They tested both response against GM corn, and corn with the roundup herbicide, and reported THE EXACT SAME EFFECT for two vastly different treatments. There are ZERO mechanisms that can explain this, but they came up with it anyway.
They didn't control for the total amount of food consumed, so the rats could easily gorge themselves, which can effect the rate of tumor incidence (
Effects of Diet and Overfeeding On Body Weight and Survival in the Rodent Bioassay: the Impact On Pharmaceutical Safety Assessment, Keenan et al.. (
Effects of Diet and Overfeeding On Body Weight and Survival in the Rodent Bioassay: the Impact On Pharmaceutical Safety Assessment). They didn't control for fungal contaminants on the food, which can effect the rate of tumor incidence(
Toxicity and carcinogenicity of the Fusanum monilzforine metabolite, fumonisin B1, in rats, Thiel et al, (
Toxicity and carcinogenicity of the Fusanum monilzforine metabolite, fumonisin B1, in rats).
They didn't control for the fact that standard rat chow can frequently contain GM corn, so even their controls were probably exposed to it (
Proof Perfect That The Seralini Paper On GM Corn And Cancer In Rats Is Rubbish - Forbes)
Their graphs are needlessly complex, which obfuscates them to a massive amount. This means that it is very hard to try to draw your own conclusions from them, so you need to rely on theirs.
They didn't do any dose response analysis to see if the different doses actually mattered. For all they can tell, 11%GM food causes the exact same response as 33%GM food.
In short, this study was
bad. It doesn't matter what your ideological position is, drawing any conclusions from this study would be wrong.
Other places to read more, they I used heavily for this writeup:
Bad science about GMOs: It reminds me of the antivaccine movement – Respectful Insolence
NeuroLogica Blog » The GM Corn Rat Study
Study linking GM crops and cancer questioned - health - 19 September 2012 - New Scientist
