DIASURV · ESMO

Validation of DiaSurv an AI-Based Algorithm for Stage II Colon Risk Stratification

Alexandra Montpetit, Chloé Lévesque, Baptiste Gourdin, Geneviève Soucy, Masa Peric, Céline Bossard, Yahia Salhi, Jérôme Chetritt, Vincent Quoc-Huy Trinh, Abdelhakim Khellaf, Bich Ngoc Nguyen

2026

Abstract

Background

Adjuvant chemotherapy decisions in stage II colon cancer (IIA T3N0; IIB T4aN0; IIC T4bN0) remain challenging, as conventional risk factors poorly predict recurrence. This results in both overtreatment and undertreatment. AI-based analysis of routine H&E slides may enable more accurate risk stratification to guide personalized treatment decisions.

Methods

DiaSurv Colon is an unsupervised deep learning-based prognostic tool that extracts morphological features from H&E-stained WSI and predicts 5-year overall survival (OS). Each patient is assigned an individual survival risk score. The model was trained on the TCGA cohort (n=463) and previously validated on two additional cohorts. In this study, we evaluated its performance on a new independent cohort from CHUM (n=127; 5-y OS 78%, 95%CI 69-84%). Prognostic accuracy was assessed using concordance index (c-index). Prognostic performance was further evaluated through multivariate Cox proportional hazards analyses, Kaplan-Meier survival analyses with log-rank testing, and hazard ratio (HR) estimation, and compared against conventional clinicopathological prognostic factors.

Results

DiaSurv Colon achieved a concordance index (c-index) of 0.64 and demonstrated significant prognostic stratification capacity. Patients classified as low-risk showed a 5-year OS of 85% (95% CI: 76-91%), compared with 58% (95% CI: 39-73%) in the high-risk group (log-rank p<0.01). The AI-derived risk score was significantly associated with overall survival, with an HR of 2.18 (95% CI: 1.23-3.85; p<0.01), highlighting its strong independent prognostic value. In multivariate analysis, the AI-derived risk group remained associated with OS (HR=2.18, 95% CI: 1.23-2.85; p=0.008), whereas conventional clinicopathological factors were not significant.

Conclusion

DiaSurv Colon demonstrated consistent prognostic performance on an independent external cohort, with robust hazard ratios and concordance indices. This AI-based tool may improve risk stratification in stage II colon cancer and support personalized adjuvant therapy decisions with more confidence.

Try it on your cases ↗︎