Abstract
Objectives
We previously reported identifying three categories of HPV16-positive head and neck tumors based on The Cancer Genome Atlas (TCGA) RNA and DNA sequence data. Category 1 had truly integrated HPV16 genomes, category 2 had simple episomal genomes, and category 3 had novel episomes that were a hybrid between viral and human DNA. Using our categorization, we investigated in this study survival of patients with integrated HPV16 tumors versus patients with episomal HPV16 tumors.
Materials and Methods
The TCGA RNA-Seq sequence reads were used to quantify HPV E2 and E7 gene expression, which was used as a marker for HPV integration.
Results
The results demonstrate that integration is associated with poor survival; those patients with integrated HPV tumors fared no better than non-HPV tumors in their five-year survival. Integrated HPV in tumors was found strikingly to be prevalent in patients born earlier while episomal HPV was prevalent in patients born later. We also observed a fairly constant incidence of all HPV forms among head and neck cancer patients over the last eight years of this study (2006-2013).
Conclusion
We propose our characterization of HPV integrated and episomal state is more accurate than previous studies that may have mischaracterized the hybrid HPV-human DNA episomes as integrated. The state of integrated HPV is associated with a poor clinical outcome. Results suggest that the incidence of integrated HPV among all HPV forms peaked and is decreasing. We discuss the importance of our findings for the management of HPV positive head and neck cancer.
Keywords: Head and neck cancer, human papillomavirus, survival, incidence, age, year of birth
Introduction
Human papillomavirus (HPV) and in particular HPV16 is a causative agent in a variety of cancers that include cervical cancer, ano-genital cancers, and head and neck cancers. The incidence of HPV-mediated head and neck cancer (HNC), has been on the rise in the US, in contrast to the reduction of non-HPV HNC that parallels reduction in tobacco use [1]. HPV16 is an 8 Kb DNA virus that is found in one of two physical states in cancer cells, episomal or integrated. The episomal HPV genome is a circular autonomously replicated DNA structure. The integrated state of HPV has the HPV genome stably maintained as part of the human genome, and has been shown to correlate with poor patient performance and advanced stage for cervical cancer in comparison to episomal HPV [2]. HPV HNC patients overall respond to therapy very well compared to non-HPV HNC patients, however, a subset of ~20% of HPV HNC patients have demonstrated poor response to therapy [3]. There are ongoing clinical trials carrying out de-escalation therapy for HPV positive individuals. The purpose of these trials is to investigate reduced treatments with corresponding reduced side effects while maintaining the optimum regimen for combating HPV+HNC [4]. However, identifying the individuals with HPV+HNC that will have a worse clinical outcome is important for these trials as it is predicted such individuals will not respond optimally to the de-escalated therapy. Thus, it would not be prudent to de-escalate treatment for patients within this subset. Identifying this subset by a correlation between integrated HPV and poor clinical outcome has not been consistently shown in HNC in which integrated HPV appears to be in the majority of HPV HNC patients [5]. This apparent lack of correlation may only be a result of the current models and interpretation of the integrated state of HPV in HNC. Our recent studies provide evidence supporting an alternative model and interpretation of HPV integration that may allow a more accurate view of the integrated state of HPV and how it relates to clinical outcome.
Integrated HPV has been detected in many studies by the presence of HPV-human DNA junctions [5–7]. Our genomic mapping of HPV with junctions to human DNA in HNC samples from The Cancer Genome Atlas (TCGA; [8, 9]) has generated evidence of two forms for HPV genomes with human DNA junctions [10]. One form is truly integrated HPV in which the HPV genome(s) are linear as part of the human genome, which is the accepted interpretation from other studies [5–7]. The other form is an HPV genome that may have been transiently integrated in the human genome at some point in the past, but has since been excised carrying with it a stretch of human DNA forming a novel HPV-human DNA circular structure that is maintained episomally. These excised HPV-human DNA episomes are capable of autonomous replication and in most cases, from the HPV origin. The HPV-human hybrid DNA episomes have been considered as being stably integrated HPV because they have HPV-human DNA junctions [5]. Other studies have shown that as much as 71% HNC samples have integrated HPV [5], while our analysis showed that only 23% of samples have integrated HPV, and samples with simple episomal HPV16 were at 44% and HPV-human DNA hybrid episomal HPV16 samples were at 33% [10].
