Ranking targeted therapies in advanced pancreatic adenocarcinoma: insights from a network meta-analysis

We read with great interest the manuscript by Dr. Ma and colleagues presenting a Bayesian network meta-analysis (NMA) of 13 phase III randomized trials involving 5,759 patients with advanced pancreatic adenocarcinoma to compare gemcitabine-based targeted combinations versus gemcitabine monotherapy (1). NMA provides a structured approach to comparing multiple cancer treatments when direct head-to-head randomized trials are unavailable. By integrating both direct and indirect evidence, NMA can inform comparative effectiveness across an increasingly complex oncology treatment landscape.

A key strength of NMA is its ability to evaluate several therapeutic options within a single analytical framework, thereby maximizing the use of available randomized evidence. This is particularly relevant in oncology, where rapid therapeutic innovation and multiple approved agents often outpace direct comparative trials. NMA results may support clinical guideline development, health technology assessment, and policy decisions, especially when treatment choices are numerous and evidence is fragmented. However, the validity of NMA depends on strong underlying assumptions that are frequently challenged in cancer research. Indirect comparisons require clinical and methodological similarity across trials, yet oncology studies often differ substantially in patient selection, disease stage, biomarker status, prior therapies, and outcome definitions. Such heterogeneity may compromise the transitivity and consistency assumptions essential to reliable network estimates. Additionally, treatment rankings derived from NMA may imply a degree of precision that is not clinically justified, particularly when effect sizes are modest or uncertainty is substantial. Importantly, NMA cannot overcome limitations inherent to the included randomized trials, nor does it address gaps between trial populations and real-world cancer patients.

While the authors have compiled a comprehensive dataset over two decades, several critical methodological and clinical issues warrant discussion to contextualize the findings:

The analysis spans trials conducted from 2010 to 2024, a period in which diagnostic techniques, staging criteria, supportive care, and treatment paradigms evolved. Early trials often enrolled mixed populations of locally advanced (stage III) and metastatic (stage IV) disease, and outcomes in these two populations are fundamentally different (2). Stage III patients may benefit from locoregional therapy or chemoradiation, whereas stage IV patients typically require systemic therapy, leading to intrinsic heterogeneity in survival outcomes. Pooling these groups without stratification introduces bias and limits interpretability.

Moreover, pancreatic cancer itself is a biologically heterogeneous disease (3). Differences in tumor genetics (basal vs. classical vs. quasi-mesenchymal), stromal composition, angiogenesis, and chemosensitivity contribute to variable responses to therapy (4). Even within the same stage, patients can respond differently to identical regimens. Trials conducted before widespread adoption of molecular profiling could not account for these variations, further complicating indirect comparisons in an NMA.

Additionally, trials differed in lines of therapy, performance status, prior treatments, and eligibility criteria, adding further heterogeneity. This complex mix underscores the difficulty of making meaningful treatment comparisons, especially when older studies included small cohorts such as the gemcitabine-plus-sorafenib trial (n=102) or nimotuzumab trials (n=82). Small sample sizes in these studies likely contributed disproportionately to surface under the cumulative ranking curve (SUCRA) rankings, despite a lack of statistical significance.

Several targeted therapies have historically failed to provide meaningful clinical benefit. For example, the addition of bevacizumab (CALGB 80303) to gemcitabine did not improve median overall survival (OS) (5.8 vs. 5.9 months; P=0.95) or progression-free survival (PFS) (3.8 vs. 2.9 months; P=0.07), demonstrating the limitations of anti-vascular endothelial growth factor (VEGF) therapy in a disease characterized by dense stroma and poor vascularization (5). Similarly, the anti-VEGF sorafenib failed to enhance chemotherapy outcomes (6,7). Furthermore, cetuximab (SWOG S0205) combined with gemcitabine showed no significant improvement in OS (6.3 vs. 5.9 months) or PFS, even in epidermal growth factor receptor (EGFR)-expressing tumors (8). Finally, Nimotuzumab, a humanized anti-EGFR monoclonal antibody, has shown numerically favorable outcomes in some small trials, particularly in KRAS wild-type tumors (9,10). However, these trials were limited by small sample sizes and heterogeneous patient populations.

