Professional Documents
Culture Documents
The prevalence and risk factors for oral human papillomavirus (HPV) infection are unknown, despite evidence
for an etiological role for HPV in oral cancers. Oral samples from human immunodeficiency virus (HIV)–
seronegative (n p 396) and HIV-seropositive (n p 190 ) adults were tested for HPV DNA. High-risk HPV
infections were present in 2.1% of tonsil and 6.3% of oral-rinse specimens. The prevalence of oral high-risk
HPV infection was greater in HIV-seropositive individuals (13.7% vs. 4.5%; P ! .001 ). In multiple logistic
regression, odds of oral HPV infection increased with age, male sex, and herpes simplex virus (HSV)–2
seropositivity in HIV-seronegative individuals and with CD4 cell count !200 cells/mL, HSV-2 seropositivity,
oral mucosal abnormalities, and 11 oral sex partner during the previous year (odds ratio, 12.8; 95% confidence
interval, 3.1–52.7) among HIV-seropositive individuals. HPV type 16, which is present in most HPV-associated
tonsillar cancers, was the most prevalent high-risk oral HPV infection.
Human papillomavirus (HPV) is etiologically associ- logical or population differences. Prevalence estimates
ated with a subset of head and neck squamous-cell of 9.2%–18.6% in exfoliated oral cells have been re-
carcinomas (HNSCC), particularly those that arise from ported in studies with disparate specimen collection
the lingual and palatine tonsils [1–6]. However, the methods, study populations, and sensitivity of poly-
prevalence of oral HPV infection in individuals without merase chain reaction (PCR) methods used for HPV
oral cancer has been uncertain, and the risk factors for detection [1, 8, 9]. HPV infection in the tonsillar ep-
and natural history of oral infection are largely unex- ithelium, the most biologically relevant site with respect
plored. Estimates of oral HPV prevalence are highly to cancer, has not been specifically evaluated, likely be-
variable [7]. Because prevalence estimates in many cause the sampling of tonsils in healthy individuals is
studies have been reported without concomitant analy- technically challenging.
ses of demographic and behavioral data, it is difficult Numerous studies have established that cervical HPV
to discern whether such variability reflects methodo- infection is necessary for the development of cervical
cancer and that it is acquired through sexual contact [10].
Similarly, a population-based study of oral cavity and
Received 25 June 2003; accepted 26 August 2003; electronically published 2
oropharyngeal cancers determined that a high number
February 2004.
Presented in part: 20th International Papillomavirus Conference, Paris, October of sex partners, younger age at first intercourse, and a
2002, (abstract O007).
history of genital warts were associated with the oral
Financial support: National Institute of Dental and Craniofacial Research (grant
DE13121); State of Maryland Cigarette Restitution Fund (to M.L.G.); Damon Runyon cancer risk in men [1]. Similar associations of increased
Cancer Research Foundation (Damon Runyon-Lilly Clinical Investigator award to
sexual behavior with oral cancer risk have been ob-
M.L.G.).
a
Present affiliation: National Cancer Institute, National Institutes of Health, served in some [11] but not in other [12–14] studies.
Department of Health and Human Services, Bethesda, Maryland. Although oral-genital contact is a likely method of
Reprints or correspondence: Dr. Maura L. Gillison, Sidney Kimmel Comprehensive
Cancer Center at Johns Hopkins, G91, Cancer Research Bldg., 1650 Orleans St., transmission of HPV to the oral mucosa, such contact
Baltimore, MD 21231 (gillima@jhmi.edu). has not been associated with oral cancer risk in case-
The Journal of Infectious Diseases 2004; 189:686–98
2004 by the Infectious Diseases Society of America. All rights reserved.
control studies. However, in a case-case comparison,
0022-1899/2004/18904-0017$15.00 the odds of a tumor being HPV-positive increased in
Oral HPV Infection and Sexual Behaviors • JID 2004:189 (15 February) • 687
Table 1. Characteristics of the study population, by HIV serostatus.
a High-risk HPV types assayed for but not detected in oral samples: 26, 31, 33, 51, 68, and 70 [28].
b HPV type classified as probable high risk [28].
c HPV-82 includes the MM4 and IS39 subtypes.
d Value different from the sum of columns because of multiply infected individuals.
e Low-risk HPV types assayed for but not detected in oral samples: 11, 40, 42, 54, and 57 [28].
