Mo1629 OPTICAL EVALUATION FOR PREDICTING CANCER IN LARGE COLORECTAL LATERALLY SPREADING LESIONS IS DEPENDENT ON LESION MORPHOLOGY

Abstract

Diagnosing submucosal invasive cancer (SMIC) prior to endoscopic resection of large (≥ 20mm) colorectal laterally spreading lesions (LSLs) is critical to inform therapeutic decisions. It is unknown whether lesion morphology influences optical evaluation performance. To evaluate whether lesion morphology affects the real-time performance of optical evaluation to diagnose SMIC prior to endoscopic resection. Colorectal LSLs ≥ 20mm referred for endoscopic resection, within a prospective, multi-center, observational cohort were evaluated. Optical evaluation was performed prior to endoscopic resection and included lesion morphology, topography and surface pattern characterization. Optical diagnosis of SMIC was based on established features consistent with invasive neoplasia. Sensitivity, specificity, accuracy and SMIC miss rates were calculated. Histology was the reference gold standard. Outcomes were reported by nodular (Paris 0-Is/0-IIa+Is) vs. flat (Paris 0-IIa/0-IIb) morphology. Multiple logistic regression with backward stepwise variable selection, was used to identify independent predictors of the outcome of interest. An alternative classification tree analysis was used to crosscheck the results of the logistic regression analysis. From July 2013-July 2019, 1583 LSLs (median size 35mm; IQR 25-50 mm; 855 flat, 728 nodular) were assessed. SMIC was identified in 146 (9.2%, 95% CI 7.9-10.8%). Overall sensitivity, specificity, and accuracy were 67.1% (95% CI 59.2-74.2%), 95.1% (95% CI 93.9-96.1%) and 92.5% (95% CI 91.2%-93.7%), respectively. The overall SMIC miss rate was 3.0% (95% CI 2.3-4.0%). Significant differences in sensitivity (90.9% vs. 52.7%), specificity (96.3% vs. 93.7%) and SMIC miss rate (0.6% vs. 5.9%) between flat and nodular LSLs were identified (all p < 0.027). Multiple logistic regression analysis identified lesion size ≥ 40mm (OR 2.0; 95% CI 1.0-3.8), rectosigmoid location (OR 2.0; 95% CI 1.1-3.7) and nodular morphology (OR 7.2; 95% CI 2.8-18.9) as independent predictors of missed SMIC (all p < 0.039). On classification tree analysis, lesion morphology (flat vs. nodular) was confirmed to be the critical variable for missed SMIC risk stratification. Optical evaluation performance for colorectal LSLs ≥ 20mm is dependent on lesion morphology. Flat lesions without SMIC features are overwhelmingly benign and can be removed by endoscopic mucosal resection. Nodular lesions have impaired optical evaluation performance and require further risk stratification to effectively identify those with a heightened risk of invisible or “covert” SMIC.