Efficacy and safety signals from early-phase studies of KRAS inhibition in pancreatic cancer.

Introduction
Pancreatic ductal adenocarcinoma (PDAC) is characterized by aggressiveness and dismal prognosis, particularly in the metastatic setting1. Standard systemic therapy consists of multi-agent chemotherapy, such as fluorouracil-based combinations or gemcitabine plus nab-paclitaxel2,3. While these regimens modestly prolong survival, their toxicity often results in dose reductions or early discontinuation4,5. Activating mutations in RAS, most notably KRAS, are present in approximately 90% of cases and are the major oncogenic drivers of PDAC6,7. For decades, KRAS was considered “undruggable” because of its high nucleotide affinity and lack of accessible binding pockets, restricting therapy to largely ineffective indirect pathway inhibition8. This paradigm shifted with the development of small molecules capable of directly disrupting KRAS signaling, a long-sought therapeutic breakthrough9,10.
KRAS is a small GTPase that cycles between inactive GDP-bound (OFF) and active GTP-bound (ON) states. Oncogenic mutations impair GTP hydrolysis, sustaining KRAS activation and persistent downstream signaling11. Classes of KRAS inhibitors have emerged, each targeting distinct conformational states, including OFF-state, ON-state, and tri-complex inhibitors, with unique mechanisms and, most importantly, differential susceptibility to resistance11–17. This mechanistic diversity indicates that efficacy, toxicity, and resistance may need to be evaluated at the individual drug level, rather than inferred across agents within the same class.
Given this context, understanding the clinical efficacy and safety of these novel agents in PDAC is essential. To synthesize the available evidence, we performed a systematic review and meta-analysis of KRAS-targeted therapies. The primary endpoint was the objective response rate (ORR), with safety assessed by the frequency of common adverse events. Seven non-randomized phase I/II cohorts encompassing 463 patients with metastatic or locally advanced PDAC treated with KRAS inhibitors in the second-line or beyond met the inclusion criteria (Tables 1 and 2) (PRISMA flow diagram and study characteristics are shown in the Supplementary Appendix). Of the 695 patients included in the study, only 293 were evaluated for the objective response rate (ORR). The remaining patients were excluded from the analysis due to the absence of adequate imaging assessments, loss to follow-up, or other reasons that precluded response assessment.

The pooled ORR across studies was 29% (Fig. 1), reflecting measurable antitumor activity in otherwise treatment-refractory disease18. Although KRAS-targeted agents have demonstrated activity across classes, the magnitude of their benefit in PDAC remains modest compared to that in other malignancies. Nevertheless, for pancreatic cancer, particularly KRAS G12D-mutated tumors, even incremental improvements are clinically meaningful, given the historically poor prognosis. Current studies remain in early development, with immature survival outcomes; however, these data collectively provide proof-of-concept that direct KRAS inhibition is feasible in PDAC.

Consistent efficacy estimates (I2 = 5.7%) and a narrow prediction interval reinforced the robustness of the pooled ORR, suggesting that future trials may yield similar results18. Meta-regression further showed no association between study size and response rate (Fig. 2).

Safety is an equally critical aspect of evaluating these therapies. Gastrointestinal toxicities were common, with diarrhea (40%) and nausea (41%) affecting more than one-third of the patients (Fig. 3), however, when analyzing the data specifically for patients with PDAC who experienced these adverse events, the proportions were 43% for diarrhea and 40% for nausea (Fig. 4). Confidence and prediction intervals were wide and leave-one-out analyses identified that no single study explained the heterogeneity effects, which reinforces the variability among the included studies (Supplementary Appendix). In multi-tumor cohorts, the absence of PDAC-specific adverse event reporting precluded granular toxicity analyses, especially for distinguishing class-related patterns (e.g., ON-state vs. OFF-state inhibitors or pan-RAS vs. mutation-specific agents). Class-dependent effects may further contribute to variability, as pan- or multi-selective RAS inhibitors are associated with higher rates of dermatologic toxicities, whereas mutation-specific agents more commonly produce gastrointestinal events19. To address these limitations, we conducted a PDAC-specific toxicity analysis when stratified data were available and supplemented it with a pooled assessment of all eligible trials. This two-tiered approach provided both a focused evaluation of toxicity in PDAC and a broader overview of the safety profile of KRAS inhibitors across heterogeneous study populations.

