Consistency of quantitative methods in ovarian tumor histopathology

Two mitotic activity indices, volume fraction of neoplastic epithelium, nuclear area, nuclear perimeter, and shortest nuclear axis were estimated in 46 ovarian tumors by three observers in two independent laboratories. The mitotic activity index, the volume corrected mitotic index (M/V index) and subjective volume fraction estimates showed a very good correlation from the same fields by two observers in the same laboratory (r = .999, .995, .950). Repeat estimates from different fields by the same observer showed coefficient values of .949, .939, and .813, and estimates by two different observers within one laboratory values of .949, .936, and .789, respectively. The correlation between two independent observers in different laboratories was also good (r = .894, .834, .834, respectively). Grading based on morphometric measurements was uniformly performed in 90-100% of cases in an interfield interobserver intralaboratory situation, and in 79-97% (M/V index) or 76-93% (mitotic activity index) in an interlaboratory situation. Because the cases near the grade limits were more often differently graded than other cases, the methods allowed the pathologist to locate the probably incorrectly graded cases. Morphometry of ovarian tumors was shown to be easy, which makes morphometric malignancy grading systems potentially able to support diagnostic and therapeutic decisions in practice.