Background and purpose: Pancreatic cancer is a highly aggressive solid tumor of the digestive system, with radical resection being unfeasible in approximately 80% of patients due to the absence of specific clinical manifestations in the early stages. The use of gemcitabine as a first-line chemotherapeutic agent has not significantly improved patient prognosis, primarily due to the development of chemoresistance. The precise mechanisms underlying gemcitabine resistance in pancreatic cancer remain unclear. This study aimed to explore potential biomarkers associated with gemcitabine chemoresistance in pancreatic cancer by utilizing gemcitabine-resistant cell lines and pathological pancreatic cancer tissues, in conjunction with data from online databases. Additionally, we analyzed follow-up data from pancreatic cancer patients to assess the impact of relevant targets on patient prognosis. Methods: In this study, gemcitabine-resistant cell lines were developed through stepwise induction using a gemcitabine concentration gradient. Second-generation high-throughput RNA-seq sequencing was conducted on these resistant cells, and bioinformatics analysis was employed to identify four pancreatic cancer genes from the Gene Expression Omnibus (GEO) datasets (GSE106336, GSE110580, GSE35141, and GSE140077). Co-expressed genes were screened using real-time fluorescence quantitative polymerase chain reaction (RTFQ-PCR), Western blot and immunohistochemistry to verify the expression levels of target molecules. Surgical specimens from 70 patients diagnosed with pancreatic adenocarcinoma at the First Affiliated Hospital of Xi'an Jiao Tong University between June 2018 and June 2021 were analyzed. These included 30 specimens resistant to gemcitabine, 16 non-resistant specimens, and 24 normal pancreatic tissues as controls. Ethical approval was obtained (Ethical approval: 2021 LunxuanKeZi No. 457, No. XJTU1AF2021LSK-457). Clinical and prognostic information was collected, and the log-rank test was used to evaluate the relationship between target molecule expression and patient prognosis. Results: The half maximal inhibitory concentration (IC₅₀) for gemcitabine was significantly higher in the gemcitabine-resistant cell strain (Mia GR) than in the wild-type cell line (Mia WT) (258.10 μmol/L vs 0.18 μmol/L), with a resistance index (RI) of 1 443.9. Transcriptome sequencing identified 3 985 differentially expressed genes, of which 25 were shared with the GEO datasets. Further analysis highlighted INPP4B as a key gene. RTFQ-PCR and Western blot confirmed that INPP4B mRNA and protein levels were significantly elevated in drug resistant cells compared to wild-type cells (P<0.05). Immunohistochemical analysis revealed that INPP4B expression was significantly higher in drug resistant pancreatic cancer tissues compared to non-drug resistant tissues, and lower in normal tissues than in both cancerous tissue types. Kaplan-Meier curves demonstrated that patients with low INPP4B expression had significantly better progression-free survival (PFS) than those with high expression (HR=2.874, 95% CI: 1.262-6.544, P=0.013). Although patients with low INPP4B expression also showed better overall survival (OS), the difference was not statistically significant (HR=1.484, 95% CI: 0.518-4.250, P=0.465). Conclusion: INPP4B may serve as a potential biomarker for gemcitabine chemoresistance in pancreatic cancer and is associated with poor prognosis in drug resistant patients. Developing targeted assays and treatments for INPP4B could facilitate early identification of patients likely to exhibit resistance to gemcitabine therapy, thereby improving their prognosis.