Two of the authors (JFS, JB) declare a financial conflict of interest. Both are employed by Merck Research Laboratories, which markets a prophylactic vaccine against anogenital HPV.
Infectious Causes of Cancer
Human papillomavirus 16 and head and neck squamous cell carcinoma
Article first published online: 21 FEB 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 120, Issue 11, pages 2386–2392, 1 June 2007
How to Cite
Furniss, C. S., McClean, M. D., Smith, J. F., Bryan, J., Nelson, H. H., Peters, E. S., Posner, M. R., Clark, J. R., Eisen, E. A. and Kelsey, K. T. (2007), Human papillomavirus 16 and head and neck squamous cell carcinoma. Int. J. Cancer, 120: 2386–2392. doi: 10.1002/ijc.22633
- Issue published online: 27 MAR 2007
- Article first published online: 21 FEB 2007
- Manuscript Accepted: 11 DEC 2006
- Manuscript Received: 10 OCT 2006
- NIH. Grant Numbers: CA100679, CA78609, T32ES07155
- human papillomavirus (HPV);
- head and neck squamous cell carcinoma (HNSCC);
Evidence suggests that human papillomavirus (HPV)16 seropositivity reflects past HPV16 exposure and is associated with risk for head and neck squamous cell carcinoma (HNSCC). Our objectives were to test the hypothesis that HPV16 seropositivity is associated with risk for HNSCC, to correlate HPV16 seropositivity with HPV16 tumor DNA, and to correlate HPV16 seropositivity and HPV16 DNA with sexual history and patient survival. In a case–control study of approximately 1,000 individuals, we assessed serology to the HPV16 L1 protein and in cases only, assayed tumors for HPV16 DNA. HPV16 seropositivity was associated with 1.5- and 6-fold risks for tumors of the oral cavity and pharynx, respectively. There was a dose response trend for HPV16 titer and increasing risk of HNSCC (p < 0.0001) and HPV16 tumor DNA (p < 0.0001). In cases, HPV16 DNA and seropositivity were significantly associated with sexual activity; odds ratios (ORs) of 12.8 and 3.7 were observed for more than 10 oral sexual partners and ORs of 4.5 and 3.2 were associated with a high number of lifetime sexual partners, respectively. Finally, HPV16 seropositivity and HPV16 tumor DNA were associated with hazard ratios of 0.4 and 0.5, respectively, indicating better survival for HPV positive individuals. HPV16 seropositivity was associated with risk for HNSCC, with greatest risk for pharyngeal cancer. We observed dose response relationships between serology titer and both risk for HNSCC and HPV16 tumor DNA. In cases, HPV16 tumor DNA and positive serology were associated with sexual history and improved disease free survival. © 2007 Wiley-Liss, Inc.
Each year, ∼40,000 new cases of head and neck squamous cell carcinoma (HNSCC) are diagnosed in the United States. Estimates for 2006 suggest 27,880 new cases in males and 12,620 in females, with 8,000 and 3,170 deaths, respectively (American Cancer Society. Cancer Facts and Figures 2006. Atlanta: American Cancer Society; 2006). Well known risk factors for HNSCC include tobacco and alcohol exposure, which account for roughly 75% of all HNSCCs in the United States.1, 2 Recently, human papillomavirus (HPV) has also been implicated in the pathogenesis of HNSCC.3
HPV, a known tumor virus, has been identified as the causative agent in 99.7% of cervical cancers.4 While over 200 different types of HPV have been identified, relatively few have been demonstrated to be associated with the development of cancer.5 Those viruses that lead to cervical cancer and precursor lesions have been labeled as “high risk” HPV.5 Roughly 25% of HNSCC are positive for high risk HPV DNA6 and among those positive, 90–95% are positive for HPV16.6, 7 Studies have shown that the presence of HPV16 antibodies is a strong marker of past exposure to HPV16 and like detection of HPV16 DNA in tumors, HPV16 antibodies may be used to estimate risk for HNSCC7, 8 and cervical cancer.9
The last few years have seen great strides toward the production of HPV vaccines based upon noninfectious virus-like particles comprised of the L1 protein of HPV.10, 11 A quidravalent vaccine against HPV6, 11, 16 and 18, protecting against genital warts, cervical cancer and cervical precancerous lesions, recently received FDA approval.12 A competing bivalent vaccine against HPV16 and HPV18, protecting against cervical cancer and precancerous lesions, is currently in phase III clinical trials.13, 14 These vaccines elicit an antibody response against the L1 protein and are efficacious in preventing the development of HPV-related diseases of the cervix.11 In a large controlled trial of young women receiving an HPV16 L1 virus-like particle vaccine, administration of this HPV16 vaccine reduced the incidence of both HPV16 infection and HPV16-related cervical intraepithelial neoplasia.11
Studies involving HPV serology have generally focused on detecting antibodies to HPV16 and have assessed the E6, E7 or the L1 protein.7, 8, 9 The detection of HPV antibodies has been positively associated with risk of both cervical cancer and HNSCC.7, 15 Antibodies are largely type-specific and correlate with the presence of viral DNA.8, 9 Detection of antibodies to L1 have been shown to correlate with increased risk for HNSCC, with the greatest risk for cancers of the tonsil/oropharynx.7, 16 High viral loads are associated with HPV16 seropositivity for patients with tumors of the oropharynx,17 but it remains to be determined whether increasing titer is associated with increasing overall risk for HNSCC and/or an increasing likelihood of detecting viral DNA in the tumor.
