Gene Selection
Conducting Cox analysis on genes within the eight cohorts, performing meta-analysis, and filtering out 15 candidate genes with meta of p < 0.001, including 14 with HR < 1 and 1 with HR > 1 (Fig. 1).
IRGS Model Construction and Validation
Calculation of the Immune-related Gene Set score (IRGS) was performed by subtracting immpoor from immsign. Using the median value in the training set as the cutoff, the 945 samples were divided into two groups: high IRGS and low IRGS. Patients in high IRGS group demonstrated significantly better overall survival (OS) compared to low IRGS group (HR = 0.56, 95% CI 0.46–0.68, P < 0.001) (Fig. 2A). Similarly, in the validation set, the 405 samples were stratified into two groups based on the same cutoff. Patients in high IRGS group exhibited significantly improved OS compared to low IRGS group (HR = 0.45, 95% CI 0.33–0.6, P < 0.001) (Fig. 2B).
To further validate the prognostic performance of the IRGS model, validation was conducted in five external cohorts. The results from the five cohorts consistently showed that patients in high IRGS group had significantly better OS than low IRGS group (GSE31210: HR = 0.31, 95% CI 0.15–0.64, P < 0.001; GSE68465: HR = 0.76, 95% CI 0.59–0.99, P = 0.039; Chen_2019: HR = 0.52, 95% CI 0.28–0.95, P = 0.031; TCGA_LUAD: HR = 0.62, 95% CI 0.46–0.68, P = 0.002; CPTAC_LUAD: HR = 0.41, 95% CI 0.18–0.97, P = 0.036) (Fig. 2C-G).
IRGS independence analysis
Single-factor and multi-factor analyses were conducted on the risk associated with Immune-related Gene Set (IRGS), IRGS scores, and clinical features. The results revealed significant correlations between age, gender, stage, IRGS risk, and IRGS scores with prognosis in the training set, internal validation set, and multiple external validation sets. In some cohorts, single-factor analysis showed a significant association between EGFR mutation and prognosis, but multi-factor analysis did not demonstrate significant correlations (supplementary Table 1–7).
Functional enrichment analysis
To explore the mechanistic correlation between IRGS and prognosis, differential expression analysis was conducted by grouping samples based on high and low IRGS. Differential genes (High vs Low) were identified, and Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotations were performed separately for upregulated and downregulated genes. The results revealed the upregulated expressed genes in the High group are mainly associated with B-cell and T-cell differentiation signal pathways. And the upregulated expressed genes in the Low group are mainly associated with the DNA damage response (DDR) pathway. (Supplementary Fig. 1, supplementary Table 8–11).
Subtype analysis
In the analysis of clinical features, IRGS showed significant correlations with age (P = 0.021), gender (P = 0.0016), and stage (P < 0.001). Among genomic instability-related biomarkers, patients with low levels of TMB, HRD, NTAI, and LOH had significantly higher IRGS than those with high levels (P < 0.001), whereas patients with low TLS showed significantly lower IRGS than those with high TLS (P < 0.001). And subgroup analysis of immune checkpoint blockade (ICB) gene markers indicated significantly higher levels of PDCD1 (PD-1), CTLA4, CD274 (PD-L1), LAG3, TIGIT, and HAVCR2 in the IRGS high group (all P < 0.001) (Fig. 3).
TME analysis
To explore the correlation between IRGS and the tumor immune microenvironment, the TIMER algorithm was utilized to calculate the correlation between IRGS and immune cells. B cells, T cells, granulocytes, and dendritic cells (DCs) all exhibited positive correlations with IRGS (Fig. 4A). Further analysis of the differences in immune cell levels between the high and low IRGS groups revealed that patients in the IRGS low group had significantly lower levels of B cells, T cells, granulocytes, and DCs compared to patients in the high IRGS group (P < 0.001) (Fig. 4B).
ICI cohort analysis
Incorporating a cohort of 21 patients from a public LUAD population who underwent immunotherapy, their IRGS was calculated, and a prognostic analysis was conducted. The results indicated that patients in the IRGS high group exhibited significantly improved progression-free survival (PFS) following immunotherapy compared to the low group (P = 0.013) (Fig. 5A). Although the objective response rate (ORR) showed no significant difference, the proportion of patients achieving complete response (CR) or partial response (PR) was higher in the IRGS high group (P = 0.63) (Fig. 5B).