Data analysis and constructing prognostic gene signature
Following the selection of the only known and available data on MM, 414 patients were extracted from the TCGA database, which is shown in Supplementary Table 1. When the transcriptomic gene expression was matched by the ICI genes (CKTTD), the univariate hazard model regression analysis was performed, which indicated only 40 genes with a statistically significant difference in low vs high expression (considering only P value = 0.05); (CD27,VSIR,ICOS,HAVCR2,IDO1,CD80,CTLA4,TNFRSF18,TIGIT,TNFRSF4,CD274,TNFRSF9,PDCD1,ISYNA1,CSF1,CD47,TLR8,ARG1,TREM2BMI1,TNFRSF1B,BTK,LILRB1,PLK1,CD52,CCR5,TLR3,TLR7,CREBRF,TLR9,TLR4,LILR2,CXCR4,CCR2,NOX4,CD96,CD38,AXL,NR2F6). (Supplemetary1 A). Further multivariate hazard model analysis revealed 13 ICI genes which were supposed to be as risk gene signature (ICOS, CD80,CTLA4,TNFRSF4,PDCD1,CSF1,TLR8, ARG1,BMI1,TLR3,TLR4,NOX4, AXL) and the top significant 6 genes was selected as risk score genes (considered only P value < 0.05) (ICOS, CD80, NOX4,ARG1,TNFRSF4, PDCD1) (Supplementary Fig. 1).
The KM survival analysis was undertaken to evaluate overall survival (OS) results to identify the potential of each ICI gene signature-related OS to predict the prognosis of MM patients. The overall survival rate for patients in the high-risk group was considerably lower than the rate in the low-risk group in the risk score. (log-rank P-value < 0.0001) (Supplementary Fig. 2 A).
This study evaluated the 6-gene signature's predictive potential using a nomogram (Supplementary Fig. 2B). Thus, patients' risk scores are divided into two groups: those at great risk and those who are not; the higher a patient's risk score, the greater the possibility of death (Fig. 1). It was determined that all of the risk score genes are oncosuppressors (their high expression was related to greater survival than their low expression). (P < 0.05) (Supplementary Fig. 3).
Ageing-related genes
An effective strategy for predicting prognosis was discovered after a multivariate cox hazard model analysis for clinical factors associated with a risk score. In addition to the risk score as the independent variable (HR = 1.58, P < 0.001, CI = 1.45–1.73), the age variable was found to be statistically significant (HR = 1.79, P < 0.001, CI = 1.32- 2.45) (Fig. 2).
According to the independent test, the expression association with relevant clinical factors showed TNFRSF4 gene expression was significantly linked with aging in patients with MM (P = 0.01) (Fig. 3). The strong significance between younger age group and the eldest one by one-way ANOVA on ranks results was clearly seen (15-year interval) ( P < 0.001) (Fig. 4). In addition to the correlation analysis in the training group, validation of gene correlation from GEO (GSE59455) and (GSE22153) were obtained as external validation. ((P < 0.0001) and r = -0.20, P = 0.01, r = -0.24) respectively (Fig. 5).
We investigate the prediction of TNFRSF4 gene and age by novel nomogram that highly detects the risk of death and associated age with higher sensitivity testing in one, three, and fifth-year AUC (79%,76%,76%), respectively (Fig. 6). By examining the correlation between risk score genes resulting from multivariate cox hazard model and age, we discover four genes that were likewise shown to be negatively correlated with age (CSF1, TLR8, TLR4, NOX4) (P < 0.05) (Fig. 7).
GO, KEGG, and construction of the PPI network
GO analysis revealed that genes associated with CKTTD were prominent in the biological processes of aging and cell aging with immune system. KEGG analysis revealed that pathways of cytokine-cytokine receptor interaction and cell adhesion molecules were associated (Supplementary Table 2). ICI genes were illustrated and described in the context of the PPI network for the CKTTD database (Supplementary Fig. 4).
Immune cells infiltration
Immune cells are distributed among risk classes were selected the most significant positive correlation of the risk genes concerning the abundance of immune cells that measured by Timer (http://timer.cistrome.org) for infiltrative immune cells between high and low-risk groups. In addition, it allows visualizing the correlation of its expression with immune infiltration level in MM. The scatterplots generated and displayed Spearman’s rho value and statistical significance of the risk score gene signature by using R package (Supplementary Fig. 5).
Experimental validation: The protein expression levels of TNFRSF4 using IHC
We tested the TNFRSF4 protein levels by IHC and examined the protein expression in 14 melanoma samples where the normal control slides showed a negative detection of TNFRSF4 protein (Fig. 8, Supplementary Table 3). The TNFRSF4 expression was observed to be lower expressed in the older of melanoma tissues, and higher in the younger age group (P = 0.02) (Fig. 9).