HPV DNA presence in the tumor has been associated with improved patient survival18, 19 as well as with less pronounced lymph node involvement.20 The presence of HPV DNA is associated with type-specific serology, but it is not known if antibody presence is also associated with better patient survival and whether antibodies may have an additional role in patient survival among those who are HPV16 DNA positive.
In a large case–control study of roughly 1,000 participants, we assessed the relationship between HPV16 seropositivity and cancers of the oral cavity and pharynx. In HNSCC cases, we studied HPV16 seropositivity as well as HPV16 tumor DNA, examining the relationship between these markers of HPV infection, sexual history and patient survival. We hypothesized that number of oral sex partners would correlate with both HPV16 DNA and HPV16 seropositivity and that HPV16 seropositivity would be associated with better overall survival among cases. We also hypothesized that increasing serologic titer would be associated with both increased risk for HNSCC and increased risk for detectable HPV DNA, as HPV antibodies are considered markers of HPV infection.
Material and methods
We conducted a case–control study in Boston, MA from December 1999 to December 2003. Cases and controls were drawn from the Greater Boston Metropolitan Area, a population of roughly 3.5 million people in 249 cities and towns within a 1 hr drive of Boston. The institutional review boards at all participating institutions approved this study, and all volunteer participants provided informed consent.
We enrolled individuals with incident HNSCC identified through the multidisciplinary head and neck clinics, otolaryngology and radiation oncology departments at 9 large teaching hospitals located in Boston, MA. The participating institutions included Beth Israel Deaconess Medical Center, Boston Medical Center, Boston Veterans Administration, Brigham and Women's Hospital, Dana-Farber Cancer Institute, Harvard Vanguard Medical Associates, Massachusetts Eye and Ear Infirmary, Massachusetts General Hospital and the New England Medical Center.
Eligibility requirements included being at least 18-years old, a resident in the Boston area, first diagnosis of HNSCC within the past 6 months, and being alive at the time of initial contact. We defined HNSCC using the International Classification of Disease Codes, Ninth Revision (ICD-9) codes 141, 143-6, 148, 149 and 161. Patients with carcinoma in situ, lip, salivary gland, or nasopharyngeal cancer or recurrent cancer of the head and neck region were excluded. No prevalent cases were studied. Histopathology reports and diagnostic specimens were reviewed at each participating hospital. These slides were then reviewed and confirmed by a study pathologist. Cases were further classified by oral cavity (ICD-9 codes 143–145), pharynx (ICD-9 codes 146, 148, 149), tongue (ICD-9 141) and larynx (ICD-9 code 161) to facilitate analysis of disease by site.
Population-based controls were frequency matched (1:1) to cases by age (±3 years), gender and town of residence. These controls were identified through systematic random selection from the Massachusetts Resident Lists for the 249 cities and towns within the study area. These annually compiled Resident Lists are mandated by state law and include all residents more than 17 years of age. The resident lists include name, gender, year of birth, usual occupation and last address.21
Residents were listed in the book by street address and precincts. A potential control of the same gender and age as the case was sought from this list, alternating the search direction through the book starting at the case address. Controls (but not cases) were offered US $80 as an incentive for participation.
823 eligible cases were invited to participate and 57 refused to participate. Of the 766 consented cases, another 44 did not complete the questionnaire, resulting in an overall 88% case participation rate. 1,643 subjects were identified and were eligible for participation as controls. Of these subjects, 828 refused to participate, 815 subjects were consented, and 771 finally enrolled in the study. Six of the controls were withdrawn as they were matched to a case that became ineligible, such that 765 controls were enrolled and completed. This represents an overall participation rate of 47% for controls. Our serology study was limited to 486 cases and 550 controls based on the availability of serum. We were limited to 270 case patients for whom we were able to obtain DNA from paraffin-embedded tumors.
