Prevalence and biological impact of clinically relevant gene fusions in head and neck cancers

Abstract

Head and neck cancer (HNC) is the seventh most common cancer worldwide. Currently-approved systemic therapies include chemotherapy, anti-EGFR antibodies, and PD-1 immunotherapy, with few genomic-based targeted therapies. Gene fusions involving cancer-driving kinase genes such as FGFR, NTRK, and ALK are clinically targetable in other solid tumors; however, there is limited knowledge about their prevalence in HNC. Here, we describe the genomic landscape and the biological impact of oncogenic fusions in a combined dataset of over 13,000 HNC tumors (excluding salivary gland tumors). We identified 66 cases (2.8%) harboring oncogenic fusions, including previously-reported FGFR3 fusions (n = 19) and gain-of-function EGFR fusions (n = 6). Fusion-positive HNC had significantly higher gene expression and higher prevalence of human papillomavirus than fusion-negative HNC (p < 0.001). Tumors with FGFR alterations were associated with enriched cell proliferation and higher abundance of NK cells and CD8+ T cells compared to wildtype. Our results provide expanded therapeutic opportunities for patients with HNCs.

Fig. 1. Large data curation revealed clinically relevant gene fusions in head and neck cancers.

A Flow chart of data curation process and sources. From these sources, we curated a total of 13,655 head and neck cancer cases. ORIEN oncology research information exchange network, TCGA the cancer genome atlas, CPTAC clinical proteomic tumor analysis consortium B Pie chart of histology types in the fusion-positive cohort.

Fig. 2. Distribution of gene fusions in diverse anatomical sites in head and neck cancers.

A Count plot of the number of samples harboring a gene fusion involving a clinically relevant gene (y-axis) in a given anatomical site (x-axis). B Distribution of clinically relevant gene fusions present in each anatomical site. Left (orange) denotes the percentage of gene fusions present in the given anatomical site and right (blue) denotes the total number of cases harboring cancer in the given anatomical site.

Fig. 3. Schematic of EGFR fusions observed in head and neck cancer.

Gene fusion constructs of EGFR. A Schematic of 5′ EGFR fusions; breakpoints occurred in intron 24 (top 3) and exon 28 (bottom). B Schematic of 3’ EGFR fusions; breakpoints occurred in intron 17 (top) and exon 9 (bottom).

Fig. 4. Gene fusion-positive cases show upregulated expression, no mutational differences, and higher incidences of HPV positivity.

Violin plots of the z scores for each given gene (denoted in the y-axis) for samples in (A) ORIEN (B) TCGA and (C) curated studies (dbGaP and Array Express). Violin plots of tumor mutation burden (TMB) comparing fusion-positive versus fusion wildtype cases in (D) ORIEN and (E) TCGA. E, F stacked bar charts of the proportion of samples in each group positive for human papillomavirus (HPV) in (F) ORIEN and (G) TCGA.

Fig. 5. Schematic of gain-of-function FGFR alterations in head and neck cancer.

A Unique gene fusion constructs involving FGFR2. All 5′ FGFR breakpoints occurred at intron 17 and the 3’ breakpoint occurred at intron 4. B Unique gene fusion construct of FGFR3 with canonical breakpoints. Two unique 5′ FGFR3 breakpoints occurred in intron 14 (top), and the end of exon 15 (middle). The 3’ FGFR3 breakpoint occurred at exon 9. C Lollipop plot of gain of function mutations observed in FGFR3 in ORIEN, TCGA, and CPTAC.

Fig. 6. FGFR alterations have a distinct immune phenotype and increased proliferation.

Violin plots of immune cell abundance estimates as measured by CIBERSORT assessing (A) CD8 T cells, (B) Activated CD4 T cells, and (C) Resting CD4 T cells. D Violin plot of NK cell abundance measured by Quantiseqr. E Violin plot of cell proliferation signature measured by ImSig.