Serous tubal intraepithelial carcinoma: diagnostic reproducibility and its implications

Abstract

Serous tubal intraepithelial carcinoma (STIC) is detected in between 5% and 7% of women undergoing risk-reduction salpingooophorectomy for mutations in the BRCA1 or 2 genes (BRCA+), and seems to play a role in the pathogenesis of many ovarian and "primary peritoneal" serous carcinomas. The recognition of STIC is germane to the management of BRCA+ women; however, the diagnostic reproducibility of STIC is unknown. Twenty-one cases were selected and classified as STIC or benign, using both hematoxylin and eosin and immunohistochemical stains for p53 and MIB-1. Digital images of 30 hematoxylin and eosin-stained STICs (n=14) or benign tubal epithelium (n=16) were photographed and randomized for blind digital review in a Powerpoint format by 6 experienced gynecologic pathologists and 6 pathology trainees. A generalized kappa statistic for multiple raters was calculated for all groups. For all reviewers, the kappa was 0.333, indicating poor reproducibility; kappa was 0.453 for the experienced gynecologic pathologists (fair-to-good reproducibility), and kappa=0.253 for the pathology residents (poor reproducibility). In the experienced group, 3 of 14 STICs were diagnosed by all 6 reviewers, and 9 of 14 by a majority of the reviewers. These results show that interobserver concordance in the recognition of STIC in high-quality digital images is at best fair-to-good for even experienced gynecologic pathologists, and a proportion cannot be consistently identified even among experienced observers. In view of these findings, a diagnosis of STIC should be corroborated by a second pathologist, if feasible.