SEPTIN9-SDC2-VIM methylation signature as a biomarker for the early diagnosis of colorectal cancer

The accurate detection of colorectal cancer (CRC) at its initial stage can reduce mortality. However, the broad application of endoscopy has been limited due to the invasive procedure and patient noncompliance. Liquid biopsy with subsequent mapping of methylation in specific cell-free DNA (cfDNA) may represent an alternative approach for early diagnosis. In this study, we have developed a minimal-invasive blood-based test for detection of precancerous lesions and early-stage CRC. Using TCGA M450K methylation data, we identified candidate methylation sites with the highest Fold Change (FC) for three genes (SEPTIN9, SDC2 and VIM), which were selected from previous studies. Based on logistic regression models, we developed a 3-gene methylation signature for CRC diagnosis with high accuracy (Sensitivity =0.959, Specificity =1, AUC =0.997). Using independent public databases and data from blood samples, this model has demonstrated superior performance. The AUC was 0.919-1 and 0.905-0.916 in public tissue database for CRC and blood sample data, respectively. Thus, our proposed 3-gene methylation signature has a more reliable performance than other methods. Furthermore, signal enhancement effect of 3-gene methylation signature can improve the accuracy of early diagnosis for CRC, which demonstrates the potential for clinical application.

Automated summary

This study developed a minimally invasive blood-based test for detection of precancerous lesions and early-stage colorectal cancer (CRC). By analyzing methylation data, a 3-gene methylation signature model with high accuracy for CRC diagnosis was established. The model demonstrated superior performance in public databases and blood samples, showing potential for clinical application in improving early diagnosis of CRC.