Bad science about GMOs: It reminds me of the antivaccine movement (revisited)

I never used to write much about genetically modified organisms (GMOs) before. I still don’t do it that often. For whatever reason, it just hasn’t been on my radar very much. That seems to be changing, however. It’s not because I went seeking this issue out (although I must admit that I first became interested in genetic engineering when I was in junior high and read a TIME Magazine cover article about it back in the 1970s), but rather because in my reading I keep seeing it more and more in the context of anti-GMO activists using bad science and bad reasoning to justify a campaign to demonize GMOs. Now, I don’t have a dog in this hunt (Forgive me, I have no idea why I like that expression, given that I don’t hunt.) I really don’t. I was, not too long ago, fairly agnostic on the issue of GMOs and their safety, although, truth be told, because I have PhD in a biomedical science and because my lab work has involved molecular biology and genetics since I was a graduate student in the early 1990s. I found the claims of horrific harm attributable to GMOs not particularly convincing, but hadn’t bothered to take that deep a look into them. It was not unlike my attitudes towards the the claims that cell phones cause cancer a few years ago, before I started finding dubious studies and looking into them and noted despite the utter lack of a remotely plausible mechanism and uniformly negative studies except for a group in Sweden with a definite ax to grind on the issue. None of this stops activists from likening cell phone companies to tobacco companies, the way antivaccine loons liken vaccine manufacturers to tobacco companies. Back then, I realized that there wasn’t really a plausible mechanism by which radio waves from cell phones could cause cancer in that the classic mechanisms by which ionizing radiation can break DNA molecular bonds and cause mutations don’t apply, but I didn’t rule out a tiny possibility that there might be an as yet unappreciated mechanism by which long term exposure to radio waves might contribute to cancer.

As was the case for the nonexistent cell phone-cancer link, there has now been a steady drip-drip-drip of bad studies touted by anti-GMO activists as “evidence” that GMOs are the work of Satan that will corrupt or kill us all (and make us fat, to boot). Not too long ago, I came across one such study, a truly execrable excuse for science by Gilles-Eric Séralini at the University of Caen purporting to claim that Roundup-resistant genetically modified maize can cause horrific tumors in rats. I looked at the methods and conclusions and what I found was some of the worst science I had ever seen, every bit as bad as the quack “science” used by the antivaccine movement, as anti-GMO activists worry about GMOs sapping and impurifying their precious bodily fluids. Then, not too long ago, I discovered a truly quacktastic bit of fear mongering by Jack Heinemann about GMOs in which, or so it is claimed, GMOs produce silencing RNAs that not only survive transit through the gut, get into the bloodstream and thence into cells to inhibit the expression of specific genes, and even get passed down to the next generation to kill your children.

The GMO fear mongering can even reach ridiculous extremes, such as this little bit dug up by GMO Pundit:


In the comments, the stupid truly burns:


So it wasn’t for nothing that I made the comparison between the antivaccine movement and the anti-GMO, because the anti-GMO movement is very much like the antivaccine movement and the cranks who claim that cell phone radiation causes cancer. Indeed, there’s a lot of—shall we say?—cross pollination between the groups. As if to demonstrate that very point, last week I came across an article by the all-purpose crank to rule all cranks, Mike Adams, at entitled GMO feed turns pig stomachs to mush! Shocking photos reveal severe damage caused by GM soy and corn:

If you have stomach problems or gastrointestinal problems, a new study led by Dr. Judy Carman may help explain why: pigs fed a diet of genetically engineered soy and corn showed a 267% increase in severe stomach inflammation compared to those fed non-GMO diets. In males, the difference was even more pronounced: a 400% increase. (For the record, most autistic children are males, and nearly all of them have severe intestinal inflammation.)

The study was conducted on 168 young pigs on an authentic farm environment and was carried out over a 23-week period by eight researchers across Australia and the USA. The lead researcher, Dr. Judy Carman, is from the Institute of Health and Environmental Research in Kensington Park, Australia. The study has now been published in the Journal of Organic Systems, a peer-reviewed science journal.

