Abstract
BACKGROUNDCurrent methods for detecting esophageal cancer (EC) are generally invasive or exhibit limited sensitivity and specificity, especially for the identification of early-stage tumors.METHODSWe identified potential methylated DNA markers (MDMs) from multiple genomic regions in a discovery cohort, and a diagnostic model was developed and verified in a model-verification cohort of 297 participants. The accuracy of the MDM panel was validated in a multicenter, prospective cohort (n = 1,429). The clinical performance of identified MDMs were compared with current tumor-associated protein markers.RESULTSFrom 31 significant differentially methylated EC-associated regions identified in the marker discovery, we trained and validated a 3-MDM diagnostic model that could discriminate among patients with EC and volunteers without EC in a multicenter clinical prospective cohort with a sensitivity of 85.5% and a specificity of 95.3%. This panel showed higher sensitivity in diagnosing early-stage tumors, with sensitivities of 56% for stage 0 and 77% for stage I, compared with the performance of current biochemical markers. In population with high risk for EC, the sensitivity and specificity were 85.68% and 93.61%, respectively.CONCLUSIONThe assessment of tumor-associated methylation status in blood samples can facilitate noninvasive and reliable diagnosis of early-stage EC, which warrants further development to expand screening and reduce mortality rates.TRIAL REGISTRATIONChiCTR2400083525.FUNDINGScience and technology funds of Beijing Municipal Science & Technology Commission, Administrative Commission of Zhongguancun Science Park. Project number: Z201100005420007.