Participating cases and controls were given a self-administered questionnaire to collect medical history and demographic information as well as information on tobacco and alcohol consumption. Each questionnaire was reviewed with each participant by a trained research coordinator. Smoking history was ascertained with a standardized instrument that assesses the number of years smoked, the number of cigarettes smoked per day, age at which an individual started smoking, number of years since quitting and the duration of smoking. Similar information was obtained regarding lifetime consumption of beer, wine and liquor. Control participants received their questionnaires by mail and returned them in person to the research assistant. Supplemental questionnaires, which included questions on lifetime number of sexual partners and lifetime number of oral sexual partners' were sent to participants' homes for completion.
For cases, clinical and pathological information on the tumor including size, location, stage, differentiation and treatment were collected by medical record review.
Exposure assessment for HPV
Venous blood samples were obtained from cases and controls. Serum was separated within 12–24 hr of blood drawing and samples were stored at −80°C. Frozen samples were shipped to Merck & Co., testing laboratory (West Point, PA). Samples were thawed under refrigeration and subsequently heat inactivated for 30 ± 2 min at (56 ± 2)°C.
Formalin-fixed, paraffin-embedded sections were obtained from participating hospitals, and tissues were processed for DNA extraction and detection of HPV16 DNA as described below.
HPV competitive Luminex® immunoassay (cLIA) for detectionof HPV 16 serological titer
The detailed procedures of the assay have previously been described by Dias et al.22 Results were read from a standard curve, corrected for dilution and reported in arbitrary units (milli-Merck units). A fixed cut-off of 12 mMU per milliliter (derived by repeatedly testing a panel of positive and negative samples against the standard curve) was used to determine HPV16 serologic status.11
DNA was extracted from paraffin embedded tissues according to the method described previously.23
Detection of HPV16 tumor DNA from HNSCC cases
The short fragment PCR (SPF) assay was performed to amplify a 65 base pair region of the L1 gene of HPV16, according to the method previously described24 with an exception that only SPF1A (forward) and SPF2B (reverse), primers specific to HPV16, were used in the assay. Amplification of beta-actin was performed as a control for total DNA presence while DNA from Siha and Ca33 cells was used as a positive and negative control for HPV16, respectively.
Detection of HPV16 E6 from HNSCC fresh tumor DNA
For a subset of HNSCC fresh tumors (matched to paraffin samples), we amplified the HPV16 E6 gene according to the method described previously.25
All statistical analysis was performed with SAS software (version 9.1; SAS Institute, Cary, NC). Race was dichotomized as white and nonwhite, with nonwhite including Asians, African-Americans, Native Americans and Hispanics because of limited number of nonwhite individuals. Education and household income were both dichotomized at the median level which corresponded to community college and a range of $50,000 to $79,999, respectively. Lifetime cigarette smoking was measured as pack-years (years smoked times packs per day) and drinking was measured by average alcohol consumption over a lifetime, i.e., average drinks per week.
Tumor location was classified as cancers of the oral cavity, pharynx or larynx based on recommendations by the American Joint Committee on Cancer (AJCC), 4th edition. Tumor location was later dichotomized as tumors of the oral cavity or pharynx. Cancers of the oral cavity included those of the oral cavity (ICD-9 codes 143–145) and those of the larynx (ICD-9 code 161) while cancers of the pharynx included those of the tongue, tonsil, and pharynx (ICD-9 codes 141, 146, 148 and 149). According to the AJCC, base of tongue tumors are classified as tumors of the pharynx while tumors in the anterior two-thirds of the tongue are classified as tumors of the oral cavity. Since we did not have exact tongue location, we combined all tumors of the tongue with pharyngeal cancers as a high percent of cancers of the tongue, like cancers of the oropharynx, are associated with positive HPV16 serology.
Oral sexual behavior was dichotomized at the 75th percentile; those with greater than or equal to 10 partners (the top quarter number of partners among cases and controls) were compared to those with less than 10 partners. Lifetime sexual behavior was similarly dichotomized based on the 75th percentile; those with the top quarter of number of partners (among cases and controls) were considered high while those in the bottom 75% were considered to have a low number of partners. The 75th percentile corresponded to greater than or equal to 12 partners for patients with HPV16 serology data and greater than or equal to 7 partners for patients with HPV16 tumor DNA data (cases only).
