A, Kaplan-Meier analysis of overall survival stratified by HPV status. For comparison of HPV-16 with HPV-Other, log-rank P = .003. B, Genomic features (TP53 and CDKN2A, mutation, deep deletion; PIK3CA, EGFR, MYC, and FGFR1, mutation, amplification; FADD, gain, amplification; FHIT, shallow loss, deep loss) stratified by HPV genotype; visualization performed using cBioPortal (http://www.cbioportal.org). C, Tukey box plots of HPV gene expression among HPV-associated tumors. HNSC indicates head and neck squamous cell carcinoma; HPV, human papillomavirus; TCGA, The Cancer Genome Atlas.3
Customize your JAMA Network experience by selecting one or more topics from the list below.
Bratman SV, Bruce JP, O’Sullivan B, et al. Human Papillomavirus Genotype Association With Survival in Head and Neck Squamous Cell Carcinoma. JAMA Oncol. 2016;2(6):823–826. doi:10.1001/jamaoncol.2015.6587
The presence of human papillomavirus (HPV) has recently been recognized as a strong positive prognostic factor for head and neck squamous cell carcinoma (HNSC) and is used to select patients for risk-adapted treatment regimens. A positive immunohistochemical (IHC) finding for p16 is used as a surrogate for HPV based on the high concordance between these 2 biomarkers1,2; however, p16 IHC analysis cannot distinguish between HPV-16 (the most common HPV genotype found in HNSC) and other HPV genotypes. We investigated the prognostic consequences of distinct HPV genotypes within HNSC.
Patient data available through The Cancer Genome Atlas (TCGA)3 was collected after obtaining informed consent in accordance with institutional review board–approved studies at each participating TCGA partner institution. We downloaded RNA-seq bam files from the University of California, Santa Cruz, Cancer Genomics Hub. VirusFinder (version 2.0)4 was used to identify 179 distinct HPV genotypes. Viral gene expression was normalized to total HPV-mapped reads and gene size (reads per kilobase of transcript per million [RPKM]). Clinical data were downloaded from Broad Institute’s Firehose database (http://gdac.broadinstitute.org; std_data run 2015_06_01).
The Kaplan-Meier method was used to estimate survival; survival between cohorts was compared using the log-rank test. Both χ2 and t tests were used to compare clinical and genomic variables. Cox proportional hazards were used for multivariate analysis. The Wilcoxon rank-sum test was used to compare viral gene expression. All tests were 2-tailed, P < .05 was considered statistically significant.
The presence of HPV and its genotype were determined for 515 HNSCs tumors from TCGA. Seventy-three tumors harbored HPV transcripts, among which 61 (84%) were HPV-16, and 12 (16%) were HPV-Other (8 HPV-33, 3 HPV-35, and 1 HPV-56). There were 32 HPV-positive cases assessed using p16 IHC analysis, including 4 associated with HPV-Other genotypes; all were positive IHC staining. Among the 73 HPV-positive cases, 3- year overall survival was significantly worse for the HPV-Other cohort (49%) than for the HPV-16 cohort (88%) (log-rank P = .003; Figure, A).
Clinical covariates were similarly distributed between HPV-16 and HPV-Other patients (Table) and did not significantly alter the association of HPV genotype with survival (Table). Furthermore, genomic aberrations commonly seen in HNSC were balanced (Figure, B). Expression levels of several viral genes, however, differed between the 2 HPV-positive cohorts (Figure, C).
Using TCGA data,3 we have shown for the first time to our knowledge that determination of HPV genotype in HNSC provides more precise risk stratification than p16 IHC findings or HPV-16 detection alone. Genotyping of HPV is not widely implemented for HNSC because (1) prior studies have indicated that only about 8% of oropharyngeal cancers in North America harbored genotypes other than HPV-161,2; and (2) distinct HPV genotypes have not previously been shown to affect clinical outcome. In this study, however, we observed that within the TCGA HNSC cohort, 16% of HPV-associated cancers (13% for oropharynx) harbored genotypes other than HPV-16 and were associated with inferior survival rates similar to those of HPV-negative HNSC.
Our findings are in agreement with prior work suggesting superior survival for cervical cancer and HNSC associated with HPV-16 vs other HPV genotypes. Most studies have focused on the poor prognosis of HPV-18 tumors5,6; however, HPV-18 is infrequently observed in HNSC1 and was not detected in this TCGA HNSC analysis. Clinical and genomic features as well as p16 IHC results were similarly distributed between the HPV-16 and HPV-Other cohorts, suggesting that the difference in survival may be attributed to features inherent to viral oncogenesis such as variable activity or expression of viral genes. Other clinical and demographic factors not uniformly available in TCGA data (eg, socioeconomic status) may also affect survival and cannot be entirely excluded as potential sources of bias. Validation studies are needed to confirm HPV genotype as an independent prognostic feature.
In conclusion, HNSCs associated with HPV genotypes other than HPV-16 have inferior survival. Patients with these genotypes might therefore be inappropriate candidates for treatment deintensification. Prospective validation in larger data sets, including specifically in patients with oropharyngeal cancer, is required before HPV genotype can be used to inform treatment decisions.
Corresponding Author: Scott V. Bratman, MD, PhD, Princess Margaret Cancer Centre, 101 College St, TMDT/MaRS, 14-313, Toronto, ON M5G 1L7, Canada (email@example.com).
Published Online: March 24, 2016. doi:10.1001/jamaoncol.2015.6587.
Author Contributions: Drs Bratman and Bruce contributed equally to this work. Dr Bratman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bratman, Bruce, O’Sullivan, Xu.
Acquisition, analysis, or interpretation of data: Bratman, Bruce, Pugh, Xu, Yip, Liu.
Drafting of the manuscript: Bratman, Bruce, O’Sullivan, Xu, Liu.
Critical revision of the manuscript for important intellectual content: Bratman, Bruce, Pugh, Xu, Yip, Liu.
Statistical analysis: Bruce, O’Sullivan, Pugh, Xu.
Obtained funding: Bruce, Yip.
Administrative, technical, or material support: Bruce, O’Sullivan, Pugh, Yip, Liu.
Study supervision: Bratman, O’Sullivan, Pugh, Yip, Liu.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by funds from the Dr Mariano Elia Chair in Head and Neck Cancer Research; the Campbell Family Institute for Cancer Research; the Princess Margaret Cancer Foundation; and the Ministry of Health and Long-Term Care. Drs Bratman and Pugh were supported by the Gattuso-Slaight Personalized Cancer Medicine Fund at Princess Margaret Cancer Centre. Dr Bruce was supported by a Knudson postdoctoral fellowship through the Princess Margaret Cancer Foundation. We also gratefully acknowledge the support from the Princess Margaret Cancer Center Head & Neck Translational Program, with philanthropic funds from the Wharton Family, Joe’s Team, and Gordon Tozer.
Role of the Funder/Sponsor: These sources of funding and support had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We appreciate additional interpretation of the data and critical reading of the manuscript provided by John N. Waldron, MD, of the Princess Margaret Cancer Centre and Department of Radiation Oncology, University of Toronto; no financial compensation was provided for this valuable contribution. The results shown here are based on data generated by the TCGA Research Network (http://cancergenome.nih.gov/).