Risks to Health From The Consumption Of Genetically Modified Foods

Submission by Dr Eva Novotny, SGR, to the Royal Society, April 2001

[Note that the figures referenced in this report are identical to those in Chardon Report I, “Unsuitability of Genetically Engineered Feed for Animals”; hence they are not reproduced here. The texts of the sections on the chicken and rat experiments are very similar, and the reader may prefer to read those sections in Report I instead.]


Most of the evidence presented below refers to the effects of genetically modified food on animals, rather than on human beings. However, since animals are often used in experiments as models for human responses, it seems appropriate to include these items.

Most of this evidence is available elsewhere. However, sections 3 and 4 contain new analyses by Scientists for Global Responsibility of data on experiments on rats and chickens. The original analyses and conclusions had been accepted in support of an application for placing a GM maize on the National Seed List. Our review, however, leads to different conclusions.


‘In 1989 there was an outbreak of a new disease in the US, contracted by over 5,000 people and traced back to a batch of L-tryptophan food supplement produced with GM-bacteria. Even though it contained less than 0.1 per cent of a highly toxic compound, 37 people died and 1,500 were left with permanent disabilities.’ (Quotation from Dr Michael Antoniou, gene therapist, Guy’s Hospital, London, as printed in GM-Free, vol. 1., no. 1, April 1999, p. 3 and again on p. 10.) The article on page 3 continues, without attribution to Dr Antoniou: ‘More may have died, but the American Centre for Disease Control stopped counting in 1991.

‘The US government declared that it was not GM that was at fault but a failure in the purification process. However, the company concerned, Showa Denko, admitted that the low-level purification process had been used without ill effect in non-GM batches. Scientists at Showa Denko blame the GM process for producing traces of a potent new toxin. This new toxin had never been found in non-GM versions of the product.’ (ibid., p. 3)


3.1 The experiment

The study is entitled ‘The Effect of Glufosinate Resistant Corn on the Growth of Male Broiler Chickens’. Like the study in section 4 on rats, this was submitted by Aventis (and accepted) in support of its application for a National Seed Listing. The purpose was apparently ‘to detect differences in nutrient quality of corn samples’ (p. 1 of the study). The duration of the experiment was 42 days.

In this study, 280 young broiler chickens of a commercial strain were used, divided into two groups. One group ate AgrEvo’s glufosinate-resistant maize, the other ate University of Guelph maize, a non-GM variety. Both diets were a conventional maize-soya type of diet; and both were adjusted on the same days, as appropriate for different stages of growth. The GM maize and the non-GM maize appear to be different varieties, and so the study is flawed in this respect.

All chickens were allowed to eat at will.

3.2 The stated conclusions of the study

The stated results and conclusion are:

‘Results of live bird traits … show that source of corn … had no effect on body weight, feed intake, … or percent mortality over the experimental period …’

‘Glufosinate tolerant corn from the U.S.A. is comparable in feeding value, for 0-42 day broilers, relative to the commercially available corn hybrid. Therefore, the nutritive value of glufosinate tolerant corn hybrid is equivalent to a commercially available corn hybrid.’

The mortality rate was judged to be normal.


3.3 Re-examination of body weights

Fig. 1a shows the body weights of the chickens in the two groups as a function of time. By the end of the study, the chickens on the genetically modified diet (represented by the black curve) have, on average, weights only 1 % below the average weight in the control group (red curve), which is insignificant. However, the error bars, shown by dotted lines for both groups, are very much greater for the chickens on the glufosinate-resistant maize. (The study does not state what is measured by the error bars, another flaw of the study.) Initially the weights and the errors bars are nearly identical, but the percentage error grows much faster for the glufosinate-fed group, as is evident from Figs. 1a and 1b, the latter being a larger-scale plot of the last 11 days of the experiment. Numerical values are given in Table 1 of the study. In spite of the fact that the body weights themselves never differ by as much as 1%, the errors quoted, which are initially the same, grow much more rapidly for the test group. On day 32 the errors are nearly twice as great for the test group; and on day 42 they are 2.5 times greater. Data for individual chickens have not been supplied, and the meanings of the error bars are not clear in terms of how extreme the individual weights of the birds might have been.

3.4 Re-examination of food consumption

Fig. 2 gives total food consumption during the intervals between measurements.

Again, the error bars are much greater for the test group: over the first interval they are only 1.3 times greater, but this ratio grows to 2.6 for the second interval and to 3.4 for the third and last interval.

