|Home | About | Journals | Submit | Contact Us | Français|
Human papillomavirus (HPV) is an etiologic agent for both oropharyngeal and cervical cancers, yet little is known about the interrelationship between oral and cervical HPV infections. Therefore, we compared the prevalences and type distributions of oral and cervical HPV infections and evaluated infection concordance in a cross-sectional study within the Women's Interagency HIV Study cohort. Oral rinse and cervical-vaginal lavage samples were concurrently collected from a convenience sample of 172 human immunodeficiency virus (HIV)-positive and 86 HIV-negative women. HPV genomic DNA was detected by PGMY09/11 L1 consensus primer PCR and type specified by reverse line blot hybridization for 37 HPV types and β-globin. Only 26 of the 35 HPV types found to infect the cervix were also found within the oral cavity, and the type distribution for oral HPV infections appeared distinct from that for cervical infections (P < 0.001). Oral HPV infections were less common than cervical infections for both HIV-positive (25.2% versus 76.9%, P < 0.001) and HIV-negative (9.0% versus 44.9%, P < 0.001) women. Oral HPV infections were more common among women with a cervical HPV infection than those without a cervical HPV infection (25.5% versus 7.9%, P = 0.002). The majority of women (207; 93.7%) did not have simultaneous oral and cervical infections by the same HPV type; however, the number of women who did (14; 6.3%) was significantly greater than would be expected by chance (P = 0.0002). Therefore, the oral and cervical reservoirs for HPV infection are likely not entirely independent of one another.
Recent molecular and epidemiological studies substantiate a role for human papillomavirus (HPV) in the etiology of oropharyngeal cancers (11, 31). High-risk HPV, predominantly type 16, has been consistently detected in a distinct subset of these cancers (11). Oral HPV infection has been associated with oropharyngeal cancer risk in case-control studies (33). Furthermore, a temporal association between exposure and risk has been supported by the strong increase in risk for incident oropharyngeal cancer among HPV16-seropositive individuals (25).
Despite the recent recognition of an HPV-associated oral malignancy, little is known about the epidemiology of oral HPV infection. It is unclear to what extent it is possible to extrapolate data from cervical HPV infection to oral HPV infection. Initial studies indicate that oral HPV infection, analogously to cervical infection, is associated with sexual behavior and immunosuppression (8, 22). However, some characteristics of oral HPV infection prevalence appear distinct from cervical infection, such as associations with age. Preliminary data suggest that the prevalence of oral HPV infection among high-risk, human immunodeficiency virus (HIV)-negative individuals either increases steadily (22) or remains unchanged (5) as a function of age, whereas a bimodal curve is observed for those with anogenital infection (7). Valid comparisons between studies of oral and anogenital infection performed to date are limited, however, by differences in sampling, HPV detection methods, and risk behaviors among study populations. In the few studies in which oral and anogenital HPV infections were analyzed either concurrently (1, 5) or sequentially (21) in the same study population, oral HPV infection prevalence appeared lower than and poorly type concordant with anogenital infection in high-risk sex workers (5) and in women with a history of cervical HPV infection (21). However, oral and cervical HPV prevalences were similar in a small group with high prevalence of oral or anogenital condylomata (1). With the exception of these few studies in very-high-risk groups, the relationship of oral infection to cervical HPV infection remains unexplored.
Several aspects of the relationship of oral HPV infection to cervical HPV infection remain poorly described. These include differences by anatomic site in HPV prevalence and HPV type distribution. Also unknown are the prevalences of isolated and concomitant oral or cervical infections and whether concomitant infections are type concordant. This study was therefore designed to evaluate these factors by performing concurrent oral and cervical sampling in a well-studied longitudinal cohort of HIV-positive and risk-matched HIV-negative women, the Women's Interagency HIV Study (WIHS) cohort (2). This population was specifically chosen because of the higher-than-expected risk for oropharyngeal cancer (10) and for oral and tonsillar HPV infections previously reported among individuals with HIV infection (8, 22). To enhance the soundness of the comparison, we chose to sample the oral region with a validated oral rinse sampling method (9) that is somewhat analogous to the cervical-vaginal lavage sampling performed for the WIHS cohort and to process the sample similarly. Concurrent evaluation of oral and anogenital HPV concordance will allow us to begin to explore differences and similarities between oral and anogenital HPV infections and how they may be interrelated.