HPV16 seropositivity was defined as greater than or equal to 12 mMU, above the threshold considered background reactivity. HPV16 titer was grouped into 3 categories to evaluate a possible dose response between titer and DNA and titer and risk for HNSCC. Low titer was defined as 12–38 mMU with 39 mMU being the median for positive sera in the control patient group. Medium titer was defined as greater than or equal to 39 mMU but less than 253 mMU (50–90% range) and high titer defined the top 10% of all positive titers, greater than or equal to 253 mMU.
Unconditional logistic regression was performed to calculate odds ratios and 95% confidence intervals, evaluating the associations between markers of HPV infection, including HPV16 tumor DNA and HPV16 seropositivity, and cancer risk and between these markers and sexual history.
Survival data were obtained from the social security death index and from active search of public databases, including telephone confirmation of vital status, retirement records and work history when available. For all patients, follow-up time was calculated as from the date of diagnosis until death or from date of diagnosis until February 1, 2005. Patients were censored if they were lost to follow-up or were still alive as of February 1, 2005.
Kaplan–Meir survival analysis was performed to determine the crude association between HPV16 seropositivity and overall patient survival and survival curves were generated with Sigma plot, version 9.0. To evaluate effect modification and adjust for confounding, Cox regression was performed, with risk factors associated with HNSCC entered into the model.
Because of the missing data, our total number of subjects in each portion of the study differed. Our data set were complete for age, ethnic, gender, smoking and drinking behaviors but was incomplete for tumor location, survival information, HPV tumor DNA status and sexual history. We had tumor location data on 478 of the 486 cases we studied based on availability of ICD-9 codes. Of our 486 HNSCC cases, we had HPV DNA data on 264 because of the availability of paraffin tumors. Our survival data were limited to 387 cases for whom we had data on HPV16 seropositivity and 270 cases for whom we had HPV16 DNA data. Finally, our sexual history data were limited. For the serology portion of our study, 99 and 103 cases and 165 and 170 controls reported their oral sex and lifetime sex partners, respectively. For individuals with HPV16 DNA data, 62 and 65 reported their number of oral sex partners and lifetime sex partners, respectively.
We studied 486 HNSCC cases and 550 controls (Table I). Roughly three quarters of our study population was male (74% of cases and 73% of controls). The majority of the population was white (92% of cases and 91% of controls). These similarities between cases and controls demonstrate successful matching of the study participants.
|Demograph characteristic1||Controls||All cases||Tumor of oral cavity||Tumor of pharynx|
|N (%)||N (%)||OR (95%CI)||N||OR (95%CI)||N||OR (95%CI)|
|3 Female||147 (27)||125 (26)||1.0 (referent)||46||1.0 (referent)||77||1.0 (referent)|
|Male||403 (73)||361 (74)||0.6 (0.4,0.8)||144||0.5 (0.3,0.8)||211||0.7 (0.5,1.0)|
|<45||42||40||1.0 (referent)||9||1.0 (referent)||31||1.0 (referent)|
|45–34||116||113||0.7 (0.4,1.2)||45||1.1 (0.5,2.7)||66||0.5 (0.3,1.0)|
|55–64||169||170||0.6 (0.3,1.0)||65||0.9 (0.4,2.0)||102||0.5 (0.3,0.9)|
|≥65||222||163||0.5 (0.3,1.9)||71||0.8 (0.4,1.9)||89||0.4 (0.2,0.7)|
|White||502 (91)||445 (92)||1.0 (referent)||174||1.0 (referent)||264||1.0 (referent)|
|NonWhite||48 (9)||41 (8)||0.8 (0.5,1.4)||16||0.8 (0.4,1.5)||24||0.8 (0.5,1.5)|
|Drinking (average drinks/week)|
|>5||241||133||1.0 (referent)||49||1.0 (referent)||83||1.0 (referent)|
|5–14||171||107||1.0 (0.7,1.5)||34||0.9 (0.6,1.6)||72||1.1 (0.7,1.7)|
|15–29||77||98||2.1 (1.4,3.1)||46||2.2 (1.3,3.8)||50||1.8 (1.1,3.0)|
|>29||61||148||3.6 (2.3,5.5)||61||3.4 (1.9,5.9)||83||3.7 (2.2,6.0)|
|P for trend||<0.0001||<0.0001||<0.000l|
|Smoking (pack years)|
|0||182||90||1.0 (referent)||25||1.0 (referent)||65||1.0 (referent)|
|1–20||157||98||1.1 (0.8,1.7)||25||1.1 (0.6,2.0)||72||1.2 (0.8,1.9)|
|21–45||124||128||1.8 (1.2,2.7)||57||2.7 (1.5,4.7)||68||1.4 (0.9,2.3)|
|>45||87||170||3.5 (2.3,5.3)||83||5.8 (3.3,10.3)||83||2.4 (1.5,4.0)|
|P for trend||<0.0001||<0.0001||0.0006|
|Negative||491||341||1.0 (referent)||162||1.0 (referent)||172||1.0 (referent)|
|Positive||59||145||4.4 (3.1,6.3)||28||1.4 (0.8,2.5)||116||6.4 (4.3,9.4)|
Among both cases and controls, individuals for whom we had serology data did not differ from those with missing data with respect to education level and household income. Among cases, those for whom we had DNA status did not differ from those without DNA status with regard to education level and household income (data not shown).
