Comparison of random and serial sections in assessment of ovarian toxicity.

Assessment of ovarian toxicity by follicle quantitation and morphometric analysis of serial sections is time consuming and expensive. This report compares the estimation of follicle number obtained from counting oocytes in serial sections or 5 random sections of mouse ovaries. Ovaries were obtained from C57BL/6N and B6C3F1 mice treated with ovarian toxicants. C57BL/6N mice were treated with cyclophosphamide (0, 75, 200, and 500 mg/kg, ip) and killed at 24, 72, and 168 h. B6C3F1 mice were treated daily, ip, with 4-vinylcyclohexene (0, 100, 400, and 800 mg/kg/day for 30 days), vinylcyclohexene diepoxide (0, 10, 40, and 80 mg/kg/day for 30 days), or benzo(a)pyrene (100 mg/kg, single dose) and killed on day 31. Ovarian serial sections were prepared and oocytes counted in every tenth section. When serial sections were evaluated, 30 to 60 sections were counted. Random section counting involved randomly selecting 5 sections from the 30 to 60 sections previously counted by the serial method. Chemically-induced follicle loss was evident by the reduction in follicle counts relative to control animals when estimating follicle number using serial or random section counting. Furthermore, a linear regression analysis of follicle counts over all treatment groups showed that the highest correlation between random and serial section counting and was for primordial follicles. Correlation coefficients (R2) for each follicle type were: primordial--C57BL/6N 0.82, B6C3F1 0.86; growing--C57BL/6N 0.16, B6C3F1 0.46; antral--C57BL/6N 0.07, B6C3F1 0.11. Although other screens may need to be developed for growing and antral follicles, primordial follicle toxicity is adequately defined by counting 5 random sections.