A convenience sample of individuals from five WIHS sites (Chicago, Ill.; San Francisco, Calif.; Brooklyn, N.Y.; Bronx, N.Y.; and the District of Columbia), excluding the Los Angeles, Calif., site, were recruited to participate in a study of oral HPV infection. WIHS core study visits occur every six months, and the samples for this cross-sectional study were obtained between August and December 2004, which was during WIHS visits 20 and 21. A detailed description of the WIHS study population and methods is published elsewhere (2). Eligibility criteria for this convenience sample were WIHS participation and the ability to provide informed consent. The research protocol was approved by the Institutional Review Board of the Johns Hopkins Hospital and all participating WIHS sites. Written informed consent was obtained from all participants.
In addition to our utilizing the previously established WIHS data collection, each participant completed a self-administered questionnaire focused on sexual risk behaviors, including recent (≤6 months) and lifetime sexual behaviors and history of sexually transmitted infection (other than HIV). Oral rinse samples (ORSs) were collected as previously described (9). Briefly, each participant vigorously swished and gargled 10 ml of Scope mouthwash for 30 seconds and expectorated into a specimen collection cup. ORSs were stored at 4°C for a maximum of seven days prior to shipment on ice to the Gillison virology laboratory at Johns Hopkins Medical Institutions for further processing. Samples were pelleted by centrifugation, washed, resuspended in phosphate-buffered saline, and stored at −80°C. A cervical-vaginal lavage sample (CVL) was collected from each participant during the same visit as previously described (24) and stored at −80°C. A one-milliliter aliquot of CVL was shipped on dry ice to the virology laboratory for further analysis. As a control for specimen collection and processing, a 10-ml sample of stock Scope was collected at each participating site after enrollment of each block of 20 subjects and was processed in a manner identical to that for the ORSs.
DNA was purified from all ORSs and CVLs, as previously described (9). Briefly, an aliquot of ORS (1.5 ml) or CVL (0.5 ml) was pelleted by centrifugation and resuspended in 1.0 ml Puregene cell lysis solution and incubated at 37°C for 15 min. The sample was digested with DNase-free RNase A (5 μg/ml) for 30 min at 37°C. Proteinase K (Sigma-Aldrich, St. Louis, MO) was added to a final concentration of 0.5 mg/ml, and digestion was performed overnight at 55°C, followed by heat inactivation at 95°C for 10 min. Samples were cooled to room temperature, protein precipitation solution (340 μl) was added, and samples were further processed per the manufacturer's protocol for DNA purification from buccal cells in mouthwash (Puregene DNA purification kit; Gentra Systems, Minneapolis, MN). DNA was resuspended in 100 μl of DNA hydration solution and stored at −80°C until further analysis.
Purified DNA (10 μl of a 100-μl sample) from ORSs and CVLs was analyzed for the presence of HPV genomic DNA by PCR amplification with the PGMY09/11 L1 consensus primer system, as previously described (14, 15). Coamplification of the β-globin gene was performed as a positive control for the presence of amplifiable DNA in the specimen. All reactions were performed in a 96-well-plate format in an Applied Biosystems (Foster City, CA) 9600 thermal cycler, with water blanks included as negative controls in each row of the plate. Positive controls with known high (~100)- and low (~10)-viral-copy-number HPV16 and HPV18 were included on each plate. For type specification, the PCR product was hybridized to an HPV probe array for genotyping of 37 HPV types and β-globin (Roche Molecular Systems, Inc., Alameda, California) (13, 28). Samples that were β-globin negative were considered to be of insufficient quality for analysis. Samples were reported as positive or negative for HPV genomic DNA, and the HPV types detected were reported for positive samples. HPV types iso39 and cp6108 were considered to be isoforms of HPV type 89.
