Interobserver Variability in Assessment of Depth of Submucosal Invasion for Colonic Endoscopic Resections Among Subspecialized Gastrointestinal Pathologists

Abstract

Context.— Recent data support that low-risk submucosally invasive (pT1) colonic adenocarcinomas (ie, completely resected tumors that lack high-grade morphology, tumor budding, and lymphovascular invasion) are considered cured via endoscopic resection, provided that the submucosal invasion is less than 1000 μm. Hence, the pathologists' assessment of depth of submucosal invasion may guide further management (ie, surveillance versus colectomy). Objective.— To assess interobserver concordance among gastrointestinal pathologists in measuring submucosal depth of invasion in colonic endoscopic resections. Design.— Six gastrointestinal pathologists from 5 academic centers independently measured the greatest depth of submucosal invasion in micrometers on 52 hematoxylin-eosin–stained slides from colonic endoscopic specimens with pT1 adenocarcinomas, per published guidelines (round 1 scoring). Two separate measurements (round 2 scoring) were subsequently performed by each pathologist following a consensus meeting, (1) from the surface of the lesion and (2) from the muscularis mucosae, and pathologists were asked to choose their (3) “real-life (best)” assessment between the first 2 measurements. Interobserver agreement was assessed by the intraclass correlation coefficient (ICC) and Cohen κ statistics. Results.— Round 1 had poor ICC (0.43; 95% CI, 0.31–0.56). Round 2 agreement was good when measuring from the surface (ICC = 0.83; 95% CI, 0.76–0.88) but moderate (ICC = 0.59; 95% CI, 0.47–0.70) when measuring from the muscularis mucosae and became poor (ICC = 0.49; 95% CI, 0.36–0.61) for the best-assessment measurement. Conclusions.— Our findings indicate that clearer and reproducible guidelines are needed if clinical colleagues are to base important management decisions on pathologists' estimate of the depth of submucosal invasion in colonic endoscopic resections.