Oral Human Papillomavirus in Healthy Individuals: A Systematic Review of the Literature

Background: Human papillomavirus type 16 (HPV16) is a common infection in the anogenital tract. HPV16 DNA detected in oral specimens has recently been identified as a risk factor for some oropharyngeal cancers. The reported prevalence of oral HPV infection from individual studies is highly variable. Methods: We systematically reviewed and abstracted data from published studies (n = 18) that detected oral HPV DNA in 4581 cancer-free subjects to determine the pooled prevalence (and 95% confidence intervals [CI]) of HPV16, carcinogenic HPV, and any HPV. Results: 1.3% (95% CI: 1.0–1.7%) of 3977 healthy subjects had oral HPV16, 3.5% (95% CI: 3.0–4.1) of 4441 subjects had carcinogenic HPV, and 4.5% (95% CI: 3.9–5.1) of 4070 subjects were positive for any HPV. Oral HPV16 accounted for 28% of all HPV detected in the oral region. Men (47 of 1017) and women (117 of 3690) had nearly exactly the same prevalence of any oral HPV detected (4.6% vs. 4.4%, respectively). Conclusions: HPV-16, a common anogenital infection, was rarely detected in oral specimens. However, a small but noteworthy proportion of healthy individuals have oral HPV infections with types known to cause cancer in the oral region.

I nfection with human papillomavirus (HPV) is a necessary cause of cervical cancers worldwide, and is increasingly recognized as an important cause of a subset of other anogenital cancers, including some vaginal, vulvar, penile, and anal cancers. 1 HPV is also associated with a subset of head and neck cancers occurring specifically in the oropharynx. 1 HPV type 16 (HPV16), the most prevalent of the approximately 13 cancerassociated types in cervical cancer, is also the most common type present in HPV-positive oropharyngeal cancers (of which, approximately 90% are positive for HPV16). 2,3 Recent case-control data suggest that detection of prevalent HPV16 infection in oral exfoliated cells increased the odds of oropharyngeal cancer more than 13 fold. 3 There is limited information about the natural history of oral HPV infection, but since oral HPV16 infection is associated with this cancer, it is important to estimate the proportion of healthy individuals with oral HPV infection. The aim of this work was to review existent literature to estimate the prevalence of HPV DNA detected in oral specimens collected from cancer-free individuals.

Study Selection
The National Institutes of Health "PubMed" search engine was employed to search for citations published from January 1997 to June 2009, using the keywords "Papillomavirus" and "Oral" and limiting to English publications in humans. Using these terms, we identified 729 manuscripts. After reviewing their titles and abstracts for relevance, we identified 47 studies that appeared to evaluate HPV DNA in oral specimens collected from healthy individuals. We excluded studies that focused on individuals with precancerous lesions (n ϭ 12) and studies of immunosuppressed populations (renal transplant or HIV-positive) (n ϭ 1); although if a study included HIVnegative and HIV-positive individuals and HPV data were provided stratified by group, we utilized available data. We also excluded case-control studies, commentaries and systematic reviews (n ϭ 11), studies that focused exclusively on infants/ children (n ϭ 2), studies with HPV test results on less than 50 people (n ϭ 4), and studies that did not use DNA-based testing methods on oral specimens (i.e., serologic assays and secretory immunoglobulin) (n ϭ 1). Using these criteria, 16 manuscripts were identified; two additional manuscripts were identified from manuscript references for a total of 18 manuscripts included in the final data abstraction. Informed consent was assumed to have been obtained by the individual studies from their respective subjects. HPV16 alone were measured only among those tested for the specific HPV type in question, and therefore, the sample size and number of contributing studies varied between analyses. Composite estimates (percent prevalence) and exact binomial 95% confidence intervals (CI) were calculated. To begin to look at oral HPV by world region, we stratified by economic status using the definition put forth by the United Nations Public Administration Network. 22 The following countries or regions were considered developed: Europe (excluding the east European transition economies), Canada, the United States of America, Japan, Australia, and New Zealand.
Categorizations were also made based on the study population to determine if the 'healthiness' of the population influenced the findings. Specifically, healthier populations "low-risk," such as those from dental clinics and the general population, were compared to higher-risk groups "high-risk," such as female sex workers. An unknown risk category was used for studies with mixed populations or for those whose 'riskiness' could not be determined.
Evaluating the age-specific oral HPV distribution and comparing oral HPV prevalence among men and women were aims of this work. While most of the studies provided data on any HPV prevalence stratified by gender, many did not provide data in a manner that could be used for the age-analysis. The raw data from individual studies is displayed in tabular format (Table 1) and a gender-stratified analysis of any HPV was conducted.
Although understanding the influences of methods of specimen collection and laboratory analysis (specifically DNA extraction and HPV testing methods) is important for comparing results between studies, such methods work is best conducted within individual studies using head-to-head comparisons, as have been previously published. 10,23 The relationship between study sample size and oral HPV prevalence was evaluated. We tested for heterogeneity between studies using the I 2 statistic, which represents the approximate proportion of total variability in point estimates that can be attributed to heterogeneity. 24 Formal testing using both the Begg and Mazumdar 25 and Egger et al 26 tests indicated publication bias was present (P ϭ 0.04 and P Ͻ 0.001, respectively).