Both education and household income were not significantly associated with HPV16 seropositivity in controls. These variables also did not significantly change β estimates for HPV16 seropositivity in case–control models. Consequently, they were not included as confounders in final models.
Both alcohol and tobacco consumption were associated with increased risk for HNSCC, with dose response relationships observed between these exposures and risk for HNSCC (Table I). Compared to those who drank less than 5 drinks per week, individuals who consumed 15–29 drinks/week had a 2.1-fold (95% CI: 1.4, 3.1) overall increased risk for HNSCC and those who consumed ≥30 drinks/week had a 3.6-fold (95% CI: 2.3, 5.5) increased risk for HNSCC. Compared to nonsmokers, those who had a lifetime exposure of 21–45 packyears were at 1.8-fold (95% CI: 1.2, 2.7) increased risk for HNSCC and those with greater than 45 packyears were at 3.5-fold (95% CI: 2.3, 5.3) increased risk for HNSCC. The greatest risk for HNSCC was associated with heavy smoking, where individuals with greater than 45 pack years of smoking had a 5.8 (95% CI: 3.3, 10.3) fold increased risk for a tumor of the oral cavity.
HPV16 seropositivity was classified according to ICD-9 site (Table II). About 57.7% of patients with a tumor of the oropharynx had positive serology, followed by patients with a tumor of other sites (oral cavity/pharynx) (36.4%), tongue (35.8%), larynx (20%), floor of the mouth (13.5%), hypopharynx (12.2) and unspecified mouth (10%). No patients with a tumor of the gum had positive HPV16 serology. Controlling for age, ethnic, gender, smoking and drinking behaviors, individuals with tumors of the oropharynx, tongue and other oral cavity site/pharynx had increased risk of HPV seropositivity [(OR = 4.9, 95% CI: 2.4, 10.0), (OR = 2.0, 95% CI:1.1, 3.9), (OR = 2.1, 95% CI: 0.6, 8.3), respectively] relative to those with a tumor of the larynx.
|ICD-9 site||N||Positive HPV16 serology N (%)||OR (95% CI)1||Positive HPV16 DNA2 N (%)|
|141: Tongue||165||59 (35.8)||2.0 (1.1,3.9)||27 (29.0)|
|143: Gum||4||0 (0)||N/A||1 (33.3)|
|144: Floor of mouth||37||5 (13.5)||0.6 (0.2,1.8)||3 (16.7)|
|145: Unspecified mouth||60||6 (10)||0.4 (0.2,1.2)||7 (19.4)|
|146: Oropharynx||78||45 (57.7)||4.9 (2.4,10.0)||15 (34.9)|
|148: Hypopharynx||33||33 (12.2)||1.1 (0.4,3.1)||6 (33.3)|
|149: Other sites (orel cavity/pharynx)||11||4 (36.4)||2.1 (0.6,8.3)||3 (42.9)|
|161: Larynx||90||18 (20.0)||1.0 (referent)||14 (31.1)|
Among 550 controls, there were only 59 (11%) with detectable serologic evidence of HPV16 infection. However, in the 486 cases, 145 (29%) had a positive serologic response to HPV16 infection. Hence, HPV16 seropositivity was associated with a highly significant 4.1-fold (95%CI: 2.9, 5.8) increased risk for HNSCC. When we characterized the serologic response into 3 groups (below the median among cases and controls, the 50th to 90th percentiles, and the upper 10th percentile) there was marked evidence of a dose response between increasing titer and increasing risk (p < 0.0001, Table III). There was a 3.8-fold (95% CI: 2.3, 6.5) overall risk for HNSCC associated with a medium HPV16 titer and a 19.2-fold (95% CI: 8.1, 45.6) overall risk associated with a high titer. The dose response relationship between titer and HNSCC was most significant for tumors of the pharynx where there was a 5.9-fold (3.4, 10.1) risk associated with a medium titer and a 28.6-fold (11.9, 68.7) increased risk associated with a high titer (p < 0.0001, Table III).
