Example 3

A multi-strain design with a qualitative end-point

A screening experiment using a single strain, whether isogenic or outbred, can be criticised because the strain may be resistant to the test chemical. There are many examples. The figure below shows the percentage of rats with prostate tumours following treatment with 3,2'-dimethyl-4-aminobiphenyl (Shirai et al 1990). Note that in this case the outbred Wistar stock is completely resistant and the CD (an outbred Sprague-Dawley stock) is relatively resistant. In contrast, the inbred F344 strain is highly susceptible, with 48% of the animals developing tumours. With other compounds strain F344 may be more resistant that other strains. Thus, if an inbred strain or outbred stock is chosen at random for carcinogenesis screening, there may be a false negative result because the strain happens to be resistant. 


The alternative is to use several strains but the same total number of animals (Festing 1995, references given below). Carcinogenesis bioassays usually have a control and three dose levels related to the maximum tolerated dose with 50 animals of each sex in each dose group. Felton & Gaylor (1989) investigated the statistical implications of using more than one stain in this type of study and concluded that it is generally more powerful than one using only a single strain, and that the more strains that are used, the more powerful the experiment becomes. It is only less powerful if the experiment would have been done using the most sensitive strain, which is unlikely as there is no a priori way of determining whether a strain will be sensitive.

The reasons for this can be seen in the Figure. For simplicity, consider only a single group of 50 control and 50 treated rats. Assuming that these five rat strains and stocks are representative of all rat strains and stocks, and that a single strain or stock is chosen at random for the screening experiment. Then there would be greater than a 1/5 chance that the carcinogenic effect would be missed (7% in the CD colony, or 3-4 rats out of 50 is on the edge of being undetectable).  However, if the test population consisted of ten rats of each of the five strains, then the incidence of tumours would be 21% in the treated group, and this would easily be picked up.

It is easy to see that the multi-strain study using the same total number of animals would be statistically more powerful than one using a single outbred or isogenic strain.


Felton,R.P. and Gaylor,D.W. (1989): Multistrain experiments for screening toxi substances.  Journal of Toxicology and Environmental Health, 26:399-411.

Festing,M.F. (1995): Use of a multistrain assay could improve the NTP carcinogenesis bioassay.  Environ.Health Perspect., 103:44-52.

Shirai,T., Nakamura,A., Fukushima,S., Yamamoto,A., Tada,M., and Ito,N. (1990): Different carcinogenic responses in a variety of organs, including the prostate, of five different rat strains given 3,2'-dimethyl-4-aminobiphenyl. Carcinogenesis, 11:793-797.