Judy Carman? that name certainly sounds familiar. Oh, I remember. She was quoted extensively supporting Jack Heinemann’s claims about GMOs. Here’s just one example:

We have not yet seen the worst damage that genetic engineering may do. Australia’s governmental agency, Commonwealth Scientific and Industrial Research Organisation (CSIRO), is developing a wheat species that is engineered to turn off genes permanently.

Professor Jack Heinemann at the University of Canterbury’s Centre for Integrated Research in Biosafety has studied the wheat’s potential. Digital Journal reports that he says1:

What we found is that the molecules created in this wheat, intended to silence wheat genes, can match human genes, and through ingestion, these molecules can enter human beings and potentially silence our genes. The findings are absolutely assured. There is no doubt that these matches exist.

The implications are clarified by Professor Judy Carman of Flinders University:

If this silences the same gene in us that it silences in the wheat—well, children who are born with this enzyme not working tend to die by the age of about five.

Silencing the equivalent gene in humans that is silenced in this genetically modified wheat holds the potential of killing people. But it gets worse. Silenced genes are permanently silenced and can be passed down the generations.

That’s right. To Judy Carman, siRNA from GMOs has the potential to kill your future children before they turn five!

But back to Carman’s most recent study. As described by Mike Adams, it sounds pretty damning, doesn’t it? It sounds truly horrific, just as the Séralini study did. Adams is useful in that he takes the messages of anti-GMO activists (well, actually, he takes the messages of just about all cranks and quacks) and, as they said in This Is Spinal Tap, turns them up to 11. On the surface, it does, anyway. But what about the actual study. There was really only one thing for me to do, and that’s the same thing I did with the Séralini study: Go and see for myself. So I did.

Judy Carman’s study was, fortunately, published in an open access journal, and there was a direct link to the study itself. The first thing I did was to look at the journal. I had never heard of it before. The journal seems to cater to the organic crowd, being sponsored by groups like the Organic Federation of Australia and CSAFE, while the guidelines for authors state that “topics are to be consistent with current principles of organic farming and its associated industries, especially those in Australia, New Zealand, Asia, and the Pacific Islands.” The journal itself appears not to be indexed on PubMed, which tends to indicate either that it’s a new journal or not a very good journal. On the other hand, to be fair, there are plenty of CAM journals indexed in PubMed, and many of them are pure pseudoscience; so I can no longer conclude that lack of indexing in PubMed automatically means a journal is dodgy. It is, however, often an indications that it is. Moreover, if you wander over to Judy Carman’s website,, you’ll see that it’s chock full of anti-GMO activism.

After having seen this study, I think that the editors of this open access journal have made a massive mistake and have, either wittingly or unwittingly, allowed their journal to become a tool of anti-GMO activist groups, a couple fo which which gleefully announced the results of the study with press releases (for example here and here) calling the study “groundbreaking,” asserting that it was evidence of “adverse effects” due to GMO feed, and claiming that the results “clear evidence that regulators need to safety assess GM crops containing mixtures of GM genes, regardless of whether those genes occur in the one GM plant or in a mixture of GM plants eaten in the same meal, even if regulators have already assessed GM plants containing single GM genes in the mixture.”

Here’s a hint: It’s none of the above.

As I read the study itself, the first thing that became apparent to me is that it’s a massive fishing expedition. What do I mean by that? I mean that there’s no clear hypothesis. Basically, the only seeming hypothesis was “GMOs bad,” and the study was designed to find bad things associated with GMOs. At first glance, the design seems simple enough. The investigators used 168 just-weaned pigs at a commercial piggery in the US. The pigs were fed a standard diet, but half the pigs were fed widely used varieties of GM soy and GM corn, while the control group fed an equivalent non-GM diet. Basically, one protein made the plant resistant to a herbicide and two proteins were insecticides. The specific GM varieties used were as follows:

