Can patients with mild non-neoplastic lesions diagnosed at baseline screening be safely exempt from surveillance: evidence from multicenter community-based cohorts.

Surveillance recommendations for gastric cancer (GC) in current guidelines focused on advanced precancerous lesions and were based on precise diagnosis of severity/extent of baseline lesions. We aimed to develop a less endoscopy-related equipment-dependent risk-stratification tool, and assessed whether mild-precursor-lesion patients can be safely exempt from surveillance. In the multicenter community-based cohort, 75,051 participants receiving baseline endoscopy were enrolled during 2015-2017 and followed-up until 2021. Cumulative incidence rates (CIRs) of GC for precancerous-conditions were calculated by Kaplan-Meier method and compared by Log-rank tests. Mixed-effects Cox regression models were used to detect potential factors for progression towards GC. A risk score was calculated as counts of selected factors. An independent cohort, including 26,586 participants was used for external validation. During a median follow-up of 6.25 years, CIRs of GC were 0.302%, 0.436%, and 4.756% for normal group, non-neoplastic (atrophic gastritis/intestinal metaplasia) and neoplastic lesions (low-grade/high-grade dysplasia), respectively (P_trend<0.001). Four predictors, including male, ⩾60 years, smoking, and limited vegetable consumption, were selected for risk-stratification. High-risk patients (⩾3 risk factors) with non-neoplastic lesions showed higher GC risks (adjusted HR=7.73, 95%CI: 4.29-13.92), and their four-year CIR reached the one-year CIR of neoplastic lesions. Further categorizing non-neoplastic lesions by histological grade, both patients with moderate-to-severe lesions (aHR=3.07, 95%CI: 1.67-5.64) and high-risk patients with mild lesions (aHR=7.29, 95%CI: 3.58-14.86) showed higher risks. Consistent trends were observed in validation cohort. High-risk mild-precursor-lesion patients should receive surveillance within 3-5 years after baseline screening. Our study provides evidence on supplementing current guideline recommendations.