Integration of HPV is thought to be stable if the integration disrupts E2, or E2 and E1 expression by dislocating the upstream HPV promoter from the downstream E1 and E2 genes [10]. This is a critical event because E1 and E2 are involved in replication from the HPV origin [11]. If replication from the HPV origin occurs for an integrated HPV genome, the resulting aberrant replication structures will likely kill the cell or be excised to form HPV-human DNA hybrid episomes [10]. We have described mechanisms for the formation of excised HPV joined to human DNA as a result of stalled replication forks within aberrantly re-replicated DNA that results in DNA breakage and non-homologous end joining to stabilize HPV-human DNA as circles. The loss of E2 expression in integrated HPV HNC has also been proposed to increase expression of the oncogenes E6 and E7, but data does not support this hypothesis [10]. We have proposed that loss of expression of the E2 gene, and downstream E4 and E5 genes could be used as markers for the stable integrated state of HPV [10].
We investigated differences in outcomes in the TCGA data between samples with integrated and episomal HPV based on our new assessment, using loss of E2 expression as a marker for integrated HPV. The results demonstrate that by our definition, patients with stably integrated HPV genomes have worse clinical outcomes than patients with episomal HPV genomes. From these results we conclude that there is value in our categorization, and propose simple clinical tests to identify patients with truly integrated tumors and therefore with worse prognosis; such tests, if validated, would assist in the stratification of HPV+HNC patients for de-escalation treatment.
Material and Methods
Integrated and episomal HPV determination
Stably integrated HPV based on genomic mapping was shown to have lost expression of the HPV E2 gene while episomal HPV was shown to express the E2 gene [10], therefore, we used E2 expression as a marker for stably integrated HPV. The HPV16 genome sequence coordinates were based on GenBank K02718.1. The level of E2 gene expression was based on the number of reads from RNA-Seq data from TCGA mapped to within coordinates specific to E2. These coordinates were toward the middle of the E2 gene at positions 2900 to 3200, not overlapping E4. The normalized E2 expression was calculated by dividing the E2 read count by the size of the probe region (300 bases), divided by the E7 read count divided by the size of the probe region (200 bases). The E7 reads were from those mapped within coordinates 600 to 800. Samples with no or low expression of E2 and categorized as having integrated HPV were determined as having a normalized E2 level below 0.02. All other samples were assessed as having episomal HPV. The mean normalized E2 expression for samples with assessed episomal HPV was 0.15. RNA-seq quality control, alignment, and read quantitation was done as previously described [10].
Analysis of survival
The clinical data from the TCGA Head and Neck Cancer samples was used for analysis of overall survival. Survival results were determined from vital status, the days to last contact, and the days to death. Analysis was performed using R. P-values were based on the log-rank test.
Analysis of age, year of birth, and year of diagnosis
The clinical data from the TCGA Head and Neck Cancer samples was used for analysis of age, year of birth, and year of diagnosis. Statistical analysis was based on one-way ANOVA.
Results
Survival analysis of integrated and episomal HPV
A total of 56 patients with HNC and their samples known to have HPV16 were analyzed from TCGA, along with 341 patients with non-HPV HNC. A minority of patients had HNC samples that contained other high-risk HPV types and were excluded from our analysis. The HPV16 HNC samples and patients were divided into two groups characterized by either 1) extrachromosomal HPV either as simple HPV episome or as an HPV-human hybrid episome, or 2) integrated HPV. We used loss of E2 expression as a marker for integrated HPV. Figure 1 shows Kaplan-Meier curves for analysis of overall survival, comparing patients with episomal HPV to patients with integrated HPV, with patients with non-HPV as a reference. Patients with episomal HPV (both simple HPV episomes and HPV-human DNA episomes) HNC showed the best survival. Patients with integrated HPV HNC showed a statistically significant reduced survival compared to patients with episomal HPV (p = 0.034). While statistical significance of the p-value, 0.034, is marginal, there is no statistically significant difference between integrated HPV and non-HPV HNC patients (p = 0.265). The 5-year survival rate was 72% for episomal HPV HNC patients versus 30% for patients with integrated HPV HNC, and 40% for patients with non-HPV HNC. Therefore, the reduced survival for patients with integrated HPV HNC compared to patients with episomal HPV HNC suggests a difference that should be studied further with a larger sample size based on our assessment methods for the HPV status. The patients with episomal HPV HNC (the majority of the HPV positive patients) showed a significantly better survival than patients with non-HPV HNC with a p-value of 0.0005, which is comparable to results from other studies comparing patients with HPV HNC to patients with non-HPV HNC [12]. There was insufficient data to analyze and control for therapy, tobacco or alcohol use.
