Establishment and Validation of the Novel Necroptosis-related Genes for Predicting Stemness and Immunity of Hepatocellular Carcinoma Machine-learning Algorithm.

Background: Necroptosis, a recently identified mechanism of programmed cell death, exerts significant influence on various aspects of cancer biology, including tumor cell proliferation, stemness, metastasis, and immunosuppression. However, the role of necroptosis-related genes (NRGs) in Hepatocellular Carcinoma (HCC) remains elusive.

Methods: In this study, we assessed the mutation signature, copy number variation, and expression of 37 NRGs in HCC using the TCGA-LIHC dataset. We further validated our results using the ICGC-LIRI-JP dataset. To construct our prognostic model, we utilized the least absolute shrinkage and selection operator (LASSO), and evaluated the predictive efficacy of the NRGs-score using various machine learning algorithms, including K-M curves, time-ROC curves, univariate and multivariate Cox regression, and nomogram. In addition, we analyzed immune infiltration using the CIBERSOFT and ssGSEA algorithms, calculated the stemness index through the one-class logistic regression (OCLR) algorithm, and performed anti-cancer stem cells (CSCs) drug sensitivity analysis using oncoPredict. Finally, we validated the expression of the prognostic NRGs through qPCR both and .

Results: About 18 out of 37 NRGs were found to be differentially expressed in HCC and correlated with clinical outcomes. To construct a prognostic model, six signature genes (ALDH2, EZH2, PGAM5, PLK1, SQSTM1, and TARDBP) were selected using LASSO analysis. These genes were then employed to categorize HCC patients into two subgroups based on NRGs-score (low vs. high). A high NRGs score was associated with a worse prognosis. Furthermore, univariate and multivariate Cox regression analyses were performed to confirm the NRGs-score as an independent risk factor. These analyses revealed strong associations between NRGs-score and critical factors, such as AFP, disease stage, and tumor grade in the HCC cohort. NRGs-score effectively predicted the 1-, 3-, and 5-year survival of HCC patients. Immune infiltration analysis further revealed that the expression of immune checkpoint molecules was significantly enhanced in the high NRGs-score group. Stemness analysis in the HCC cohort showed that NRGs-score was positively correlated with mRNA stemness index, and patients with high NRGs-score were sensitive to CSCs inhibitors. The findings from the external validation cohort provided confirmation that the NRGs-score presented a trait with universal applicability in accurately predicting the survival of HCC. Additionally, the six prognostic genes were consistently differentially expressed in both the HCC cell line and the mouse HCC model.

Conclusion: Our study demonstrated the pivotal role of NRGs in promoting stemness and immune suppression in HCC and established a robust model which could successfully predict HCC prognosis.