The demographic characteristics of the study population were summarized using means and confidence intervals (CIs) for continuous variables and counts and percentages for categorical variables. Comparisons between the HIV-positive and -negative subjects were made by use of a chi-square test for categorical variables, Student's t test for continuous variables, or the Mann-Whitney test for equality of medians as appropriate. HPV DNA detection results, stratified by sample type (ORS or CVL), HIV serostatus, and HPV type classification (high and low risk), were reported (26). Individuals coinfected with high- and low-risk types were included in each category. HPV DNA prevalence estimates and exact 95% CIs were calculated for all evaluable ORSs and CVLs and then restricted to paired samples. McNemar's test was used to compare infection prevalences in paired ORSs and CVLs, stratified by HIV serostatus. Prevalence ratios and 95% CIs were used to compare prevalence estimates for HIV-positive and -negative women. The HPV type distributions in the oral cavities and cervices were compared by using a chi-square test of distribution.
The analysis of concordance was restricted to subjects with paired ORSs and CVLs. Samples were considered to be paired if the ORS and CVL were concurrently collected and suitable for laboratory analysis. An infection was considered concordant if the same HPV type was detected in the ORS and CVL. To determine whether the agreement of oral and cervical infections was greater than would be expected by chance, a permutation analysis was performed. The null hypothesis was that type-specific infections detected in both the oral cavity and cervix of one woman occurred due to chance. We assumed that HPV infections of each anatomic site were independent events and that the number of infections at each site was constant for each individual, because risk of infection is related to individual characteristics (e.g., sexual activity and HPV status). Infection types were randomly permuted among individuals for each anatomic site, keeping the marginal distribution of the frequency of each HPV type constant. After each permutation, the number of type-specific matches between the oral and cervical cavity, as well as the number of individuals with a match, was computed. The process was repeated 10,000 times, and the P values were generated by counting how often the simulated counts fell below the observed counts.
All P values reported are two sided and considered significant if they are <0.05. Data analysis was performed by use of Intercooled STATA, version 8.0 (StataCorp, College Station, Texas); however, R (freeware version of S-plus) was used for the permutation analysis.
A total of 258 women were enrolled in this study of oral and cervical HPV infection concordance during WIHS visits 20 and 21, from August through December 2004. This included 172 (66.7%) HIV-positive women and 86 (33.3%) HIV-negative women.
Of 258 enrolled women, ORSs were collected from 255 (98.8%) and CVLs were collected from 235 (91%). Eleven (4.3%) of 255 ORSs were discarded because of damage during shipping. Three (1.2%) of 244 ORSs and 1 (0.4%) of 235 CVLs were β-globin negative by PCR and were excluded from analysis. Therefore, 241 ORSs and 234 CVLs were available for analysis, and 221 individuals provided paired, concurrently collected samples.
Negative and positive controls for specimen collection, processing, and analysis were tested for HPV and β-globin; all were appropriately negative or positive, respectively.
The characteristics of the HIV-positive and -negative women enrolled in this study with paired ORSs and CVLs are compared in Table Table1.1. HIV-positive and -negative women did not differ by median age (41 versus 40 years; P = 0.09). The subgroups were similar with regard to age, race, ethnicity, education, number of lifetime sexual partners, recent oral sexual history, recent sexual partners, and age at coitarche (Table (Table1).1). Despite these similarities, HIV-positive women were more likely than HIV-negative women to have a history of sexually transmitted infection (P < 0.001), genital warts (P < 0.001), or an abnormal Pap smear reported during a previous WIHS visit (P < 0.001). HIV-positive women were more likely than HIV-negative women to use barrier contraception during vaginal intercourse (P < 0.001) or when performing oral sex (P = 0.02) in the previous six-month period.
Among all enrolled women who provided an evaluable ORS (n = 241), approximately 14% (95% CI, 8.9 to 20.3) of HIV-positive women had an oral high-risk HPV infection and 15.8% (95% CI, 10.5 to 22.5) had a low-risk infection. High-risk oral HPV infections were detected in 3.6% (95% CI, 0.7 to 10.2) of the HIV-negative women, and low-risk oral HPV infections were detected in 6% (95% CI, 2.0 to 13.5) of the HIV-negative women.
