Further evaluation of reproducibility and prognostic value of histologic typing and grading in FIGO stage I ovarian cancer patients without systemic locoregional adjuvant treatment.

Abstract

In order to analyse the reproducibility and prognostic value of histologic typing and different methods of histologic grading, 102 hematoxylin and eosin stained slides of primary FIGO stage I ovarian cancers of the common epithelial types were evaluated. Patients were treated by surgery only. Histologic typing was done using the FIGO criteria and overall total agreement was 61%. Survival was not statistically significant (P-value range: 0.31-0.66). For grading, three different methods were used: (a) the 'intuitive' method; (b) the FIGO method; and (c) the so-called Pathology Research and Practice (PRP) method, which is based on invasion, architecture and nuclear atypia. The best intra- and interobserver results were obtained with the PRP criteria (interobserver agreement rate: 84.3%, varkappa 0.67; intraobserver agreement rate: 87.3%, K 0.76). Univariate survival analyses showed significant differences between grades for all methods and observers. Results of the PRP method were clearly better than for the FIGO method (PRP, Mantel-Cox range 6.70-24.52,P-value range 0.07 - < 0.0001; FIGO, Mantel-Cox range 5.31-6.77,P-value range 0.07-0.034). The 5-year survival per grade showed considerable variation within and between the observers, and was higher with higher grade. We conclude that the assessment of histologic type was reproducible but prognostically not significant in this set of FIGO stage I ovarian cancer patients. The intra- and interobserver reproducibility of the same grading method was fair to good, and showed highly significant results for survival from the PRP method. However, application of different grading methods could result in considerable prognostic variations for the 'same' grades.