RESULTS
Eighteen studies including 4581 healthy individuals were identified and are presented in the Studies that occurred in developing nations (n ϭ 4) had a greater prevalence of HPV16 compared to those from developed nations (n ϭ 12) (4.3% vs. 0.7%, respectively); similar patterns were observed for carcinogenic and any HPV detected ( Table 2).
By risk strata, studies (n ϭ 6) that recruited lower-risk populations, such as people from the general population or college students, had a lower HPV16 prevalence (0.4%) compared to the unknown (n ϭ 5) and higher (n ϭ 2) risk categories (4.1% vs. 2.1%, respectively); a similar trend was observed for any HPV. An anomaly was noted for carcinogenic HPV, where low and high risk populations had similar prevalence of  do Sacramento et al, 7 Giraldo et al, 9 Kreimer et al, 10 Montaldo et al, 15 and Summersgill et al. 20 ‡ Low-risk includes publications by D'Souza et al, 23 Kujan et al, 11 Kurose et al, 12 Lambropoulous et al, 13 Marais et al, 14 Ragin et al, 16 Rintala et al, 17 Smith et al, 18 Smith et al, 19 and Winer et al. 21 95% CI indicates 95 percent confidence interval.

Oral Human Papillomavirus in Healthy Individuals
Sexually Transmitted Diseases • Volume 37, Number 6, June 2010 oral HPV (2.7% vs. 2.6% respectively). For all categorizations of HPV, studies of individuals of unknown risk had the highest oral HPV prevalence ( Table 2). Thirteen studies provided oral HPV data stratified by gender. Men (47 of 1017) and women (117 of 3690) had similar prevalence of any oral HPV detected (4.6% vs. 4.4%, respectively).
The relationship between any HPV prevalence and study size was assessed (Fig. 1). The majority of studies had around 100 to 200 individuals and showed broad variation in detection of any HPV (ranging from 2.6%-20.7%). The long right tail showed that six studies included over 200 people; the largest study had a sample size of 662 people. These larger studies tended to show overall HPV prevalence lower than the average (Fig. 1). By economic status, the few studies that occurred in developing countries tended to be smaller (all but one had less than 200 individuals).
Significant heterogeneity was present between studies: I 2 for HPV16 of 89.3%, for carcinogenic HPV of 92.1%, and for overall HPV of 88.7% (all P Ͻ 0.001).