(from 50,000 to 0.5) of a diploid human cell line, CCD-18LU obtained for the negative control serum samples (excluding
(ATCC CCL-205). The cell number was defined as the number outliers).
of ERV-3 copies (in the diploid genome) divided by 2. HIV serological testing. The presence of HIV-1 infection
HPV serological testing. Serum samples were tested for was determined according to Centers for Disease Control rec-
antibodies to HPV-16, -18, and -33 in a viruslike particle–based ommendations [24]. Serum samples were tested for HIV-1 an-
ELISA [23]. Serum samples were dichotomized as positive or tibodies using the Vironostika HIV-1 Microelisa System
negative for each antibody. The assay cutoff was determined by (bioMerieux), according to the manufacturer’s protocol. All sam-
comparison with the distribution of values obtained for 74 chil- ples with reactive ELISA results were subsequently evaluated by
dren, ages 2–5 years, after the exclusion of outliers [23]. Sero- use of the Cambridge Biotech HIV-1 Western Blot Kit (Calypte
positivity was defined as 3 SD above the mean optical density Biomedical).
Oral HPV Infection and Sexual Behaviors • JID 2004:189 (15 February) • 689
Table 3. Univariate analysis of associations with tonsillar human papillomavirus (HPV) infection in the
total study population.
(continued)
Table 3. (Continued.)
NOTE. CI, confidence interval; HSV, herpes simplex virus; OR, odds ratio; STD, sexually transmitted disease.
a Columns are not equal to total population because of subject nonresponse or unsuccessful phlebotomy.
Herpes simplex virus (HSV)–1 and -2 serological testing. The final model was created by the inclusion of variables with
Serum samples were analyzed for the presence of antibodies to potential biological significance, as well as those that remained
HSV-1 and HSV-2 by use of Focus Technologies HerpeSelect- statistically significant after adjustment. All P values reported
1 ELISA IgG and HerpeSelect-2 ELISA IgG, respectively [25]. are 2-sided and were considered to be statistically significant
Statistical analysis. HPV DNA prevalence estimates and at P ! .05.
exact 95% confidence intervals (CIs) were calculated. The pres-
ence of HPV DNA in tonsillar epithelial and oral-rinse speci-
RESULTS
mens was analyzed separately and combined to represent HPV
DNA in the “oral region.” the presence of HPV DNA was The characteristics of the study population, stratified by HIV
stratified by type and high- and low-risk classifications. Indi- serostatus, are shown in table 1. Compared with the HIV-sero-
viduals coinfected by high- and low-risk types were placed in negative group, the HIV-seropositive group was more likely to
the high-risk category. The k statistic was used to compare be older, black, current alcohol drinkers, in the lowest income
agreement between cytobrush and oral-rinse data. Logistic re- group, and to have histories of more extensive cigarette smok-
gression was used to model the relationship between age as a ing and abnormal Pap smear results (table 1). Age at first in-
continuous variable and oral HPV infection. To evaluate the tercourse and total number of lifetime sex partners did not
data from ERV-3 cell quantitation, the median cell number and differ between HIV-seropositive and -seronegative individuals
interquartile range were determined. Matched pairs of tonsillar (table 1). With respect to history of oral sex, HIV-seropositive
and oral rinse specimens from the same individual were com- individuals reported fewer lifetime partners, were less likely to
pared by use of a 2-sided sign test. The rank sum test was used have had 11 recent partner, and had an older mean age at first
to compare median cell number, stratified by specimen collec- performing oral sex (23.0 [HIV-positive] vs. 21.2 years; P p
tion method and HIV serostatus or the presence of HPV-DNA. .02). However, HIV-seropositive individuals were more likely
Associations observed in data when they were stratified by to be seropositive to HPV-16, -18 or -33. The seroprevalence
study site and questionnaire administration method were sim- of HSV-2 in the study population increased with increasing
ilar in univariate and multivariate analyses; therefore, data were years of sexual activity (OR, 1.03; 95% CI, 1.01–1.05), history
pooled for further analysis. The HIV-seronegative and -sero- of STDs (OR, 2.5; 95% CI, 1.8–3.6), and increased number of
positive groups were compared by use of the x2 or Fisher’s lifetime sex partners (0–5, 6–10, and 110 partners; P p .07 for
exact test for categorical variables and by Student’s t test (with trend), as reported elsewhere [26].
unequal variance) for continuous variables. Logistic regression Prevalence of HPV DNA by specimen collection method.
models were used to estimate odds ratios (ORs) (and 95% CIs) All subjects had either (or both) a tonsillar epithelial or oral
for associations between demographic and exposure variables rinse specimen positive for b-globin by PCR. Tonsillar epithelial
and the presence of HPV DNA. Trend tests were conducted samples from 2.3% and oral rinse specimens from 1.2% were
across ordered groups. Variables that were important in HIV- negative for b-globin by PCR.
