Oral health and human papillomavirus‐associated head and neck squamous cell carcinoma

Indicators of poor oral health, including smoking, have been associated with increased risk of head and neck squamous cell carcinoma, especially oropharyngeal squamous cell carcinoma (OPSCC), yet few studies have examined whether this association is modified by human papillomavirus (HPV) status.


INTRODUCTION
Cancers of the head and neck comprise a heterogeneous group of malignancies, of which >90% are squamous cell carcinomas (SCCs) arising from the mucosal lining of the upper aerodigestive tract. 1,2 Over the last 20 years, the epidemiology of head and neck SCC (HNSCC) has dramatically changed as a result of its association with human papillomavirus (HPV). Historically, cancers of the head and neck have been most strongly associated with tobacco and alcohol use. 3 More recently, evidence has estimated that from 60% to 70% of oropharyngeal SCCs (OPSCCs) in the United States are associated with HPV infection. 4 In the United States, the incidence of HPV-positive OPSCC increased by 225% between 1988 and 2004. 5 Compared with patients who have HPV-negative OPSCC, those who have HPV-positive OPSCC have distinct risk-factor profiles and improved oncologic outcomes. 6 In addition, HPV-positive OPSCC has a different genetic profile and a different pathway to malignancy than HPV-negative OPSCC, 7 suggesting that HPV-positive OPSCC is a different disease. Recent data from the Centers for Disease Control and Prevention estimate that >15,000 HPVassociated OPSCCs are diagnosed annually in the United States, and the oropharynx is the only head and neck cancer site to have increased in incidence. 8 Because the majority of the reported risk factors for HNSCC were established before HPV status was commonly tested, it is imperative to assess the traditional risk factors with reference to HPV status.
Previous studies have examined the association between oral hygiene, dental health, and HNSCC. Oral health indicators, including poor dental health, tooth loss, lack of routine dental care by a dentist, and a diagnosis of periodontitis, have been associated with HNSCC. [9][10][11][12][13] Two studies reported that periodontitis is associated with an increased risk of OPSCC without taking into account HPV status. 10,14 Only 2 case-only studies have examined the association between oral health and malignancy by comparing HPV-positive OPSCC with HPV-negative OPSCC. Both of those studies reported a positive association between periodontitis and HPV-positive OPSCC compared with HPV-negative OPSCC. 15,16 It has been postulated that the association between poor oral health and cancer risk may be driven by a chronic inflammatory state that alters the natural course of HPV infection, as has been demonstrated in cervical cancer. 11,12,15 In the current study, we used a large, populationbased head and neck cancer case-control study to evaluate the association between oral health, frequency of dental examinations, and HNSCC. We also explored this association by HPV status in OPSCC to determine whether the association with oral health is modified by HPV. We hypothesized that poor oral health indicators, including smoking, would be associated with an increased risk of OPSCC regardless of tumor HPV status.

Study Population
The Carolina Head and Neck Cancer Study (CHANCE) is a population-based case-control study in North Carolina. 12 Patients were eligible for CHANCE if they had been diagnosed with a first primary SCC of the oral cavity, pharynx, or larynx between January 1, 2002, and February 28,2006; were ages 20 to 80 years at diagnosis; and resided in a 46-county region in central North Carolina ("cases"). Patients who had benign tumors, carcinomas in situ, thyroid papillary carcinomas, and adenoid cystic carcinomas were excluded. Patients who had lip and hypopharynx cancers, those for whom the hospital would not release tumor blocks, and those who had completed only proxy interviews were excluded from p16 immunohistochemistry. All patients who had oropharyngeal cancers (N 5 248) and a random sample of patients who had nonoropharyngeal cancers (N 5 244; because the relevance of HPV in non-oropharyngeal cancer has not been established) 17 were selected for the evaluation of p16. A sex, age, and race frequency-matched control group was identified through the North Carolina Department of Motor Vehicle records, and those individuals were frequency-matched with cases on age, race, and sex. The study was approved by the Institutional Review Board at the University of North Carolina at Chapel Hill.

