Enhanced Diagnostic Efficiency of a Novel Fecal Methylated Gene Model for Early Colorectal Cancer Detection

Abstract

Background and Aims Methylation of stool DNA (sDNA) is a reliable noninvasive early diagnostic marker for colorectal cancer (CRC). Our study aimed to identify a new gene panel for the early diagnosis of CRC. Methods We conducted methyl-CpG binding domain isolated genome sequencing (MiGS) on 3 CpG island methylation phenotype (CIMP)-positive and 3 CIMP-negative CRC tissues and their corresponding normal adjacent tissues. Subsequently, by utilizing both the aforementioned data and public datasets, we identified a set of promising methylated sDNA markers for CRC. Finally, we developed a combined diagnostic model (CDM) for CRC based on the methylation status of PRDM12, FOXE1, and SDC2 and evaluated its performance in an independent multicenter validation cohort. Results A total of 1,062 participants were included in this study. The area under the curve (AUC) of the CDM was 0.979 (95% CI: 0.960–0.997), and the optimal sensitivity and specificity were 97.35% and 99.05%, respectively, in the training cohort (n = 231). In the independent validation cohort (n = 800), the AUC was 0.950 (95% CI: 0.927–0.973), along with the optimal sensitivity of 92.75% and specificity of 97.21%. When CRC and advanced adenoma (AAD) were used as diagnostic targets, the model AUC was 0.945 (95% CI: 0.922–0.969), with an optimal sensitivity of 91.89% and a specificity of 95.21%. The model sensitivity for nonadvanced adenoma patients was 68.66%. Conclusion The sDNA diagnostic model CDM, developed from both CIMP-P and CIMP-N, exhibited exceptional performance in CRC and could serve as a potential alternative strategy for CRC screening.