Inappropriate use of tumor markers for GI cancer screening in asymptomatic patients: A systematic review and proposed follow-up framework.

Abstract

804 Background: Serum tumor markers such as carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA 19-9), and alpha-fetoprotein (AFP) are inexpensive and widely available, leading to their frequent and often inappropriate use in asymptomatic individuals. Their role in gastrointestinal (GI) cancer screening remains uncertain, with concerns about false positives, low sensitivity, and absence of mortality benefit. Methods: We systematically searched PubMed, Cochrane Library, and LILACS from inception to May 2025 for studies evaluating CEA, CA 19-9, or AFP in asymptomatic populations. Eligible studies reported diagnostic accuracy or clinical outcomes. Two reviewers independently performed screening, data extraction, and risk-of-bias assessment using ROBINS-I, Newcastle–Ottawa, QUADAS-2, and RoB-2. Results: Of 1,179 records, 38 studies were included, 14 of which specifically evaluated GI malignancies. Across large cohorts (>70,000 participants), sensitivities were consistently poor: CEA ≤11% for colorectal cancer, CA 19-9 ~1% for pancreatic cancer, and AFP <20% for hepatocellular carcinoma. Specificities were higher (94–98%), but positive predictive values (PPVs) rarely exceeded 5% in average-risk groups. False positives were common, prompting additional colonoscopies, imaging, or surgeries, with complication rates up to 15%. No randomized trial demonstrated reduced GI cancer mortality with biomarker-based screening. Risk-of-bias assessments revealed most observational studies had moderate to serious concerns, particularly in patient selection and confounding, while only a minority of RCTs and case-control studies were judged at low risk. Based on this evidence, we developed a structured framework for the evaluation of incidentally elevated tumor markers, emphasizing confirmation with repeat testing, interpretation of marker kinetics, and escalation to targeted imaging only for persistent or markedly elevated results. Conclusions: Routine use of serum tumor markers (CEA, CA 19-9, AFP) for screening asymptomatic individuals is not supported by evidence. Despite moderate specificity, their low sensitivity and PPV lead to frequent false positives, unnecessary procedures, and no demonstrated survival benefit. Risk-of-bias analysis underscores the limitations of existing literature, and structured frameworks for incidental elevations may help minimize harm and guide clinicians toward cost-effective, evidence-based care.