1154 - A Lymph Node Immune Score (LNIS) Identifies Prognostic Immune Phenotypes Shared across Head and Neck Cancers

T. Karpinets¹, A. S. Garden², D. I. Rosenthal², C. D. Fuller³, S. J. Frank², G. B. Gunn², J. Phan², W. H. Morrison², A. Lee², A. C. Moreno³, X. Song⁴, Y. Mitani⁴, R. Ferrarotto², N. D. Gross², J. Zhang¹, J. N. Myers⁵, A. K. El-Naggar², and M. T. Spiotto²; ¹Department of Genomic Medicine, The University of Texas MD Anderson Cancer Center, Houston, TX, ²The University of Texas MD Anderson Cancer Center, Houston, TX, ³Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, ⁴MD Anderson Cancer Center, Houston, TX, ⁵University of Texas MD Anderson Cancer Center, Houston, TX

Purpose/Objective(s): Lymph node metastases (LNMs) reflect the interplay between cancer cells and immune cells in the lymph node microenvironment. Given that the immune microenvironment likely reflects common host responses against cancer, we hypothesized that molecular classification of immune expression profiles in LNMs provides biological markers that are shared across different subtypes of head and neck squamous cell cancers (HNSCCs). Materials/

Methods: We performed exomic capture mRNAseq in 69 LNMs from 59 patients with Human Papillomavirus (HPV)-positive HNSCCs. Unsupervised clustering was performed using immunology-related genes from the InnateDB. 33 non-overlapping genes reflecting immune rejection and tolerance clusters were used to generate a Lymph Node Immune Score (LNIS). LNIS was calculated as the difference between clusters from average total gene expression in an individual lymph node. Gene set enrichment analysis was performed using QuSAGE. Immune cell abundance was calculated using MCPcounter. Associations of LNIS with relapse free survival, extranodal extension (ENE) and existing prognostic gene signatures were tested using univariate Cox proportional hazard models or the Mann-Whitney test, respectively. The LNIS was validated using two previously published datasets of HPV-negative HNSCC LNMs (EGAS00001003233) or non-small cell lung cancer (NSCLC) LNMs (GSE197929).

Results: Unsupervised clustering identified two immune gene clusters in HPV-positive HNSCCs representing Immune Rejection and Immune Tolerance subtypes. Confirming these phenotypes, the Immune Rejection subtype was enriched for several HLA genes while the Immune Tolerance subtype was enriched for checkpoint genes, including CTLA4 and PDCD1 (PD-1). A LNIS was generated that quantified the relative dominance between the Immune Rejection subtype, which had a positive LNIS, and the Immune Tolerance subtype, which had a negative LNIS. A positive LNIS was associated with increase in T cell and B cell populations (P <0.0001), the absence of ENE (P=0.03) and correlated with favorable prognostic gene signatures (P < 0.0001). By contrast, a negative LNIS was associated with neutrophil (P=0.04) and fibroblast (P=0.02) populations In HPV-negative HNSCC patients, the LNIS correlated with a similar immune cell infiltrates, prognostic gene signatures (P <0.0001), locoregional control (P=0.01) and trended to an association with ENE (P=0.09). Finally, in a NSCLC cohort, a positive LNIS was associated with significantly improved relapse free survival (P=0.01).

Conclusion: Differences in immune phenotypes in metastatic lymph nodes were associated with distinct biological features and outcomes in HNSCCs and NSCLCs. Identification of immune microenvironments in LNMs can elucidate cancers with a propensity for immune evasion that benefit from treatment intensification.