Subtypes of HPV-Positive Head and Neck Cancers Are Associated with HPV Characteristics, Copy Number Alterations, PIK3CA Mutation, and Pathway Signatures

Abstract

Purpose: There is substantial heterogeneity within human papillomavirus (HPV)-associated head and neck cancer (HNC) tumors that predispose them to different outcomes; however, the molecular heterogeneity in this subgroup is poorly characterized due to various historical reasons.

Experimental design: We performed unsupervised gene expression clustering on deeply annotated (transcriptome and genome) HPV(+) HNC samples from two cohorts (84 total primary tumors), including 18 HPV(-) HNC samples, to discover subtypes and characterize the differences between subgroups in terms of their HPV characteristics, pathway activity, whole-genome somatic copy number alterations, and mutation frequencies.

Results: We identified two distinct HPV(+) subtypes (namely HPV-KRT and HPV-IMU). HPV-KRT is characterized by elevated expression of genes in keratinocyte differentiation and oxidation-reduction process, whereas HPV-IMU has strong immune response and mesenchymal differentiation. The differences in expression are likely connected to the differences in HPV characteristics and genomic changes. HPV-KRT has more genic viral integration, lower E2/E4/E5 expression levels, and higher ratio of spliced to full-length HPV oncogene E6 than HPV-IMU; the subgroups also show differences in copy number alterations and mutations, in particular the loss of chr16q in HPV-IMU and gain of chr3q and PIK3CA mutation in HPV-KRT.

Conclusions: Our characterization of two subtypes of HPV(+) HNC tumors provides valuable molecular level information that point to two main carcinogenic paths. Together, these results shed light on stratifications of the HPV(+) HNCs and will help to guide personalized care for HPV(+) HNC patients. Clin Cancer Res; 22(18); 4735-45. ©2016 AACR.

Figure 1. Identification of two HPV(+) subgroups and pathway differences between them

(A) Hierarchical clustering (shown here) and consensus clustering (Fig S1A) revealed two distinct HPV(+) clusters in the UM cohort. (B) Multi-Dimensional Scaling plot displaying the relationship among combined UM (n=36) and TCGA (n=66) samples by subgroup. The HPV-KRT subgroup is more similar to the non-HPV samples than is the HPV-IMU subgroup. (C) Heatmap showing representative genes/pathways different between HPV-KRT and HPV-IMU. (D) Heatmap showing the top differentially expressed genes (and pathways) among the three clusters from UM cohort. The genes were grouped by their expression signatures across the three clusters. The genes used to generate this heatmap are listed in Table S5.

Figure 2. HPV(+) subgroups correlate with several HPV characteristics

(A) Barplot showing HPV-KRT tumors were more likely to have a detected genic-integration than HPV-IMU. (B) Boxplot of HPV E5 expression levels; HPV-KRT had lower E5 expression than HPV-IMU. (See Fig S2A,B for plots of E2 and E4.) (C) Boxplot of HPV E6 expression levels. (D) Boxplot of HPV E6 full-length percent for HPV type 16 samples. HPV-KRT had significantly lower E6 full-length percent than HPV-IMU. (E) Heatmap of the E6-regulated genes from Duffy et al. (38), showing genes repressed by E6 were lower expressed in HPV-IMU. E6 activity score is displayed at the bottom. (F) Boxplot of E6 activity score shows overall higher E6 activity in HPV-IMU. (See supplementary methods for how E6 score is calculated.)

Figure 3. HPV(+) subgroups differ by copy number alterations

(A,B) Heatmaps of somatic copy numbers for UM (A) and TCGA (B) samples. The dark dashed lines highlight the regions that differ by HPV(+) subgroup (chr3q gain and chr16q loss). The color legends show absolute copy number estimates for UM samples and log2 ratios of tumor to normal intensity for TCGA samples. (C,D) Plots showing the expression fold changes of genes on chr3q (C) and chr16q (D). There were more up-regulated genes (shown in red) than down-regulated genes (shown in blue) in HPV-KRT compared to HPV-IMU due to the CNA difference. Known differentially expressed cancer-related genes are noted for each. Significance was defined by q-value < 0.05.

Figure 4. The HPV-KRT subgroup has more PIK3CA mutations and amplifications

(A) Scatterplot showing the difference in mutation percent between the HPV(+) subgroups for each gene. PIK3CA was the only gene that had > 20% difference in both the UM and TCGA cohorts. The difference was calculated by subtracting the mutation percent in HPV-KRT from that of HPV-IMU. (B) Barplot showing the PIK3CA mutation percent for each subgroup and cohort (see supplementary methods for details). (C) Boxplot showing that in both cohorts, HPV-KRT had more PIK3CA copy number amplifications than HPV-IMU.

Figure 5. Summary of characteristics that differ by HPV(+) subgroup, and prognosis

(A) Heatmap of variables that correlated with HPV(+) subgroup. The columns represent samples, sorted by cluster, cohort, viral integration and PIK3CA mutation. E5 expression values are the log10 transformed CPM values. (B) Observed features and associated publications suggesting better prognosis for HPV-IMU for all but EMT. (C) Overall survival for TCGA tumors with high and low T-cell activation scores. (D) Overall survival for HPV-KRT and HPV-IMU tumors from TCGA cohort. (C,D) P-values were calculated using a univariate Kaplan-Meier log rank test.