Reproducibility of the Diagnosis of Malignant Colorectal Polyps
Benoit Terris, Genevieve Belleannee, Denis Chatelain, Joel Cucherousset, Marie-Daniele Diebold, Jean-Francois Flejou, Anne Fricker, Genevieve Monges, Francois Piard, Jeanne Ramos, Marie-Christine Saint-Paul, Jean-Yves Scoazec, Nahla Yazigi, Vivian Viallon. Hôpital Cochin, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes, Paris, France; CHU, Bordeaux, France; CHU, Amiens, France; CH Inter Communal, Le Raincy Montfermeil, France; CHU, Reims, France; Hôpital Saint-Antoine, Paris, France; CHG Mulhouse, Mulhouse, France; Institut Paoli-Calmettes, Marseille, France; CHU, Dijon, France; CHU, Montpellier, France; CHU, Nice, France; CHU, Lyon, France; Hôpital Kremlin Bicêtre, Paris, France
Background: Due both to introduction of screening of colorectal cancer and improvement of endoscopically techniques, the resection of malignant polyps (Tis and T1) is increasing. On a limited specimen, the pathologist has the critical task of determining whether there is a significant risk of recurrence or metastasis which may result in complementary surgical resection of the colorectum. To date, however, the level of agreement among pathologists in diagnosing malignant polyps has not been well established.
Design: Using a web-based virtual microscopy, we examined the observer variation in the diagnosis of 100 malignant colorectal polyps (Tis and T1) endoscopically resected by 13 gastrointestinal pathologists. Different pathological parameters were assessed as tumor differentiation, angiolymphatic vessel invasion, tumor budding, Haggitt's or Kudo's levels, measurement of submucosal invasion and base resection margin status. Results were analyzed by kappa (k) statistics and for percentage agreement.
Results: k analysis indicated that the strength of agreement was substantial for resection margin (k=0.61), moderate for pTis/pT1 classification (k=0.54), tumor budding (k=0.44), and slight for tumoral differentiation (k=0.14). Interobserver agreement was fair and moderate for Haggitt's and Kudo's classifications respectively. A better agreement was observed when measurement of the width or depth of submucosal invasion was performed (k=0.58 and k=0.54). According to the presence or not of pejorative criteria, there was a moderate agreement among pathologists on patient management strategies (k=0.54). A better agreement was observed among pathologists working in most active endoscopic centers and for specimens well orientated.
Conclusions: Gastrointestinal pathologists demonstrate moderate agreement for differentiate intramucosal carcinomas (pTis) from invasive adenocarcinomas (pT1) in endoscopically colonic resected specimens. The high interobserver variability concerning the tumoral differentiation parameter is clinically irrelevant. Measurement of the extent of invasion in submucosa appears more reproducible than the Haggitt's and Kudo's classifications.
Tuesday, March 20, 2012 8:15 AM
Platform Session: Section D, Tuesday Morning