Construction and validation of senescence risk score signature as a novel biomarker in liver hepatocellular carcinoma: a bioinformatic analysis.

Background: Globally, liver cancer as one of the most frequent fatal malignancies, hits hard and fast. And the lack of effective treatments for liver hepatocellular carcinoma (LIHC), activates the researchers to promote promising precision medicine. Interestingly, emerging evidence proves that cellular senescence is involved in the progression of cancers and is recognized for its hallmark-promoting capabilities. Hence, efforts have been made to construct and validate the senescence risk score signature (SRSS) model as a novel prognostic biomarker for LIHC.

Methods: The existing databases were mined for the following bioinformatics analyses. GSE22405, GSE57957, and senescence-related genes (SRGs) from public databases were utilized as a training set and the validation set was constituted by LIHC and pancreatic adenocarcinoma (PAAD) from The Cancer Genome Atlas (TCGA). After overlapping differentially expressed genes (DEGs) with SRGs, differentially expressed SRGs were identified with the progression of liver cancer through univariate and multivariate Cox regression and enrichment analyses. The model that utilized three SRGs was constructed using the least absolute shrinkage and selection operator (LASSO) regression algorithm. Next, to evaluate the predictive performance of the SRSS model, the overall survival (OS) and survival rates were assessed through Kaplan-Meier (KM) and the receiver operating characteristic (ROC) curves. The predictive value for LIHC prognosis was further evaluated by capitalizing on risk score, nomograms, decision curve analysis (DCA) curves, and clinical information including tumor stages, gender, age, and race.

Results: DEGs were revealed as enriching in multiple tumor-related biological processes (BPs) and pathways. , , and were identified as the three considerable SRGs for the model. The high-risk group had a worse prognosis [both hazard ratio (HR) >1, P<0.001] and ROC curves showed a reliable predictive model with area under the curve (AUC) predictive values ranging from 0.673-0.816 for different-year survival rates respectively. The univariate and multivariate Cox regression analyses exhibited that risk score was the only credible prognostic predictor (HR >1, P<0.001) among clinical features such as tumor stage, age, etc., in LIHC. The nomograms, and DCA curves, combined with multiple clinical information, proved that the predictive ability of SRSS was strongest, followed by nomogram and traditional tumor node metastasis (TNM) stage was the weakest.

Conclusions: In summary, comprehensive analyses supported that the SRSS model can better predict survival and risk in LIHC patients. Promisingly, it may point out a brand-new direction for LIHC therapy.