A High-Resolution Microendoscope Improves Esophageal Cancer Screening and Surveillance: Implications for Underserved Global Settings Based on an International Randomized Controlled Trial

Abstract

Background & aims: Lugol's chromoendoscopy (LCE)-based detection of esophageal squamous cell neoplasia (ESCN) is limited by low specificity. High-resolution microendoscopy (HRME) was shown to improve specificity and reduce unnecessary biopsies when used by academic endoscopists. In this international randomized controlled trial, we determined the clinical impact, efficiency, and performance of HRME in true global health contexts with a range of providers.

Methods: Individuals undergoing screening or surveillance for ESCN by expert and novice endoscopists were enrolled in China and the United States from diverse clinical settings. Participants were randomized to LCE (standard of care) or LCE + HRME (experimental). The primary outcomes were the efficiency and clinical impact of LCE vs LCE + HRME using gold-standard consensus pathology.

Results: Among 916 consented participants, 859 (93.8%) were recruited in China and 36 (3.9%) in the United States; 21 (2.3%) were excluded due to incomplete procedure or data. In the screening arm, 217 participants were randomized to LCE and 204 to LCE + HRME; in the surveillance arm, 236 were randomized to LCE and 238 to LCE + HRME. HRME increased efficiency in screening: diagnostic yield (neoplastic/total biopsies) improved from 20.0% (95% confidence interval [CI], 12.7-29.2) to 51.7% (95% CI, 32.5-70.6) with 65.2% (95% CI, 54.6-74.9) of biopsies potentially saved and 59.7% (95% CI, 47.5-71.1) of participants potentially spared any biopsy. Six participants (0.7%) had neoplasia missed by the endoscopist on HRME (false negatives); of these, 3 were moderate or high-grade dysplasia missed by novices.

Conclusions: A low-cost microendoscope improves the efficiency and clinical impact of ESCN screening and surveillance when combined with LCE. HRME may spare unnecessary biopsies, leading to cost savings in underserved global settings where the disease is prevalent. (ClinicalTrials.gov, Number NCT02029937).

Keywords: Artificial Intelligence; Computer-Assisted Diagnosis; Esophageal Cancer; Esophageal Squamous Cell Neoplasia; High-Resolution Microendoscopy.