Sir

It was gratifying to see M. Festing1 address one of the most vexing problems in toxicology, that of rodents so obese that the validity of carcinogenesis screening is brought into question2,3.

Festing recommmended the use of isogenic rather than outbred strains of rat. But he did not mention some work that had already been done using isogenic strains. The problem in carcinogenicity screening tests of increasing tumour rates (in the anterior pituitary gland, for example) and decreasing survival, accompanied by an increase in body weight, was actually first reported in an isogenic strain, the F-344 rat4.

We have related body weights to survival and neoplasia (such as leukaemia) in the same F-344 rat5. We6 and others7 have also correlated body weights to liver (and other) tumour incidences as well as increases in body weights to increases in tumour incidences over the same time period in another isogenic strain used in carcinogenicity tests, the B6C3F1 mouse. These studies have identified a major source of the variability in the tumour incidences of control animals used in carcinogenicity tests as the differences in body weights in the various studies.

Rather than incurring the problems that ensue from changing strains, or trying to breed small animals, one approach is to use dietary control6, which maintains body weights at predetermined practical ranges through control of dietary intake.

This approach not only prevents the ‘fat rat’, but also minimizes the variability in tumour endpoints and survival associated with the usual wide range in body weight seen in toxicology studies. Control of this variability is important because, besides being too large, rodents can also be too small in carcinogenicity bioassays, because of either an inadvertent restriction of feed or exposure to an agent, and thus be fairly insensitive to the action of toxicants.

We strongly support the evaluation of new models in biological studies. However, new models require painstaking evaluation and comparison with those now accepted by the toxicological community so that their results can be put into a proper context. We have a wealth of results in the strains used in toxicity testing that allow the kinds of comparison necessary for the development of new procedures that can provide improved means to assess human risk.

The complicating effects of uncontrolled body weight in toxicity tests is a consequence of the practice, common in rodents although unusual for most test species, of ad libitum feeding.

These complications do not appear restricted to either isogenic or outbred strains, or only to certain rodent species. Indeed, they are not confined to carcinogenicity tests per se. The doses required to produce short-term toxicity vary two- to fourfold, based on the dietary intake and resulting body weight of the animal8.

So dietary control could be a better solution than the genetic approach to resolve the problems currently observed in toxicity tests resulting from ‘fat rats’ and the other sequelae of uncontrolled food consumption and growth.