Figure 1.
Overall survival Kaplan Meier curves for episomal HPV tumor patients, integrated HPV tumor patients, and non-HPV tumor patients. No indicators are shown for censored data. Episomal HPV is shown in red (dark), integrated HPV is shown in green (medium), and non-HPV is shown in blue (light). An overall p-value is shown. Pairwise p-values are described in the text. Sample size (n) for each group is shown.
Age and year of birth differences between integrated and episomal HPV
Studies have shown patients with HPV+ HNC are younger at diagnosis than patients with non-HPV HNC [13, 14], with a difference ranging from 4 to 10 years. We observed patients with episomal HPV HNC had a mean age of 55.9 years while patients with integrated HPV HNC had an age 5.8 years older at 60.6 (p = 0.042). The age for patients with integrated HPV HNC was comparable to patients with non-HPV HNC at 61.9 (p = 0.685). We then explored one possible contribution to this difference in age. The era when patients were exposed to HPV and the events specific to that era could potentially affect the physical genomic state of HPV, so we gauged this potential effect by investigating the year of birth for patients. Data within the TCGA was collected on patients with diagnoses from 1992 to 2013, and year of birth from 1909 to 1990, therefore, it’s possible the year of birth could be more informative than age of diagnosis (for example, it’s possible that a cohort aged 60 diagnosed in 1992 is different from one aged 60 diagnosed in 2013 with regards to integration status). Patients with episomal HPV HNC had a mean year of birth of 1954.5 while the mean for patients with integrated HPV was 1946.3, 8.2 years earlier (p = 0.0035). The mean year of birth for integrated HPV HNC patients was comparable to that for patients with non-HPV HNC at 1945.7 (p = 0.880). Figure 2 shows a bar chart of the incidence of integrated HPV and episomal HPV versus the year of birth for patients. Integrated HPV is found predominantly in patients born earlier. No patient born after 1958 had HNC with integrated HPV while 33% of patients born after 1958 had episomal HPV. After controlling for age, we found a statistically significant p-value of 0.0052 (versus p = 0.0035, when age was not controlled) for the difference in year of birth between integrated and episomal HPV. Therefore, the genomic state may have shifted from integrated HPV in patients born earlier to episomal HPV in patients born later.
Figure 2.
A bar graph of the incidence of episomal and integrated HPV in head and neck cancer over year of birth for patients. A trend in the incidence for patients with episomal and integrated HPV is shown based on when the patients were born. Integrated HPV is shown by orange (dark) and episomal HPV is shown by blue (light). The data is shown for five-year intervals. The incidence is calculated based on the total for each interval. The sample size is shown for each interval. There are 4 patients born before 1914 and 2 patients born after 1983 with no HPV and not shown in this plot.
Incidence of HPV and how it’s changing
The incidence of HPV has been reported to be increasing within HNC [15], so we investigated the incidence for HPV within all TCGA HNC patients over the years of patient diagnosis, as well as looked at the dynamics of integrated and episomal HPV. Figure 3 shows a bar chart of the incidence of integrated HPV and episomal HPV within all HNC patients from 1992 to 2013. Each value is summed for a 2-year interval. In this dataset, HPV presence appears to start in the 1990s, peaks in the 2002-2005 period at ~35% of HNC samples and levels out at an incidence of around 13% of all HNC samples. The incidence of integrated HPV among total HPV samples in the last 4 years is ~17%. The last 8 years (2006-2013) show a relatively constant incidence of HPV. The greatest number of samples analyzed and thus the greatest confidence is within these last 8 years, and thus we feel the data within these 8 years show the highest confidence in prediction and trend for HPV in HNC from this data. Therefore, within the limits of the TCGA dataset, there’s a constant contribution of HPV to head and neck cancer.
Figure 3.
A bar graph of the incidence of episomal and integrated HPV in head and neck cancer over year of patient diagnosis. A trend in the incidence for patients with episomal and integrated HPV is shown based on when the patients were diagnosed. Integrated HPV is shown by orange (dark) and episomal HPV is shown by blue (light). The data is shown for two-year intervals. The incidence is calculated based on the total for each interval. The sample size is shown for each interval.
