Interobserver Agreement in the Diagnosis of Ovarian Carcinoma Types: Impact of Sub-Specialization
Chirag Patel, Brian Harmon, Robert Soslow, Karuna Garg, Deborah DeLair, Sonya Hwang, Jingxuan Liu, Sui Zee, Kenneth Shroyer, Stephanie Burke, Carmen Tornos. Stony Brook University Medical Center, Stony Brook, NY; Memorial Sloan Kettering Cancer Center, New York, NY
Background: Ovarian carcinomas are a diverse group of tumors with different morphologic and molecular features. Correct classification has a potential impact on therapy since not all lesions need adjuvant chemotherapy, and not all tumors respond equally to standard chemotherapy. The aim of this study was to determine interobserver agreement in the diagnosis of ovarian carcinomas between specialized gynecological (GYN) pathologists and general (GEN) surgical pathologists.
Design: We selected 58 ovarian carcinomas surgically resected at SBUMC between 1/1/1999 and 6/1/2011, including 23 high grade serous carcinomas, 19 clear cell carcinomas, and 16 endometrioid carcinomas. All cases were reviewed by a senior GYN pathologist, and all diagnoses were confirmed with immunostains for WT1, p53, ER, and hepatocyte nuclear factor 1-beta. Immunostains were done in tissue microarrays containing three random cores per tumor. Only cases in which all three cores had the same immunoprofile were included in the study. One representative slide from each case was sent to 3 GYN pathologists and 3 GEN surgical pathologists, who were asked to classify the lesions according to their own criteria. The reliability of agreement between the pathologists was assessed using Fleiss' kappa value.
Results: The reliability of agreement was "substantial" for the GYN pathologists (Fleiss' kappa value = 0.67), and "moderate" for the GEN pathologists (Fleiss' kappa value = 0.54). The most common problem was to confuse serous carcinoma with either clear cell carcinomas or endometrioid carcinomas. This happened in 13% of cases in the GEN group, but only in 0.7% in the GYN group. The accuracy rate for diagnosing serous carcinoma was 84.5% in the GYN group and 58.33% in the GEN group (p= 0.0003).
Conclusions: Interobserver agreement was better among GYN pathologists than general surgical pathologists. Diagnosis of ovarian carcinoma tumor type is more accurate when done by GYN pathologists than GEN pathologists. However, there is a small percentage of cases in which classification by light microscopy alone is difficult even for GYN pathologists. If tumor type becomes an important factor in treatment decisions cases should be reviewed by a gynecological pathologist, and confirmatory immunostains may be necessary when the H&E morphology is ambiguous.
Category: Gynecologic & Obstetrics
Wednesday, March 21, 2012 9:30 AM
Poster Session V # 148, Wednesday Morning