Responses to the Royal Societyūs (RS) six referees' reviews on the Audit and Alternative Report (my comments) placed on the internet by the Rowett Research Institute on 16.02.99. Reviewer No. 1. I am afraid, this report is rather poor; the reviewer seems to be out of touch with present day nutritional science and the relevance of his/her comments is doubtful. Thus, his suggestion of using rabbits instead of rats is rather quaint. Also as no human studies have ever been done, it is impossible to satisfy his request for references to such studies. Furthermore, his/her complaint that no details of the experimental diets are given in the two Rowett Reports and that other details of the experimental protocols are also omitted should not have surprised him/her because internal reports usually do not contain such details. Reviewer No. 2. 1. Our study was carried out with genetically modified (GM) potatoes and we have never inferred or claimed from these studies that GM-food is unsafe. The referee should have read our introduction properly that clearly stated: "This report will therefore only describe the results of our ... work on GNA-GM- potatoes". We have never extrapolated from these studies to other GM-material. As a general comment, the referee must know that his statement that only one gene was inserted is not true because in the genetic modification most of the construct DNA are inserted including the promoter, the transcription terminator, the selection marker antibiotic resistance and the reporter genes. He is just trying to give the usual and misleading statement of biotechnologist that in genetic modification only a single gene is inserted. Moreover, the method of gene transfer we used is nearly identical with that used by the biotech companies for over 90% of GM-crops on the market! 2. There is a curious remark by the referee: the various reports "are well taken and I do not have substantial further comments to add". Is there another previous review by this referee that has not been disclosed to me? Furthermore, although the reviewer admitted that there were a number of statistically significant differences in growth, organ development and immune reactivity but because these experiments were "defective" in some unspecified way, these differences should not be taken seriously. I am afraid, without specifying exactly the details of these "defects", such a negative criticism cannot be taken seriously. 3. The reviewer failed to grasp the significance of our showing in this work that our potato lines, regardless whether parent or GM-lines, were compositionally not substantially equivalent. Had our concern been to ask the regulatory authorities to license these potato lines, almost certainly they would have been rejected. However, our task was to establish novel testing methods and therefore we could turn the argument around to find out whether feeding rats on diets containing compositionally different potatoes did also lead to differences in their metabolism, organ development and immune responses. Had I been given an opportunity by the RS to explain the aims and the design of our experiments, the reviewer could have avoided his/her mistakes and misunderstandings. 4. The reviewer asks for something that was obvious not only to him/her but also to us. Unfortunately, we were stopped by the Rowett to go beyond these four "preliminary" experiments. Thus, a potato line transformed using a construct without GNA gene has already been generated at Durham at my request. Moreover, the multivariant statistical analysis has given clear indications of possible construct and/or genome positioning effects which could have been developed further by more direct methods as also suggested by the referee. 5. Again the fault lies with the RSūs sending out internal reports for peer-reviewing. As these were prepared for collaborators and not outside "referees" detailed information on rats, experimental protocols, methodology, etc were not given. The referee should have directly asked me for these details but, of course, then he would have become known to me and that could have jeopardised his/her "impartiality" (according to the RS letter of 24 March 99). 6. Same applies to this comment as above. 7. The lack of hypotheses described in the reports does not necessarily mean that we had none to test. The RS, as a minimum, should have provided the referees with a copy of our full project (SOAEFD 1995; RO 818) that described in minute detail the aims, milestones, deliverables, etc that incidentaly had been fully per-reviewed by BBSRC before SOAEFD accepted and funded our proposal. It is difficult to understand why an experienced referee selected by the RS as my peer did not ask for that document. Reviewer No. 3. A general comment: This was the most unfair, unprofessional and biased reviewing I ever had the misfortune to be exposed to. Undoubtedly, this was aided and abetted by the RSūs assurance of anonimity. Examples are abound but here are just a few. The reviewer says: "I found the data impossible to review in the usual sense since proper descriptions of methodology are missing". However, this has not deterred him/her from saying: "...my overwhelming impressions are of extremely poor experimental design, for whatever was intended to be the outcome of the work (?! our intention was to find out and not to impose our prejudices on the experimental results !) chaos and confusion, with hopelessly confounded issues". How could someone come to such a damning conclusion as the reports did not describe the design and methodology used (as what he/she read (?) was an internal report) is beyond me, unless of course the reviewer actually came to this conclusion without reading the reports. 