Objective:

To develop and validate a machine-learning model that integrates portal-venous phase CT features from both intratumoral and peritumoral regions for non-invasive prediction of TP53 mutation status in pancreatic ductal adenocarcinoma (PDAC), addressing the urgent need for reliable non-invasive biomarkers.

Key Findings:

• The Intra-Peri Model (IPM) combining intratumoral and peritumoral features achieved the best performance.
• The XGBoost classifier yielded an AUC of 0.893 in the external test set, significantly outperforming single-region models (P < 0.05).
• Intratumoral gray-level skewness and peritumoral texture correlation were identified as the most influential predictors, with statistical significance noted.

Interpretation:

Greater intratumoral asymmetry and lower peritumoral correlation indicated a higher likelihood of TP53 mutation.

Limitations:

• The study is retrospective and may have inherent biases, particularly in patient selection and data interpretation.
• The requirement for high-quality imaging data may limit generalizability, especially in diverse clinical settings.

Conclusion:

Integrating intratumoral and peritumoral radiomics enables accurate, non-invasive prediction of TP53 status in PDAC, warranting further prospective validation.