Discussion
The survival rate observed for the episomal HPV HNC patients appeared to be consistent with the higher survival rate others have observed for patients with HPV HNC in general. The patients with integrated HPV HNC, which by our analysis is a minority of the patients with HPV HNC, showed a survival rate different from patients with episomal HPV HNC and similar to the lower survival rate shown for patients with non-HPV HNC. This observational study suggests a need for further investigation with greater number of HPV HNC patients, and experimentally designed studies. We used the loss of expression of the HPV gene E2 as a means to distinguish between episomal and integrated HPV, and it’s possible that E2 expression could be a causative factor in the survival. The role of E2 in this process needs to be further investigated. Expression of the genes E4 and E5 were also lost along with that of E2, so loss of expression of any of these three genes could be causative. A third type of variable to consider is HPV genome copy number, in which integrated HPV samples have a low copy number while episomal HPV samples have a high copy number. We have shown that HPV genome copy number does not influence HPV gene expression [10] but other effects are possible. The use of E2 expression has already been used to assess survival in HNC with somewhat similar results [16]. Ramqvist et al. showed a significant improvement in a 3-year disease-free survival for patients with E2 expression compared to patients with low or no E2 expression, though no difference for overall survival. Since E2 expression loss is a marker for integration, further investigation would be required to determine if E2 expression loss could possibly be separated from integration. In addition, there is a substantial difference in HPV DNA copy number between integrated and episomal HPV samples that could account for biological differences between integrated and episomal HPV samples and patients. In our analysis, we were not able to control for treatment variables that could in part account for survival differences. A larger study is required to fully address how the integrated and episomal HPV states assessed by our analysis and HPV gene expression relate to survival.
Our analysis of how the episomal and integrated HPV states related to age of diagnosis and to year of birth revealed a marginally significant difference in age between the two HPV states and more striking difference in year of birth. Patients with integrated HPV HNC were born a mean of ~8 years earlier than patients with episomal HPV HNC. The difference could be explained by the different era in which the two groups were exposed to HPV. A person born in 1946 could have been exposed to HPV starting in the early to mid 60s, while a person born in 1954 could have been exposed to HPV starting in the early to mid 70s. Many changes in society were occurring during this time-period that potentially could affect the distribution of integrated and episomal HPV, such as sexual behavior and tobacco use.
Our analysis of the year of diagnosis for the HPV HNC patients shows that HPV became prevalent among the HNC patients in the late 90s to the early 2000s with a peak of around 2004-2005. Both integrated and episomal HPV were most prevalent in the same period. However, the drop and leveling off to a steady state of 13% for all HPV for the last 8 years of the TCGA population suggests that there was no increase in HPV-related cancers among all HNC during 2006 to 2013. The trend for the prevalence of integrated HPV among all HPV also suggests that it was maintained at a relatively low level for the last 8 years of the study.
Much of our observations are encouraging for the future of HNC patients. The incidence of HPV HNC does not appear to be rising in proportion to non-HPV HNC. Also, based on the year of birth analysis, if we project into the distant future, we might see integrated HPV tumors very rarely. The primary form of HPV is the episomal HPV state that confers a better prognosis. The ability to identify poor responders to therapy, possibly by using the assessment of integrated HPV presented in this study, is important for the study of de-escalation treatment, and further study and larger studies should prove more definitive. We propose that a simple RNA in situ hybridization for E7 and E2, E4, or E5 using formalin fixed paraffin embedded samples could be used in retrospective studies to determine whether HPV+HNC patients who are E7 positive but E2, E4, and E5 negative have worse clinical outcomes. This analysis could be done on patients treated with standard chemo-radiation therapy and also those who have joined de-escalation clinical trials. This is a relatively straightforward assay that, if validated, could be adopted into stratification studies for patients in de-escalation clinical trials. This could prevent harm to those patients with a poor prognosis and whom we predict would not fare well in de-escalation clinical trials.
Highlights.
Patients with integrated HPV16 in head and neck cancer show poor survival.
The state of HPV16 has shifted from integrated to episomal.
The overall incidence of HPV in head and neck cancer has leveled off.
Acknowledgments
This work was supported by the grant R03 DE026230 from the National Institute for Dental and Craniofacial Research and by P30 CA016059 from the National Cancer Institute. We thank Amy Olex of VCU’s Wright Center for Clinical and Translational Research for assistance with data files and storage, (funded by the CTSA award No. UL1TR000058 from the National Center for Advancing Translational Sciences).