The corn used in this study contained 90% DK 42-88 RR YG PL (a triple stack of NK603, MON863 and MON810 genes) with the remainder being equal quantities of Pannar 5E-900RR (containing NK603), Pannar 4E-705RR/Bt (a double stack of NK603 and MON810) and Producers 5152 RR (containing NK603). Therefore, the GM corn that was used was genetically modified to produce three new proteins. Two were Bt proteins that protected the plant against insect attack, while the third protein provided the plant with tolerance to the herbicide glyphosate (Testbiotech, 2012; Monsanto, 2012). Because Roundup ReadyTM (RR) soy is predominant in the GM soy market, this was used. This crop contains a gene that provides tolerance to the herbicide glyphosate. GM DNA analysis (Genetic ID, Fairfield, Iowa, US) confirmed that the GM corn contained a combination of NK603, MON863 and MON810 genes (expressing the CP4 EPSPS, Cry 3Bb1 and Cry 1Ab proteins respectively), that the RR soy was 100% RR soy (expressing the CP4 EPSPS protein), that the non-GM feed contained a median of 0.4% GM corn and that the non-GM soy contained a median of 1.6% GM soy. Such GM contamination of apparent non-GM material is common in the US.

So the investigators fed piglets a diet of GMO grain versus non-GMO grain, let the pigs mature according to the normal methodology, and then after slaughter looked at a variety of outcomes. Worse, the authors measured these variables without any sort of control for multiple comparisons. Of course they found differences! Actually, what surprised me is how few differences they found between the groups, not how many. I’m going to hone in on the main finding of the paper first. It’s the finding that seemed the most dramatic and was the most highly publicized, the one mentioned by Mike Adams in his breathless description of he results, as though they were slam-dunk evidence that GMOs are evil. I’m referring, of course, to the claim that more stomach inflammation was observed in the pigs fed a GMO diet, specifically a 267% increase in severe stomach inflammation in the GMO group, with a whopping 400% increase in male pigs. It’s the result that produced pictures like this one in the paper (and, not surprisingly the same picture posted to many an anti-GMO website):


These images certainly look striking, but what do they mean? Well, not much. First of all, as many have pointed out, the photos chosen are deceptive in that not enough of the groups are shown, nor can we be sure that these are representative. Also, as Mark Hoofnagle points out, the assay for inflammation in the gastric mucosa of the piglets was only based on gross pathology. Basically, there was no histological study and pathological examination of the tissue to detect and quantify actual inflammation. Basically, the assay was based just on a gross visual inspection of the the tissue by a veterinarian (not even a veterinary pathologist, even, as far as I can tell). Unfortunately, such inspections can be highly misleading, particularly after animals have been slaughtered in an abattoir, as described by Professor Robert Friendship, University of Guelph:

Dr. Robert Friendship, a professor in the Department of Population Medicine at the Ontario Veterinary College, University of Guelph and a swine health management specialist, reviewed the paper [see reference below]. He concluded that “it was incorrect for the researchers to conclude that one group had more stomach inflammation than the other group because the researchers did not examine stomach inflammation. They did a visual scoring of the colour of the lining of the stomach of pigs at the abattoir and misinterpreted redness to indicate evidence of inflammation. It does not. They would have had to take a tissue sample and prepare histological slides and examine these samples for evidence of inflammatory response such as white blood cell infiltration and other changes to determine if there was inflammation. There is no relationship between the colour of the stomach in the dead, bled-out pig at a slaughter plant and inflammation. The researchers should have included a veterinary pathologist on their team and this mistake would not have happened. They found no difference between the two experimental groups in pathology that can be determined by gross inspection.”

What I found particularly suspicious was Table 3. Notice how the level of inflammation is divided into no inflammation, mild inflammation, moderate inflammation, severe inflammation, erosions, pin-point ulcers, frank ulcers, and bleeding ulcers. This is not really a standard way of scoring inflammation. I don’t know about pigs, but in humans there are a variety of scoring systems for the endoscopic assessment of inflammation (for example, this one), particularly chronic gastritis (which is what we’re talking about, although such redness as described would, if associated with gastritis, be more associated with acute gastritis). Worse, gross visual assessment of gastric mucosa is subject to high inter-observer variability, and, although the personnel caring for the pigs and doing the autopsies were blinded to the experimental group (which is good), I don’t see any attempt to control for inter-observer variability, and, again, no control for multiple comparisons.

