Identification of an Immune-Related Prognostic Signature Associated With Immune Infiltration in Melanoma

Abstract

Melanoma is the leading cause of cancer-related death among skin tumors, with an increasing incidence worldwide. Few studies have effectively investigated the significance of an immune-related gene (IRG) signature for melanoma prognosis. Here, we constructed an IRGs prognostic signature using bioinformatics methods and evaluated and validated its predictive capability. Then, immune cell infiltration and tumor mutation burden (TMB) landscapes associated with this signature in melanoma were analyzed comprehensively. With the 10-IRG prognostic signature, melanoma patients in the low-risk group showed better survival with distinct features of high immune cell infiltration and TMB. Importantly, melanoma patients in this subgroup were significantly responsive to MAGE-A3 in the validation cohort. This immune-related prognostic signature is thus a reliable tool to predict melanoma prognosis; as the underlying mechanism of this signature is associated with immune infiltration and mutation burden, it might reflect the benefit of immunotherapy to patients.

Keywords: IRGs; TMB; immune cells infiltration; immunotherapy; melanoma; prognostic signature.

FIGURE 1

Recognition of candidate DE IRGs. (A) 55 DE IRGs were selected on LASSO coefficient profiles. The colored curves correspond to DE IRGs; horizontal axis represents the L1 Norm; vertical lines show the values of coefficients. (B) Feature selection used ten-times cross-validation to prevent overfitting and a confidence interval was obtained for optimal parameters. Red points represent log (lambda) values and two gray vertical lines represent the confidence intervals. (C) Feature selection based on the fivefold CV accuracy rate via the SVM-RFE algorithm. (D) Feature selection based on the fivefold CV error rate via the SVM-RFE algorithm. (E) Venn diagram for candidate DE IRGs.

FIGURE 2

Predictive value of the immune-related signature. (A) Risk score distribution, survival status, and expression profiles of the signature. (B) Kaplan-Meier survival analysis of the 10-IRG prognostic signature for patients with melanoma. Red line indicates the high-risk group; blue line indicates the low-risk group. (C) Time-dependent ROC analysis of the sensitivity and specificity of the risk signature. Red plot represents the 3-year OS rates (AUC = 0.731); blue plot represents the 5-year OS rates (AUC = 0.774); green plot represents the 10-year OS rates (AUC = 0.76).

FIGURE 3

Nomogram model establishment. (A) The forest plot of univariate Cox regression analysis was used to show the HR, 95% CI of each variable and P-value. (B) Nomogram model construction for predicting the probability of melanoma patients with mortality risk odds. (C) Nomogram model construction for predicting the probability of melanoma patients with OS rates. (D) Nomogram calibration curves for predicting 5- and 10-year OS in melanoma patients. Nomogram-predicted OS is plotted on the x-axis; observed OS is plotted on the y-axis; a plot along the 45° line represents a perfect calibration model.

FIGURE 4

Biological function associated with the risk-related DEGs. (A) Volcano plot for DEGs between the high- and low-risk groups. Among these DEGs, immune-related genes are highlighted. GO enrichment analysis for the upregulated (B) and downregulated (C) gene clusters. (D) GSVA demonstrates upregulated immune-related pathways in the low-risk group.

FIGURE 5

Immune characteristics of melanoma patients estimated by the ssGSEA algorithm. (A) The landscape of immune cells using the ssGSEA scores. Tumor site, mutation status of BRAF and NRAS, gender, AJCC-T, AJCC-N, AJCC-M, survival, anatomic location, stage, and Clark status are shown as patient annotations in the lower panel. Two distinct immune infiltration clusters, termed high infiltration and low infiltration, were identified by the risk groups. (B) Cellular interaction of the TME immune cell types. Three immune cell clusters were defined as cluster-A, cluster-B, and cluster-C. The thickness of the line between immune cells indicates the strength of the correlation.

FIGURE 6

Immune characteristics of melanoma patients estimated by CIBERSORT and ESTIMATE algorithms. (A) The distribution of 22 immune cell infiltrations in each melanoma sample. The lengths of the bars indicate the levels of immune cell populations and different colors represent different types of immune cells. (B) The association of 22 immune cells infiltration abundances and risk scores. The blue and red violins represent the 10-IRG signature low- and high-risk group, respectively. The white points inside the violin represent median values. (C) Heatmap of the tumor infiltrating cells with statistical significance. (D) Correlation matrix for all 22 immune cell proportions. A fraction of immune cells was negatively related and is represented in blue, whereas others were positively related and are represented in red. (E) Distributions and comparisons of Estimate, stromal, and immune scores among melanoma patients with different prognosis. The red box represents the low-risk group and the green box represents the high-risk group.

FIGURE 7

The mutation profile and clinical information of risk subgroups. (A) The heatmap shows the correlation between risk score and gene mutations (P < 0.05). Red marks the mutations, gray marks the non-mutation. (B) Mutation frequency between the high- and low-risk groups. (C) Mutational landscape and clinical characteristics of melanoma patients including event, stage, Clark status, gender, sites, and location. The right bar plot shows the mutational frequency of each gene.