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Sturgis EM, Cinciripini PM. Incidence trends for human papillomavirus-related and -unrelated oral squamous cell carcinomas in the United States. Cancer. 2007;110:1429–1435. [Google Scholar]
- 2.Das P, Thomas A, Kannan S, Deodhar K, Shrivastava SK, Mahantshetty U, et al. Human papillomavirus (HPV) genome status & cervical cancer outcome - A retrospective study. Indian J Med Res. 2015;142:525–532. doi: 10.4103/0971-5916.171276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Fakhry C, Westra WH, Li S, Cmelak A, Ridge JA, Pinto H, et al. Improved survival of patients with human papillomavirus-positive head and neck squamous cell carcinoma in a prospective clinical trial. J Natl Cancer Inst. 2008;100:261–269. doi: 10.1093/jnci/djn011. [DOI] [PubMed] [Google Scholar]
- 4.Mirghani H, Amen F, Blanchard P, Moreau F, Guigay J, Hartl DM, et al. Treatment de-escalation in HPV-positive oropharyngeal carcinoma: ongoing trials, critical issues and perspectives. Int J Cancer. 2015;136:1494–1503. doi: 10.1002/ijc.28847. [DOI] [PubMed] [Google Scholar]
- 5.Parfenov M, Pedamallu CS, Gehlenborg N, Freeman SS, Danilova L, Bristow CA, et al. Characterization of HPV and host genome interactions in primary head and neck cancers. Proc Natl Acad Sci U S A. 2014;111:15544–15549. doi: 10.1073/pnas.1416074111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Olthof NC, Speel EJ, Kolligs J, Haesevoets A, Henfling M, Ramaekers FC, et al. Comprehensive analysis of HPV16 integration in OSCC reveals no significant impact of physical status on viral oncogene and virally disrupted human gene expression. PLoS One. 2014;9:e88718. doi: 10.1371/journal.pone.0088718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Olthof NC, Huebbers CU, Kolligs J, Henfling M, Ramaekers FCS, Cornet I, et al. Viral load, gene expression and mapping of viral integration sites in HPV16-associated HNSCC cell lines. Int J Cancer. 2015;136:E207–E218. doi: 10.1002/ijc.29112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Network, C.G.A. Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature. 2015;517:576–582. doi: 10.1038/nature14129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Network T.T.R. The Cancer Genome Atlas. 2015 http://cancergenomenihgov.
- 10.Nulton TJ, Olex AL, Dozmorov M, Morgan IM, Windle B. Analysis of The Cancer Genome Atlas sequencing data reveals novel properties of the human papillomavirus 16 genome in head and neck squamous cell carcinoma. Oncotarget. 2017;8:17684–17699. doi: 10.18632/oncotarget.15179. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kadaja M, Silla T, Ustav E, Ustav M. Papillomavirus DNA replication - from initiation to genomic instability. Virology. 2009;384:360–368. doi: 10.1016/j.virol.2008.11.032. [DOI] [PubMed] [Google Scholar]
- 12.Ang KK, Harris J, Wheeler R, Weber R, Rosenthal DI, Nguyen-Tan PF, et al. Human papillomavirus and survival of patients with oropharyngeal cancer. N Engl J Med. 2010;363:24–35. doi: 10.1056/NEJMoa0912217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Grisar K, Dok R, Schoenaers J, Dormaar T, Hauben E, Jorissen M, et al. Differences in human papillomavirus-positive and -negative head and neck cancers in Belgium: an 8-year retrospective, comparative study. Oral Surg Oral Med Oral Pathol Oral Radiol. 2016;121:456–460. doi: 10.1016/j.oooo.2015.10.035. [DOI] [PubMed] [Google Scholar]
- 14.Heath S, Willis V, Allan K, Purdie K, Harwood C, Shields P, et al. Clinically significant human papilloma virus in squamous cell carcinoma of the head and neck in UK practice. Clin Oncol. 2012;24:18–23. doi: 10.1016/j.clon.2011.05.007. [DOI] [PubMed] [Google Scholar]
- 15.Chaturvedi AK, Engels EA, Pfeiffer RM, Hernandez BY, Xiao W, Kim E, et al. Human papillomavirus and rising oropharyngeal cancer incidence in the United States. J Clin Oncol. 2011;29:4294–4301. doi: 10.1200/JCO.2011.36.4596. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ramqvist T, Mints M, Tertipis N, Nasman A, Romanitan M, Dalianis T. Studies on human papillomavirus (HPV) 16 E2, E5 and E7 mRNA in HPV-positive tonsillar and base of tongue cancer in relation to clinical outcome and immunological parameters. Oral Oncol. 2015;51:1126–1131. doi: 10.1016/j.oraloncology.2015.09.007. [DOI] [PubMed] [Google Scholar]