Allele-stratified activity of RAS inhibitors in pancreatic ductal adenocarcinoma: a systematic review with model-validation-stratified synthesis of preclinical evidence and exploratory meta-analysis of early-phase trials
Abstract
Background: PDAC is a highly aggressive malignancy driven predominantly by activating KRAS mutations, and allele-specific and pan-RAS inhibitors have emerged as candidate targeted therapies. The evidence supporting them is dominated by experimental studies whose models differ widely in how faithfully they reproduce human disease, and clinical maturity is uneven across alleles.
Objectives: To evaluate allele-stratified RAS-inhibitor activity across KRAS subtypes in PDAC, synthesizing experimental and clinical evidence separately, and to determine whether the principal conclusions of the review survive when the experimental evidence is stratified by the level of model validation achieved.
Methods: PubMed, Cochrane Library, Scopus, and Google Scholar were searched from inception to 26 June 2026, with a pre-specified updated search in June 2026. Experimental studies were appraised with a modified JBI Critical Appraisal Checklist for Quasi-Experimental Studies supplemented by preclinical-specific reporting domains, and were independently classified by the highest level of model validation achieved, from conventional two-dimensional cell lines through to multi-model validation spanning patient-derived and immunocompetent in vivo systems; clinical studies were appraised with ROBINS-I. Experimental and clinical evidence were synthesized separately. Single-arm proportions from KRAS G12C trials were pooled on the logit scale using inverse-variance random-effects meta-analysis with Hartung-Knapp confidence intervals, and certainty was rated using GRADE.
Results: Twenty-four studies were included (20 experimental and 4 early-phase, single-arm clinical trials). Four of the 20 experimental studies (20%) rested on conventional two-dimensional cell lines alone, 9 (45%) incorporated patient-derived material, 7 (35%) used immunocompetent models, and 14 (70%) reproduced their principal finding in at least two independent model classes. The allele-dependent pattern of single-agent activity persisted when cell-line-only studies were set aside, whereas 6 of the 17 studies reporting combination strategies provided no confirmation in any higher-order system. Across three single-arm KRAS G12C trials, the pooled objective response rate was 36.9% (95% CI, 9.8-75.7%; I² = 66.2%; individual trials 21.1%, 45.5%, and 46.9%), the pooled disease control rate was 85.9% (95% CI, 62.1-95.8%; I² = 0%; individual datasets 81.8%, 84.2%, and 93.8%), and the grade ≥3 treatment-related adverse-event rate was 23.8% (95% CI, 11.2-43.5%; I² = 0%). GRADE certainty was very low for all three outcomes.
Conclusions: The available evidence is consistent with allele-dependent activity of RAS-targeted therapy in PDAC and supports biomarker-guided treatment selection, but the clinical evidence base is confined to small, single-arm, early-phase trials of KRAS G12C inhibitors. The pooled estimates are exploratory descriptions of observed activity in selected early-phase populations, not comparative estimates of treatment efficacy. KRAS G12D-directed therapy remains at a substantially earlier stage of clinical validation than KRAS G12C-directed therapy despite a much larger preclinical literature.
