Purpose

Tumor genomic features have been of particular interest because of their potential impact on the tumor immune microenvironment and response to immunotherapy. Due to the substantial heterogeneity, an integrative approach incorporating diverse molecular features is needed to characterize immunologic features underlying primary resistance to immunotherapy and for the establishment of novel predictive biomarkers.

Experimental design

We developed a pan-cancer deep machine learning model integrating tumor mutation burden, microsatellite instability, and somatic copy-number alterations to classify tumors of different types into different genomic clusters, and assessed the immune microenvironment in each genomic cluster and the association of each genomic cluster with response to immunotherapy.

Results

Our model grouped 8,646 tumors of 29 cancer types from The Cancer Genome Atlas into four genomic clusters. Analysis of RNA-sequencing data revealed distinct immune microenvironment in tumors of each genomic class. Furthermore, applying this model to tumors from two melanoma immunotherapy clinical cohorts demonstrated that patients with melanoma of different genomic classes achieved different benefit from immunotherapy. Interestingly, tumors in cluster 4 demonstrated a cold immune microenvironment and lack of benefit from immunotherapy despite high microsatellite instability burden.

Conclusions

Our study provides a proof for principle that deep learning modeling may have the potential to discover intrinsic statistical cross-modality correlations of multifactorial input data to dissect the molecular mechanisms underlying primary resistance to immunotherapy, which likely involves multiple factors from both the tumor and host at different molecular levels.

Treatment with immune checkpoint blockade (ICB) has revolutionized cancer therapy. Until now, predictive biomarkers^1-10 and strategies to augment clinical response have largely focused on the T cell compartment. However, other immune subsets may also contribute to anti-tumour immunity^11-15, although these have been less well-studied in ICB treatment¹⁶. A previously conducted neoadjuvant ICB trial in patients with melanoma showed via targeted expression profiling¹⁷ that B cell signatures were enriched in the tumours of patients who respond to treatment versus non-responding patients. To build on this, here we performed bulk RNA sequencing and found that B cell markers were the most differentially expressed genes in the tumours of responders versus non-responders. Our findings were corroborated using a computational method (MCP-counter¹⁸) to estimate the immune and stromal composition in this and two other ICB-treated cohorts (patients with melanoma and renal cell carcinoma). Histological evaluation highlighted the localization of B cells within tertiary lymphoid structures. We assessed the potential functional contributions of B cells via bulk and single-cell RNA sequencing, which demonstrate clonal expansion and unique functional states of B cells in responders. Mass cytometry showed that switched memory B cells were enriched in the tumours of responders. Together, these data provide insights into the potential role of B cells and tertiary lymphoid structures in the response to ICB treatment, with implications for the development of biomarkers and therapeutic targets.