Inter-pathologist reproducibility in classifying extraovarian implants in borderline serous ovarian tumor and low-grade serous carcinoma

One of the key differences between ovarian borderline serous tumors and low-grade serous carcinoma is the pattern of extraovarian tumor spread. Therefore, accurate diagnostics of the peritoneal implant's histotype plays a crucial role in treatment and disease prognosis. Despite established histological criteria for differential diagnosis, the classification of implant type remains subjective, and the interobserver reproducibility among pathologists is still understudied. OBJECTIVE: To assess the level of reproducibility in implant type classification among pathologists for ovarian borderline serous tumors and low-grade serous carcinoma. MATERIAL AND METHODS: A series of 33 slides from resected omentum and peritoneum specimens obtained from 23 patients with ovarian borderline serous tumors and low-grade serous carcinoma were independently evaluated by three gynecologic pathologists and three oncopathologists with varying experience in gynecologic pathology to determine implant type. A consensus diagnosis was established by majority agreement among gynecologic pathologists. RESULTS: The consensus diagnosis classified 42.4% of cases as low-grade serous carcinoma metastases and 57.6% as non-invasive implants of borderline serous tumors. The Fleiss' kappa was 0.61, indicating substantial reproducibility among all pathologists, but Cohen's kappa varied significantly (0.348-0.817). Reproducibility was perfect between gynecologic pathologists and between them and the consensus diagnosis. However, reproducibility among general pathologists was only moderate, while their agreement with the consensus diagnosis ranged from minimal to substantial. CONCLUSION: This study confirms significant diagnostic challenges in distinguishing non-invasive implants of borderline serous tumors from low-grade serous carcinoma metastases among pathologists, highlighting the need for developing and implementing standardized diagnostic algorithms.