3.5 Re-examination of mortality

In justification of the conclusion of the study that percentage of mortality is unaffected by the feeding of the GM maize, it is stated that ‘we normally see values of 5 to 8 % in male broilers.’ Nevertheless, it may be significant that twice as many deaths occurred amongst the chickens eating the glufosinate-resistant maize (7.14 ± 5.47 % between days 0 and 42) as compared with those fed commercial hybrid corn (3.57 ± 5.47 % between days 0 and 42).

3.6 Conclusions on the chicken study

Average body weights and feed intakes do not vary significantly, as concluded in the study. Nevertheless, the much larger error bars for both these quantities give concern that the weight gains and the feeding patterns were erratic in the treated group, indicating that at least some of the chickens were not thriving on the glufosinate-resistant maize.

There were twice as many deaths amongst chickens fed the GM maize as there were amongst those fed non-GM maize, although the study considered the number of deaths to be normal.


4.1 The experiment

This experiment was sponsored by HOECHST SCHERING AgrEvo GmbH (note that AgrEvo is now Aventis) in support of an application in the United Kingdom to have the genetically modified maize Chardon LL added to the National Seed List. The experiment was accepted as evidence of safety of the maize for the feeding of cattle.

The report on this experiment is entitled ‘PAT-PROTEIN – Repeated Dose Oral Toxicity (14-Day Feeding) Study in Rats’. This study on rats, like that on chickens, has little relevance to cattle, as the digestive systems of these animals are very different: cattle are ruminants and have four stomachs. Furthermore, it was not the maize itself, Chardon LL, but the isolated PAT-protein it contains that was tested; and the effects of feeding the isolated protein must be expected to differ from the effects of feeding the whole maize. Also, the very short time during which the experiment was pursued gives no indication of possible long-term effects of feeding over a lifetime, especially when the maize is to be fed to a very different animal species. Only five male rats and five female rats were used in each of the four groups, and the individual rats had substantial differences in weight even at the start of the experiment. While we believe that the experiment was faulty and that no firm conclusions can be drawn from it, we have re-examined the measurements to confirm, or otherwise, the internal consistency of the some of the conclusions drawn in the study.

The original proposal submitted by the laboratory performing the experiment was for a 10-day study; however, this was rejected by the sponsor and it was agreed that a 14-day study would be undertaken. Although it is stated (p. 15, para. 2) that ‘This study should provide a rational basis for toxicological risk assessment in man’, the conclusions are somewhat pre-empted (p.18, middle) by the statement that ‘As PAT-PROTEIN consists of normal amino acids it was not expected to cause any remarkable toxicity. Therefore, a treatment period [i.e., length of experiment] of 14 days was considered to be sufficient.’ . (In fact, measurements of body weight and food consumption spanned only 13 days.) Measurements were made only on Days 0 (pre-test), 1, 3, 7, 9 and 13. ‘Day 0’ actually occurred several days prior to Day 1, which accounts for the discontinuities in slope of some figures. Only body weights and food consumption will be discussed in this report. Presentations at the Chardon LL hearing held in London in October and November 20001 considered further aspects of this study, such as biochemical effects: see the statements of Dr Vyvyan Howard, Senior Lecturer and Head of the Foetal and Infant Toxico-pathology Group at the University of Liverpool, on 18 October; of Dr Bob Orskov, Honorary Professor in Animal Nutrition at Aberdeen Universtiy and Director of the International Feed Resource Unit, also on 18 October; and Dr Arpad Pusztai on 24 October. (We are aware that the Society dismissed previous evidence by Dr Pusztai, on the effects of GM potatoes on rats; nevertheless, we include this reference, should the Society wish to consult it.) All these presentations appeared on the MAFF website. [They can now be found by visiting http://www.defra.gov.uk and searching for ‘Chardon’. Transcripts are listed according to ‘year month day’; e.g., ‘001016’ means ‘2000 October 16’.

Although the purpose of the study was to test for toxicity, the data provide evidence that the animals may not be thriving on a diet including the PAT-protein. The evidence for this suggestion, from body weights and food consumption, will be examined below. Firstly, however, it is necessary to describe the experiment.

A total of 40 rats took part. They were delivered to the laboratory at the age of about 4 weeks, and so were very young animals that were growing rapidly. There were two control groups and two test groups, each group containing 5 males and 5 females. Body weights at the beginning of the experiment varied widely between 53 and 82 g for males and between 50 and 74 g for females. The groups were divided as follows:

Group 1: CONTROL group given a diet normal for laboratory rats throughout the experiment. In the 5-day acclimatisation period preceding the experiment, all rats were given this diet, which provided the standard of total protein content of 50’000 ppm [sic] to which the diets of the other three groups were adjusted.