stratified univariate analysis and for which P ! .20 were eval- HPV DNA was present in the tonsillar epithelium of 18
uated in a multiple logistic-regression model, as were variables (3.1%) of 583 (95% CI, 1.8%–4.8%) evaluable individuals.
that were considered to be relevant on the basis of the literature. HPV-16 was the most common type detected (6 infections)
Oral HPV Infection and Sexual Behaviors • JID 2004:189 (15 February) • 691
and was present in the tonsillar epithelium of 1.0% (95% CI, with a tonsillar HPV infection, including a history of same-sex
0.4%–2.2%) of individuals. Coinfection of tonsils by multiple oral sex, 15 lifetime casual sex partners, and a history of an
HPV types was observed in 1.4% (95% CI, 0.6%–2.7%), and STD. The odds of a tonsillar HPV infection were elevated in
2.2% (95% CI, 1.2%–3.8%) had infections with high-risk types. current tobacco users, compared with nonusers, and a signif-
The prevalence of HPV DNA did not differ in individuals with icant trend was noted for increased odds with increased pack-
or without tonsils by physician examination (13 [2.9%] of 447 years of use (P p .02 for trend) and with number of packs per
vs. 5 [3.9%] of 128; P p .6). day (P p .03 for trend) among current cigarette smokers. All
HPV DNA was detected in the oral rinse of 72 (12.2%) of associations found to be significant in this univariate analysis
590 (95% CI, 9.6%–14.9%) evaluable individuals. This was sig- of HPV DNA in the tonsillar epithelium were also significant
nificantly higher than the detection rate in the tonsillar epithe- in an analysis of HPV DNA in oral-rinse specimens, but as-
lium (P ! .001). Classification of an individual as HPV infected sociations were not as strong (data not shown).
Figure 1. The prevalence of oral human papillomavirus (HPV) infection in the study population, stratified by HIV serostatus and age. The prevalence
of an oral HPV infection did not increase with age in individuals with HIV infection (P p .95 for trend) but did in those without (P ! .001 for trend).
HIV-seronegative individuals are represented in black and HIV-seropositive individuals in gray. Sample size within each age category is indicated below,
stratified by HIV serostatus. +, positive; ⫺, negative.
Oral HPV Infection and Sexual Behaviors • JID 2004:189 (15 February) • 693
Table 4. Univariate analysis of associations with an oral human papillomavirus (HPV) infection, by HIV serostatus.
(continued)
Table 4. (Continued.)
NOTE. Data are no. (%) unless otherwise indicated. CI, confidence interval; NA, not applicable; OR, odds ratio; STD, sexually transmitted disease.
a Columns are not equal to total population because of subject nonresponse or unsuccessful phlebotomy.
b The mean value of drinks per week among current drinkers was chosen as the cutoff.
tistically significant associations may not be independent mark- titer infection more likely to be detected in the setting of im-
ers of risk for tonsillar HPV infection because of the low prev- munosuppression. A prospective investigation of the factors
alence of tonsillar infection. associated with an incident oral HPV infection may distinguish
Our study builds on evidence that oral HPV infection can from among these possibilities.
be acquired through sexual behavior, particularly recent oral- Oral-genital contact was not significantly associated with oral
genital contact. Oral HPV infection in individuals with and HPV infection in the HIV-seronegative population. Surpris-
without HIV infection was associated with HSV-2 seropositiv- ingly, in univariate analysis, the odds of an oral HPV infection
ity, an accepted surrogate measurement of several sexual be- were significantly reduced in HIV-seronegative individuals who
haviors [27]. Recently, HSV-2 infection was demonstrated to reported 11 recent oral, vaginal, or casual sex partner (table
increase the risk of an incident HPV infection [32], and it has 4), but this finding was not corroborated in the multivariate
been suggested that HSV-2 may act to promote cervical disease model. This finding could be real and could indicate that the
progression [33–35], even though HSV-2 sequences have not transmission of HPV to the oral cavity is significantly different
been found in tumors [36]. In the present study, however, oral in HIV-seronegative adults, but we think this unlikely. Other
HPV infection was significantly associated with seroreactivity possible explanations would include insufficient sample size
to HSV-2 but not with HSV-1, the herpes virus that more caused by a lower prevalence of infection in this group; a finding
commonly infects the oral cavity. Therefore, we interpret the caused by chance, given the multiple comparisons performed
association of oral HPV infection with HSV-2 as evidence of in the study; or bias in selection of the study population. In
an association with sexual behavior. particular, racial and ethnic heterogeneity was greater in the
The strongly positive association of oral-genital contact with HIV-seronegative group. We were unable to evaluate the effects
an HPV infection in HIV-seropositive but not in HIV-sero- of race or ethnicity because of colinearity with several other
negative adults would suggest that the same behavior may have exposure variables. For example, the HIV clinic served a pre-
different consequences in the setting of immunosuppression. dominantly black community. Because of the low prevalence
Evidence for abnormal oral mucosal immunity is provided by of oral HPV infection, a larger study is warranted to clarify the
reports of high prevalence of oral infectious complications and associations of race and sexual behaviors with infection among
a rise in prevalence with advancing immunosuppression in HIV-seronegative individuals.
