Mitotic Index of Breast Carcinoma: Understanding and Remedying Irreproducibility
JS Meyer, E Cosatto, HP Graf. St. Luke's Hospital, Chesterfield, MO; NEC Laboratories, Princeton, NJ
Background: Histologic grading of invasive breast carcinoma has been practiced for more than 50 years. The Nottingham modification of the Bloom Richardson (NBR) system is endorsed by the College of American Pathologists and World Health Organization. While consistently prognostic, NBR grading has not been accepted as a standard on which therapeutic decisions can be based. Shortfalls in inter-observer reproducibility and inability to define a group of patients with sufficiently low probability of relapse (less than 10%) to justify omission of surgical adjuvant therapy are serious problems. Consensus favors the mitotic index (MI) as the most important of the three NBR variables.
Design: We undertook characterization of errors in measurement of the MI (mitotic count per 10 high power fields [HPF]) in a group of 328 specimens studied en passant in course of grading breast carcinomas for a tissue resource for researchers (N=214) and in clinical practice (N=114). MI was determined in two to six sets of 10 HPF, and standard error of mean and coefficient of variation (CV) were calculated for each specimen. Observed confidence limits and CVs were compared with theoretical values calculated by the binomial distribution model. Tertial cutoffs were determined.
Results: Mean MI of zero was observed in 40 specimens. In the other 288 specimens (Fig), observed CV means were 147% for lower third, 72.0% for mid third, and 34.6% for upper third.
Plots of observed and theoretical CVs for the 288 specimens were closely superimposed. Low and high tertial cutoffs for MI were 1.14 and 5.32.
Conclusions: Error in reproducibility of MI can be explained nearly entirely by sampling. The very high CVs of MI based on one set of 10 HPF is unacceptable. The NBR lower third cutoff of 7 mitotic figures per 10 HPF is far too high for our population, and could not be expected to separate good from poor prognosis. The CV can be reduced in proportion to the square root of the number of sets of HPF counted. At least 5 sets should be counted to establish that a low MI measurement is below a low tertial cutoff of 1 or 2 per 10 HPF for a given population. Counting multiple sets of HPF when the MI is low should increase the prognostic efficacy of the MI.
Tuesday, March 10, 2009 1:00 PM
Poster Session IV # 19, Tuesday Afternoon