An Unfolded Protein Response-Related mRNA Signature Predicting the Survival and Therapeutic Effect of Hepatocellular Carcinoma

Abstract

Background: Tumorigenesis, metastasis, and treatment response of hepatocellular carcinoma (HCC) are regulated by unfolded protein responses (UPR) signaling pathways, including IRE1a, PERK, and ATF6, but little is known about UPR related genes with HCC prognosis and therapeutic indicators. Objective: We aimed to identify a UPR related prognostic signature (UPRRPS) for HCC and explore the potential effect of the current signature on the existing molecular targeted agents and immune checkpoint inhibitors (ICIs). Methods: We used The Cancer Genome Atlas (TCGA) database to screen candidate UPR genes (UPRGs), which are expressed differentially between hepatocellular carcinoma and normal liver tissue and associated with prognosis. A gene risk score for overall survival prediction was established using the least absolute shrinkage and selection operator (LASSO) regression analysis, which was validated using data from the International Cancer Genome Consortium (ICGC) database and evaluated by the C-index. Then immune and molecular characteristics stratified by the current UPRRPS were analyzed, and the corresponding drug sensitivity was conducted. Results: Initially, 42 UPRGs from the TCGA database were screened as differentially expressed genes, which were also associated with HCC prognosis. Using the LASSO regression analysis, nine UPRGs (EXTL3, PPP2R5B, ZBTB17, EIF2S2, EIF2S3, HDGF, SRPRB, EXTL2, and TPP1) were used to develop a UPRRPS to predict the OS of HCC patients in the TCGA set with the Cindex of 0.763. The current UPRRPS was also well-validated in the ICGC set with the C-index of 0.700. Multivariate Cox regression analyses also confirmed that the risk score was an independent risk factor for HCC in both the TCGA and ICGC sets (both P<0.05). Functional analyses showed that low-risk score was associated with increased natural killer cells, T helpers, tumor immune dysfunction and exclusion score, microsatellite instability expression, and more benefit from ICIs; the high-risk score was associated with increased active dendritic cells, Tregs, T-cell exclusion score, and less benefit from ICIs. Gene set enrichment analyses showed that the signaling pathways of VEGF, MAPK, and mTOR were enriched in high UPRRPS, and the drug sensitivities of the corresponding inhibitors were all significantly higher in the high UPRRPS subgroup (all P<0.001). Conclusion: With the current findings, UPRRPS was a promising biomarker for predicting the prognosis of HCC patients. UPRRPS might also be taken as a potential indicator to guide the management of immune checkpoint inhibitors and molecular targeted agents.