A deep learning histology biomarker, PANCprAId, was developed in 231 patients with resected pancreatic ductal adenocarcinoma and externally validated in 313 patients from PRODIGE-24/CCTG PA6 to predict who may benefit more from adjuvant gemcitabine versus modified FOLFIRINOX. In the randomized validation cohort, PANCprAId predicted differential benefit for disease-free survival (interaction P = .02) and cancer-specific survival (interaction P = .001), suggesting tumor morphology may help guide Chemotherapy selection beyond performance status.

Why It Matters To Your Practice

• Adjuvant regimen choice in resected PDAC is usually driven by general fitness, despite major differences in toxicity and efficacy between gemcitabine and mFOLFIRINOX.

• This study suggests routine histology slides may contain predictive signals that identify relative benefit from one regimen versus the other.

• If validated prospectively, this could support more personalized postoperative treatment selection.

Clinical Implications

• PANCprAId was built from treatment-specific whole-slide image models trained on disease-free survival in a retrospective multicenter cohort.

• In PRODIGE-24/CCTG PA6, the underlying histology scores stratified outcomes within both treatment groups: gemcitabine HR 1.69 (95% CI, 1.04-2.73; P = .03) and mFOLFIRINOX HR 2.02 (95% CI, 1.4-3.0; P < .001).

• The biomarker was predictive rather than merely prognostic, with significant treatment interaction for disease-free survival and cancer-specific survival.

Insights

• Predicted sensitivity to each regimen was linked to distinct epithelial and stromal morphologic features, supporting biologic plausibility.

• The approach uses standard pathology material, which may make deployment more feasible than assays requiring new tissue processing.

• Even so, the training set was modest and retrospective, so clinical adoption should await further validation.

The Bottom Line

• PANCprAId is an AI pathology tool that may help predict whether a patient with resected PDAC is more likely to benefit from adjuvant gemcitabine or mFOLFIRINOX.

• For now, it is a promising decision-support biomarker, not a practice-changing standard.