Beyond efficacy and safety profiles, an important consideration is the risk of bias, which was found to be substantial. The ROBINS-I assessments indicated a moderate-to-high risk of bias across domains, including confounding, selection, and outcome reporting. In the absence of control arms, response rates cannot be reliably contextualized against the natural history of PDAC [Supplementary Appendix]. Visual inspection of the funnel plot did not reveal clear asymmetry, although the limited number of available studies restricts definitive assessment of publication bias. Studies were largely symmetrically distributed around the pooled estimate, with greater dispersion among smaller trials, as expected. [Figure 5].

Taken together, the results of this meta-analysis indicate that direct KRAS inhibition in PDAC is both feasible and biologically active, although the supporting clinical evidence is limited, heterogeneous, and prone to bias, warranting cautious interpretation. Even so, these studies contribute to defining an evolving therapeutic landscape. After decades in which KRAS was deemed “undruggable,” the emergence of covalent inhibitors and mutation-directed delivery systems represents a paradigm shift. The modest but measurable activity validates the biological rationale and establishes a foundation for further development. Moving forward, progress will hinge on allele-specific inhibitors, rational combinations with immunotherapy and targeted agents, mutation-directed vaccines, and, critically, the standardization of trial design, endpoints, and biomarker integration, such as ctDNA dynamics and resistance profiling, to achieve durable benefits.

Methods

Search strategy and study selection
We performed a systematic review and meta-analysis to evaluate the efficacy and safety of KRAS inhibitors in pancreatic ductal adenocarcinoma (PDAC). A comprehensive literature search was conducted in MEDLINE (via PubMed), Embase, and the Cochrane Library using predefined Medical Subject Headings (MeSH) and free-text keywords. The search terms included: (“pancreatic neoplasms” OR “carcinoma, pancreatic ductal” OR “pancreatic” [tiab]) AND (“Proto-Oncogene Proteins p21(ras)” OR “ras proteins” OR KRAS OR G12D OR GFH375 OR G12C) AND (Sotorasib OR Adagrasib OR Pyridines OR Krazati OR Piperazines OR MRTX849 OR “RNAi therapeutics” OR Daraxonrasib). The search was last updated in November/2025. Reference lists of relevant articles were manually screened to identify additional eligible studies. Full-text publications, conference abstracts, and presentations were included. Additionally, an updated screening was conducted to incorporate recently published reports from two clinical trials presented at European Society for Medical Oncology (ESMO) Conference 202520,21, which met the predefined eligibility criteria and were included in the quantitative synthesis. The latest updates for all included studies were sought to ensure greater robustness and the most up-to-date evidence22.

Eligibility criteria
Studies were included if they met the following criteria: (1) enrolled patients with KRAS-mutated PDAC, (2) included patients with previously treated metastatic or locally advanced disease, and (3) evaluated the use of KRAS inhibitors as second-line and beyond therapy. The exclusion criteria were as follows: (1) duplicate reports, (2) use of KRAS inhibitors as first-line therapy, and (3) studies on non-pancreatic primary tumors.

Study selection and data extraction
Two independent reviewers (K.O.M.T. and O.C.) screened the titles, abstracts, and full texts against predefined criteria. Disagreements were resolved through discussion and consensus. Data were extracted from each eligible study on the objective response rate (ORR), defined as partial or complete response, and safety outcomes, focusing on the most frequently reported adverse events.

Risk of bias and publication bias
The risk of bias in non-randomized studies was independently assessed by two reviewers (K.O.M.T. and O.C.) using the Risk of Bias in Non-randomized Studies of Interventions tool. Discrepancies were resolved through consensus. Publication bias was evaluated by visually inspecting the funnel plot, and the Freeman–Tukey double arcsine transformation was used to stabilize the variance of the proportions before calculating the pooled effect.

Statistical analysis
All statistical analyses were performed using R software (version 4.5.1; R Foundation for Statistical Computing, 2025) in accordance with the PRISMA guidelines. Patients were dichotomized as those who achieved or did not achieve ORR. A single-arm meta-analysis of proportions was conducted to estimate the pooled ORR and incidence of adverse events, each with corresponding 95% confidence intervals (CI). The studies were weighted according to the inverse of their variance to account for the study size and precision. Between-study heterogeneity was assessed using Cochran’s Q test and quantified with the I2 statistic, with thresholds of I2 >25% or p < 0.10 indicating significant heterogeneity. A random-effects model was applied unless significant heterogeneity warranted an alternative approach. Sensitivity analyses were performed to assess the robustness of these findings.

Supplementary Information
Below is the link to the electronic supplementary material.