The Construction of a Prognostic Model Based on a Peptidyl Prolyl Cis–Trans Isomerase Gene Signature in Hepatocellular Carcinoma

Objective: The aim of the present study was to construct a prognostic model based on peptidyl prolyl cis-trans isomerase (PPIase) gene signature and explore the prognostic value of this model in patients with hepatocellular carcinoma (HCC). Methods: The transcriptome and clinical data of HCC patients were downloaded from TCGA and ICGC database as training set and validation set, respectively. PPIase gene sets were obtained from the molecular signatures database. The differential expression (DE) of PPIase genes were analyzed by R software to screen out the DE-PPIase genes. A prognostic model based on PPIase gene signature was established by Cox, Lasso and stepwise regression methods. Kaplan-Meier survival analysis was used to evaluate the prognostic value of the model and validate it with an independent external data. Finally, nomogram and calibration curves were developed in combination with clinical staging and risk score. Results: Differential gene expression analysis of HCC and adjacent tissues showed that there were 16 up-regulated PPIase genes, and 0 gene was down-regulated. A prognostic model of HCC was constructed based on three gene signature (FKBP6, CWC27 and FKBP1A) by Cox, Lasso and stepwise regression analysis. The Kaplan-Meier curve showed that HCC patients in high risk score group had a worse prognosis (P<0.05). The ROC curve revealed that the AUC value of predicting survival rate at 1, 2 and 3 years was 0.725, 0.680, 0.644, respectively. In addition, the evaluation results of the model by the validation set were basically consistent with those of the training set. A nomogram incorporating TNM stage and risk score was established, and the calibration curve matched well with the diagonal. Conclusion: A prognostic model based on 3 PPIase gene signatures is expected to provide reference for prognostic risk stratification in patients with HCC.