A Prognostic Survival Model of Pancreatic Adenocarcinoma Based on Metabolism-Related Gene Expression

Accurately predicting the survival prospects of patients suffering from pancreatic adenocarcinoma (PAAD) is challenging. In this study, we analyzed RNA matrices of 182 subjects with PAAD based on public datasets obtained from The Cancer Genome Atlas (TCGA) as training datasets and those of 63 subjects obtained from the Gene Expression Omnibus (GEO) database as the validation dataset. Genes regulating the metabolism of PAAD cells correlated with survival were identified. Furthermore, LASSO Cox regression analyses were conducted to identify six genes (XDH, MBOAT2, PTGES, AK4, PAICS, and CKB) to create a metabolic risk score. The proposed scoring framework attained the robust predictive performance, with 2-year survival areas under the curve (AUCs) of 0.61 in the training cohort and 0.66 in the validation cohort. Compared with the subjects in the low-risk cohort, subjects in the high-risk training cohort presented a worse survival outcome. The metabolic risk score increased the accuracy of survival prediction in patients suffering from PAAD.

Automated summary

This study analyzed RNA matrices of PAAD patients and identified genes related to PAAD cell metabolism regulation. A metabolic risk score was created using LASSO Cox regression analysis. The proposed scoring framework showed good predictive performance and improved the accuracy of survival prediction in PAAD patients.