|HPV16 Serology||Controls N = 550||All cases N = 486||Tumor of oral cavity N = 190||Tumor of pharynx N = 288|
|N (%)||N (%)||OR (95% CI)1||N (%)||OR (95% CI)||N (%)||OR (95% CI)|
|Nondetectablc||491 (89.3)||341 (70.2)||1.0 (ref)||162 (85.3)||1.0 (ref)||172 (59.7)||1.0 (ref)|
|Any Positive||59 (10.7)||145 (29.8)||4.0 (2.8,5.7)||28 (14.7)||1.4 (0.9,2.4)||116 (40.3)||6.0 (4.1,8.7)|
|Low titer2||29 (5.3)||36 (7.4)||1.5 (0.8,2.6)||12 (6.3)||0.9 (0.4,1.9))||24 (8.3)||2.0 (1.1,3.6)|
|Medium titer||24 (4.4)||54 (11.1)||3.8 (2.2)||8 (4.2)||1.2 (0.5,2.9)||45 (15.6)||5.8 (3.3,10.0)|
|High titer||6 (1.1)||55 (11.3)||20.1 (8.4,48.1)||8 (4.2)||6.1 (2.0,18.4)||47 (16.3)||30.3 (12.4,73.6)|
Among the patients with available tumor tissue (N = 264), we assessed the tumor for evidence of HPV16 DNA in the tumor itself (assaying this blindly from the serologic data). 28% of tumors had detectable HPV16 DNA. The presence of positive serology was significantly associated with the detection of HPV16 DNA in the tumor (Table IV). Any positive serology was associated with a 3.2-fold (95% CI: 1.8, 5.8) increased risk of detecting HPV16 DNA in the tumor. Additionally, there was a dose response relationship between increasing titer and increasing risk for detection of HPV16 DNA in the tumor, with a 3.8-fold (95% CI: 1.8, 8.3) increased risk associated with a medium titer and a 5.0 (1.9, 12.9) fold increased risk associated with a high titer.
|HPV16 Serology||All cases N=264|
|Any positive||36||40||3.2 (1.8,5.8)|
|Low titer2||4||14||1.2 (0.4,3.9)|
|Medium titer||18||17||3.8 (1.8,8.3)|
|High titer||14||9||5.0 (1.9,12.9)|
From a subset of fresh tumor samples matched to paraffin samples in our study (N = 150), we amplified the HPV16 E6 gene. Of HPV16 DNA positive samples, 96% (44 of 46) of individuals had an intact E6 gene (data not shown).
In an effort to examine the mechanism of transmission of the virus, we also assessed the sexual history of the study participants. Although a limited number of participants completed this section of the questionnaire, the individuals for whom we had sexual history data did not significantly differ from the remaining patients with regard to age, ethnic and gender (data not shown). HPV16 seropositivity was associated with sexual behavior, including number of lifetime sexual partners and lifetime oral sexual partners, for HNSCC cases (Table V). Cases with 10 or more oral sex partners had a 3.7-fold (95%CI: 1.1, 12.3) increased risk for HPV16 seropositivity and a 12.8 (95%CI: 2.2, 74.4) fold increased risk for HPV16 DNA in their tumor. Cases with a high number of lifetime sex partners had a 3.2-fold (95% CI: 1.1, 9.3) increased risk for HPV16 seropositivity and a 4.5 (95% CI: 1.0, 20.7) fold increased risk for having detectable HPV16 DNA in their tumors (Table V). In controls, there was a relationship of borderline significance between HPV16 seropositivity and number of lifetime partners, where those with a high number of partners were 3.3-fold (95% CI: 0.8, 12.7) more likely to have positive serology. Number of oral sex partners was also related to HPV16 seropositivity in controls, where those with a high number of partners were more likely to have positive serology (OR = 2.3, 95% CI: 0.4, 13.9).
|Predictor||HPV16 L1 Serology||HPV16 DNA|
|Oral Sex||N = 99||N = 165||N = 62|
|<10 partners||1.0 (referent)||l.0 (referent)||1.0 (referent)|
|≥ 10 partners||3.7 (1.1, 12.3)||2.3 (0.4, 13.9)||12.8 (2.2, 74.4)|
|Lifetime sex2||N = 103||N = 170||N = 65|
|Low# partners||1.0 (referent)||1.0 (referent)||1.0 (referent)|
|High# partners||3.1 (1.1, 9.0)||3.3 (0.8, 12.7)||4.5 (1.0, 20.7)|
Finally, HPV16 seropositivity and evidence of detectable HPV16 DNA in the tumor were associated with better survival in cases (Table VI). In Cox regression models, HPV serology positive individuals had significantly increased survival compared to HPV serology negative individuals (HR = 0.4, 95% CI: 0.2, 0.7). Similarly, cases with detectable HPV16 DNA in their tumors were half as likely to die at any given point post diagnosis as those without (HR = 0.5, 95% CI: 0.3, 1.0). Among HPV16 DNA positive individuals, those who were also positive for HPV16 serology had significantly improved survival compared to those negative for HPV16 serology (HR = 0.2, 95% CI: 0.0, 0.9).
