Nature Biotechnology 25, 1356 - 1358 (2007)
doi:10.1038/nbt1207-1356b
NOTICE: THIS WORK MAY BE PROTECTED BY COPYRIGHT
YOU ARE REQUIRED TO READ THE COPYRIGHT NOTICE AT THIS LINK BEFORE YOU READ THE FOLLOWING WORK, THAT IS AVAILABLE SOLELY FOR PRIVATE STUDY, SCHOLARSHIP OR RESEARCH PURSUANT TO 17 U.S.C. SECTION 107 AND 108. IN THE EVENT THAT THE LIBRARY DETERMINES THAT UNLAWFUL COPYING OF THIS WORK HAS OCCURRED, THE LIBRARY HAS THE RIGHT TO BLOCK THE I.P. ADDRESS AT WHICH THE UNLAWFUL COPYING APPEARED TO HAVE OCCURRED. THANK YOU FOR RESPECTING THE RIGHTS OF COPYRIGHT OWNERS.
Response to GM soybeans—revisiting a controversial format
Bruce Chassy, Vivian Moses, Alan McHughen & Val Giddings respond:
We limit our comments here primarily to the issues relating to Ermakova's experiments and findings in the order in which she raises them in her letter.
Although Ermakova states that she received “soy clearly labeled as GM and non-GM soy,” she still has not established the identity of the material tested, which is of paramount importance to an animal feeding study. The methodology and materials described by Ermakova are fatally flawed in several additional respects and as a consequence invalidate the experimental results. One of the basic issues is the content of the feed. The Archer Daniels Midland (ADM) catalog states and B.C. contacted ADM on October 20 and November 5, 2007, to verify that they do not sell—and have never sold—a 100% GM-soy product containing the RR-40-3-2 line to which Ermakova refers.
We must nevertheless apologize to Ermakova (and to readers) for any confusion that may have resulted from a typographical error in the statement she quoted from page 981—we can understand her confusion about our intended meaning. Our statement should have read: “the best that can be said is that commercial products sold by ADM would have been an indeterminate and variable mixture of conventional and GM soybeans.” Our point was that commercial products not specifically labeled GMO free are unsegregated mixtures of numerous varieties of conventional and GMO beans; Ermakova therefore had no measure of how much GM content was fed to the animals. The market has many varieties, each of which has its own unique composition and properties. This is why it is necessary to ensure that comparisons are between isogenic or even near-isogenic varieties; comparison of like varieties is a prerequisite for animal studies.
The PCR results she reports in Figure 1 do not demonstrate that the so-called GM soybean was 100% transgenic because Ermakova claims only that all samples (100%) of Arcon SJ tested positive by PCR. This is not the same as demonstrating that the positively testing sample is 100% transgenic soybean; all the samples might have had a small GM content so that all would have tested positive. It is essential to determine the percentage of GM-soybean content of the test materials both to allow others to attempt to replicate the procedures reported as well as to allow the actual exposure to GM-soybeans to be calculated.
Ermakova is comparing results obtained using different and uncharacterized soybean fractions. In one case, she fed ground soybean flour and compared that with results obtained using a protein concentrate (Arcon SJ). She refers to Arcon SJ as 100% transgenic soybean flour, which is incorrect; it is a concentrate, which is inconsistent with the claim that concentrates produced much less dramatic results.
There is also a much larger issue here: the composition of a sample of soybeans (or of any crop plant for that matter) is highly dependent on the location and conditions under which it was grown and harvested. Good practice dictates the cultivation of GM and non-GM soybeans in the same or adjacent fields to reduce soil and positional differences that might affect the composition. To overcome seasonal variation, the soybeans should be cultured in the same year.
It is of particular importance to note that isoflavone content varies between varieties, site of cultivation and growth year1. A point that we noted previously is that isoflavones have estrogenic activity that can dramatically affect the outcome of animal studies; neither was tested or controlled in her study.
As Ermakova thanks us for our detailed analysis of her work, we would also like to make the following suggestion. Guidelines describing the proper methods of preparing crop materials for animal studies were published this year by the International Life Sciences Institute (Washington, DC, USA) and are free online2.
We thank Ermakova for clarifying that extra soy was provided to males and females during mating; however, we would be interested in her response to a more fundamental problem that we noted in the original article. In all of her experiments, she housed three female rats together and fed them animal chow and soy product in separate dishes. That experimental design does not allow one to measure how much soy and chow each animal consumed. This information is essential, without which no scientific conclusion can be drawn. As in the original Feature, we refer Ermakova to the internationally accepted guidelines for performing animal feeding studies published by the Organization for Economic Cooperation (OECD; Paris; http://www.olis.oecd.org/olis/2003doc.nsf/
43bb6130e5e86e5fc12569fa005d004c/4502bee1ca16c943c1256d520028e259/
$FILE/JT00147696.PDF).
