Predetermined Search Methods Can Increase the Yield in Counting Mitotic Figures in Uterine Leiomyosarcoma (ULMS)
Henry R Mahler, Matthew R Lindberg, Charles M Quick. UAMS, Little Rock, AR
Background: Diagnosis of ULMS is based on cytologic atypia, presence of coagulative tumor cell necrosis and mitotic index, all features with a variable degree of subjectivity. Frequently, no guidance or search plan is used when performing a mitotic count and a drop and run (linear) or drop and weave (random) search pattern is used. This study assessed accuracy and reproducibility in utilizing specific search patterns for mitotic figures in ULMS.
Design: 11 cases of ULMS (chosen to assure abundant mitotic figures and desired tumor patterns) were culled from divisional files. Each case was evaluated at low power (4X) and examples of each tumor pattern were selected for counting. The tumor patterns were designated random, linear, along fascicles, fascicle cross-section, reactive (including areas near necrosis), hypercellular, and interface (tumor & normal). Slides were screened by 3 observers at 20X for the presence of mitotic figures, and, at the first mitotic figure seen, one 40X mitotic count (10 high power fields) was performed for each available search pattern. The data was then compiled as an average count per pattern along with a maximum inter-observer range (MIR) for each case.
Results: Counting in hypercellular areas and along fascicles yielded the highest counts, while fascicle cross section, linear, and random patterns yielded the lowest. 13 Non-diagnostic counts, i.e. less than 10 mitosis per 10 40X fields, were obtained (5 into fascicles, 3 random, 3 linear, 2 reactive\necrosis). Pattern based counting did not improve reproducibility based on MIR scores, with counting along the interface of ULMS and normal tissue faring the worst (see table 1).
|Random||Linear||Hypercellular||Interface||Along Fascicle||Into Fascicle||Reactive \ Necrosis|