A novel natural killer-related signature to effectively predict prognosis in hepatocellular carcinoma

Abstract

Background: Hepatocellular carcinoma (HCC) is a prevalent tumor that poses a significant threat to human health, with 80% of cases being primary HCC. At present, Early diagnosis and predict prognosis of HCC is challenging and the it is characterized by a high degree of invasiveness, both of which negatively impact patient prognosis. Natural killer cells (NK) play an important role in the development, diagnosis and prognosis of malignant tumors. The potential of NK cell-related genes for evaluating the prognosis of patients with hepatocellular carcinoma remains unexplored. This study aims to address this gap by investigating the association between NK cell-related genes and the prognosis of HCC patients, with the goal of developing a reliable model that can provide novel insights into evaluating the immunotherapy response and prognosis of these patients. This work has the potential to significantly advance our understanding of the complex interplay between immune cells and tumors, and may ultimately lead to improved clinical outcomes for HCC patients.

Methods: For this study, we employed transcriptome expression data from the hepatocellular carcinoma cancer genome map (TCGA-LIHC) to develop a model consisting of NK cell-related genes. To construct the NK cell-related signature (NKRLSig), we utilized a combination of univariate COX regression, Area Under Curve (AUC) LASSO COX regression, and multivariate COX regression. To validate the model, we conducted external validation using the GSE14520 cohort.

Results: We developed a prognostic model based on 5-NKRLSig (IL18RAP, CHP1, VAMP2, PIC3R1, PRKCD), which divided patients into high- and low-risk groups based on their risk score. The high-risk group was associated with a poor prognosis, and the risk score had good predictive ability across all clinical subgroups. The risk score and stage were found to be independent prognostic indicators for HCC patients when clinical factors were taken into account. We further created a nomogram incorporating the 5-NKRLSig and clinicopathological characteristics, which revealed that patients in the low-risk group had a better prognosis. Moreover, our analysis of immunotherapy and chemotherapy response indicated that patients in the low-risk group were more responsive to immunotherapy.

Conclusion: The model that we developed not only sheds light on the regulatory mechanism of NK cell-related genes in HCC, but also has the potential to advance our understanding of immunotherapy for HCC. With its strong predictive capacity, our model may prove useful in evaluating the prognosis of patients and guiding clinical decision-making for HCC patients.

Keywords: HCC; Natural killer cell; Prognostic signature; Tumor microenvironment.

Fig. 1

Working flow chart, TCGA-LIHC: The Cancer Genome Atlas-Live Hepatocellular Carcinoma, ROC: Receiver operating characteristic

Fig. 2

Construction of the NKRLSig model based on the target genes A A Venn diagram depicts 115 overlapping mRNAs considered NK cell-related, including 85 down-regulated genes and 30 up-regulated genes. B Heat map plot differentially expressed NKRLSig. C Volcano plot differentially expressed NKRLSig. D Forest plot showing 17 mRNAs with hazard ratios (95%confidence intervals) and P values based on the result of univariate Cox regression analysis. E–F mRNAs screened by the LASSO-Cox regression model

Fig. 3

Evaluation of the predictive efficacy of the prognostic model. A The multivariate Cox relapse coefficient. B Circus plot show the correlation between risk scores and target genes. Kaplan–Meier survival curves in the high- and low-risk groups stratified by risk scores for overall survival in the training set C and test set D. E The risk score distribution and patient status for the TCGA-LIHC cohort

Fig. 4

Verified the accuracy of the prognostic model. Time-dependent ROC curves analysis in the train set A and test set B. C GO and KEGG functional enrichment analysis of NKRLSig

Fig. 5

GSEA enrichment analysis identifies KEGG pathways associated with high-risk A and low-risk groups B in the training set

Fig. 6

Clinical application of the 5-NKRLSig model in HCC. The difference in risk score by Female A-B, Male C-D, Age < 65 E–F, Age ≥ 65 G-H, Stage I/II I-J, Stage III/IV K-L, Grade I/II M–N, Grade III/IV O-P of HCC

Fig. 7

Construction of the nomogram. A Univariate and multivariate Cox regression analysis in TCGA-LIHC. B Nomogram integrating the Risk score and Stage. C Calibration curves for predicting 1, 3and 5 years OS in the train set and test set. D Concordance index curves depicting risk scores and other clinical parameters relevant to predicting HCC patient prognosis. E DCA of the nomogram and AJCC stage

Fig. 8

Genetic alterations and tumor microenvironment. Top 10 gene mutations in high-risk A and low-risk B groups. C The analysis of TMB scores of high-risk and high-risk groups. D Comparison of the immune score, matrix score and ESTIMATE score in high- and low-risk groups

Fig. 9

Application of risk score in immunotherapy, Chemotherapy and target therapy. A Prediction of immunotherapy response based on the TIDE algorithm. IPS score for immunotherapy. B CTLA4- PD1 − . C CTLA4 + PD1 − . D CTLA4 − PD1 + . E CTLA4 + PD1 + . Risk score predicts chemotherapy sensitivity. Cetuximab F, Erlotinib G, Trametinib H, XAV939 I, Docetaxel J, MLN4924 K, OSU-03012 L, YM155 M, Salubrinal N, Vorinostat O