Ermakova mentions that her protocol was different from that of Brake and Evenson3. Indeed, Brake and Evenson started with 18 animals and sacrificed them in groups of 3 over an 87-day period. Their protocol observed all international guidelines and norms and would have detected effects of the magnitude that Ermakova observed. Their paper can be used as a model of how to conduct a reproductive toxicology study: Brake and Evenson had a known field source of soybeans, reported the exact composition of the diet and, because they fed the animals a single preparation containing test or control materials and measured weight gain, they could have interpreted any differences in weight gain. No differences in weight gain and no pup mortality were observed by them. Brake and Evenson studied four generations and, contrary to Ermakova's claim, animals were exposed to soy throughout the life cycle.
Ermakova compared her animal study with human-based clinical trials. Humans are not genetically homogenous and as a rule produce results showing considerable variation. The inbred laboratory rat is quite the opposite of a human in this regard; it has been developed to perform studies that will result in small variances of the measured variables. It is not good practice to pool results from separate animal studies because individual lots of animals can and do differ, and reproducing diet and environmental conditions is difficult at best. Thus, results from animal studies are normally not pooled; instead, statistics derived from each group are compared. Doing so increases the variance of measured variables.
The numbers of animals used in the study and the means and variances Ermakova now report for the body weights of males and females do not correlate with the data she reported in Table 3 in the original Feature published by Nature Biotechnology. The data in the Feature showed a wide variation in weight gain for all three groups; the new means and variances she now reports cannot be produced from the original data in Table 3. In the Feature, Ermakova's three conclusions for the GM soy-fed rates she observed were: (i) higher pup mortality; (ii) lower weight gains; and (iii) poor reproductive performance. The current data seem to contradict her original claim of reduced weight gain.
The weight gains reported in the controls are uncharacteristic of well-established literature averages for the Wistar rat. We interpret this as an indication of diet or environmental problems. Such wide variance in growth rates and the high control–death-rates are red flags signaling problems in experimental procedure.
The reporting of external variables that can affect behavior is good practice. The use of accredited facilities with standardized parameters not only ensures optimal health and development of the subject animals, it facilitates the comparison of results between experiments. We are still not sufficiently reassured by Ermakova's responses that environmental conditions were homogeneous throughout the animal facility where she carried out her experiments.
Ermakova now presents behavioral data from her feeding experiments, but doesn't mention if the studies were blinded. Because we still have serious concerns about the nourishment and treatment of the animals used, we cannot comment on her results. Ermakova's claims that GM soybeans have higher isoflavone notwithstanding, we cite published research demonstrating that, whereas soy isoflavone content varies considerably between varieties and harvests, GM soybeans have the same content of isoflavones as conventional soybeans4, 5.
In responding to our point in the Feature that several previous papers3, 6, 7 contradict her results, Ermakova claims they “had a different aim...and thus they are not comparable.” We respectfully disagree and do not see this as a basis for rejecting the feeding studies that we cited. These are well-conducted, peer-reviewed studies that exposed animals to diets containing a high content of soy or GM soy. Ermakova goes on to cite three papers from one group8, 9, 10 that have reported adverse effects of GM soy on testes and livers. We feel it is important to stress here that unlike the studies we cited3, 6, 7, the reports from Malatesta and colleagues8, 9, 10 do not conform with established international standards and protocols and fail to document the source, the composition or the identity of the soybeans under study. But in contrast to Ermakova, these authors8, 9, 10are scientifically cautious about the biological significance of their observations. We suggest that readers compare the literature we have cited with the three papers to which Ermakova refers and make a judgment for themselves about the effects of GM soy.
Ermakova goes on to state she carried out the experiments under conditions that were “as close to natural” as possible and concedes that no evidence is available as to the cause of pup death. It is pro forma in animal studies to determine the cause of death. Laboratory animal studies are not intended to mimic nature. The white laboratory rat does not exist in nature: it was bred in a laboratory to be used in very standardized studies designed to reduce variability and minimize uncontrolled variation that might confound the results.
Ermakova contends that the Brake and Evans study3, which contradicts her results, is not relevant because the feeding regime was “completely different” from the one used in her experiments, and suggests this is of significance because “only embryonic cells in the womb” would be affected by the EPSPS gene sequences, not the “sexual cells and/or organs before and during mating.” As we note above, the whole-life, four-generational nature of the Brake and Evenson3 study and its rigorous design cannot be disregarded. National and supranational regulatory agencies working in the public interest all over the world have examined extensive animal study data on GM soy and concluded it is as safe as, or safer than, conventional soy.
