2443 - Feasibility of an Artificial Intelligence Model for Prognostic Risk Stratification Using Pathology Images after Neoadjuvant Chemoradiotherapy in Rectal Cancer
Presenter(s)
Z. Dong1, S. Chen2, Z. Lyu1, M. Zhang3, J. Wang1, W. Hu1, and Z. Zhang1; 1Department of Radiation Oncology, Fudan University Shanghai Cancer Center; Department of Oncology, Shanghai Medical College, Fudan University; Shanghai Clinical Research Center for Radiation Oncology; Shanghai Key Laboratory of Radiation Oncology, Shanghai, China, 2Department of Radiation Oncology, Fudan University Shanghai Medical College, Shanghai, China, 3Fudan University Shanghai Cancer Center, Shanghai, China
Purpose/Objective(s):
To test the hypothesis that artificial intelligence–extracted features from post-treatment whole-slide pathology images can stratify recurrence risk and predict disease-free survival (DFS) in patients with locally advanced rectal cancer after neoadjuvant chemoradiotherapy.Materials/Methods: A retrospective cohort study was conducted including patients with locally advanced rectal cancer who underwent neoadjuvant chemoradiotherapy followed by surgery. A total of 534 whole-slide images were used to develop an artificial intelligence–based model integrating pathology image features with DFS outcomes. Model performance was evaluated in an independent validation cohort of 229 patients. Kaplan–Meier survival analysis with log-rank testing was performed to compare DFS between predicted risk groups. Model discrimination was assessed using the concordance index (C-index). Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using Cox proportional hazards regression.
Results: In the validation cohort (n=229), the model stratified 116 patients as low risk and 99 as high risk. Predicted high-risk patients demonstrated significantly worse DFS compared with low-risk patients (log-rank p=0.001). The model achieved a C-index of 0.68 for DFS prediction. On univariate Cox analysis, high-risk classification was independently associated with inferior DFS (HR=2.27, 95% CI: 1.36–3.79, p=0.001).
Conclusion: Artificial intelligence–extracted features from post-treatment whole-slide pathology images were associated with disease-free survival and enabled risk stratification in patients with locally advanced rectal cancer after neoadjuvant chemoradiotherapy. These findings suggest that AI-based pathology analysis may support individualized postoperative risk assessment and inform adjuvant treatment decision-making.