Tumor microenvironment-based signatures distinguish intratumoral heterogeneity, prognosis, and immunogenomic features of clear cell renal cell carcinoma

Background

The tumor microenvironment (TME) performs a crucial function in the tumorigenesis and response to immunotherapies of clear cell renal cell carcinoma (ccRCC). However, a lack of recognized pre-clinical TME-based risk models poses a great challenge to investigating the risk factors correlated with prognosis and treatment responses for patients with ccRCC.

Methods

Stromal and immune contexture were assessed to calculate the TMErisk score of a large sample of patients with ccRCC from public and real-world cohorts using machine-learning algorithms. Next, analyses for prognostic efficacy, correlations with clinicopathological features, functional enrichment, immune cell distributions, DNA variations, immune response, and heterogeneity were performed and validated.

Results

Clinical hub genes, including INAFM2, SRPX, DPYSL3, VSIG4, APLNR, FHL5, A2M, SLFN11, ADAMTS4, IFITM1, NOD2, CCR4, HLA-DQB2, and PLAUR, were identified and incorporated to develop the TMErisk signature. Patients in the TME^high risk group (category) exhibited a considerably grim prognosis, and the TMErisk model was shown to independently function as a risk indicator for the overall survival (OS) of ccRCC patients. Expression levels of immune checkpoint genes were substantially increased in TME^high risk group, while those of the human leukocyte antigen (HLA) family genes were prominently decreased. In addition, tumors in the TME^high group showed significantly high infiltration levels of tumor-infiltrated lymphocytes, including M2 macrophages, CD8⁺ T cells, B cells, and CD4⁺ T cells. In heterogeneity analysis, more frequent somatic mutations, including pro-tumorigenic BAP1 and PBRM1, were observed in the TME^high group. Importantly, 19.3% of patients receiving immunotherapies in the TME^high group achieved complete or partial response compared with those with immune tolerance in the TME^low group, suggesting that TMErisk prominently differentiates prognosis and responses to immunotherapy for patients with ccRCC.

Conclusions

We first established the TMErisk score of ccRCC using machine-learning algorithms based on a large-scale population. The TMErisk score can be utilized as an innovative independent prognosis predictive marker with high sensitivity and accuracy. Our discovery also predicted the efficacy of immunotherapy in ccRCC patients, indicating the intimate link between tumor immune microenvironment and intratumoral heterogeneity.