Conclusions from new GMO feeding study by Carman et al. (2013) are not reliable

Ghent, 12 June 2013. The results of a new study in which pigs were fed genetically modified crops are currently being distributed via social media and news agencies. An analysis of the proposed results shows that the author’s conclusions cannot be derived from the study.

Food safety is a basic right for all consumers and all initiatives to guarantee food safety must be encouraged. The VIB is a research institute that uses genetically modified plants for fundamental plant research. The insights that the VIB achieves in this manner can in some cases be used in the development of genetically modified crops. Therefore, the VIB considers it an obligation to science and society to closely monitor new information about possible health effects of GMOs.

The “Journal of Organic Systems” published a long-term nutritional study in which 72 pigs were fed a GMO diet for 22.7 weeks (the normal life expectancy of commercial pigs from weaning to slaughter) and 73 pigs were fed a non-GMO diet.^[1] The GMO diet consisted of a non-specified mixture of GMO corn (insect-resistant + herbicide-tolerant) and GMO soy (herbicide-tolerant). The non-GMO control diet consisted of a mixture of conventional soy and corn, which is supposedly comparable to the GMOs used but does not meet the definition of the only correct control, namely the isogenic line. Only 1 factor may differ from the control in a comparative study, in this case the presence of GMOs. As isogenic lines were not used, the diets differed on more levels and this means that the cause of the differences cannot be identified. This situation is like comparing a GMO-jonagold apple to a conventional Granny Smith apple.

The investigators reported no differences at all between pigs fed a GMO diet or a non-GMO diet as far as feed uptake, feed conversion, disease, visits of the veterinarian, body mass, mortality and blood analysis. All the organs were checked for abnormalities after slaughter. These data are summarized in Table 3 of the study. A critical examination demonstrates that this study contains a lot of unexplained variation. A total of 15.1 % of the non-GMO fed pigs were found to have heart abnormalities, whilst this figure was only 6.9 % for the pigs on the GMO diet. Six of the 73 animals (8.2 %) that received conventional feed had liver abnormalities compared to 3 of the 72 pigs (4.2 %) that ate GMO feed. In the non-GMO group, 12.3 % of the pigs had lung abnormalities, whilst this figure was less than half (5.9 %) in the GMO group.
It was found that virtually all the pigs had stomach infections. If we look at the number of animals that did not have any stomach infections we see that 11.1 % of the pigs in the GMO group were healthy compared to only 5.4 % in the non-GMO group. This raises questions about the living conditions and care.
The non-GMO group had more mild and average infections, whilst the pigs in the GMO group had severe infections. As this is the only statistically significant value, the scientists only refer to the severe infections. If the diet would be the real cause of the stomach inflammations one would expect that there would be fewer pigs without infection, which is not the case. Moreover one would expect that there would be more animals in the mild and moderate categories, which is not the case.
The categorization (mild, average and severe infection) is also very arbitrary, meaning that the experiment should have been repeated to confirm the conclusion of increased severe infection.
A second aspect that the authors present is that – on average – more animals in the GMO group had a heavier uterus. Since the statistical difference is rather small, it would have been better if the authors would have repeated the experiment with the correct control before they came up with their conclusions. The most obvious explanation for the data by Carman et al. is that by halving the test population (only the female animals) the natural variation is emphasized more clearly.

To summarize, because of the absence of a correct control the experimental set-up of Carman et al. is weak. Together with the small statistical differences between GM-fed and non GM-fed pigs the conclusions drawn by Carman et al. cannot be substantiated.