>

The Monsanto data: a (semi) medical perspective

NOTE: This post has some serious inaccuracies and overwroughtness; I’ll be altering it in the next couple of days.

There’s some buzz about a new study in the International Journal of Biological Sciences, which states that the studies Monsanto used to justify the use of three varieties of genetically modified corn are flawed.

Thing is, in another life I’m a medical writer (mostly whoring for the drug companies, but that’s a story for another time), which means long bitter years of experience reading medical articles, so I figured I’d take a look and see if I could make sense of what’s going on.

The basic facts:

  • To meet EU requirements, Monsanto did studies on three strains of genetically modified corn (MON 810, MON 863, NK 603). Rats were given the GM corn (11% or 33% of their diet) or control corn for three months.
  • The studies were done by an independent lab, but the data were analyzed and interpreted by Monsanto researchers, who concluded that the corn was safe.
  • Greenpeace funded a reanalysis of the data of one study, and then another reanalysis of all three, which is the study that’s getting the attention. These reanalyses found problems with Monsanto’s analysis and conclusions, and found several statistically significant effects of the GM corn.
  • The first reanalysis was in turn contradicted in Douro et al by an expert panel, which didn’t include Monsanto employees but may have been paid for by Monsanto (I don’t have access to the full article).

Some criticisms of the Greenpeace reanalyses, in descending order of compellingness (imho), are:

The Greenpeace authors cherrypick statistical significance. “Statistical significance” is when a given result is very unlikely and therefore can be treated as if it didn’t happen by chance. (If you lose to a royal flush, that’s bad luck. If you lose to three royal flushes in a row, it’s time to start stabbing). The conventional cutoff between “random noise” and “real effect” is the .05 significance level, which means that a result is considered significant if it would only happen by chance 5% of the time or less. The corollary is that 5% of truly random results will be statistically significant, so if you look at 500 variables, around 25 of them will be statistically significantly different just by chance. Looking only at these 25 variables (which is, in effect, what the Greenpeace researchers do) is misleading.

The effects they note are often not dose dependent. For instance, female rats with 11% of their diets made up of MON 810 corn lost kidney weight, but female rats whose diet was 33% MON 810 corn didn’t. Intuitively, if you’re eating a poison, more should be worse, and if it’s not, maybe it’s not a poison. The Greenpeace authors say:

  • Many of the effects are in fact dose dependent
  • Effects on metabolism are not always expected linear with dose
  • With only two doses studied, it’s hard to establish a dose effect

The effects are often not time dependent. The rats were evaluated at 5 and 14 weeks; many effects that were seen at week 5 were not seen at week 14. Although intuitively, 14 weeks of a poison should be worse than 5 weeks, it’s actually entirely conceivable (imho) that early effects would be more dramatic than later ones. If you suddenly started drinking a fifth of vodka a day, it would lay you flat on your back the first day, but after a while your body would adjust until the vodka finally destroyed your health years later. (For that matter, if I remember Super Size Me correctly, Morgan Spurlock’s vital signs went way out of whack early on and then improved somewhat as he got used to a crap diet.)

Many of the results were not demonstrated in both sexes. This is where the Greenpeace researchers make a reasonably compelling case: they say there are in fact clear sex-specific effects, which is what you would expect from a substance that has effects on sex hormones, as many substances do (although the original study didn’t look for hormonal effects themselves).

Still, some of the criticisms are valid; the Greenpeace researchers are clearly, in at least some cases, torturing the data until it confesses. So who do we believe?

Drum roll please:

The Greenpeace authors.

The Greenpeace authors’ main conclusion isn’t that Monsanto’s corn is harmful. It’s that the data don’t prove Monsanto’s corn is not harmful. Which is 100% true. Monsanto’s studies were too small, and too short, to prove anything except that GM corn can be fed to a small number of rats for three months without immediately killing them. Long-term effects? Cancer? Reproductive effects? Birth defects? These studies prove nothing either way. (Heck, if you gave a small number of rats cigarettes for three months, you might not find any cancers.)

And the burden of proof should be on Monsanto. Two of the three strains tested—and I find this simply incredible—have been modified to express a toxin as an insecticide. So instead of dumping pesticide on the food (which is generally not the best idea), Monsanto has put pesticide in the food. The fact that many countries have allowed this to happen at all is bizarre on the face of it. The fact that they let it happen on the basis of one small rat study per strain is insane.

As I said, I’m familiar with pharmaceutical studies; when a pharmaceutical is tested, small studies on animals are just the first step. Everyone in medicine understands that a drug that seems safe at first may turn out to be deadly, a result in one small study may disappear in the next, effects in rats may not hold true for humans, and so on. In real medicine, anyone who drew broad conclusions from a small rat study, or even a few rat studies, would be laughed out of the profession. And yet Monsanto does just that. Here’s the conclusion of the original MON 810 study:

This study complements extensive agronomic, compositional and farm animal feeding studies with MON 810 grain, confirming that it is as safe and nutritious as grain from existing commercial corn varieties.

There is simply no way that conclusion is warranted by the data. (For the record, the expert panel didn’t go nearly so far—they merely said that neither the original studies or the first reanalysis gave conclusive evidence of problems, which is true as far as I can tell.)

Now: there’s another study, undertaken by the Austrian government, that did much better—it studied several generations of rats, and while it found that GM corn had an effect on reproduction in the third and fourth generations (a result that has been criticized), the study is remarkable for all the problems it didn’t find—for instance, rats fed GM corn actually lived slightly longer than the other rats (although that was probably statistical noise, it still means that GM corn is unlikely to shorten a rat’s life).

So the point is not that GM corn is horrible; the point is that things like this should really be studied a lot better before we release them into the environment and the food supply. If these strains really are harmless, that only means we got lucky this time.

And anyway, if Monsanto really believes that GM corn’s safety has been proven, why not do a legit study to confirm it? One obvious possibility is that Monsanto is not nearly as confident as it pretends to be; and simply doesn’t want to know what a real study would reveal.

But I do.

PS: I was basically done with this post when I read this, which makes many of the same points with a different take on them. My only disagreement is with the idea that the funding source calls the Greenpeace results into question. Or rather, I don’t disagree with that at all (which is why I called them the “Greenpeace authors”), but the fact is, the same thing applies when Monsanto is funding a study. What’s needed is truly independent analysis.

Leave a Reply

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>