Tumor Microenvironment Heterogeneity-Based Score System Predicts Clinical Prognosis and Response to Immune Checkpoint Blockade in Multiple Colorectal Cancer Cohorts

Abstract

Despite immune checkpoint blockade (ICB) therapy contributed to significant advances in cancer therapy, only a small percentage of patients with colorectal cancer (CRC) respond to it. Identification of these patients will facilitate ICB application in CRC. In this study, we integrated multiple CRC cohorts (2,078 samples) to construct tumor microenvironment (TME) subtypes using TME indices calculated by CIBERSORT and ESTIMATE algorithms. Furthermore, a surrogate quantitative indicator, a tumor microenvironment immune gene (TMEIG) score system, was established using the key immune genes between TME clusters 1 and 2. The subsequent analysis demonstrated that TME subtypes and the TMEIG score system correlated with clinical outcomes of patients in multiple CRC cohorts and exhibited distinct immune statuses. Furthermore, Tumor Immune Dysfunction and Exclusion (TIDE) analysis indicated that patients with low TMEIG scores were more likely to benefit from ICB therapy. A study on two ICB cohorts (GSE78220 and IMvigor210) also validated that patients with low TMEIG scores exhibited higher ICB response rates and better prognoses after ICB treatment. The biomarker evaluation module on the TIDE website revealed that the TMEIG score was a robust predictive biomarker. Moreover, differential expression analysis, immunohistochemistry, qPCR experiments, and gene set prioritization module on the TIDE website demonstrated that the five genes that constitute the TMEIG score system (SERPINE1, FABP4, SCG2, CALB2, and HOXC6) were closely associated with tumorigenesis, immune cells, and ICB response indices. Finally, TMEIG scores could accurately predict the prognosis and ICB response of patients with CRC. SERPINE1, FABP4, SCG2, CALB2, and HOXC6 might be potential targets related to ICB treatment. Furthermore, our study provided new insights into precision ICB therapy in CRC.