1. It is highly unlikely that this referee could have read the reports because the only factual information quoted by him/her, the protein content of the diets in the 4 experiments, which were according to the referee were 5, 5, 7.5 and 10%, were all incorrect (they were 6, 6, 8.4 and 16%). So what weight should one attach to his/her "conclusions" which followed ? 2. It appears that the referee is unaware that something like 30% of the human diet is eaten uncooked (or lightly cooked). Thus, rejecting the validity of our studies with raw potatoes is ill advised, particularly as almost one quarter of the gene product GNA survived in functional form the 1 h boiling process that is three times as long as it would have been necessary to fully cook the potatoes. A final comment: I cannot share the confidence of this reviewer that any GM- potatoes destined for the market would first be tested by appropriate methodology, particularly because presently there is no requirement to do biological testing of any GM-crops if they are deemed to be substantially equivalent. In any case no appropriate methodology exists. Reviewer No. 4. Introduction. I am somewhat perplexed by the reviewer first comment. Is it his experience that the results of compositional analyses described by internationally recognised authors in reports or papers are checked by a group of outside experts whether these are valid or not ? Is it our professionalism or our honesty that are questioned ? I am afraid, because of this referee's further comments imply bias in our measurements it is likely that he/she probably questions our honesty. Obviously, as he/she is hiding behind a cloak of anonimity, he/she can get away with it. Overview. Experiment D211 was not described by the Coordinator and is not a GM-experiment. Thus, the number of rats used is irrelevant. I am also not sure why the referee thinks that we had no particular hypotheses to test. However, what is even more important, the design of the experiments and the number of rats/group used (which is not described in our internal report but based on our over 20 years of experience with this type of work), was more than adequate to observe significant differences between test and control groups. Thus, although based on our extensive testing carried out for 6 years previous to 1995 in which GNA was added to control diets at dietary concentrations which exceeded those expressed in GM-potatoes by about 1000-fold, we did not expect differences in rat organ weights between the groups. However, when these occurred the experimental design ensured that these differences could be established with certainty. The number of such papers we have published using essentially the same methodology for GM-crops as that used for testing non-GM ingredients with rats, is our best testimonial. We do not necessarily agree with the reviewer's rather curious view that there is something inherently wrong with post-hoc interpretations in science. I think that in a pioneering work one MUST NOT PRESUME anything because then it will be difficult to avoid bias. In fact the great value of our study was that we had, unlike most of the biotech fraternity, an open mind about the outcome of our studies and this was in addition to our having the necessary experience and all the expertise to observe the true effects of feeding rats diets containing GM-potatoes. The assertion that there was a lack of consistency in the results and references to the BioSS' report is misleading as the statisticians had no biological input into their number crunching. The fact is that the design of the four experiments and most of the experimental variables were different in the four studies. Most importantly they were carried out with two different GM-lines which were not substantially equivalent and therefore not comparable. The reviewer's expectation of a presumably linear dose-response curve in immune responses shows clearly his lack of knowledge of the method used in which the response is always a maximum curve, peaking at some characteristic mitogen concentration. 1. The assertion by the reviewer that bias must have been one of the reasons for the larger number of significant results than expected is not only outrageuos and uncalled for but it also shows that this particular referee had no idea how such an experiment is conducted and therefore demonstrates his/her unsuitability to referee our work. In an experiment where 42 rats are dissected by a team of 4-5 people producing anything up to a 1000 tissue pieces per experiment, the suggestion that one can tell technicians what figures to put down in the notebooks should be laughable if it was not so offensive. I am afraid, the RS must have had real difficulties to find referees if they had to select someone like this inexperienced particular referee. 2. In contrast to the referee's opinion, the tabular material is clearly set out (within the terms of reference of an internal report) and could therefore only be misread with some difficulty. All appropriate parent (control) lines are listed in the Coordinator's report but admittedly not in the Audit Report. I fully agree with his comment if this referred to the Audit. The protein, starch and sugar contents of the lines were given without SD values but not the lectin and antinutrient contents. However, in a proper paper everything would have been given with SD. The protein values were the ones used for diet formulation and their validity was always checked by N analysis of the diets. In his/her consideration of the significance of the differences in the contents of glycoalkaloids (not our work!) and other antinutrients the reviewer failed to grasp the most important point of our findings i.e. that because of the identical growth conditions and the large individual samples size the differences had biological significance and most likely were the result of the gene transfer. 