DISCUSSION
HPV16 infection has been established as a cause for a subset of oropharyngeal cancers 1,3 ; therefore, quantitating the prevalence of oral HPV16 among healthy individuals is important. The present review included 18 studies that investigated HPV DNA, as measured by PCR-based assays, in over 4500 oral specimens collected from healthy individuals: more than one percent had HPV16 detected in their oral region. HPV16 accounted for 28% of all HPV infections detected in the oral region amongst individuals. Correspondingly, HPV16 accounts for approximately 25% of cervical infections detected among cytologically normal woman. 27 Paradoxically, the proportion of HPV16 to all HPV types detected appears similar in the oral and cervical regions yet HPV16 causes a greater proportion of HPV-associated oropharyngeal cancers (approximately 90%) 2,3 compared to cervical cancers (approximately 50%). 28 While detection of oral HPV DNA does not translate to productive HPV infection, these data argue for large, well-designed studies that aim to understand the natural history and epidemiology of oral HPV infection. Given that HPV persistence is necessary for cervical cancer development, 29 it is likely important for the oral region as well. Further, as differences were observed in the prevalence between nations with differing economic statuses, inclusion and comparison of multiple countries within one study will help determine if these differences are real or due to differences in study methodology.
To obtain an estimate of the burden of oral HPV infection in the general population, we focused on studies of healthy adults and chose to exclude certain populations. For instance, healthy controls from case-control studies were excluded because more than half of the control populations from these studies were recruited from either a hospital or clinic based setting and were typically matched to the case-distribution by at least age and gender, thereby potentially skewing results towards older ages and male gender, given the distribution of head and neck cancers. Additionally, studies conducted exclusively among children were excluded as the route of transmission is in most instances nonsexual, and the prevalence may accordingly differ as well. While our work is not necessarily generalizable to the population at large, especially considering the inclusion of higher-risk populations such as female sex workers and STD clinic attendees, we wanted to avoid further skewing the work in known ways. Lastly, because we restricted to healthy adults, we excluded populations at increased risk of infection, including HIV-positive, renal transplant patients, and cancer patients. Each of these special populations is of interest and warrants further investigation.
The methods employed in these studies were heterogeneous. Sources of variability between studies that we were unable to address with the current review included method of specimen collection and processing methods. In fact, inadequate sample purification due to PCR inhibition has been shown to importantly underestimate the prevalence of oral HPV infection. 23 In addition, studies of oral HPV among cancer-free people have been, with rare exception, small (less than 200 cases); the HPV prevalence seemed to be inversely proportional to the study sample size. Hence, this work appeared to be susceptible to publication bias, in which certain studies may have been published because of higher oral HPV prevalence.
We determined from existing studies that a small but significant proportion of healthy individuals have HPV types detected in their oral region that are known to cause cancer in the oral region. While the fraction of these infections in healthy people that will lead to cancer will be small, HPV16 appears to cause approximately 40% to 60% of oropharynx cancers. 3,30 There theoretically exists the possibility that the prophylactic HPV16 vaccine could be protective in this context. While a direct evaluation of vaccine efficacy against oral HPV16 infection remains necessary because there are no published data on this topic, if the HPV vaccine protects against oral HPV infections akin to the near complete efficacy observed for cervical infections among HPV naïve women, 31,32 a percentage of oropharynx cancers could be prevented. Little is known about transmission and natural history of oral HPV infections, but prophylactic vaccines directed against oral HPV16 would need to be administered before exposure and would need to provide long-term protection. However, before any programmatic vaccine implementation, cost-effectiveness analyses are necessary and should aid in the decision of whether to promote vaccination to protect against some oropharyngeal cancers.
Prevention of even a subset of head and neck cancers is important, and is highlighted by recent data suggesting that the incidence of oropharynx cancer is increasing and may be due to changes in HPV endemicity. 33 Prospective, epidemiologic studies that focus on the age-specific and type-specific oral HPV distribution, oral HPV prevalence by gender, and evaluation of rates of oral HPV incidence and persistence will be an important next step in understanding the natural history of oral