Gastric cancer kills roughly one million people annually worldwide, yet in the United States it is rarely screened for systematically — partly because the economics have never been rigorously modeled across demographic subgroups. A new Markov modeling analysis changes that calculus in meaningful ways, particularly for several populations who carry disproportionate risk but receive little structured surveillance.
Using detailed natural-history models of gastric cancer progression, researchers evaluated multiple screening and surveillance strategies involving esophagogastroduodenoscopy (EGD), including standalone screening and a bundled approach pairing EGD with colonoscopy at routine colorectal cancer screening visits. The analysis applied standard willingness-to-pay thresholds of $100,000 per quality-adjusted life-year gained (QALYG). Screening EGD beginning at age 50 with five-year surveillance intervals upon discovery of gastric intestinal metaplasia (IM) proved cost-effective in Asian individuals, yielding an incremental cost-effectiveness ratio (ICER) of $83,600/QALYG. For Black individuals, cost-effectiveness emerged at age 55 initiation (ICER $99,500/QALYG), as did screening for Hispanic individuals ($78,700/QALYG), those with a first-degree family history of gastric cancer ($76,200/QALYG), and immigrants from high-incidence regions ($95,900/QALYG). Standalone EGD in average-risk White populations did not meet cost-effectiveness thresholds.
This modeling work arrives at a moment when U.S. gastroenterology guidelines acknowledge risk factors for gastric cancer but stop short of prescribing formal screening thresholds — a gap that leaves high-risk patients without clear surveillance pathways. The bundling concept is practically significant: piggybacking an upper endoscopy onto an already-scheduled colonoscopy reduces marginal procedural costs substantially, potentially tipping the economics in favor of screening populations that would otherwise fall below the cost-effectiveness bar. Key limitations include the inherent assumptions embedded in Markov modeling, particularly around IM progression rates and endoscopy adherence, which may not reflect real-world clinical settings. These are also population-level estimates that cannot account for individual clinical nuance. Still, as an analytical framework informing future guideline development, this study represents a meaningful step toward formalizing risk-stratified gastric cancer surveillance in the U.S. — an incremental but potentially practice-shaping contribution.