Development and Validation of a 7-Gene Prognostic Signature to Improve Survival Prediction in Pancreatic Ductal Adenocarcinoma

摘要

Abstract Background: Previous prognostic signatures of pancreatic ductal adenocarcinoma (PDAC) are mainly constructed to predict overall survival (OS) and predictive accuracy needs to be improved. Gene signatures that efficaciously predict both OS and disease-free survival (DFS) are of great clinical significance but are rarely reported. Methods: Univariate Cox regression analysis was adopted to screen common genes that were significantly associated with both OS and DFS in three independent cohorts. Multivariate Cox regression analysis was subsequently performed on the identified genes to determine an optimal gene signature in MTAB-6134 training cohort. Kaplan–Meier (K-M), calibration and receiver operating characteristic (ROC) curves were employed to assessed the predictive accuracy. Biological process and pathway enrichment analysis were conducted to elucidate the biological role of this signature. Results: Multivariate Cox regression analysis determined a 7-gene signature contained ASPH, DDX10, NR0B2, BLOC1S3, FAM83A, SLAMF6 and PPM1H. The signature had ability to stratify PDAC patients with different OS and DFS, both in the training and validation cohorts. ROC curves confirmed the moderate predictive accuracy of this signature. Mechanically, the signature was related to multiple cancer-related pathways. Conclusion: A novel OS and DFS prediction model was constructed in PDAC with multi-cohort and cross-platform compatibility. This signature may foster individualized therapy and appropriate management of PDAC patients.