Abstract
Pancreatic cancer (PC) is an aggressive malignant disease with poor prognosis, often diagnosed late, progressing rapidly, and resistant to chemotherapy. Although small molecule inhibitors (SMIs) show promise in preclinical PC models, translation into clinics remains challenging. In this systematic review and meta-analysis, we screened literature according to predefined criteria to identify preclinical PC mouse models used for SMI therapy in primary tumors, assess reporting quality, and evaluate tumor reduction, heterogeneity and publication bias. Following a pre-registered PROSPERO protocol (CRD42022314932), literature searches in PubMed and Embase yielded 2972 articles, of which 297 were included for data extraction. Most studies used PDX models or MiaPaCa-2 and PANC-1 cell lines as heterotopic xenografts, with Foxn1(nu) mice representing the predominant genetic background. Reporting quality, assessed using the ARRIVE guidelines, revealed substantial gaps, particularly in blinding (94% not reported), inclusion/exclusion criteria (49% not reported), and randomization (34% not reported). Meta-regression accounted for part of the observed heterogeneity and funnel plot asymmetry was consistent with publication bias. This study emphasized the large variability among preclinical PC mouse models used to investigate SMI candidates and the complexity of model choice highlighting the critical need for improved reporting practices to enhance reproducibility and reliability of preclinical tumor models. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1038/s41598-025-25191-1.