When analysis was restricted to the 221 women with paired specimens from oral and cervical sites, oral HPV infection was found to be less common than cervical infection, regardless of HIV serostatus (P < 0.001) (Table (Table2).2). Among HIV-positive women, 25.2% (95% CI, 18.3 to 33.1) had an oral HPV infection and 76.9% (95% CI, 69.1 to 83.6) had a cervical HPV infection. By contrast, 9.0% (95% CI, 3.7 to 17.6) of HIV-negative women had an oral HPV infection and 44.9% (95% CI, 33.6 to 56.6) had a cervical HPV infection.
HIV-positive women were significantly more likely to have an oral HPV infection than HIV-negative women (25.2% versus 9.0%, P < 0.001; prevalence ratio [PR], 2.8 [95% CI, 1.3 to 6.0]). Cervical infections were also more prevalent among HIV-positive women (Table (Table2).2). High-risk HPV infections were more common in the oral cavities and cervices of HIV-positive women than in those of HIV-negative women. Similarly, multiple concurrent oral or cervical infections were more common among HIV-positive women (Table (Table22).
A total of 69 oral HPV infections were detected in all evaluable samples, including 39 (57%) low-risk infections and 30 (43%) high-risk infections. Twenty-six of the 37 HPV types present on the Roche prototype line blot were detected in the oral cavity. HPV45 and HPV83 were the most prevalent high-risk oral infections (5.8% for both; n = 4). High-risk types 51, 52, 53, 67, 68, and 73 and low-risk types 6, 40, 54, 57, and 64 were not detected in the oral cavity.
The number of cervical infections was substantially higher than the number of oral infections, in part due to the number of multiple infections detected in cervices being greater than the number of infections in oral cavities (Table (Table2).2). A total of 479 cervical HPV infections, including 245 (51%) low-risk infections and 224 (49%) high-risk infections, were detected. In contrast to the oral cavity, 35 of the 37 HPV types specified by the Roche prototype line blot were detected in cervical samples. The most prevalent high-risk HPV type detected in the cervix was type 83 (15%; n = 34). Only high-risk type 69 and low-risk type 64 were not represented in the cervical infections detected.
The type-specific prevalences of oral and cervical HPV infections are compared among paired samples in Table Table3.3. The prevalence of HPV infection by type appeared to be consistently lower in the oral cavity than in the cervix, with the exceptions of the prevalences of types 26 and 69. Type-specific differences in prevalence between oral and cervical sites were not evaluated because of the low prevalence estimates for most types.
The type distributions of the HPV infections that were detected in the oral cavity and cervix are compared in Fig. Fig.1A.1A. Ten HPV types were detected in the cervix and not in the oral cavity, and HPV69 was found in the oral cavity but not in the cervix. The type distribution of HPV infections in the oral cavity appeared to be distinct from that of infections in the cervix (P < 0.001).
To understand the relationship of oral HPV infection to cervical HPV infection, the prevalences of concomitant or isolated oral and cervical infections were investigated in the study population. This analysis was restricted to women with paired oral and cervical samples (n = 221). Oral infections were more common among women with a cervical HPV infection than among those without a cervical infection (25.5% versus 7.9%, P = 0.002; PR, 3.2 [95% CI, 1.4 to 7.3]). Six (2.7%) of 221 women had only an oral infection(s), 37 (16.7%) had both oral and cervical infections, 108 (48.9%) had only a cervical infection(s), and 70 (31.7%) of 221 women had neither an oral HPV infection nor a cervical HPV infection.
Isolated oral HPV infections were not more common among HIV-positive women than among HIV-negative women (2.8% versus 2.6%, respectively, P = 0.9; PR, 1.1 [95% CI, 0.20 to 5.8]). However, HIV-positive women were more likely than HIV-negative women to have either an oral or cervical HPV infection (79.7% versus 47.4%, P < 0.001; PR, 1.7 [95% CI, 1.3 to 2.2]), an isolated cervical infection (54.5% versus 38.5%, P = 0.02; PR, 1.4 [95% CI, 1.03 to 1.9]), or concomitant oral and cervical infections (22.3% versus 6.4%, P = 0.002; PR, 3.5 [95% CI, 1.4 to 8.6]).
