Genomic profiling clarifies prognosis in resected pancreatic cancer, but not adjuvant selection
In resected pancreatic ductal adenocarcinoma (PDAC), the pressure to individualize adjuvant therapy remains intense because recurrence is common and the therapeutic window is narrow. The PRODIGE-24/CCTG PA6 biomarker analysis examined whether tumor genomics could identify patients more or less likely to benefit from mFOLFIRINOX versus gemcitabine after surgery. The answer was more modest than many precision-oncology efforts would hope. Molecular subtyping added prognostic information, yet it did not produce a biomarker that justified changing the current adjuvant standard.
Tumor DNA sequencing was successful in 317 of 350 tumors, including 168 from the mFOLFIRINOX arm and 149 from the gemcitabine arm. Transcriptomic classification with the PurIST assay layered a biologic readout onto the DNA findings, while investigators also assessed the four major PDAC driver genes, 24 homologous recombination repair–associated genes, and single-base substitution mutational signatures. Disease-free survival served as the primary endpoint, with cancer-specific survival as the secondary endpoint. The study design gave the analysis enough depth to ask a precise question: can molecular features distinguish who benefits from intensive platinum-containing therapy and who does not?
PurIST subtype carried prognostic weight inside the mFOLFIRINOX arm
The clearest signal emerged from the PurIST classifier. Within the mFOLFIRINOX-treated cohort, classical tumors had better disease-free survival than basal-like tumors, with a stratified hazard ratio of 0.48 and a 95% confidence interval of 0.31 to 0.77. In practical terms, the transcriptional state of the tumor still mattered after resection and adjuvant therapy, likely because basal-like PDAC biology is tied to more aggressive behavior, therapeutic resistance, and a tendency toward early dissemination. Classical tumors, by contrast, appear to retain a less refractory phenotype even in the postoperative setting.
The finding is biologically credible without being practice-changing. Transcriptomic subtype captures tumor behavior rather than simply cataloging mutation counts. It reflects programs of differentiation, stress response, and metastatic potential, all of which may shape postoperative recurrence risk. A more favorable disease-free survival in classical tumors does not mean these cancers should receive less treatment; it means the assay may help explain prognosis better than it helps choose between currently available regimens.
KRAS status separated the cohort, but only one subgroup showed a treatment interaction
KRAS mutation status drew the most attention among the DNA-based biomarkers. In patients with KRAS-mutated tumors, mFOLFIRINOX significantly improved disease-free survival compared with gemcitabine, with a stratified hazard ratio of 0.60, a 95% confidence interval of 0.45 to 0.79, and a P value below .001. The interaction test was significant as well, with Pint. = 0.010, indicating that the relative treatment effect differed by KRAS status. The signal is clinically interesting because KRAS-mutant PDAC represents the dominant biologic class in this disease, and the benefit of an intensive multi-agent regimen in that population is consistent with the established standard of care.
The pattern changed completely in KRAS wild-type tumors, where no benefit from mFOLFIRINOX over gemcitabine emerged. This is the sort of result that invites caution as much as excitement. Wild-type KRAS PDAC is uncommon and biologically heterogeneous; it can include tumors with alternate oncogenic drivers, unusual lineage programs, or distinct repair defects that do not map cleanly onto the usual PDAC framework. A lack of observed benefit in this subgroup is hypothesis-generating, not a mandate to abandon current therapy.
One sentence captures the clinical tension: KRAS wild-type disease did not behave like the rest of PDAC.
Homologous recombination repair markers did not identify a predictive subset
Homologous recombination repair biology has long attracted interest in pancreatic cancer because defective DNA repair can plausibly increase sensitivity to platinum-based chemotherapy. This analysis, however, did not support HRR status as a predictive biomarker for adjuvant treatment choice. The interaction test for HRR status was not significant, with Pint. = 0.568, and BRCA status also failed to predict differential benefit, with Pint. = 0.785. Those results matter because they limit the immediate clinical use of broad DNA repair panels for selecting between mFOLFIRINOX and gemcitabine in the postoperative setting.
The lack of predictive value does not mean DNA repair alterations are irrelevant. They may still affect prognosis, future response to PARP inhibition in other settings, or sensitivity to platinum in metastatic disease. In this resected adjuvant population, though, the data do not support using HRR or BRCA findings as a substitute for the current regimen selection logic. The biology is plausible; the predictive signal simply did not materialize here.
Mutational signatures did not split the treatment effect
The analysis also examined single-base substitution mutational signatures, and the mFOLFIRINOX advantage remained consistent across SBS-positive and SBS-negative subgroups. This finding narrows the argument for using mutational signature testing to steer adjuvant therapy in PDAC. Signatures can reflect prior DNA damage and underlying repair defects, but the current data do not show that they identify a group whose postoperative outcome is meaningfully altered by choosing one standard adjuvant regimen over another.
For bedside decision-making, that is an important restraint. Precision oncology often advances by finding one biomarker after another that works only in a narrow molecular niche. Here, the biomarker layer refined prognosis more than it changed treatment selection. Even where the tumor genome differed, the clinical hierarchy remained stable: mFOLFIRINOX continues to anchor adjuvant management after resection for eligible patients.
Why the standard did not change after the biomarker analysis
The central clinical conclusion is straightforward. These results do not support changing current adjuvant treatment strategies. mFOLFIRINOX remains the standard adjuvant regimen in PDAC, and the apparent absence of benefit in KRAS wild-type tumors should not be used in isolation to redirect therapy. The evidence is not strong enough to replace a regimen that has already demonstrated survival benefit in the broader resected population.
What this study does offer is a sharper map of tumor behavior after surgery. PurIST appears prognostic, KRAS status may enrich for treatment effect, and HRR/BRCA testing did not provide the predictive separation many clinicians hoped for. Those distinctions matter because they shape the next round of trial design: better enrichment strategies, more careful molecular stratification, and prospective testing of whether rare biologic subsets truly need different adjuvant approaches.
For now, the practical lesson is restraint. Genomics has clarified some of the biology of resected pancreatic cancer, but it has not yet displaced the regimen that remains the backbone of postoperative care.