With regard to the data in Table 5 of the Feature: the litter size we computed from Ermakova's Table 5 as printed is that 12 dams produced 72 pups, which computes to six pups per dam, as we stated. A control group litter size value of eight does not improve the situation, because this is 50% below the normal litter size and a sign of animals in distress. With such high mortality and stunted growth, we must ask how normal reproductive experiments could have been performed with the GM soy-fed mice.
We leave Andrew Marshall to respond to the questions raised about the publication process, but we strongly object to Ermakova's characterization of us as 'pro-GM' scientists and in particular Brian John's slander that we are “apologists for the GM industry.” It is a matter of public record that we declare no conflict of interest, save for V.M., who maintains a GM information website that does receive some funding from industry and L.V.G. who works as a consultant with some industry clients (none of which are involved in transgenic soy). Contrary to the correspondence presented here, the scientists with whom we have spoken and from whom we have received letters in regard to this matter have expressed their appreciation to us for trying to correct the misinformation contained in Ermakova's 2005 report. B.C., A.M. and V.M. are, or have been, university faculty whose mission is the apolitical and objective teaching of science. None of us characterize ourselves as 'pro-GMO' or 'anti-GMO' as a matter of philosophy. It is an issue on which we remain agnostic; rather, we characterize ourselves as 'pro-science', 'pro-environment' and 'pro-humanity'.
All scientific work can and should be subject to the full force of reasoned criticism. Ermakova's remarks that there is an industry conspiracy to criticize and suppress articles containing evidence of the negative effects of GMOs is refuted by Ermakova herself when she cites published work on GMOs (albeit flawed) that shows negative effects. Rather than a worldwide conspiracy, we deduce there are few publications showing harm because GM soy is safe and does not cause harm.
We conclude then, that Ermakova's research relied on experimental designs that fall short of internationally accepted norms, with animals handled in such a way that even control lines were negatively affected. The feeding studies used materials that were characterized inadequately, incorrectly or not at all. Thus, no scientific conclusions can be drawn from the work.
We must stress again that GM soy has been thoroughly studied in the peer-reviewed literature, by regulators around the globe and by the cruel testing place of the real world. More than 500 million hectares were cultivated over the past decade. Much of this has been fed at high concentration to domestic animals, poultry and fish. There have been no reports of stunted growth or reproductive failure as one might expect if Ermakova were correct.
Competing interests statement
Declaration of competing financial interests
From the following article
Response to GM soybeans—revisiting a controversial format
Nature Biotechnology 25, 1356 - 1358 (2007)
doi:10.1038/nbt1207-1356b
back to article
Declaration: V.M. notes he is the chariman of CropGen (London), an information service directed at informing the UK public about agricultural biotechnology, which receives limited support from the biotech industry but acts entirely independently (http://www.cropgen.org/). V.G. was employed for eight years by the Biotechnology Industry Organization in the past and now works as a consultant with some industry clients, none of which are involved in transgenic soy.
References
1. Eldridge, A.C. & Kwolek, W.F. J. Ag. Food. Chem. 31, 394–396 (1983). | Article | ChemPort |
2. <http://www.ilsi.org/NR/rdonlyres/D84A9349-AC08-4CA2-BAC8-B79EBE92FC41/0/OutPutTraitsFinalforWeb.pdf>
3. Brake, D.G. & Evenson, D.P. Food Chem. Toxicol. 42, 29–36 (2004). | Article | PubMed | ISI | ChemPort |
4. Padgette, S. et al. J. Nutr. 126, 702–716 (1996). | PubMed | ISI | ChemPort |
5. Mebrahtu, T., Mohamed, A.I., Wang, C.Y. & Andebrhan, T. Plant Foods for Human Nutr. 59, 55–61 (2004). | Article | ChemPort |
6. Teshima, R. et al. J. Food Hyg. Soc. Japan. 41, 188–193 (2000). | Article | ISI | ChemPort |
7. Zhu, Y., Li, D., Wang, F., Yin, J. & Jin, H. Arch. Anim. Nutr. 58, 295–310 (2004). | Article | PubMed | ISI | ChemPort |
8. Malatesta, M. et al. Cell Struct. Funct. 27, 173–180 (2002). | Article | PubMed |
9. Malatesta, M. et al. Eur. J. Histochem., 47, 385–388 (2003). | PubMed | ChemPort |
10. Vecchio, L., Cisterna, B., Malatesta, M., Martin, T.E. & Biggiogera, B. Eur. J. Histochem. 48, 449–453 (2003).