Comprehensive characterization of the tumor microenvironment for assessing immunotherapy outcome in patients with head and neck squamous cell carcinoma

Abstract

The tumor microenvironment (TME) constitutes a complex milieu of cells and cytokines that maintain equilibrium between tumor progression and prognosis. However, comprehensive analysis of the TME and its clinical significance in head and neck squamous cell carcinoma (HNSCC) remains to be unreported. In this study, based on large-scale RNA sequencing data pertaining to single nucleotide variants (SNVs) and copy number variations (CNVs) in HNSCC patients from The Cancer Genome Atlas database, we analysed subpopulations of infiltrating immune cells and evaluated the role of TME infiltration pattern (TME score) in assessing immunotherapy outcome. TME signature genes involved in several inflammation and immunity signalling pathways were observed in the TME score subtype, which were considered immunosuppressive and potentially responsible for significantly worse prognosis. In comparison with SNV- and CNV-mediated tumor mutation burden, TME score can significantly differentiate between high- and low-risk HNSCC and predict immunotherapy outcome. Our data provide clarity on the comprehensive landscape of interactions between clinical characteristics of HNSCC and tumor-infiltrating immune cells. TME score seems to be a useful biomarker that can predict immunotherapy outcome in HNSCC patients.

Keywords: copy number variations; head and neck squamous cell carcinoma; immunotherapy; single nucleotide variants; tumor microenvironment.

Figure 1

Overview of TME characteristics. (A) Relative percentage of each immune cell type in 502 patients with HNSCC from TCGA database. (B) Tumor–immune cell interactions. The size of each cell represents the impact of each TME cell type on survival and was calculated using log10 (log-rank test P value). Risk factors for overall survival are indicated in pink, and favourable factors are in green. The lines connecting TME cell types represent cellular interactions. The thickness of the lines represents the strength of correlation, which was estimated using Spearman correlation analysis. Negative correlation is indicated in grey and positive correlation in black. (C) The elbow criterion determines the optimal number of TME clusters (K = 3). (D) Consensus clustering analysis identification of the three TME clusters (samples, n = 500). The white (consensus value = 0, samples never clustered together) and blue (consensus value = 1, samples always clustered together) heatmap display sample consensus. (E) Kaplan–Meier curves for survival probability of the three clusters. Log-rank test was used for data analysis.

Figure 2

Signature and functional annotation of the TME clusters. (A) Relative percentage of each immune cell type in the three TME clusters. (B) Unsupervised hierarchical clustering of the clusters. (C) Relative populations of TME cells present in the three clusters. Within each group, the scattered dots represent expression values of TME cells. We also plotted the Immunoscore for the three clusters. The thick line represents the median value. The lower and upper ends of the boxes are the 25^th and 75^th percentiles. The whiskers encompass 1.5 times the interquartile range. Statistical differences in the three clusters were compared using the Kruskal–Wallis test. The range of P values are labelled above each boxplot with asterisks (* P < 0.05, ** P < 0.01, *** P < 0.001, ****P < 0.0001, ns = not significant). (D–F) Gene ontology enrichment analysis of TME signature genes in the cellular component (D), biological process (E) and molecular function (F) categories. The x-axis indicates the number of genes within each gene ontology term.

Figure 3

Clinical characteristics of the TME phenotypes. (A) TME score distribution in patients with HNSCC. (B) Overall survival status of patients with HNSCC. (C) Alluvial diagram of TME gene clusters in groups with different TME clusters, DEG clusters, TME scores and survival outcomes. (D) Gene set enrichment analysis of hallmark gene sets between the high and low TME score groups. (E) Kaplan–Meier curves for the high and low TME score groups. As evident, the high TME score group was associated with better outcomes than the low TME score group (log-rank test, P < 0.001).

Figure 4

Somatic mutations in HNSCC. (A, B) Distribution of highly variant mutated genes correlated with TME score groups. The upper bar plot indicates overall survival (OS), TMB and TME score for each patient, whereas the left bar plot shows the mutation frequency of each gene in separate TME score groups [high (A) and low (B) TME score groups]. TME score, grade, overall survival status, gender, age, smoking, alcohol frequency, daily alcohol, HPV P16 status and HPV ISH status are shown as patient annotations. (C) Mutation percentage of common mutated genes in the TME score groups. (D) Genome variant classification. (E) Genome variant type. (F) Single nucleotide variant class.

Figure 5

Mutational signature of the TME score groups. (A, B) Distribution of mutation type frequency in the high (A) and low (B) TME score groups. (C–H) Mutational signatures identified in the high (C–F) and low (G–H) TME score groups, respectively. The y-axis indicates exposure of 96 trinucleotide motifs to overall signature. The plot title indicates best match against validated COSMIC signatures and cosine similarity value along with the proposed aetiology.

Figure 6

CNV analysis in HNSCC. (A–D) CNV at arm level. The bar graphs show the frequency of arm-level CNV amplification (A, B) and deletion (C, D), the vertical axis denotes chromosome arms. (E–H) CNV at focal regions detected by GISTIC v2·0. Regions of recurrent focal amplifications (E, F) and focal deletions (G, H) in the high and low TME score groups are plotted by false discovery rate (x-axis) for each chromosome (y-axis). Dashed lines represent the centromere of each chromosome.

Figure 7

Clinical and integrated genomic landscape of HNSCC with the TME score phenotype. (A) Comprehensive genomic landscape of HNSCC. (B) Prediction immunotherapy effect of TME score by ROC analysis.