Seralini: GMO Maize and Giant Rat Tumors

Some very dramatic pictures have been making the rounds, purporting to illustrate the extreme danger of GMO food.  Images like this one:

which comes from this article claiming that the Seralini study shows how dangerous Monsanto's Round-Up ready maize is, and that this should be a wake-up call to the world to ban GMO foods in general.

I mean, you don't want a tumor larger than your head, do you?

I have some sympathy for the anti-GMO movement.  I think it is reasonable to proceed with some caution here.  I am not eager to consume new-to-our-species substances without compelling reason because this does at least open the door to unexpected detrimental effects.  Even less so if there are not long-term studies for it.  Worries about contamination of non-GMO crops seems legitimate to me, not because the GMO crops are necessarily a problem but because I am wary of actions that are difficult to reverse.  I just have a conservative (literally, not politically) mindset here.  But here I have laid out a steel-man of the anti-GMO movement, and not a description of what it really is.

I'm also not very well educated on the topic, and am weighing in seriously for the first time because one thing at least is really clear:

This Seralini study is bullshit.  I really couldn't say it better than this analysis by VIB in Belgium.  Please go read it and come back.  It's 11 pages, very layman accessible, and a good general education on how to conduct scientific experiments and why this one in particular is worthless.

I'm not going to go through all of those criticisms blow by blow.  I do want to draw particular attention to this:

The results of Séralini et al.’s experiment show that there were fewer

deaths among the male animals whose diet comprised food with 22 or 33% genetically modified maize (= negative control) (the left hand side of the figure below). This is remarkable given that the genetically modified maize was herbicide tolerant, and no new properties that may have had health advantages for the rats. We see the same result after Roundup was added to the drinking water. There were fewer deaths among the male rats that had drunk the highest concentration of Roundup than among those who had drunk pure water (the right hand side of the figure below). And this while Roundup certainly does not contain any known life-extending properties. The researchers should have taken these observations as a warning that there was something wrong with the experiment, because if these results were correct it would mean that consuming large amounts of genetically modified NK603 maize or Roundup would be a way to live longer. These strange findings are not interpretable because as noted previously there is something fundamentally wrong with the research design.

I built a computer program that mimics the trial structure using a random 57% chance of tumor development in any rat in the control group or any of the 9 experimental groups.  Just pure random noise, structured to duplicate just the female trials in the Seralini experiment (so 100 virtual rats.)   I ran this 10 times, and never once failed to have an experimental group with a higher tumor rate than the control group.  Sometimes the tumor rate in an experimental group was double that of the control.  I am, of course, failing to report negative results and ignoring all of the same problems that Seralini has been criticized for.  If this is how you run and report your experiments, you can make them say anything.  It would be dumb luck if you didn't get this result.  If your findings are not distinguishable from random fluctuations, you do not have findings.  Here is a randomly selected and typical result from the program, based on the naturally occurring tumor rate in these particular rats:

Trial:
Rat group 1: 6 tumors.
Rat group 2: 8 tumors.
Rat group 3: 6 tumors.
Rat group 4: 6 tumors.
Rat group 5: 4 tumors.
Rat group 6: 8 tumors.
Rat group 7: 5 tumors.
Rat group 8: 6 tumors.
Rat group 9: 5 tumors.
Rat group 10: 5 tumors.
-----------------------

Unless you have selected Group 2 or 6 as your control (I am using 1), you have experimental groups that show at least a 33% increase in the incidence of tumors, supposedly caused by whatever distinguishes those experimental groups from the control.  This is meaningless. It is a lie with numbers.

To put it all another way, if you design an experiment like this and selectively report your results like this, you could prove that literally anything causes giant tumors in rats during long-term trials, because the reason for these results is random variation in a population of rats with an already very high incidence of tumors.  Playing Mozart?  Giant rat tumors!  Beige walls on their cages?  Giant rat tumors!  Lots of affection?  Giant rat tumors!

This combined with the way negative results were ignored, criticism has been dodged, and only dramatic photos released really leaves me with the impression that the experimenters had an axe to grind and are being deliberately deceptive.  This is what it looks like when you cannot trust a source.

Here, from the Seralini website, is their 10 things you need to know about the Seralini study.

Not a single one of those 10 points is actually addressing any of the criticisms presented in the analysis I linked, though they try hard to make it sound like the criticism of their study is baseless.  This is an inadequate response at best, and I am inclined to read it as dishonest.

The only thing I find compelling about their arguments is the need for long-term studies and a healthy skepticism for what a negative result in a 90-day trial can really tell us.  But here, I must confess my personal ignorance of how to do this science properly, and note that the source agreeing with me has shown itself willing to compromise the truth to make a point.


From the Natural Independent article:

"The new Monsanto GMO study is considered to be the most comprehensive to date involving the feed/crops and Roundup, which is perhaps why it is drawing such a large reaction across the web from concerned consumers."

If that is a true representation of the results out there (and it may not be, I don't know), then it is an acknowledgement that the anxieties about this crop have no basis in evidence.  If you're anti-GMO and care about the truth, you need to stop referencing this study.