Objective: The total number of cancer cell nuclei, N(nuc), and of mitoses, N(mit), in the primary lesion are potentially important indicators of tumor biology. In the present study, such estimates were obtained on breast cancers by an unbiased stereologic method.
Study design: The total number estimates are the product of two variables: (1) the volume of tumor, V(T), estimated by the Cavalieri principle, and (2) the densities of cancer cell nuclei and of mitoses obtained in small, three-dimensional samples (i.e., optical disectors) of 40-micron-thick methacrylate sections, which were selected systematically at random from the whole specimen.
Results: In 93 prospectively collected tumors, N(nuc) ranged from 0.06 to 7.9 10(9) (median, 0.6 10(9)), and N(mit) ranged from 0.02 to 64 10(6) (median, 1.5 10(6)). Both N(nuc) and N(mit) correlated significantly with V(T) (r = .77 and .60, respectively); however, the steep slopes of the regression lines indicated that densities of nuclei and mitoses increased as a function of tumor size. On average, N(mit) and estimates of mitotic frequency tended to be larger in lymph node-positive patients as compared with lymph node-positive patients as compared with lymph node-negative ones (2P < or = .08), whereas no such relation was found for nuclear counts (2P > or = .40). By counting a median number of 195 nuclei and 28 mitoses per tumor, the average coefficients of error of N(nuc) and N(mit) were 17% and 32%, respectively; this gave seemingly sufficient precision as compared with the huge interpatient variation in estimates, 180% and 490%. Moreover, the intraobserver reproducibility of density estimates was excellent (r > or = .88).
Conclusion: The present study showed the feasibility, efficiency and reproducibility of the unbiased optical disector principle applied to human breast cancer and provided data on new parameters of biologic relevance. The technique seems suitable for use in experimental oncology, but further studies are needed to investigate its clinical value.