Size of Lugol-unstained lesions as a predictor for risk of progression in premalignant lesions of the esophagus

Abstract

At present, the surveillance strategy for premalignant esophageal lesions in China is based solely on the pathologic diagnosis in Lugol's chromoendoscopy (LCE). In this study, we sought to determine the degree to which various unstained features under LCE may lead to improved ability to predict the risk of progression in esophageal lesions. We re-examined and followed up on 1058 subjects who had Lugol-unstained lesions (LULs) together with a pathologic diagnosis that was lower than severe dysplasia at baseline screening based on a population-based randomized controlled trial over a median time of 5.8 years. We established a logistic regression model and calculated the adjusted cumulative incidence of severe dysplasia or malignancy. LUL size was predictive of progression to malignant lesions in individuals with a nondysplastic diagnosis (adjusted odd ratio6-10mm vs≤5mm, 6.7; 95% confidence interval, 1.7-25.7; adjusted odds ratio>10mm vs≤5mm, 27.9; 95% confidence interval, 7.3-105.7), and the corresponding adjusted cumulative incidence of malignant lesions was 3.6 and 13.2 per 100 persons. This is higher than that of small (≤5mm) lesions, which showed mild dysplasia (2.7 per 100 persons), a condition for which surveillance every 3 years is recommended. Under the current approach, 65.3% of interval cancers missed at surveillance would be detected if individuals with medium (6-10mm) and large (>10mm) nondysplastic LULs were additionally monitored. We propose a modified surveillance strategy that combines findings under LCE examination and the pathologic analysis, where follow-up endoscopy is recommended for individuals with relatively large nondysplastic lesions.