Prognostic and predictive impact of the baseline systemic proteome in patients with <i>RAS</i> wild-type metastatic colorectal cancer: Analysis from the randomized phase II PanaMa (AIO KRK0212) trial.

Abstract

3546 Background: Systemic proteomics offers a minimally invasive approach to identifying biomarkers in metastatic colorectal cancer (mCRC). This study analyzed the prognostic and predictive potential of the systemic proteome in RAS wild-type (wt) mCRC patients from the PanaMa trial, which investigated maintenance therapy with fluorouracil/folinic acid (FU/FA) ± panitumumab (Pmab) following induction with FU/FA plus oxaliplatin and Pmab. Methods: Baseline serum samples were analyzed using liquid chromatography-mass spectrometry to identify protein markers. Their impact on overall survival (OS) and progression-free survival (PFS) was evaluated using Kaplan-Meier estimates and Cox regression. Prognostic analyses utilized hierarchical clustering and random forest models to delineate protein groups associated with survival outcomes, enhanced by sparse partial least squares discriminant analysis for feature selection. Gene ontology analysis was used to identify biological functions of these markers. ROC analysis was performed to evaluate the accuracy of prognostic protein signatures. For predictive analysis, PFS outcomes of maintenance with FU/FA ± Pmab were assessed using Kaplan-Meier estimates and Cox regression. Hazard ratios (HR), differences in hazard ratios (delta HR), and statistical significance (p-values) were used to differentiate patient outcomes based on protein marker expression. Results: Of 378 patients treated in the trial, 231 had baseline serum samples. Proteomic clustering identified two survival clusters: the high-survivability cluster showed significantly longer induction PFS (HR 0.75, 95% CI 0.56–1.00, P = 0.05) and OS (HR 0.63, 95% CI 0.45–0.88, P = 0.01). Hierarchical clustering revealed 470 proteins, with specific proteins enriched in high-survivability (e.g., ALB, APOA2) and low-survivability (e.g., SERPINA3, CRP) clusters. Gene ontology analysis highlighted distinct pathways, such as enzyme inhibitor activity in low-survivability clusters and peptidase regulator activity in high-survivability clusters. For maintenance, prognostic arm-specific proteomic signatures linked to improved PFS with strong accuracy in the FU/FA + Pmab arm (AUC 0.99), and FU/FA arm (AUC 1.00). Predictive analysis revealed a total of eight proteins that predicted benefit of Pmab addition to maintenance. A positive combined proteomic biomarker including these proteins (ITIH4, FLNC, HP, CTPS1, SERPINA1, HRG, MAN1C1, C4A) predicted significant benefit of addition of Pmab to FU/FA maintenance (PFS: HR 0.68, 95% CI 0.49–0.95, P = 6.4e-09). Conclusions: Proteomic profiling identified prognostic clusters linked to distinct survival outcomes and predictive signatures for FU/FA ± Pmab maintenance, supporting its utility in guiding personalized treatment strategies for RAS wt mCRC.