HIV-infected individuals [37, 38]. HIV-seropositive individuals The association of oral-genital contact with oral HPV infec-
were more likely to have an oral HPV infection, even though tion in our results does not exclude a role for other methods
HIV-seropositive individuals reported fewer recent oral sex of transmission of HPV to the oral mucosa. These may in-
partners. Possible explanations for this finding would include clude mother-to-child vertical transmission [39–42], nonsexual
reactivation of a latent infection, infection persistence, or high- transmission in children prior to their sexual debut [43], or
Oral HPV Infection and Sexual Behaviors • JID 2004:189 (15 February) • 695
Table 5. Associations of selected variables with an oral human analysis was limited to the 38 most common sexually trans-
papillomavirus (HPV) infection, after adjustment for covariates, mitted genital-mucosal HPV types. Unusual HPV types pre-
stratified by HIV serostatus.
viously associated with oral lesions in HIV-seropositive indi-
OR (95% CI)
viduals, such as HPV-7, -13, and -32 [52, 53], would not have
been detected. In the setting of HIV infection, unusual HPV
HIV seronegative HIV seropositive
Variablea (n p 384) (n p 182) types (e.g., cutaneous HPV-7) and unusual associations of HPV
type with particular lesions (e.g., HPV-13, -18, or -32 papil-
Age, yearsb 1.05 (1.02–1.07) 1.0 (0.95–1.05)
lomas) may be found in the oral cavity [52, 53].
Sex
Although HPV infection was strongly associated with mu-
Female 1.0 1.0
cosal abnormalities in HIV-seropositive individuals, an etiologic
Male 3.0 (1.3–7.0) 1.9 (0.8–4.3)
association cannot be asserted because we did not analyze HPV
HSV-2
Oral HPV Infection and Sexual Behaviors • JID 2004:189 (15 February) • 697
26. Cowan FM, Johnson AM, Ashley R, Corey L, Mindel A. Antibody to mosexual men. J Acquir Immune Defic Syndr Hum Retrovirol 1998;
herpes simplex virus type 2 as serological marker of sexual lifestyle in 17:314–9.
populations. BMJ 1994; 309:1325–9. 47. Palefsky JM, Holly EA, Ralston ML, Da Costa M, Greenblatt RM.
27. Dillner J, Kallings I, Brihmer C, et al. Seropositivities to human pap- Prevalence and risk factors for anal human papillomavirus infection
illomavirus types 16, 18, or 33 capsids and to Chlamydia trachomatis in human immunodeficiency virus (HIV)–positive and high-risk HIV-
are markers of sexual behavior. J Infect Dis 1996; 173:1394–8. negative women. J Infect Dis 2001; 183:383–91.
28. Munoz N, Bosch FX, de Sanjose S, et al. Epidemiologic classification 48. Ahdieh L, Klein RS, Burk R, et al. Prevalence, incidence, and type-
of human papillomavirus types associated with cervical cancer. N Engl specific persistence of human papillomavirus in human immunodefi-
J Med 2003; 348:518–27. ciency virus (HIV)–positive and HIV-negative women. J Infect Dis 2001;
29. Herrero R, Castellsague X, Pawlita M, et al. Human papillomavirus 184:682–90.
and oral cancer: the International Agency for Research on Cancer 49. Palefsky JM, Minkoff H, Kalish LA, et al. Cervicovaginal human pap-
Multicenter Study. J Natl Cancer Inst 2003; 95:1772–83. illomavirus infection in human immunodeficiency virus–1 (HIV)–
30. Ha PK, Pai SI, Westra WH, et al. Real-time quantitative PCR dem- positive and high-risk HIV-negative women. J Natl Cancer Inst 1999;
onstrates low prevalence of human papillomavirus type 16 in premalig- 91:226–36.
nant and malignant lesions of the oral cavity. Clin Cancer Res 2002; 8: