In recent decades, studies focused on the immune landscape have obtained a lot of attention in cancer biological and clinical research. Immune markers such as CD3 + or CD8 + have suggested the predictive value of prognosis in several cancer types30. The immune landscape is an important factor in the immune response in cancer. Therefore, to explore genomic and clinical characteristics in different immune infiltration pattern of LUSC patients, we collected 728 LUSC patients from common datasets and our institution and divided them into 2 groups named “TME cluster” based on the abundant of immune cells by GSEA method. Then we systemically compared the genomic significance, clinical characteristic and immune infiltration pattern in 2 TME clusters and found TME cluster was an important predictive factor in prognosis.
Several studies have suggested that the immune infiltration status in the tumor samples was significantly correlated with the prognosis in breast cancer, renal cancer, and neck cancer31–33. In our results, we also found that TME cluster A, a subgroup with lower immune cell abundant, was associated with poor survival., Some studies have reported the abundance of immune cells in tumor tissue could affect the response to immune checkpoints inhibitors such as nivolumab, which could explain the non-ideal OS in some advanced NSCLC patients with immune therapy. However, it also has been reported the recruitment of Tregs could promote the immune escape of tumors in NSCLC34, which might be conflictive to our founding. Therefore, we speculated the co-infiltration of Treg cells and the effect of T cells in cluster B could maintain the immune balance35,36. The anti-tumor effect of the immune microenvironment was decided by the ratio of effect cells and suppressor cells. We have calculated the ratio of CD8 + T cell/Treg cell and aDC/iDC and found the TME cluster B had higher activated immune cells.
The co-expression immune-related genes of defining affect T cells and IFN-γ expression including CD8A, GZMA, GZMB, IFNγ, EOMES, CXCL9, CXCL10, and TBX21 were both up-regulated in TME cluster B, which was consistent with the T cells infiltration pattern. It has been reported the high-expression of immune-related genes could increase the response rate to immunotherapy in NSCLC cancer and improve the overall survival, which might suggest TME cluster B was an immunotherapy-sensitive subgroup37,38. At the same time, the cytolytic activity score which was associated with OS as some studies reported based on the PRF1 and GZMA which reflected the anti-tumor effect also was higher in cluster B39,40. Sumana Narayanan et al41 suggested a higher CYT score was associated with higher expression of immune checkpoint molecules and higher mutation load in colon cancer. It was further verified in LUSC that cluster B had a higher mutation load in our study. Tumor mutation burden was an important predictive factor in immune therapy. Several significant mutation DNA repair genes including TP53, ATR, and LIG1 which were considered to be associated with the effect of PD-1 inhibitors42 were both up-regulated in the TME B cluster. Key targeted gene aberrations in LUSC such as PI3K, ROS and EGFR mutation were more common in TME cluster B, which means LUSC patients with those mutations may also be suitable for immunotherapy. Besides, it has been reported higher CNV could increase the expression of the cancer-related genes, which was consistent with our study13.
The low immune infiltration pattern was considered being associated with distinct tumor immune escape mechanisms43 including reduced expression of major histocompatibility complex (MHC) class I, adhesion and costimulatory molecules, loss of antigens, and increased expression of immunosuppressive components such as HLA-G, HLA-E, and PD-L1 and of other immunosuppressive factors such as cytokine and metabolites that contribute to the escape from immune recognition44. The expression of adhesion (CADM1) 45 and costimulatory molecules (CD80, CD86) were increased in cluster B while the immune checkpoint molecules such as PD-L1 and CTLA-4 were up-regulated in cluster B, which was coincident with GO analysis. TGF-β, another factor that has been identified as a key driver of tumor plasticity46,47, was up-regulated in cluster B, which might be an important therapy site.
Rachel Rosenthal et al suggested the diverse TME impact upon neoantigen presentation, as well as the tumor-specific mechanisms leading to immune escape48. The disruption of tumor antigen presentation was a significant pathway to promote immune escape. Antigen presentation cells (APC) such as dendritic cells and macrophages were up-regulated in TME cluster B and the ratio of activated DC/inhibited DC was also higher in cluster B, so, we speculated the cluster with lower infiltration pattern could lead to immune escape via the down-regulation of active APC. Some specific genes in the APM process including PSMB5, PSMB6, PSMB7, PSMB8, PSMB9, PSMB10, TAP1, TAP2, ERAP2, CANX, CALR, PDIA3, TAPBP, B2M, HLA-A, HLA-B, and HLA-C could predict the efficiency of this antigen-processing and presenting steps based on previous studies (46). Therefore, identification of such genes might lead to further insight into the complex interaction between tumor cells and the immune system and thus facilitate the development of personalized immune therapeutic regimens and enhance the response rate of immune checkpoint inhibitors in LUSC patients.
There were several limitations to our study. First, accessible gene-expression data of patients who received immunotherapy is insufficient currently, so we can’t validate our findings in patients who received immunotherapy. Second, most of the tumor samples from TCGA and GEO were single-loci. Given the spatial heterogeneity of intratumor immunoreactivity, the exploration of the multi-foci sample is still warranted. Besides, the study was a retrospective descriptive study. In summary, by machine learning methods and multi-omics profiling, our large-cohort study still described a comprehensive landscape of LUSC immune infiltration patterns and integrated biomarkers associated with distinct immunophenotypes, thus explored the interaction between tumors and immune microenvironment, which may guide a more precise and personalized immune therapeutic strategy for LUSC patients.