|Predictor||HPV16 L1 serology N = 387||HPV16 tumor DNA N = 270||HPV16 DNA tumor/serology N = 68|
|HR (95%CI)||HR (95%CI)||HR (95%CI)|
|Negative||1.0 (referent)||1.0 (referent)||1.0 (referent)|
|Positive||0.4 (0.2,0.7)||0.5 (0.3,1.0)||0.2 (0.0,0.9)1|
|Age2||1.03 (1.01,1.05)||1.03 (1.0,1.05)||1.10 (1.02,1.18)|
|Female||1.0 (referent)||1.0 (referent)||1.0 (referent)|
|Male||0.9 (0.5,1.4)||1.0 (0.5,1.7)||1.0 (0.2,6.6)|
|Oral cavity||1.0 (referent)||1.0 (referent)||1.0 (referent)|
|Oropharynx||1.5 (1.0,2.3)||1.3 (0.8,2.1)||3.4 (0.7,16.4)|
Our data confirm that heavy alcohol and tobacco exposure are associated with risk for HNSCC, with increased risk associated with increasing exposure. HPV has been implicated in the etiology of a subset of HNSCC and both the presence of HPV16 DNA and antibodies to the L1 protein of HPV16 have been associated with risk.3, 7 In our study, HPV16 seropositivity was associated with a significant 4.1-fold increased risk for HNSCC, controlling for other HNSCC risk factors, such as smoking and drinking. This was a striking finding given that this magnitude of risk was as great as that associated with heavy smoking (3.1-fold) and drinking (3.5-fold), the 2 main risk factors for HNSCC.
We observed a significant dose response trend for increasing HPV titer with increasing overall risk for HNSCC, where patients with a medium titer had a roughly 4-fold increased risk for HNSCC and patients with a high titer had a 19-fold increased risk. This finding agrees with the general consensus that HPV antibodies are markers of infection.26, 27, 28 In our data there was also a significant, increasing association between the presence of detectable virus in the tumor and higher titers. Thus, consistent with other work, our data suggests that high titers likely reflect the presence of higher type-specific viral loads in the tumor, rather than, for example, increased immune response resulting from infection with multiple types; indeed, HPV seropositivity previously has been shown to be type-specific,28 to reflect the presence of viral DNA,9 and to correlate with high viral loads.17
The risk associated with HPV16 seropositivity was greatest for tumors of the pharynx where the risk was 6.0 compared to the risk associated with tumors of the oral cavity which was 1.5. Of cases, 40.3% with a tumor of the pharynx and 14.7% of those with a tumor of the oral cavity had positive serology compared to 10.7% of controls. When HPV16 serology was classified by ICD-9 site, tumors of the oropharynx were associated with the highest seroprevalence (57.7%). These data agree with a previous case–control study that found greatest risk associated with tumors of the oropharynx.7 Our study detected higher seroprevalences compared to this study, however, where 6.0% of controls, 8.9% of cases with a tumor of the oral cavity and 13.4% of cases with a tumor of the oropharynx had positive serology to HPV16.7
The dose response relationship between increasing titer and increasing risk for HNSCC was most significant for tumors of the pharynx where there was a roughly 6-fold increased risk associated with a medium titer and a roughly 29-fold increased risk associated with a high titer. The greater risk associated with HPV serology and tumors of the pharynx may reflect the fact that HPV preferentially infects the pharynx. The lymphoid tissues opposed to the oral mucosa in the tonsils and the base of the tongue may anatomically resemble the cervical mucosa. Evidence supporting the tropism of HPV for the pharynx comes from our finding that significantly more HNSCC cases of the pharynx are positive for HPV16 DNA compared to HNSCC cases of the oral cavity. Consistent with this idea is the finding that HPV16 DNA positive cases with a high viral load have been shown to be significantly associated with oropharyngeal tumor location.17 Another explanation for the high titers associated with pharyngeal tumors is that the immune response to the virus may be greater for tumors in this region due to its close proximity to lymphoid tissue.
