Human Papillomavirus Prevalence in Oropharyngeal Cancer Before Vaccine Introduction, United States

Martin Steinau; Mona Saraiya; Marc T. Goodman; Edward S. Peters; Meg Watson; Jennifer L. Cleveland; Charles F. Lynch; Edward J. Wilkinson; Brenda Y. Hernandez; Glen Copeland; Maria S. Saber; Claudia Hopenhayn; Youjie Huang; Wendy Cozen; Christopher Lyu; Elizabeth R. Unger


Emerging Infectious Diseases. 2014;20(5):822-828. 

In This Article


Our finding of >70% HPV prevalence in a large sample of the oropharyngeal cancer patients from around the United States suggests that a substantially higher faction may be HPV-related than has been reported in many previous investigations.[14] In a systematic review of HPV prevalence studies, including several small investigations of populations in North America, Kreimer et al. estimated that 47% of OPSCC cases were HPV related.[11] Chaturvedi et al. more recently estimated weighted HPV prevalence at 72%,[12] which is comparable to our findings. A continuous increase in HPV-related OPSCC that was observed during the past 20 years and escalated after 2000[23] may explain some of the discrepancies found in the literature. More sensitive laboratory methods may also have contributed to the increased HPV prevalence relative to earlier investigations. All studies in North America considered by Kreimer et al.[11] relied on the MY09/11 or GP5+/6+ consensus primer sets that require intact HPV L1 fragments of 450 and 150 bp, respectively. By contrast, in the study by Chaturvedi et al.[12] and ours, testing incorporated the INNO-LiPA assay, which has SFP primers that target 65-bp amplicons. Assays targeting smaller L1 amplicons can achieve increased sensitivity for HPV detection in formalin-fixed, paraffin-embedded tissues known to have smaller DNA templates in their extracts than fresh or frozen samples do.[18] Although not identical in specificity to that of the Linear Array, INNO-LiPA had shown comparable performance for detecting single type HPV infections predominantly found in cancer tissues.[19]

Our current study results further confirm a trend of increasing incidence of tonsillar cancers with shifting demographic patterns.[24,25] The results further confirmed that HPV-16, detected in 84% of all positive tissues, was by far the most frequent type found in oropharyngeal cancers. Although type 16 also has nominally the highest prevalence in the "normal" oral cavity or oropharynx, other types are usually found at similar frequency.[26,27] The ability of HPV-16 to establish persistent infection and its potential to transform might be responsible for its prominence in cancers. Currently available HPV vaccines targeting HPV-16 and -18 may be highly effective against OPSCC.[9] A candidate 9-valent vaccine (currently in clinical trials) could have the potential to prevent virtually all HPV-associated oropharyngeal cancers: our data showed that 2.9% of the case-patients were positive for a high-risk type not covered in this formulation. (Figure 1).

The most noticeable differences were observed between racial groups, with notably fewer HPV-positive SCCs in non-Hispanic Black persons (50.7%) compared with non-Hispanic White persons (73.6%), Hispanic persons (74.4%), or Asian Pacific Islanders (80.0%). Other studies that noted similar differences by race/ethnicity found this to be a recent but ongoing development.[28] Settle et al.,[29] who investigated oral cancer survival, also reported reduced HPV prevalence in Black persons compared with other race/ethnicity groups and found that this difference was particular to oropharyngeal cancer and not to other cancers of the oral cavity.

In addition to differences by race/ethnicity, HPV prevalence also varied by sex, particularly for HPV-16/18. Prevalence was 66% among men, notably higher than the 53% found among women, which was a finding consistent with results of other investigations.[2] Although further data stratification might show even greater dissimilarities, for instance between Black women and White men, the sample sizes for these analyses were modest and confidence intervals were large (data not shown). The precise causes for these discrepancies are unknown and most likely complex, but may be anticipated to influence vaccine efficacy for OPSCC. Prevalence differences observed between the registry states may be, in part, caused by demographic variations. Difference in age at diagnosis between patients with HPV-positive and HPV-negative cases was borderline significant (p = 0.023). Although other studies have also shown that HPV-positive cancers occur in a younger population,[30] the role of 3 years difference in median age is unclear. It is possible that differences in behavior associated with causal pathways, such as smoking and drinking, provide a partial explanation. Persistent HPV infection at these anatomic sites may occur early, leading to more rapid and damaging alterations in cell cycle regulation and proliferation than those that occur with other carcinogenic exposures.

Of particular note, high-risk HPV types were detected in 80% of tonsillar SCCs. The microanatomy of the lymphoepithelial tissue of Waldeyer's ring, most notably the lingual and palatine tonsils, may explain this finding. Deep invaginations in this area by the tonsillar crypts may expose immature basal cells to HPV.[31] The median age of case-patients was slightly lower than that of those with cancer in other sites (55 years), but proportions of infection, when sex and race/ethnicity were considered, were not different than proportions for the other oropharyngeal sites (data not shown). One limitation of this study is that not all participating sites were able to perform systematic random sampling of case-patient specimens from their eligible pool. The sizable specimen collections from the 4 cancer registries (Michigan, Kentucky, Louisiana, and Florida) were sampled by a simple random or systematic sampling approach, on the basis of the number of eligible cases. Sampling from the SEER tissue repositories (Los Angeles, California; Hawaii; and Iowa) was dependent on what tissue specimens were available. However, the resulting sample population that was ultimately tested represented diverse geographic regions and a wide range of demographic variables regarding sex, age, and race/ethnicity.

It should be noted that the presence of HPV DNA does not confirm its causal role in carcinogenesis. Detection in tumor tissues potentially overestimates the true involvement of the virus because coincidental, transient infections and complementary transforming effects to other factors cannot be distinguished. The natural history of cases in this study could not be assessed in this retrospective cross-sectional study and behavioral data were not available from the participating cancer registries. In particular, information regarding tobacco or alcohol use and HIV status would potentially improve estimation of the proportion of OPSCC caused by HPV alone. Because it is not clear at this point if HPV alone is sufficient to cause oropharyngeal cancer, factors other than use of tobacco products should be considered. Additional HPV markers, such as viral transcription (particularly E6 and E7 mRNA) or characteristic gene expression profiling, may provide further insights in future assessments and show distinction between actively transforming HPV infections and random, transient occurrences.[32,33] Similar investigations may also be warranted to explicate the 7 cases in which only low-risk types were found. It is likely that these HPV types were present coincidentally and played no role in malignant transformation, but genomic changes that altered their pathogenic properties to bring them closer to those of high-risk types could provide an intriguing alternative explanation.