Quantitative evaluations of testicular histology can provide sensitive endpoints for determining toxicity of chemicals to the male reproductive system. But, the numbers of observations per testis or number of animals per treatment group often are selected by tradition or availability, rather than from a statistical basis. Therefore, we studied the number of observations per male (sampling intensity) and number of animals per treatment (replication) needed to detect treatment effects of given magnitude, with predictable error probabilities, using data from Sprague-Dawley rats and Dutch-belted rabbits that received 0.0, 0.94, 1.88, 3.75, 7.5, or 15.0 mg/kg body wt of 1,2-dibromo-3-chloropropane (DBCP). Data for one testis from 102 rats and 34 rabbits were available. For each testis, observations included measurement of the minor diameter of 50 seminiferous tubules, counts of the number of leptotene primary spermatocytes per 250 Sertoli cells, and counts of spherical spermatids within 20 seminiferous tubular cross sections. Tabular data are presented showing optimal numbers of observations per testis and animals per treatment group as a function of the difference to be detected and selected probabilities for Type I and II errors. In general, precise assessments required far fewer observations per testis than are used routinely. However, due to the inherent variability among animals, the number of animals required per treatment tended to be greater for experiments with the rabbit, and increased substantially for both species when detection of small differences had to be ensured. The data presented should enable investigators to design experiments of chosen sensitivity and precision while making cost-effective use of animals and labor.