NMA places sorafenib and nimotuzumab combination with gemcitabine high in SUCRA rankings, but these rankings do not represent statistically significant superiority. Early enthusiasm for combining bevacizumab with gemcitabine in advanced pancreatic cancer was driven by encouraging signals from phase II studies, where improvements in progression metrics and median survival appeared promising. However, when tested rigorously in large, randomized phase III trials, these benefits failed to materialize. The definitive studies demonstrated no improvement in overall survival, underscoring a recurring problem in pancreatic cancer research: signals of activity observed in small, non-comparative trials often dissipate when subjected to broader, more heterogeneous patient populations. Several explanations account for this discrepancy. Phase II cohorts are frequently enriched for patients with better performance status and less aggressive disease biology, which can inflate apparent benefits. In contrast, phase III trials more accurately reflect real-world disease complexity (11). Moreover, targeting angiogenesis in pancreatic cancer may be fundamentally limited by the tumor’s dense desmoplastic stroma and abnormal vasculature, which restrict effective drug penetration (12). Thus, while bevacizumab may exert measurable biologic effects, these effects are insufficient to overcome the dominant mechanisms driving disease progression. This experience reinforces the distinction between statistical signals and true clinical relevance, particularly when survival gains are negligible.

A parallel narrative unfolded with cetuximab plus gemcitabine. Despite a strong biologic rationale for EGFR inhibition and supportive preclinical data, randomized trials failed to show any meaningful survival advantage. Even when statistical differences in secondary endpoints were suggested, they did not translate into tangible benefit for patients. This failure highlights the limitations of targeting single signaling pathways in a disease characterized by ubiquitous KRAS mutations, pathway redundancy, and rapid adaptive resistance. EGFR expression alone proved insufficient as a predictive marker, emphasizing the inadequacy of non-selective treatment strategies.

Without molecular stratification, clinical significance remains uncertain.

Several methodological concerns should be noted:

• Trial heterogeneity: variation in disease stage, performance status, prior therapy, and inclusion criteria may bias results.
• Aggregate data: lack of individual patient-level data prevents adjustment for key covariates (e.g., age, comorbidities, molecular profile).
• Missing PFS data: some trials omitted PFS outcomes, limiting pooled analysis completeness.
• Probabilistic rankings (SUCRA): without statistical significance, SUCRA rankings may mislead readers regarding relative efficacy.

The analysis confirms that no gemcitabine-based targeted combination significantly improved OS or PFS in unselected populations, consistent with two decades of clinical experience. Probabilistic rankings suggesting superiority (e.g., nimotuzumab or sorafenib combinations) should be interpreted cautiously. Most agents included are no longer routinely used in clinical practice, limiting the study’s applicability.

Moving forward, the field is increasingly focused on molecularly defined subsets and targeted therapies for KRAS G12C, KRAS G12D, and other actionable mutations (13). These contemporary approaches were not represented in the historical trials, underscoring the limitations of indirect comparisons of older studies.

In summary, we congratulate the authors on their strong work. NMA is a valuable evidence synthesis tool in oncology, but should be interpreted with caution. Authors’ Bayesian NMA provides a detailed review of historical targeted therapy trials in advanced pancreatic cancer. The key finding that no combination significantly improves OS or PFS over gemcitabine alone aligns with prior clinical experience. Their findings are best viewed as complementary to direct randomized comparisons and real-world data rather than as definitive evidence for clinical decision-making. However, the inherent heterogeneity of pancreatic cancer, differences in trial populations, and small sample sizes of key studies limit the interpretability of SUCRA rankings. Authors appropriately acknowledge some limitations, but the clinical impact is tempered by outdated regimens and lack of molecular stratification. Future work should focus on precision medicine approaches and molecularly defined therapies reflective of current standards.

Acknowledgments

None.

Footnote

Provenance and Peer Review: This article was commissioned by the editorial office, Journal of Gastrointestinal Oncology. The article did not undergo external peer review.