We evaluated whether the number of type-specific HPV infections that were simultaneously present in the oral cavity and cervix of the same woman was greater than would be expected by chance. Among the 221 women with paired samples, 14 (6.3%) had a single concordant pair of type-specific HPV infections detected at both oral and cervical sites; however, 207 (93.7%) women had no concordant infections. The discordant and concordant type-specific pairs of infections are shown in Fig. Fig.1B1B.
Although the majority of women did not have concordant oral and cervical HPV infections, when a permutation analysis was performed, the number of women with simultaneous oral and cervical infections by the same HPV type was significantly greater than would be expected due to chance (P = 0.0002). Similarly, the number of concordant pairs of HPV infections was greater than would occur at random (P = 0.02).
To explore potential differences between the 14 women with concordant oral and cervical HPV infections and the remaining women with paired samples that lacked concordance (n = 207), their characteristics were compared. The two groups of women were similar with regard to age, past and current smoking habits, and current and lifetime sexual behaviors. Strikingly, all 14 women with concordant HPV infections were HIV infected. In addition, the group with concordant infections had a higher median number of oral (2 versus 0, P < 0.001) and cervical (4.5 versus 1, P < 0.001) HPV infections.
In this cross-sectional study, oral HPV infection was found to be less prevalent than cervical HPV infection in both HIV-positive and -negative women. Although they were less prevalent, oral HPV infections were detected in approximately 25% of HIV-positive women and 9% of HIV-negative women. These data suggest that the oral cavity may be a reservoir of HPV infection with a sufficiently high prevalence to affect the dynamics of HPV transmission in populations.
We may have underestimated the true prevalence of both oral and cervical HPV infection in this study by limiting detection to the 37 HPV types present on the Roche prototype line blot. Because the Roche assay was designed to detect HPV types associated with cervical dysplasia and cancer, the resulting bias may have a more significant effect on estimates for oral HPV prevalence than on estimates for cervical HPV prevalence. HPV types associated with nonmalignant oral lesions, such as 7, 13, 32 and (16, 36) would not be detected by our methods. In a recent study, oral HPV infections by low-risk HPV types not represented in the Roche prototype line blot were detected in 2.5% of population-based controls without cancer (17), suggesting that the prevalence of low-risk oral HPV infection may be underestimated by the Roche line blot. However, this study was conducted in an HIV-negative population and may not be representative of HPV detection among HIV-positive subjects. Cameron and colleagues recently reported that only a small proportion of oral HPV infections amplified from saliva of HIV-positive individuals by use of PGMY09/11 were not genotyped by a Roche linear array containing probes for 27 HPV types (4). This suggests that the majority of HPV infections in HIV-infected individuals, especially high-risk HPV infections, are accounted for by the Roche array.
Approximately one-third of the HPV types detected in the cervices of women in this study population were not detected within the oral cavities of the same women. In addition to the potential bias introduced by the Roche assay, we cannot exclude the possibility that the relatively narrow HPV type distribution in the oral cavity is due to differential HPV-type susceptibilities of the oral and cervical mucosa. Although it is unlikely, we also cannot rule out the possibility for an oral-cavity-specific HPV clade with sufficient dissimilarity of L1 sequence such that detection would not be possible with currently available consensus primers designed to amplify genital-mucosal HPV types. We acknowledge that, while distributions of HPV types detected in ORSs and CVLs appeared to differ, this study was not designed to analyze type-specific differences and that a larger sample size would be required to determine, through analysis of type-specific prevalence ratios, whether type-specific differences in prevalence at oral sites and cervical sites occur.