Group 2: TEST group given a low dose of PAT-protein, 5’000 ppm, plus 45’000 ppm of soya protein.

Group 3: TEST group given a high dose of PAT-protein, 50’000 ppm, without soya protein.

Group 4: CONTROL group without PAT-protein but containing 50’000 ppm of soya protein.

In terms of similarity of diet, the test groups should better be compared with Group 4 than with Group 1.

All animals were allowed to eat at will.

Measurements of body weights and of food consumption were made during pre-test and on days 1 (before administering the new diets) and on days 3, 7, 9 and 13, although the last two measurements had been scheduled for days 10 and 14.

4.2 The stated conclusions of the study


The results of the study are summarised on p.34 of the study:

‘Average mean food consumption over treatment was in the same range for treated groups and controls.’

‘Occasionally recorded differences between controls and treated groups were generally small, showed no dose-relationship or consistent trend. They are considered to lie within the normal range of biological variation for rats of this age and strain housed under the conditions described above.’

‘Mean body weights were similar for treated groups and controls. There were no differences which could be attributed to treatment with the test article.’

4.3 Re-examination of body weights

Fig. 3a (p.39 of the study) plots the body weights of male rats, averaged within each group, against time during the experiment. The rats in Group 2 (dotted curve), which received the low dose of PAT-protein, gained weight at nearly the same rate as control Group 1 (solid curve), which received the standard rat diet. On the other hand, the rats in Group 3 (dashed curved), eating the high-dose of PAT-protein, gradually fell below all other groups, although they had been marginally the heaviest at the beginning of the experiment.

Fig. 3b (p. 40 of the study) shows a similar plot for the body weights of female rats. The rats in Groups 2 and 3 (dotted curve and dashed curve, respectively), eating the PAT-protein, gradually fell below the control groups; Group 3 had been the heaviest group at the beginning of the experiment.

Weights of individual rats were not plotted or analysed in the study, but we investigate these below.

Figs. 4a-d and 5a-d plot the individual body weights of males and females, respectively, against time for the four Groups. These figures are included for possible reference when examining the crowded plots in Figs. 6a and 6b, which show individual curves for all males and all females, respectively.

Fig. 6a illustrates the effect of including PAT-protein in the diet of the male rats: Group 2 (green curves) received the small dose and Group 3 (red curves) received the large dose.

Three of the green curves (short-dashed, solid and dotted curves) display a small gradual rise with respect to the control curves (black curve for Group 1 and blue curve for Group 4).that were nearest on Day 1, although the latter two follow a normal slope after Day 9. The other two green curves (dash-dotted and long-dashed), however, show a distinct downward trend with respect to the control curves.

Amongst the red curves, only the solid-line curve maintains pace with the control groups. The uppermost (dotted) curve begins above any other curve on Day 1 but gradually falls to meet the highest of the control curves by Day 13. Similarly, the short-dashed curve drops progressively lower than the control curves that were nearest on Day1. From Day 9, the long-dashed curve declines with respect to the control groups. The dash-dotted curve is the lowest on Day 1 and drifts ever lower with respect to the control curves.

Fig. 6b is a similar set of curves for females. The colour scheme is the same as for Fig.6a.

In the green set, the dotted curve follows a normal pattern. The short-dashed curve rises at first with respect to the control curves, then attains a normal slope from Day 9. The solid curve displays a persistent decline with respect to the control groups, while the long-dashed curve begins its downward progression only at Day 9. The dash-dotted curve, while maintaining a normal slope after Day 9, falls below all the control curves.

In the red set, the short-dashed curve and the solid curve are normal. The long-dashed curve declines with respect to the control curves between Days 3 and 9, before settling to a normal slope. The dash-dotted curve has a downward trend with respect to the control curves after Day 9; and the dotted curve gradually falls below all other curves, although it attains a normal slope on Day 7.

Alternatively, the slopes may be examined in tabular form. Table 1 lists the slopes for individuals and also for averages within groups, between Days 1 and 13. For the males eating a small amount of PAT-protein (Group 2), average weight-gain per day was the same as that of control Group 4, although it was slightly less than that for control Group 1. For females in Group 2, the average was distinctly below that for the two control groups. For both males and females in Group 3, which consumed the large amount of PAT-protein, average weight-gain per day was distinctly lower than for either control group.

TABLE 1. Weight-gain (gm) per day, between Day 1 and Day 13



1 This hearing is now indefinitely adjourned.