I also note that the difference between pin-point ulcers, frank ulcers, and bleeding ulcers is rather arbitrarily defined and not entirely clear. Also notice how twice as many pigs had no inflammation in the non-GMO group and that there was actually a lower risk of mild and moderate inflammation, as well as erosions and pin-point ulcers. Of course, the p-values are all non-significant, except for one: that for severe inflammation. In fact, on the entire table, the only “statistically significant” result is for “severe inflammation.” In fact, as Mark Lynas points out, many more pigs fed non-GMO feed had stomach inflammation than those with GMO feed.

Lynas also points out that the data are all over the place with respect to reported levels of inflammation, asking the very apt question, “If GMO feed is causing the severe inflammation, why is the non-GMO feed causing far more mild to moderate inflammation?” One also can’t help but notice that for “moderate” inflammation, there was a difference favoring the non-GMO feed, and I echo the question, “Do Carman et al perform a test for statistical significance to see if GMO feed has a protective effect on pigs stomachs? Of course not – that’s not the result they are after.” Exactly. Even worse, they used the wrong statistical analysis to analyze categorical data. When the data are analyzed more appropriately, there appears to be no statistically significant difference between the groups, just as there was no real statistically significant difference in the tumor burden of the rats in the Séralini study. Come to think of it, Carman’s study resembles the Seralini study in that it basically looks at a whole lot of outcomes in a fairly arbitrary fashion and cherry picks the inevitable “positive” result. In fact, if you take all the groups together, there actually appears to be a non-statistically significant trend towards less stomach inflammation in the GMO group. Yes, less. As Karl Haro von Mogel put it, the authors appeared to be “trying to shoe-horn individual categories that aren’t binary data into a statistical test designed for binary data is the wrong approach.” Basically, however you look at it, there’s just no “there” there. Analyzed correctly, there is no statistically significant (or, no doubt, biologically significant) difference in stomach inflammation in this study. As for the reported increase in uterus weights, as Professor David Spiegelhalter, Winton Professor of the Public Understanding of Risk at the University of Cambridge points out, “There are also 19 other reported statistical tests, which means we would expect one significant association just by chance: and so the apparent difference in uterus weight is likely to be a false positive.”

There’s another aspect of this paper that’s very troubling, and that these animals were all very sick. Indeed, I have to wonder how they were being cared for. Over half the animals are reported in Table 3 to have pneumonia, defined as “consolidating bronchopneumonia of the cranial ventral lung lobe(s) and/or caudal lobes.” That is just not normal, and it doesn’t sound like a minor pneumonia. True, this pneumonia wasn’t histologically verified, either, as far as I can tell, although pneumonia can be viewed grossly if it’s bad enough. It is, after all, basically puss mixed with mucous in the alveolae and bronchial passages. As has been pointed out in multiple discussions of this study, such a high percentage of animals with pneumonia is an indicator of very bad animal husbandry, indeed. The bottom line is that there are many, many problems with this study, the totality of which are more than enough to render its results meaningless. There is no dose-dependent mechanism for the effects reported, no rhyme or reason consistent with a mechanism that would explain why GMOs would affect just the stomach (and then only to cause severe inflammation) and and uterus size. The study was a fishing expedition and not hypothesis-driven. It’s not surprising that it found something. I’d be shocked if it hadn’t. In the end, this study abused a fairly large number of innocent pigs to produce no useful data. She might try to defend it against criticism, but she basically fails. In particular, one notes that she can’t seem to defend against the charge of a lack of hypothesis and that she didn’t even try to defend the criticism that she didn’t bother to look at stomach histology to verify that there really was inflammation in the gastric mucosa, despite Carman’s touting that the “authors have over 60 years of combined experience and expertise in medicine, animal husbandry, animal nutrition, animal health, veterinary science, biochemistry, toxicology, medical research, histology, risk assessment, epidemiology and statistics.” Sad that they didn’t use all that experience to produce a paper whose results are believable and useful.

Scientific failures are seldom so spectacular.