ARTICLE ABSTRACT
Purpose: Genetic and morphologic heterogeneity is well-documented in solid cancers. Immune cells are also variably distributed within the tumor; this heterogeneity is difficult to assess in small biopsies, and may confound our understanding of the determinants of successful immunotherapy. We examined the transcriptomic variability of the immunologic signature in head and neck squamous cell carcinoma (HNSCC) within individual tumors using transcriptomic and IHC assessments.Experimental Design: Forty-four tumor biopsies from 16 HNSCC patients, taken at diagnosis and later at resection, were analyzed using RNA-sequencing. Variance filtering was used to identify the top 4,000 most variable genes. Principal component analysis, hierarchical clustering, and correlation analysis were performed. Gene expression of CD8A was correlated to IHC analysis.Results: Analysis of immunologic gene expression was highly consistent in replicates from the same cancer. Across the cohort, samples from the same patient were most similar to each other, both spatially (at diagnosis) and, notably, over time (diagnostic biopsy compared with resection); comparison of global gene expression by hierarchical clustering (P ≤ 0.0001) and correlation analysis [median intrapatient r = 0.82; median interpatient r = 0.63]. CD8A gene transcript counts were highly correlated with CD8 T-cell counts by IHC (r = 0.82).Conclusions: Our data demonstrate that in HNSCC the global tumor and adaptive immune signatures are stable between discrete parts of the same tumor and also at different timepoints. This suggests that immunologic heterogeneity may not be a key reason for failure of immunotherapy and underpins the use of transcriptomics for immunologic evaluation of novel agents in HNSCC patients. Clin Cancer Res; 23(24); 7641–9. ©2017 AACR.