3. Again the reviewer missed the point. All diets were iso-proteinic and iso-energetic and the rats were pairfed, so where is the "protein imbalance" the referee is talking about ? 4. The referee thinks (or genuinely believes) that he/she is reviewing a paper and therefore looking for the description of methods. He/she again misses the point about GNA-binding to the gut. The fraction of GNA that was bound to the jejunum and ileum was not significantly different regardless whether we used GM- potatoes or parent potatoes spiked with the same amount of purified GNA. This is not surprising because GNA-binding is slight in comparison with its luminal concentration and determined by the scarcity of receptors for this lectin (remember we picked out GNA in advance in the six years of our preliminary studies before our GM work because it had only slight binding to the gut !). 5. I am happy to acknowledge that this referee at least read the reports concerning experiment D227 even though he/she says that this experiment was not described in the Audit Report. As the Alternative Report was commenting on and complementing the Audit Report it is not surprising that my report only referred to those data which were significantly different. I have to keep reiterating that my report was an internal document written for people who were familiar with these experiments. In a proper paper all the results, trends, differences and significances will be given. However, such a paper would not be submitted to the RS but to a Journal as appropriate. Moreover, even the Alternative Report had to based on the data and in-house statistical analyses released back to me by the Rowett after confiscating them when I was suspended. Thus not surprisingly the Audit Report is also based on these. The full statistical analysis (BioSS) report was done in February 1999 and this was done at our request on the primary data gathered up by Dr Susan Bardocz from original notebooks and animal house datasheets. 6. I have no different and/or additional comments to make to the reviewer's assessment of the results of experiment D242. Although there were a few minor mistakes I am somewhat perplexed by his/her remark about "any dose-response trends for any of the diets tested" because we did not carry out any dose-response experiments; all our diets were compared at one particular dietary dose. In fact, all our diets as before were iso-proteinic and iso-energetic and the rats were restrictively pairfed. As regards the statistics this was done in-house, presumably, by the audit group with the possible help of one of the scientists in our former group and also used by myself in the Alternative Report in the absence of the return of our primary data by the Rowett. Incidentally, it is most likely to have been done one-way analysis of variance as the referee also suggested. 7. I have no comments on the referee's appraisal of the results of our long-term D237 experiment other than to point ot that had we carried out the experiment with the GM-line 74/2T without supplementing it with over 20% additional lactalbumin protein, the two diets would not have been iso-proteinic and therefore the growth of the rats on the GM-diet would have been significantly retarded. In other words, feed conversion efficiency (growth/potato eaten) would have been significantly less with GM-potatoes than their unmodified parent line. 8. The referee admitted that in experiment D249 despite supplementing the GM-line 71/1 diet with over 10% lactalbumin to relieve any potential protein stress that might have affected the rats in D227 experiment, most of the significant differences in organ weights remained. Thus, the high flooding dose of extra protein could not override the GM effects observed in D227. So what about the complaint that there was no consistency in the experiments? Apparently, when one compares experiments in which the same GM-line was used, there is reasonably good consistency even though the experimental conditions in the two experiments were quite different. 9. I am not sure what the referee wants to say with his/her open-ended remarks. 10. Statistical report. "The independent statistical report is of a high and reliable quality". I am at a loss to understand why he/she then says in the conclusions vii that apparently inappropriate statistical techniques were used. If the referee refers to the Audit Report (and my comments to it) which was done in-house, he/she may be right. The only statistical report that is authorative was commissioned by me after we managed to regain most of the primary data early in 1999. It is therefore not clear which of all the statistical reports the referee is talking about and some clarity in his/her presentation would have been helpful. I have already briefly referred to the reviewer's inexperience with the lymphocyte responsiveness assay above and his/her comments here just confirmed my impressions. The facts are that without exception when the GM lines were tested in the three experiments, lymphocytes responded very poorly to mitogenic stimuli. Although variability was prominent feature of the results obtained with all other lines and controls but the results with them were always higher (except in D237) than with the GM-line