GMO Pundit a.k.a. David Tribe: Toxicity tests are tricky for tyro toxicologists to try.

GMO Pundit's comment
If he were a rat, he'd prefer to live in Tokyo rather than Moscow.
He'd also eat Japanese chow, - GM or non-GM - rather than the stuff they sell at Moscow markets.

A seminar given by visiting toxicologist Dr Hiroaki AOYAMA in Melbourne June 26th 2007 (see a previous GMO Pundit post for more details) revealed why it is so very difficult to find out whether foods are unsafe by using rat feeding experiments to uncover lack of safety.

Such animal feeding tests have to be done according to very exacting protocols if they are to give meaningful results, needing hard work, a high level scientific expertise, and lots of money.

Rat feeding trials are in fact fundamentally ill suited to testing foods for safety. They are much better used where there is plausible evidence of a particular toxic effect, and to test a well defined idea about a particular known toxic chemical such as dichlorophenol.

In the latter situation, given the need to filter out unavoidable, random, and unknown background influences on rat growth and reproduction, experiments can be designed with different and appropriate chemical doses in different diets , and responses to increasing doses of test chemical can be exploited reveal actual causes for observed effects. Not so with foods. Testing of foods in rats has relatively meagre statistical power.

In other words, in the rat house, finding stuff you can confidently believe is expensive and challenging.

In his recent Melbourne seminar Dr Aoyama detailed these exacting requirements for performing and interpreting dietary toxicity in rats. The reason for his presentation was to provide better understanding of recent controversial claims that Russian investigator Dr Irina Ermakova has made at a conference on transgenic crops.

Irina Ermakova: Influence of genetically modified soya on the birth-weight and survival of rat pups, in Proceedings of the Conference Epigenetics, Transgenic Plants& RiskAssessment December 1st 2005, Literaturhaus, Frankfurt am Main, Germany (pdf file)

Ermakova's rat experiments have been used to question whether GM soy-meal is safe for humans to eat.

In his Melbourne seminar, Dr Aoyama pointed out several features of Ermakova's experimental findings, that suggest that the general rat handling conditions used in her laboratory are the the cause of rat deaths she reported, and that they are not caused by the use of GM soy as she claims. He also discussed the features of Ermakovas experiments which mean that they are badly designed.

Aoyama's Japanese government laboratory has extensive experience in toxicology testing and has reported many well designed experiments which provide a good model for how rat feeding trials should be carried out, and illustrate how they can be useful.

e.g. Hiroaki AOYAMA, Hitoshi HOJO, Ken L. TAKAHASHI, Naoko SHIMIZU, Masayuki ARAKI, Michiko HARIGAE, Naoki TANAKA, Noriko SHIRASAKA, Maki KUWAHARA, Nobuaki NAKASHIMA, Eri YAMAMOTO, Machiko SAKA and Shoji TERAMOTO: "A TWO-GENERATION REPRODUCTIVE TOXICITY STUDY OF 2,4-DICHLOROPHENOL IN RATS": J. Toxicol. Sci., Vol. 30: No. Special, S59-78. (2005) .

In this Journal of Toxicological Science study, Aoyama uses a recommended 20 rats in any one test group to test an experimental diet for toxicity. Ermakova's study used only 3 or 4 rats, making it of weak statistical value.

One telling statistic underlining the inexperienced handling of rats in Dr Ermakova's laboratory is the extraordinarily high death rates suffered by Ermakova with pups in rat litters reared on non-GM feeds - the so-called "control groups" in the study. These death rates (6.8% to 9%) were ten- to fifteen-fold higher [Sep 5th 2008: correct this to six to eight-fold] than seen in comparable litters in well managed laboratories, such as Dr Aoyama's.

Importantly, Aoyama points out that Ermakova's control or "non-GM feed" groups of rats show great variability themselves (that is, even when rats are not exposed to GM soy in their diets) and this variability provides an obvious indication of inappropriate handling conditions in her Russian laboratory.

Thus even in the control groups of rats (not fed GM soy) the range of pup weights seen by Ermkova was extremely variable. This extreme variability contrasts with the narrow range of pup weight in rat litters he sees in comparable litters (that is pup weight averadge 38.8 g, SD 2.6 for males, Average 37.2g SD 2.5 for females at 2 weeks) in his Japanese research institute investigations.

Ermakova's questionable handling of experimental variability was also a subject of critical comment. According to Aoyama, there is a need to consider at growth of male and female rats separately. Because of significant natural variability in pup weight, the average weight of a litter, rather than individual pup weight should be used as a measure of toxic effects. Ermokova pooled data of males and females, and does not use litter averages as a measure of toxic effects.

The Pundit Postcript:

At NIH labs in Washington DC, they test toxins on lawyers, because they found that lawyers will do things that even a rat won't do. But in Moscow, because lawyers are even more crafty than American attorneys, laboratory rats still bear the full brunt of the Soviet tradition for brutally cruel public accommodation.

Updates