Back

Identification of Novel mRNA Biomarkers with Improved Performance for Colorectal Cancer Screening from a Multicenter Large Gene Screen

Hansen, L.; Liu, H.; Lin, H.; Song, C.; Liang, Y.; Kirchner, J.; Chen, D.; Chen, Z.; Du, J.; Pan, W.

2026-02-07 gastroenterology
10.64898/2026.02.03.26345497 medRxiv
Show abstract

BackgroundColorectal cancer (CRC) is a leading cause of cancer mortality. While early detection improves outcomes, current non-invasive tests often lack sensitivity for early-stage CRC and advanced precancerous lesions (APL). Stool-based host messenger RNA (mRNA) biomarkers offer a promising approach, though the most clinically useful candidates remain undefined. MethodsWe screened for mRNA biomarkers by first using bioinformatic analysis of tissue RNA-seq datasets to identify candidate genes with strong and ubiquitous differential expression in CRC versus normal tissues. The top 135 computationally predicted biomarkers were evaluated using "gold standard" RT-PCR on clinical stool samples across two independent cohorts. ResultsSeveral biomarkers, including PPBP, MYC, MMP7, and TGFBI, exhibited strong predictive power. Integrating top-performing markers through machine learning yielded an AUC of 0.98 for CRC and 0.76 for APL detection. The optimized panel demonstrated 98% sensitivity for CRC and 50% for APL, with a specificity of 90%. ConclusionsThis study derives a high-performance mRNA-based stool test for non-invasive CRC screening. Our findings demonstrate that a multi-marker panel achieves exceptional sensitivity and good specificity, providing a viable tool for clinical diagnostics.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.