The implications of subjective recognition of malignant cells in aspirations for the grading of prostatic cancer using cell image analysis.

A pilot study of selective morphometry was performed in smear preparations from fine needle aspirates of six cytologically graded prostatic carcinomas. Given the instruction to select carcinoma cells with large nuclei for measurement, cytotechnologists and a cytopathologist differed in the selection of cells from grade III tumors. The cytotechnologists had discarded large bare atypical nuclei because they assumed that these were degenerated cells. Transmission electron microscopy, using an open-face embedding technique, demonstrated, however, that these cells were not degenerated but were poorly differentiated carcinoma cells that had lost their cytoplasm during smear preparation. It is obviously of great importance for the morphometric discrimination of grade I and grade III carcinomas that such bare atypical nuclei be included in the selections of cells to be measured. Our data indicate that the measurement of 20 selected cells per smear preparation is sufficient for accurate morphometric grading and that such measurements are reproducible.