Exposure Assessment
Oral health was assessed by trained nurse-interviewers using a structured questionnaire during an in-home visit for both cases and controls. 12 Although the interview was conducted after diagnosis (average time between diagnosis to interview, 5.3 months) for cases, the questionnaire specifically asked about dental health and care before cancer diagnosis. Self-reported oral health variables included: 1) the self-reported number of natural teeth lost, excluding third molars and teeth extracted for orthodontic reasons; 2) history of self-reported tooth mobility or "teeth loose in their socket" because of disease; 3) 1 or more routine (non-emergency) dental visits during the decade before HNSCC diagnosis; and 4) gum disease diagnosed by a dentist. History of smoking, dichotomized at 10 packyears, also was included, because it is an important risk factor for poor oral health and OPSCC.

Questionnaire and Clinical Assessment
Demographic, lifestyle, diet, and other risk-factor data also were collected during the in-home interview. Confounders to be adjusted for in statistical models were selected a priori based on their potential association with survival and HPV status. In addition to the matching factors (race, sex, and age), confounders obtained from the questionnaire included education, annual income, the number of sexual partners, and alcohol consumption.
Clinical information such as tumor site was abstracted from participants' medical records and reviewed independently by a pathologist and a head neck cancer surgeon. Tumors were classified by site according to International Classification of Diseases for Oncology, third edition, topography codes for the oral cavity (C02.

HPV Status
The International Agency for Research on Cancer performed the p16 immunocytochemistry evaluation according to the protocol provided with the CINtec Histology p16 INK4a kit (9511; MTm Laboratories Inc, Westborough, Mass) for qualitative detection of the p16 expression pattern on slides prepared from formalin-fixed, paraffin-embedded tumor samples. Scores for the percentage of stained cells (with a score of 0 for 0% stained cells, 1 for 1%-10% stained cells, 2 for 11%-50% stained cells, 3 for 51%-80% Original Article stained cells, and 4 for 81%-100% stained cells) and for the intensity of nuclear or cytoplasmic staining (with a score of 0 for no staining, 1 for weak staining, 2 for moderate staining, and 4 for strong staining) were multiplied to yield a composite score ranging from 0 to 12. Scores equal 4 were considered positive for p16 expression. Because p16 is the most commonly used clinical marker, tumors that exhibited p16 protein expression were considered HPVpositive. 18 HPV infection also was determined through DNA extraction and genotyping using Luminex-based multiplex polymerase chain reaction (PCR) (TS-E7-MPG, IARC, Lyon, France) for the following genotypes: HPV type 6 (HPV6), HPV8, HPV11 HPV16, HPV18, HPV26, HPV31, HPV33, HPV35, HPV39, HPV58, and HPV59. [19][20][21] For sensitivity analyses, we also considered tumors to be HPV-positive only if they were positive for both HPV PCR and p16 protein expression.

Statistical Analysis
Differences in descriptive statistics according to p16 status were estimated using the chi-square test. HNSCC sitespecific adjusted odds ratios (ORs) and p16-positive and p16-negative, OPSCC-specific ORs were estimated with unconditional multinomial logistic regression models comparing patients with p16-positive OPSCC versus controls and comparing patients with p16-negative OPSCC versus controls, adjusting for the study matching factors (age, race, and sex) as well as important confounders related to alcohol use and socioeconomic status (such as income, insurance, and education). On the basis of evidence indicating that women are at increased risk from the carcinogenic effects of tobacco and literature demonstrating a differential inflammatory response from cigarette smoking among women compared with men, [22][23][24][25] we explored multiplicative interactions of sex with all oral health variables (routine dental examination, tooth mobility, gum disease, number of lost teeth) and smoking. Because sex was a matching factor, we were unable to estimate additive interactions with the relative excess risk because of interaction. All statistical analyses were implemented using the SAS version 9.4 statistical software package (SAS Institute Inc, Cary, NC), and an a value of 0.05 was used.

Descriptive Statistics
Most oral health variables differed by tumor p16 status (Table 1). Controls were more likely than both p16positive and p16-negative patients to have markers of good oral health. Patients with p16-negative tumors were less likely to have undergone a routine dental examination in the last 10 years (P < .001) and were more likely to have lost teeth (P 5 .001) and tooth mobility (P 5 .030) than those with p16-positive tumors. The prevalence of gum disease did not differ substantially between patients with p16-positive and p16-negative OPSCC. Smoking for 10 pack-years was most common among patients who had p16-negative tumors (82.9%), followed by those with p16-positive tumors (63.1%), and controls (44.2%).