diets and the difference between non-transformed and gene-transformed lines was always significant at mitogen concentrations giving maximal peak responses. These were consistently between 0.3 - 1.0 microgram/well for Con A and 9 microgram for PHA, fully in line with expectations. Finally, the referee's suggestion of bias and "no easily interpreted pattern of effects" comes up again. As I have given my views on these points I only wish to say that the emphasis in the referee's comment should be placed on the easily interpretable. Perhaps if I had been allowed by the RS to interpret our results for the referees I might have taken out some the difficulties in their interpretation. Conclusion In contrast to the referee's views our results with the right biological input and interpretation provide reliable and convincing evidence for the existence of consistent differences in the biological and immune effects of GM- versus parent-line potato-based diets. More particularly: (i) specific hypotheses have been tested in the experiments (but not given in the reports) (ii) six rats per group were more than adequate for valid conclusions to be drawn (as we have amply demonstrated this in something like over 40 published papers in major nutritional journals) (iii) no design is given in the reports therefore the confusion is that of the referee (iv) the two GM-potato lines were significantly different (v) glycoalkaloids were measured by SCRI scientists; it is not my habit to give the unpublished results obtained by other scientists (vi) this is an offensive remark (vii) for statistics see above; a little clarity in his/her comments would have been welcome (viii) the reviewer's claim that there is no consistency in the results only demonstrates the shortcomings of his/her approach (see above!) Final comment. The referee has done a great deal of work and unlike some of the others must have read through the reports and ought to be commended for this. Although I regret greatly that he/she should have asserted a bias on our part that was totally uncalled for but I ascribe it to the lack of his/her understanding the methodology we used - after all his expertise is in statistics. Moreover, the RS carry a great deal of the blame by providing the referees with internal documents which are manifestly inappropriate for peer-review. Had I been in his situation I would have refused to do it. Reviewer No. 5 1. It is the same problem as before as the referee could not see that the differences between parent and GM-lines fell into a readily discernible pattern, particularly as he/she was not even prepared to spend sufficient time to study the reports. Although even from these internal reports it is obvious that not only the four experiments had been done with two different and not substantially equivalent GM-lines but that also the protein concentration, the duration of the experiments and other conditions in the experiments were substantially different. Thus, it is not surprising that the results are not falling into readily discernible common patterns and it would have been only with my active help that the results could have made more sense. Unfortunately, the RS was not prepared for this. 2. The referee's recommendations are so vague, general and condescending, such as to test defined hypotheses, use proper (!) statistics, and exclude experimental variables (differences in the age of rats, diets, etc) that they would be useless even for a PhD student. In any case, how does the referee know that these variables have not already been dealt with ? I am afraid the rest of his/her criticism is inappropriate because I have never extrapolated from our results with GM-potatoes to the general safety of GM-food. Reviewer No. 6 There are two versions of this referee's report: one was received on 11 May that made an attempt to be relatively objective. The second version arrived by fax to Bergen (where I was in the middle of doing an experiment) on 13 May, 35 min before the expiry of the deadline the RS gave me for commenting on all referees'reports. Incidentally, there was no explanation why this reviewer had to send a second version of his/her comments but as the second version was much harsher than the first one and totally condemned my work, the implications were clear. The referee has obviously been asked to toughen up his/her criticism of my work because the first version was not up to what was expected of him/her. For this reason I feel that I cannot reply seriously to the points raised by him/her because it would not be obvious which version I should take as representative of the referee's true opinion and criticism. However, it might be instructive if I pointed out some of the more glaring examples of the differences between the two versions. First version, point 6-7: "Clearly, the interpretation of these results is problematical for several reasons". This changes in the second version to "It is impossible from the present study to claim that transgenic potatoes caused a reduction in the immune response in rats. The tests were not statistically significant..." Thus the significant differences in the BioSS report were totally ignored. In the first version: "-all factors that would urge caution in the interpretation of these data" has been replaced by "All these factors urge that the data are unsafe". In the first version it is recommended that "the findings prompt further investigation by SOAEFD. in the second version they would need "to fund an experienced immunotoxicologist to establish a valid and proven procedure" which clearly implies that our expertise was not up to the job.