Prior studies that have detected HPV16 DNA in tumors have suggested that an etiologic role for the virus may be implied by additional detection of an intact E6 gene,25 HPV16 E6/E7 expression25 and/or p16 overexpression.25, 29 From a subset of fresh tumor samples matched to paraffin samples in our study (150 tumors), we amplified the E6 gene from 96% of patients (44 of 46 HPV positive individuals), implying a causal role for the virus in our HNSCC patients. Previously, the presence of a wild-type E6 gene was associated with a high prevalence of E6/E7 expression while a mutated E6 gene was associated with a low prevalence of E6/E7 expression, indicating that a causal role for the virus may be defined by amplification of the wild-type E6 gene.25
Among cases, we observed that HPV16 seropositivity was associated with a 3.2-fold increased risk for having an HPV16 DNA positive tumor, with a dose response between increasing titer and the presence of DNA in the tumor. 47% of DNA positive cases had positive serology, which agrees with previous findings where roughly 50% of HPV16 DNA positive individuals mount a serologic response.9, 30
Both HPV16 DNA and HPV16 seropositivity were associated with a high number of oral sex and lifetime sex partners in cases, consistent with earlier findings that HPV is associated with sexual behavior in HNSCC patients.8, 27 The trend between HPV16 seropositivity and sexual behavior was also evident in controls. A disadvantage to studies of HPV serology is that it is impossible to identify the site of viral infection based on the presence of systemic antibodies. In most cases of HPV genital infections, the virus is spread through sexual means31 and the same may hold true for oral infections. The finding that serology presence is associated with both oral sex behavior and number of lifetime partners suggests that infections in the oral cavity may reflect infections in the genitals. This is consistent with the ideas that HPV spreads through oral sex and/or self-inoculation, specifically spread of an individual's genital infection to their own oral cavity through oral sex.
We observed that HPV16 DNA and HPV16 seropositive patients had overall better survival than HPV negative patients. Both HPV16 DNA and HPV16 seropositivity were independent predictors of survival when controlling for age, gender, drinking, smoking and tumor location. Prior investigators have suggested that the survival advantage associated with positive HPV DNA may be related to the size of the cancerous field32, 33 or to the absence of p53 mutations.6 On first examination, the association between HPV16 seropositivity and better survival seemed to reflect the finding that serology reflects the presence of DNA. However, we also observed that DNA positive patients who have positive serology have significantly improved survival compared to DNA positive patients who are negative for serology. This finding may suggest a role for HPV seropositivity in survival independent of the association between HPV seropositivity and HPV DNA.
A possible limitation of our study was the detection limit of the assay. As assay cut-offs for testing HPV seropositivity are not well standardized across labs,34 problems with standardization may account for the dose response observed between high titers and risk for HNSCC. It is also possible that HPV infections with very low viral load do not elicit an antibody response. We therefore may have failed to detect positive serology in an HPV DNA positive individual. Because of poor yield of HPV DNA in buccal samples,7 we did not study HPV16 DNA status in controls, and this may represent another limitation to our study. The sample size available for analysis of sexual history was limited because of limited patient participation, and therefore, we may have lacked statistical power to completely describe the association between sexual behavior and HPV16 seropositivity.
Our data support the emerging consensus that HPV is involved in the etiology of a subset of head and neck cancers, that the serological response reflects the presence of HPV16 DNA found in the HNSCC tumor and that risk associated with viral presence is greatest for tumors of the pharynx. A possible implication of these findings would be the diagnostic use of serum markers for HNSCC. Our findings further suggest that HPV is sexually transmitted to the oral cavity and that HPV DNA and serology positive tumors are associated with improved overall patient survival. Unique findings from our study were the dose response relationships between serology titer and both risk for HPV16 DNA in the tumor and risk for HNSCC. Our data also support a differential role of HPV16 antibodies among patients who are HPV DNA positive.
We thank Lin-quian Zhao and Henry Calderon from Harvard School of Public Health for technical assistance. We also thank Dr. Zhi-Min Yuan from Harvard School of Public Health and Dr. Karl Munger from Brigham and Womens Hospital for advice. Finally, we thank Dr. Kathrin Jansen from VaxGen for initial consult in preparation of the project.
- 1European School of Oncology Advisory report to the European Commission for the Europe Against Cancer Programme: oral carcinogenesis in Europe. Eur J Cancer B Oral Oncol 1995; 31: 75–85., , , , , , , .
- 12U.S. Food and Drug Administration. FDA licenses new vaccine for prevention of cervical cancer and other diseases in females caused by human papillomavirus. Rockville, MD: U.S. Food and Drug Administration, 2006.
- 14Vaccines for the prevention of human papillomavirus infections. Skin Ther Lett 2006; 11: 1–3., .