Abstract: Objective To screen significantly differentially expressed genes in pancreatic cancer and establish a quantitative model to predict 1-year disease-free survival (DFS) of pancreatic cancer patients.Methods From January 2021 to January 2024,124 patients with pancreatic cancer pathologically confirmed and underwent radical resection into department of hepatobiliary and pancreatic surgery of the fifth affiliated hospital of Xinjiang medical university .They were followed up for 6~12 months and divided into DFS group (n=75) and recurrence group (n=49).Bioinformatics analysis was applied to screen differentially expressed genes.KRAS,TP53,CDKN2A,SMAD4,FKBP1A,PLD1,PSMA4,PES1,miR-320b and LncRNA MIR210HG in tumor tissues were detected by real-time PCR.The general datas,blood biochemistry,tumor pathological characteristics were compared between the two groups.Multivariate Cox regression analysis was used to identify predictors of DFS. A Nomogram was established using R software, and the receiver operating characteristic (ROC) curve was employed to evaluate the predictive performance of the Nomogram for DFS.Results Compared with DFS group,albumin decreased,maximum diameter≥4 cm,histological grade 3,TNM stage Ⅱb~Ⅲa and lymph node metastasis increased,the expressions of TP53,PES1 and LncRNA MIR210HG increased,and expressions of SMAD4 and miR-320b decreased in recurrence group (P<0.05).Multivariate Cox regression showed that Ⅱb~Ⅲa (HR=2.659),lymph node metastasis (HR=2.012),PES1 (HR=1.429),LncRNA MIR210HG (HR=1.758) and miR-320b (HR=1 659) were predictive factors to DFS.The nomogram was established by R software.ROC showed that AUC of nomogram for predicting DFS was 0.876 (95%CI=0.823~0.936,P<0.001).Conclusion The quantitative model established by combining tumor pathology and differential gene expressions has good application potential for predicting 1-year DFS after pancreatic cancer resection.

Key words: pancreatic cancer; disease free survival; nomogram;PES1; LncRNA MIR210HG; microRNA-320b