Current surveillance guidelines for patients with long-standing ulcerative colitis (UC) recommend repeated colonoscopy with random biopsies, which is time-consuming, discomforting, and expensive. A less invasive strategy is to identify neoplasia by analyzing biomarkers from the more accessible rectum to predict the need for a full colonoscopy. The goal of this pilot study was to evaluate whether optical markers of rectal mucosa derived from a novel optical technique, partial-wave spectroscopic microscopy (PWS), could identify UC patients with high-grade dysplasia (HGD) or cancer (CA) present anywhere in their colon.
Banked frozen nondysplastic mucosal rectal biopsies were used from 28 UC patients (15 without dysplasia and 13 with concurrent HGD or CA). The specimen slides were made using a touch prep method and underwent PWS analysis. We divided the patients into two groups: 13 as a training set and an independent 15 as a validation set.
We identified six optical markers, ranked by measuring the information gain with respect to the outcome of cancer. The most effective markers were selected by maximizing the cross-validated training accuracy of a Naive Bayes classifier. The optimal classifier was applied to the validation data yielding 100% sensitivity and 75% specificity.
Our results indicate that the PWS-derived optical markers can accurately predict UC patients with HGD/CA through assessment of rectal epithelial cells. By aiming for high sensitivity, our approach could potentially simplify the surveillance of UC patients and improve overall resource utilization by identifying patients with HGD/CA who should proceed with colonoscopy. (Inflamm Bowel Dis 2011)