Isolated oral infections were relatively unusual. Most women with an oral HPV infection also had a cervical infection, likely because the majority of women had a cervical infection. The increased prevalence of oral infections in women with simultaneous cervical infections would suggest that the behaviors that place a woman at risk for oral HPV infection could substantially overlap with cervical infection. Anogenital HPV infections in adults are predominantly sexually transmitted (3), and indeed, sexual behaviors have been previously associated with oral HPV infection (22). Concomitant type-specific infections at both anatomic sites could be obtained during the same sexual encounter or sequential sexual encounters with an infected partner or autoinoculation from one site to the other. Direct mouth-to-mouth transmission could account for the isolated oral HPV infections observed in this study. Given the fact that oral HPV infections are not unusual (~9% in HIV-negative women and 25% in HIV-positive women) (Table (Table2)2) and viral loads can be substantial (9), direct mouth-to-mouth and mouth-to-genital transmission may occur. However, mouth-to-mouth transmission may be difficult to parse in an epidemiological study because of colinearity of sexual behaviors.
The majority of HPV infections detected, even among women infected at both sites, were type discordant. However, although small in number, concurrent oral and cervical HPV infections of the same type occurred more commonly than would be expected by chance. Each woman with these concordant infections was HIV infected, suggesting that the presence of HPV infection of the same type in both the oral cavity and cervix is strongly influenced by systemic immunity. Assuming an infection at both sites is concurrently acquired, increased persistence of infection in immunosuppressed women might increase concordance in a cross-sectional study such as this. HIV-positive women were also found to have a two- to sevenfold increase in prevalence of high-risk and multiple-type HPV infections at both oral and cervical sites. The higher prevalence of cervical HPV infection among HIV-positive women in the WIHS has been previously ascribed to increases in both the incidence and duration of infection, both of which increased directly as a function of HIV viral load within the strata of the CD4 count (35). The degree to which clearance of HPV infection is affected by local mucosal immunity compared to systemic immunity and the relative impact of HIV infection on both are currently unclear. If systemic immunity plays a dominant role, then clearances at oral and mucosal sites would be expected to be similar.
While HIV-related systemic immunosuppression may affect oral and cervical sites similarly, local factors found to influence persistence at the cervix may have differential effects on the natural history of oral HPV infections. These include coinfection by Chlamydia trachomatis (29, 32) or herpes simplex virus (20, 34), smoking (6, 12, 24), age (7), HPV type (30), and use of hormones and contraceptives (6, 23, 27). Even if infections at both sites are acquired at the same time, differences in the role, strength, and direction of local factors on infection persistence could result in high discordance in a cross-sectional study. Reactivation of latent infection, presumably random with regard to HPV type, would also be expected to increase discordance of infections. Evidence for reactivation of latent cervical HPV infection was recently reported by the WIHS, and this reactivation appeared to be enhanced by HIV-related immunosuppression (35).
In conclusion, these data suggest that the oral cavity is a significant reservoir for HPV infection that may not be entirely independent of the cervical reservoir. Because the high discordance of infections may reflect differences in the risk factors for or natural history of infection at the two sites, it may not be entirely appropriate to extrapolate the vast literature on cervical HPV natural history to oral HPV infection. A prospective study to clarify the interrelationship between HPV infections at both sites and to understand possible differences in incidence and factors affecting clearance or persistence in the oral cavity and the cervix is warranted.
This work was supported in part by the Damon Runyon Cancer Research Foundation (M.G.) and the National Institute of Dental and Craniofacial Research (grant DE016631-02 to M.G.).
Data in this study were collected by the Women's Interagency HIV Study (WIHS) Collaborative Study Group with centers (principal investigators) at the New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); the Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); and Data Coordinating Center (Stephen Gange). The WIHS is funded by the National Institute of Allergy and Infectious Diseases, with supplemental funding from the National Cancer Institute, and the National Institute on Drug Abuse (UO1-AI-35004, UO1-AI-31834, UO1-AI-34994, UO1-AI-34989, UO1-AI-34993, and UO1-AI-42590). Funding is also provided by the National Institute of Child Health and Human Development (grant UO1-HD-23632) and the National Center for Research Resources (grants MO1-RR-00071, MO1-RR-00079, and MO1-RR-00083).
Published ahead of print on 4 October 2006.