Use of the mitotic counts for the prognosis and grading of breast cancer. Method evaluation study.

Reproducibility of the volume fraction-corrected mitotic index (M/VV index) was studied in 144 unselected breast cancer specimens. The influence on decision making of variation in determining the index was also analysed. In the complete series of specimens the correlation between two observers, one subjectively estimating the epithelial fraction of tumor epithelium and the other using point-counting (10 x 10 ocular grid), was good (Pearson's r = 0.82, p < 0.001, 95% CI 0.70-0.92). A subset of 30 specimens was used to evaluate the grading efficiency (GE) of the M/VV index method. The mean grading efficiency as estimated from this subset varied between 90% and 93%. The average minimum GE value was 82.8% (SD = 3.4%). The findings suggest that when the M/VV index method is used, the grading is correct on average in 90% or more of the cases, but dependent on the cutoff point. The over-all grading efficiency of the M/VV index method was comparable to that obtained from published S-phase fraction data on breast cancer specimens from three independed laboratories. We conclude that the M/VV index in breast cancer analysis is a sufficiently reproducible method in mitosis counting, and that it can be used with subjective or point count estimation of the area fraction of neoplastic epithelium.