Interobserver Agreement in Subclassifying High-grade/Poorly Differentiated Colorectal Carcinomas Among Subspecialty Trained Gastrointestinal Pathologists; Differentiating the Undifferentiated

Document Type

Conference Proceeding

Publication Date

3-1-2024

Publication Title

Lab Invest

Abstract

Background: The morphologic diagnosis of colorectal carcinoma (CRC) is typically straight forward. However, in recent years, additional CRC subtypes have been recognized posing diagnostic challenges for daily practice due to sometimes overlapping morphologic and immunohistochemical features. Entities including poorly differentiated adenocarcinoma NOS, in the absence of conventional morphology (PDA-NOS), large cell neuroendocrine carcinoma (LNEC), medullary carcinoma (MC), undifferentiated carcinoma (UC) and lymphoepithelioma-like carcinoma (LELC). Here we analyze interobserver agreement among gastrointestinal pathologists (GIPS) at our institution in subclassifying high-grade/poorly differentiated CRC. Design: Fifty-eight cases identified between July 2018 and July 2023 were included in our study. H&E and available IHC slides were reviewed by 3 subspeciality trained GIPS (2 with > 5 years’ experience and 1 with < 5 years’ experience) who were asked to assign the cases as PDA-NOS, LNEC, MC, UC and LELC. When a reviewing GIP was uncertain and provided an alternative diagnosis, the primary diagnosis was used for analysis. Relevant clinical and pathologic data were collected. Interobserver agreement was analyzed using Cohen’s Kappa. Results: For the 58 cases analyzed, 3 GIPS agreed on 42 (72%) cases. The most common diagnosis was PDA-NOS and for 33 (57%) agreement was unanimous. MC was the second most common diagnosis and considered by at least one of the GIPS on 10 (17%) cases; however, all 3 GIPS agreed on this diagnosis on 2 (3.4%) cases. All 3 GIPS agreed with the diagnosis of undifferentiated carcinoma on 1 (1.7%) case, while this diagnosis was considered by at least one of the GIPS on 8 (14%) cases. There was moderate agreement (k 0.41 – 0.60) between all 3 GIPS, fair agreement (k 0.21 – 0.40) between GIP 1 (P1) and P2, moderate agreement (k 0.41 – 0.60) between P2 and P3, and substantial agreement (0.61 – 0.80) between P1 and P3. In reviewing results, all three GIPS provided a differential diagnosis between 5.1 and 10.3% of the time, reflecting some uncertainty.

Volume

104

Issue

3

First Page

S1981

Last Page

S1982

Share

COinS