HPV Structure and Integration Impact the Genomic Profile of HPV-Positive Oropharyngeal Squamous Cell Carcinoma

Abstract

Although smoking and drinking rates have fallen, oropharyngeal cancer rates have been on the rise due to Human Papillomavirus (HPV). However, to date, few genomic studies have been conducted on primary tumors. The objective of this study is to examine the mutational, structural, and viral profiles for HPV-positive oropharyngeal squamous cell carcinomas (OPSCC). HPV-positive oropharyngeal squamous cell carcinoma samples were obtained from multiple institutions. To be included in the study, tumors must be HPV16 positive and have evidence of HPV to HPV or HPV to chromosomal integration. Tumors with more than 100 HPV16 per million reads were classified as HPV16 positive. DNA was sequenced using hybrid capture technology and Illumina brand paired-end sequencing on the HiSeq2000/2500 and NextSeq500. We focused on 24 genes most commonly mutated genes in head and neck cancer (AJUBA, CASP8, CCND1, CDKN2A, CUL3, FAT1, FBXW7, FGFR1, FGFR2, HLA-A, HRAS, KMT2C, KMT2D, KRAS, NFE2L2, NOTCH1, NOTCH2, NSD1, PIK3CA, PIK3R1, PTEN, RB1, TGFBR2, TP53, TRAF3). A total of 150 HPV-positive OPSCC tumors were included in the study. The most commonly mutated gene was PIK3CA (31%), followed by KMT2D (24%) and KMT2C (22%). There was no statistical difference in the mutational profile between HPV-positive smokers and non-smokers or by pack-year status. HPV genomic integration was observed in 53% (N = 79) of the tumors with the remaining 71 tumors having HPV to HPV integration and classified as episomal. We demonstrate a significant difference in HPV viral read count by integration status; tumors with HPV integration have lower counts of corrected HPV reads (median: 14,598 HPV16 per million reads (Q1-Q3: 2,260-32,825) compared with those with episomal HPV (median: 28,386 HPV16 per million reads (Q1-Q3: 18,723-44,208) (P < .001). Although not statistically significant, tumors with HPV integration were more likely (40.5%) to have PIK3CA mutations compared to episomal tumors (25.4%). PIK3CA mutations occurred most commonly in the helical domain (73.4%), followed by kinase domain (10.7%), ABD domain (10.7%), and C2 (5.4%). Tumors with HPV integration were significantly more likely to have helical domain mutations, with 68% of helical domain mutations occurring in integrated cases. There is no association between PIK3CA mutation location and smoking status and corrected HPV read count. HPV physical state and structure contribute significantly to the mutational profile of HPV-positive OPSCC, particularly with respect to PIK3CA mutational characteristics. These findings demonstrate the potential importance of examining HPV-host tumor interactions as prognosticators in HPV-positive OPSCC.