Spliceosome-driven transcriptomic profiling to reveal prognostic subtypes and risk stratification in hepatocellular carcinoma.

Abstract

592 Background: Recent evidence strengthened the link between alterations in spliceosome components and key cellular processes driving cancer progression. These dysregulations have been implicated in enhanced chemoresistance and the development of more aggressive cancer phenotypes. In this study, we investigate the transcriptomic alterations of spliceosome genes in Hepatocellular Carcinoma (HCC) and their association with prognosis and progression, identifying distinct prognostic molecular subtypes and risk-groups. Methods: Transcriptomic and clinical data of 366 HCC and 50 normal samples were obtained from TCGA, with a compiled 437-spliceosome gene list from reviews and the molecular signature database. Differential expression analysis between normal and tumor samples identified differentially expressed spliceosomal genes (SGs). These were fitted into LASSO, univariable, and multivariable Cox regression models to extract SGs associated with prognosis. Prognostic SGs were then used to construct a risk score model. Each patient received a risk score, and patients were stratified into high- and low-risk groups using the median score. External validation was performed on 221 HCC patients (GSE14520), 140 sorafenib-treated patients (GSE109211), and 147 TACE-treated patients (GSE104580). Results: In the TCGA cohort, 1814 DEGs were identified between tumor and normal samples (adjusted P < 0.05, |Log2FC| > 1). Intersection with the SG list revealed 15 differentially expressed SGs. Regression modeling extracted DHX34 and SF3B4 as independent survival predictors (P < 0.05). High-risk patients (n=182) had worse survival than low-risk (n=183) in TCGA (P < 0.05). Validation confirmed these findings, with high-risk patients (n=111) showing worse outcomes than low-risk (n=110) (P < 0.05). In treated cohorts, high-risk status was associated with non-response to sorafenib (n=31, P=0.0078) and TACE (n=68, P=0.0013). High-risk groups showed higher protein expression of RPTOR, MTOR, CCNB1, SERPINE1, and RAF1 (P < 0.05). Risk groups correlated with TP53 mutations, TMB, aneuploidy, tumor grade, AFP, and ECOG scores (P < 0.05). High-risk TP53-mutant patients had poorer survival than low-risk TP53-mutants (P=0.013). Conclusions: In conclusion, we developed a two-gene spliceosome-based risk score that predicts HCC prognosis and may guide personalized treatment strategies.