Grading of tumor malignancy in breast cancer should contribute essenti
al information both for the prospective outcome of the individual pati
ent as well as for TNM staging. In a series of 104 breast cancer patie
nts we tested the prognostic validity and reproducibility of mitotic f
igure counting compared with TNM staging, Bloom and Richardson grading
, DNA single cell cytometry, and morphometry. Four-micrometer thick he
matoxylin-eosin-stained routine slides were investigated. Mitotic figu
res were counted in representative areas of the tumor in 10 159-mu m(2
)-sized high power fields (HPFs) at a 400X magnification; the median v
alue was seven and the threshold for the 25th percentile was three. Th
is value should replace the common but prognostically invalid threshol
d of 10. Univariate survival analysis showed that mitotic figure count
ing allows the identification of three groups of patients (less than o
r equal to 3, 4 to 20, >20 mitoses per HPF) with significantly differe
nt survival probabilities (P < .0001; P = .0178). Depending on the num
ber of mitotic figures, length of survival was significantly different
within the group of T1N0 tumors (P=.0082) and the group of T1N1 or T2
N0 tumors (P = .0251). In a Cox stepwise regression model mitotic freq
uency counting added prognostic information to tumor size and was of h
igher prognostic significance than lymph node status, DNA ploidy, or m
ean nuclear area. The 95% confidence limit for interobserver reproduci
bility, tested in 20 cases, was plus/minus 8 mitoses. After quartiliza
tion an agreement of 75% was observed. Hmi PATHOL 26:47-52. Copyright
(C) 1995 by W.B. Saunders Company