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0209/coif). M.W.S. received research funding from SystImmune, Endevica Bio, GSK, Tyra, Teva, Marengo, OBI Pharma, IDEAYA, Binhui, Astellas, Takeda, Mersana Therapeutics, Summit, J-Pharma, Beijing Biostar; consulting fees from Genentech, Oncologic Drug Advisory Committee (US World Meds) and royalty from UpToDate. P.A.P. received honoraria from AstraZeneca; Bayer; Blueprint Medicines; Bristol-Myers Squibb/Medarex; Celgene; Daiichi Sankyo/Astra Zeneca; Guardant Health; incyte; Ipsen; Merck; Rafael Pharmaceuticals; SynCoreBio; TriSalus Life Sciences. P.A.P. has consulting or advisory role for Celgene; Daiichi Sankyo; Ipsen; Merck; SynCoreBio; Taiho Pharmaceutical; TriSalus Life Sciences. P.A.P. has served on Speakers’ Bureau for Bayer; Bristol-Myers Squibb/Medarex; Celgene; Incyte; Ipsen; Novartis. P.A.P. has received research funding from Advanced Accelerator Applications (Inst); ASLAN Pharmaceuticals (Inst); Bayer (Inst); Boston Biomedical (Inst); Caris Life Sciences (Inst); Genentech (Inst); Halozyme (Inst); Immunomedics (Inst); Incyte (Inst); Karyopharm Therapeutics (Inst); Lilly (Inst); Merck (Inst); merus (Inst); Momenta Pharmaceuticals (Inst); Novartis (Inst); Plexxikon (Inst); QED Therapeutics (Inst); Regeneron (Inst); Taiho Pharmaceutical (Inst); Tyme (Inst). P.A.P. has received travel, accommodations, and expenses from AbbVie; Celgene; Rafael Pharmaceuticals. A.S.A. received funding from Purple Biotec, FanWave Therapeutics, Colorado Chromatography, RLL and Blackstone Therapeutics; and is a council member at GLG and Guidepoint. The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Ma H, Wang S, Ya Y, et al. Landscape targeted of therapy in advanced pancreatic adenocarcinoma: a network meta-analysis of randomized controlled trials 2010-2024. J Gastrointest Oncol 2025;16:2802-13. [Crossref] [PubMed]
2. Saif MW. Pancreatic cancer: highlights from the 42nd annual meeting of the American Society of Clinical Oncology, 2006. JOP 2006;7:337-48. [PubMed]
3. Sugarman R, Patel R, Sharma S, et al. Pharmacokinetics and pharmacodynamics of new drugs for pancreatic cancer. Expert Opin Drug Metab Toxicol 2019;15:541-52. [Crossref] [PubMed]
4. Saif MW. Advanced stage pancreatic cancer: novel therapeutic options. Expert Rev Clin Pharmacol 2014;7:487-98. [Crossref] [PubMed]
5. Kindler HL, Niedzwiecki D, Hollis D, et al. Gemcitabine plus bevacizumab compared with gemcitabine plus placebo in patients with advanced pancreatic cancer: phase III trial of the Cancer and Leukemia Group B (CALGB 80303). J Clin Oncol 2010;28:3617-22. [Crossref] [PubMed]
6. Gonçalves A, Gilabert M, François E, et al. BAYPAN study: a double-blind phase III randomized trial comparing gemcitabine plus sorafenib and gemcitabine plus placebo in patients with advanced pancreatic cancer. Ann Oncol 2012;23:2799-805. [Crossref] [PubMed]
7. Saif MW. Pancreatic cancer: Sorafenib: no effect on efficacy of chemotherapy in pancreatic cancer. Nat Rev Gastroenterol Hepatol 2014;11:8-9. [Crossref] [PubMed]
8. Philip PA, Benedetti J, Corless CL, et al. Phase III study comparing gemcitabine plus cetuximab versus gemcitabine in patients with advanced pancreatic adenocarcinoma: Southwest Oncology Group-directed intergroup trial S0205. J Clin Oncol 2010;28:3605-10. [Crossref] [PubMed]
9. Schultheis B, Reuter D, Ebert MP, et al. Gemcitabine combined with the monoclonal antibody nimotuzumab is an active first-line regimen in KRAS wildtype patients with locally advanced or metastatic pancreatic cancer: a multicenter, randomized phase IIb study. Ann Oncol 2017;28:2429-35. [Crossref] [PubMed]
10. Qin S, Li J, Wang Z, et al. Nimotuzumab plus gemcitabine improves survival in KRAS wild-type advanced pancreatic cancer: a randomized phase III study. J Clin Oncol. 2023;40:LBA4011.
11. Saif MW. From Screening to Treatment of Pancreatic Cancer: A Comprehensive Review. JOP 2021;22:70-9. [PubMed]
12. Saif MW. Anti-angiogenesis therapy in pancreatic carcinoma. JOP 2006;7:163-73. [PubMed]
13. Choucair K, Imtiaz H, Uddin MH, et al. Targeting KRAS mutations: orchestrating cancer evolution and therapeutic challenges. Signal Transduct Target Ther 2025;10:385. [Crossref] [PubMed]