Update: Impact of Austrian reproduction study on the safety of GM corn lines MON810 and NK603
January 2009
Summary
- The Austrian government commissioned scientists to conduct a detailed study into the long-term effects in mice continuously fed a diet containing GM corn. The recently released report of the study claims that a small effect on fertility was found only in the third and fourth litters of mice on the GM diet who were allowed to breed continuously.
- FSANZ scientists examined the Austrian report in detail and found numerous deficiencies in the experimental methods used and in the interpretation of results. Several calculation errors have also led the authors to incorrectly conclude that they had found a statistically significant difference in fertility between mice on the GM corn diet compared with mice fed conventional (non-GM) corn.
- Having identified several major flaws in the Austrian study, FSANZ considers that the conclusions drawn in the report are not supported by the results. In fact, no differences of biological significance in reproduction or longevity were found in the mice irrespective of their dietary group.
- The European Food Safety Authority has also examined the results of the Austrian study and is strongly critical of the methodology and authors’ evaluation of results. EFSA has reached a similar conclusion to FSANZ and has dismissed the findings of the study as being of minimal scientific value.
A report titled Biological effects of transgenic maize NK603xMON810 fed in long term reproduction studies in mice was posted on the website of the Austrian Ministry of Health, Family and Youth in November 2008. The unpublished studies by Velimirovet al. were commissioned by the Austrian government and conducted predominantly at the University of Vienna.
The studies investigated whether any reproductive and long-term effects could be detected in mice following consumption of genetically modified (GM) corn. The corn used in the studies is known as MON810xNK603, a conventional cross of two separate GM corn lines. Both MON810 (insect-protected) and NK603 (glyphosate tolerant) lines are approved for food use in Australia and New Zealand, as well as in a number of other countries.
The overall scope of the Austrian studies was comprehensive. They comprised a life-long feeding study, and two types of studies to monitor effects on reproductive performance, namely reproductive assessment by continuous breeding (RACB) and a conventional multigeneration study (MGS). In addition to standard endpoints, some studies also included detailed microscopic examination of selected tissues, as well as DNA microarray analysis to examine gene expression patterns.
In each study, three experimental diets were tested each consisting of 33% corn: either MON810xNK603 corn (GM), the non-GM near isogenic line, or an unrelated commercial non-GM corn.
There were no differences in reproduction and no diet-related effect on food consumption, body weight or life span between mice fed the GM or control diets. However, the authors’ analysis of the results concluded that statistically significant reductions in the number of offspring were evident in mice in the RACB study, but only in the third and fourth litters. No such effect was evident in the MGS, and no treatment-related effects were reported in the life-long feeding study. The authors consider that similar animal reproduction studies should be routinely included in the safety assessment of GM food and feed.
FSANZ has reviewed the data and statistical analyses presented in the report to ascertain whether the findings could have an impact on the previous safety assessment of corn lines MON810 and NK603. The complete FSANZ review of the Austrian report is available here
FSANZ evaluation summary
The conclusions in the Austrian report concerning a reproductive effect in mice fed a diet containing GM corn are not supported by the results obtained in the study. Not only can significant flaws in the experimental design be easily identified, but there is also a general lack of detail and transparency in their methodology. Careful examination of the data shows other major deficiencies in the interpretation of results and obvious errors in their statistical evaluations. For example:
- An inspection of tabulated results presented in the report for various measures of fertility in the mice reveals several systematic errors that have led the authors to an erroneous conclusion. Interpreting data from reproduction studies needs to take into account two different concepts, namely fertility and fecundity. Typically, fertility in mice is determined by the relationship between the number of delivered pups and the number of pregnant mice. Instead, the authors of the Austrian study used the total number of paired mice (pregnant and non-pregnant) in their calculation of fertility.
- Calculation errors have led the authors to incorrectly report statistically significant differences between the GM and control diet groups in the size of the third and fourth litters of mice in the RACB study. While the authors interpret this finding as a diet-related effect, further scientific scrutiny of the results does not support this view. As well as the calculation errors, it appears that the apparent statistical difference in the third and fourth litters is based on an unusually large litter size in the control group. Furthermore, it is worth noting that pup losses in the GM group were actually lower than in the control group in the first, third and fourth litters, however this was not reported by the authors.
- Conventional toxicological assessments do not rely solely on statistical analysis to detect possible treatment-related effects. It is important to recognise that when many individual biological measurements are compared using statistical analysis, there will be some significant differences found which are unrelated to treatment. In the Austrian study, this is demonstrated by the statistically significant differences between the two control (non-GM) groups in the MGS with regard to food consumption, bodyweight, litter sizes, pup weights and relative testes, liver and kidney weights. Clearly, differences between independent control groups are attributable to normal biological variation. It is noted that the number of statistical differences between the two non-GM control groups exceeded the differences between the GM and isogenic control group.
- DNA microarray analyses were used to investigate gene expression in samples of small intestine obtained from selected animals in the studies. The authors claim to have identified a small number of differences in gene expression between control and GM fed groups. However, as an even greater number of genes were found to be differentially expressed when comparing the two non-GM control samples, the authors were unable to comment on the biological significance of their findings. It is worth noting that the use of these techniques for analysing changes in gene expression according to dietary components is not a validated scientific approach.
FSANZ conclusion
Despite the comprehensive nature of the studies by Velimirovet al, the results show no differences of biological significance in reproductive performance or longevity of mice fed a diet containing GM corn, compared with mice fed a conventional corn diet. In addition, the DNA microarray analyses do not provide meaningful information on gene expression in the different diet groups. Based on the available evidence, FSANZ can confirm the conclusions of the previous safety assessments of corn lines NK603 and MON810.
Other evaluations
The European Food Safety Authority (EFSA) has also reviewed the Austrian study and released its conclusions as part of the Minutes of the 46th Plenary Meeting of the Scientific Panel on Genetically Modified Organisms (GMO Panel) held on 3-4 December 2008. This is available on the EFSA website at: http://www.efsa.europa.eu/cs/BlobServer/Event_Meeting/gmo_statement_austrianstudy_en.pdf?ssbinary=true
EFSA does not consider the Austrian studies to be of scientific value due to flaws in methodology, a number of statistical errors, and inconsistent evaluations of data. The GMO Panel also considered the report lacked historical control data relating to the mouse strain used in the studies, and provided insufficient information on the genetic analysis.