Endoscopic diagnosis and treatment of colon cancer: current evidence and future directions

Colorectal cancer remains a major cause of cancer morbidity and mortality, while screening and advances in endoscopic imaging have increased detection of lesions that may be amenable to organ-preserving treatment. Accurate assessment of morphology, optical pattern, and invasion depth is central to selecting among endoscopic mucosal resection (EMR), endoscopic submucosal dissection (ESD), endoscopic full-thickness resection (EFTR), and surgery. This structured narrative review synthesizes contemporary evidence and major society guidance on endoscopic diagnosis and treatment of early colon cancer, with particular attention to comparative effectiveness, evidence limitations, regional practice differences, surveillance, recurrence management, patient-centered decision-making, and implementation of artificial intelligence. EMR remains the principal approach for most large low-risk nonpedunculated lesions because of procedural efficiency and favorable safety, although piecemeal resection limits histologic staging and historically carries a higher risk of local recurrence. ESD provides en bloc resection and more reliable pathologic assessment for lesions with suspected superficial submucosal invasion or other features requiring precise staging, at the cost of longer procedures, greater technical demands, and higher perforation risk. EFTR has a complementary role for selected small non-lifting, fibrotic, or difficult-location lesions, but long-term oncologic evidence is less mature. Cross-study comparisons are limited by heterogeneous lesion selection, operator expertise, outcome definitions, and follow-up. Major European, Japanese, and US recommendations share a risk-based approach but differ in the practical positioning of ESD, reflecting regional expertise and referral infrastructure. Artificial intelligence improves adenoma detection in randomized trials, yet evidence for early cancer detection, invasion-depth prediction, interval cancer prevention, and survival remains insufficient for uncritical implementation. A structured, multidisciplinary approach that integrates lesion biology, local expertise, pathology, patient preferences, and the possibility of additional surgery after non-curative resection is essential.