Proteomics-Driven Risk Stratification in Stage III Colon Cancer: A Validated Prognostic Signature for Recurrence Prediction using three independent cohorts.

Abstract

Despite advances in colorectal cancer (CC) management, stage-III disease lacks robust, clinically applicable prognostic biomarkers. Proteomic profiling provides a quantitative, tissue-based approach to improve recurrence risk stratification. We performed data-independent acquisition mass spectrometric proteomic analysis of tumor samples from three independent stage-III CC cohorts (n=759). Differential protein expression between tumor and matched-normal adjacent tissue was analyzed in the training cohort (Cohort-1) using the limma package, and a univariate Cox model identified candidate biomarker proteins. A risk-score based on the levels of six proteins (ITIH1, PPIE, LTBP1, KPNA2, IGFBP7, and CKAP4) was developed using multivariate Cox regression in the training cohort (n=175) and validated in two external cohorts (Cohort-2, n=386; Cohort-3, n=198). Kaplan-Meier and multivariate Cox regression analyses assessed the prognostic value of the score for recurrence. The six-protein risk-score categorized patients in the training cohort as high- or low-risk for recurrence (hazard ratio (HR) 5.7, p < 0.001), and this was validated in the two separate cohorts (Cohort-2: HR 1.8, p < 0.001; Cohort-3: HR 1.8, p = 0.02). Integrating the proteomic score with clinical risk factors further enhanced prognostic accuracy (p < 0.001 in all three cohorts). The combined proteomic-clinical risk-score consistently identified a subgroup of patients with very low recurrence risk across cohorts. The validated six-protein risk-score improves prognostic stratification beyond standard clinical factors in stage-III CC, providing a robust framework for risk-adapted clinical investigation. Prospective, treatment-stratified clinical trials are warranted to determine whether this prognostic information can inform adjuvant therapy decision-making.