All HNSCC Sites
A history of routine dental examinations was significantly associated with decreased risk across all sites except the hypopharynx, which had a reduced OR similar in magnitude to that for the other sites (Table 2). We observed that tooth mobility because of disease increased the risk of cancer across all sites compared with controls. The associations for sites in the larynx, oral cavity, and oropharynx were statistically significant. A report of gum disease was not significantly associated with any sites of HNSCC.
Smoking for 10 pack-years was significantly associated with an increased risk of HNSCC across all sites.
In the sensitivity analysis, 4.3% of our patients with p16-positve OPSCC did not have high-risk HPV DNA identified by PCR (n 5 7). When considering only tumors that were both HPV-positive by PCR and positive for p16 protein expression, the point estimates remained unchanged. However, smoking was no longer significantly associated with HPV-positive OPSCC (OR, 1.47; 95% CI, 0.95-2.26; P 5 .082).

Sex Interaction
There was little evidence of an interaction of sex with dental examinations, tooth mobility because of disease, gum Cancer disease, and the number of teeth lost with in patients with either p16-positive or p16-negative OPSCC (Table 4). We observed evidence of an interaction between smoking and sex with p16-positive OPSCC, in which women had an increased risk of HPV-positive OPSCC when their smoking history was 10 pack-years, but this association was not observed in men, and the estimates were imprecise.

DISCUSSION
The objective of the current study was to provide insights into the correlation between oral health, the frequency of dental examinations, and HNSCC stratified according to disease site and HPV status in patients with OPSCC using data from the CHANCE study. Our findings indicate that routine oral examinations are associated with a decreased risk of both HPV-positive and HPV-negative OPSCC. Furthermore, tooth mobility was associated with a significantly increased risk of HPV-negative OPSCC. Tooth mobility was non-significantly associated with increased risk of HPV-positive OPSCC. Gum disease diagnosed by a dentist was not associated with either HPV-positive or HPV-negative OPSCC. Although HPV-positive and HPV-negative HNSCC differ significantly with respect to etiology and tumorigenesis, our findings suggest that poor oral health and frequency of dental examinations similarly impact the risk of HPVpositive and HPV-negative HNSCC. The association between oral health and tobaccoassociated HNSCC is well established, and the current  Cancer study replicates these results. [11][12][13] Although those studies did not take HPV status into consideration, they did identify a positive association between poor oral health and HNSCC in sites not commonly associated with HPV, such as the larynx and hypopharynx. [11][12][13] Our study is one of the first to demonstrate an association between poor oral health indicators and the risk of OPSCC in HPVnegative patients. Periodontitis is a disease typified by bacterially induced, chronic inflammation, most often associated with Gram-negative anaerobic rods 26,27 ; it is plausible that the increased risk of HPV-negative OPSCC is caused by the microbial dysbiosis and chronic inflammatory state associated with periodontitis and poor oral health in general. 28,29 The association between oral health and the risk of HPV-positive OPSCC has not been studied as extensively. Tezal and colleagues examined the association between HPV status and periodontitis in a small sample of 30 patients who had base of tongue cancers and observed a 3fold increased risk (OR, 3.96; 95% CI 1.18-13.36) of HPV-positive tumor status for every 1 mm of alveolar bone loss compared with HPV-negative tumor status. 16 In a later study published by Tezal and colleagues that had a larger sample size and included all head and neck sites (N 5 124), those authors demonstrated a similar trend but a weaker association (OR, 2.61; 95% CI, 1.58-4.30) with HPV-positive versus HPV-negative head and neck tumors. 15 Although we did not specifically assess periodontitis in our current study, we used tooth mobility caused by disease and report of gum disease as proxies for periodontitis. Tooth mobility is a result of alveolar bone destruction and loss of periodontal attachment, which are associated with periodontitis; furthermore, the assessment of self-reported or clinically determined tooth mobility is commonly used in the periodontal assessment and has been used in previous studies as a marker for periodontitis. 12,30 The results from our study, which had a much larger sample size of patients with OPSCC (N 5 372), do not support the strong association between periodontitis indicators and HPV-positive OPSCC reported by Tezal and colleagues. The weaker associations we observed may have been the result of using tooth mobility as a proxy for a clinically determined diagnosis of periodontitis. A previous study indicated that patient self-report of gum disease was only moderately correlated with periodontitis, 31 which could explain the null association observed with gum disease. However, Hashim and colleagues observed no association with gum disease in a pooled analysis of 1855 patients who had oropharyngeal cancer and 7939 controls, suggesting that there may not be an association between self-reported gum disease and oropharyngeal cancer. 32 Oral HPV infection is necessary for the development of HPV-positive OPSCC. Poor oral health can affect cancer development by increasing either the risk of HPV infection or the carcinogenicity of HPV. Previous studies have demonstrated a correlation between oral HPV infection and poor oral health. 33,34 Research on HPV-associated cervical cancer indicates that coinfections with bacterial species like Chlamydia and HPV exhibit synergistic effects and result in an increased risk of cervical cancer. 35 The biologic pathways underlying this association may involve increased levels of inflammatory cytokines, such as interleukins and tumor necrosis factor-a, that modulate HPV gene expression. 36,37 Because periodontitis is a disease characterized by a polymicrobial dysbiosis, similar mechanisms between the inflammatory cascade, HPV gene expression, and cancer risk may explain the associations demonstrated between HPV status and periodontitis. Furthermore, a similar mechanism may also play a role in the pattern observed in our study between HPV-positive OPSCC and poor oral health in general. Future studies to elucidate the mechanisms behind these associations with HPV are warranted.
We also observed an association between the risk of OPSCC and smoking in both HPV-positive and HPVnegative patients with OPSCC. Although most HPVpositive cancers involve some form of tobacco use, this association is often less pronounced than that in HPVnegative cancers. 1,38 This was confirmed in our study, in which smoking was more strongly associated with the risk of HPV-negative than HPV-positive OPSCC. The correlation between oral HPV infection and smoking is well established. 39,40 Because smoking is farther upstream in the tumorigenesis pathway (ie, smoking increases the risk of oral HPV infection, which leads to HPV-positive oropharyngeal cancer), 41 a diminished association between smoking and HPV-positive OPSCC is expected. In an exploratory analysis, we examined the interaction between smoking and sex. Women who smoked appeared to be at higher risk of HPV-positive OPSCC compared with women who were nonsmokers, but this correlation was not observed in men. However, there were very few women with HPV-positive OPSCC (n 5 28) in our study, and these sparse data produced imprecise estimates.
There are a few limitations to our study. All oral health variables were self-reported and thus may be considered as less valid oral health indicators than clinically diagnosed disease. 42 However, previous studies have also reported a high correlation between self-reported tooth loss and a clinical examination. 43 The current study is in Original Article agreement with previous work demonstrating that routine dental visits are associated with a decreased risk of developing head and neck cancers. Although our study was large, we were unable to replicate the results for HPVpositive base of tongue cancer reported by Tezal and colleagues, because some of our strata were very sparse. Further studies are needed in this area to clarify potential associations and effect sizes. Although we adjusted for smoking in the model, we did not have adequate power to further examine the interaction of oral health and smoking with HPV status.
It is noteworthy that oral health, frequency of dental examinations, and HNSCC have been strongly associated with socioeconomic factors and risk behaviors. 44,45 Although we included indicators of socioeconomic status, such as number of sexual partners, education, annual household income, and insurance status, in our final adjusted model, there still is potential for residual confounding. However, this inverse association may be related to routine oral examinations and interventions that facilitate a healthier oral ecology and microbiome with less pathogenic microflora and lower levels of inflammation.
This study has several notable strengths. We used the CHANCE study, which is a large, population-based, casecontrol study with a diverse population. We were able to ascertain detailed information on smoking, oral health indicators, and demographics from interviews conducted by trained nurses. In addition, the current study is the largest study to date of oral health and the risk of OPSCC stratified according to HPV status using p16 immunohistochemistry, which is the most commonly used clinical marker of HPV status, as well as HPV status through PCR.
In conclusion, in this population-based, case-control study we observed a modest positive association between oral health indicators and the risk of both HPV-positive and HPV-negative OPSCC. Routine dental visits cut the risk of both types of OPSCC almost in half, whereas smoking appeared to have a weaker association with HPV-positive OPSCC than HPV-negative OPSCC. These findings underscore the importance of oral health surveillance and routine dental examinations for the prevention of HNSCC regardless of HPV status. Further research into the correlation between oral health, HPV infection, and risk of OPSCC is warranted to clarify possible mechanisms and optimize prevention strategies.

FUNDING SUPPORT
This study was supported in part by the National Cancer Institute (R01 CA90731) and by the Lineberger Cancer Control Education Program (R25 CA57726).