|Home | About | Journals | Submit | Contact Us | Français|
Background.The association between partner human papillomavirus (HPV) viral load and incident HPV detection in heterosexual couples is unknown.
Methods.HPV genotypes were detected in 632 human immunodeficiency virus (HIV)–negative couples followed for 2 years in a male circumcision trial in Rakai, Uganda, using the Roche HPV Linear Array. This assay detects 37 genotypes and provides a semiquantitative measure of viral load based on the intensity (graded 1–4) of the genotype-specific band; a band intensity of 1 indicates a low genotype-specific HPV load, whereas an intensity of 4 indicates a high load. Using Poisson regression with generalized estimating equations, we measured the association between partner's genotype-specific viral load and detection of that genotype in the HPV-discordant partner 1 year later.
Results.Incident detection of HPV genotypes was 10.6% among men (54 of 508 genotype-specific visit intervals) and 9.0% among women (55 of 611 genotype-specific visit intervals). Use of male partners with a baseline genotype-specific band intensity of 1 as a reference yielded adjusted relative risks (aRRs) of 1.14 (95% confidence interval [CI], .58–2.27]) for incident detection of that genotype among women whose male partner had a baseline band intensity of 2, 1.75 (95% CI, .97–3.17) among those whose partner had an intensity of 3, and 2.52 (95% CI, 1.40–4.54) among those whose partner had an intensity of 4. Use of female partners with a baseline genotype-specific band intensity of 1 as a reference yielded an aRR of 2.83 (95% CI, 1.50–5.33) for incident detection of that genotype among men whose female partner had a baseline band intensity of 4. These associations were similar for high-risk and low-risk genotypes. Male circumcision also was associated with significant reductions in incident HPV detection in men (aRR, 0.53 [95% CI, .30–.95]) and women (aRR, 0.42 [95% CI, .23–.76]).
Conclusions.In heterosexual couples, the genotype-specific HPV load in one partner is associated with the risk of new detection of that genotype in the other partner. Interventions that reduce the HPV load may reduce the incidence of HPV transmission.
Human papillomavirus (HPV) infection can cause genital warts, and high-risk HPV (HR-HPV) genotypes cause oral, anal, and penile cancer, as well as cervical cancer, the third most common cancer in women worldwide [1, 2]. Greater than 85% of the cervical cancer burden is in developing countries [2, 3]. HPV infection also may be associated with an increased risk of human immunodeficiency virus (HIV) acquisition in both men and women [4–7].
Models have shown that HPV transmissibility is substantially higher than that of other viral sexually transmitted pathogens [8, 9], but data on the natural history of HPV transmission between heterosexual partners are limited. Numerous studies throughout North America, Europe, and Asia have shown a concordance of HPV genotypes between heterosexual partners at a single time point [10, 11], and one study found an association between HPV load and concordance . Another study evaluated the impact of HIV infection on HPV detection and clearance among men and women in South Africa . However, data on transmission (or incident HPV detection, in couples) are primarily from studies in North America, with a limited number of heterosexual couples and limited information on risk behaviors [14–16].
Little is known about HPV transmission in heterosexual couples in African populations, which have among the highest rates of penile and cervical cancers worldwide [1, 2, 17]. While it has been shown that transmission of other viral sexually transmitted pathogens, such as HIV, are dependent on the source partner's viral load , no studies have evaluated the association between HPV load and incident HPV detection in partners. We hypothesized that a higher genotype-specific HPV load in HPV-infected men and women is positively associated with subsequent genotype-specific incident detection in their heterosexual partner. With the roll out of HPV vaccination and male circumcision programs in Africa, it is important to understand the risk factors and natural history of HPV infection among heterosexual couples in Africa. Here we report risk factors for incident HPV detection among heterosexual men and women in Rakai, Uganda.
HPV was evaluated among 1097 heterosexual couples, consisting of 1011 men and 1097 women aged 15–49 years, in Rakai District, Uganda. The number of women exceeded the number of men because of polygamy. Couples were assessed for HPV as part of a randomized trial of male circumcision for preventing transmission of HIV and other sexually transmitted pathogens in Rakai, enrolled between August 2003 and December 2006 [19–23]. Both partners reported being married or in long-term consensual relationships at baseline and were followed annually as a couple over 2 years. Men were randomly assigned to receive immediate circumcision (intervention) or circumcision delayed for 24 months (control). Written informed consent was provided by all male and female participants at enrollment. The consent form described study procedures, risks, benefits and the voluntary nature of participation.
Testing for HPV and HIV, physical examinations, and interviews to ascertain sociodemographic characteristics and sexual risk behaviors were conducted at baseline and at months 12 and 24 of follow-up. All subjects were offered free HIV counseling and testing, health education, and condoms at each visit.
This study was restricted to couples in which both partners were HIV negative at baseline. We also only included couples if HPV results were available for both partners at baseline and/or the month 12 visit and if HPV results were available 1 year later (ie, the month 12 or 24 visits) for either or both partners. We excluded visit intervals if either partner experienced HIV seroconversion during the interval or had indeterminate HIV test results at the end of an interval, because an association exists between HIV and HPV acquisition  and these events were too infrequent for stratified analyses.
The trials were approved by the HIV Subcommittee of the Ugandan National Council for Science and Technology (Kampala) and by 3 institutional review boards (IRBs): the Science and Ethics Committee of the Uganda Virus Research Institute (Entebbe, Uganda), the Johns Hopkins University Bloomberg School of Public Health IRB (Baltimore, Maryland), and the Western IRB (Olympia, Washington). The trials were overseen by independent data safety monitoring boards [20, 21] and were registered with clinicaltrials.gov (NCT00425984 and NCT00124878).
Penile swab samples were obtained for HPV detection by trained clinicians, using a standardized procedure. Moistened polyethylene terephthalate–tipped swabs were rotated around the circumference of the penis at the coronal sulcus and placed in specimen-transport medium (Digene, Gaithersburg, Maryland). Female partners were asked to provide self-administered vaginal swabs for HPV detection . Women were instructed to squat and then insert and rotate a 20-cm polyethylene terephthalate– or cotton-tipped swab high in the vaginal vault. After collection of the specimen, the women handed the swab to a field worker, who placed the swab in specimen transport medium (Digene). This approach to specimen collection was well accepted, with compliance rates of >90% . Swabs were maintained at 4°C–10°C for <6 hours, when they were frozen at −80°C.
HPV genotyping was performed using the Roche HPV Linear Array, which detects 37 HR-HPV and low-risk HPV (LR-HPV) genotypes (Roche Diagnostics, Indianapolis, Indiana) . HPV genotypes 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68 were considered to be HR-HPV genotypes. Swabs with no detectable cellular β-globin were excluded since the adequacy of the sample collection could not be ensured.
It has been previously demonstrated in both men and women that the linear array hybridization signal is linearly correlated with log-transformed HPV load (Spearman r = 0.73) [28–30]. A linear array band signal strength of 4 is approximately equivalent to a viral load of >2000 copies/5 µL. A band signal strength of 3 is approximately 200–2000 copies/5 µL. Linear array results and band intensity were evaluated independently by 2 observers, using a labeled acetate overlay provided by Roche, in which the overlay indicated the position of the genotype probes on the test strip. Observer disagreement was rare (1%–4% of results) and was resolved by reevaluation by the initial observers .
HIV status was determined using 2 separate enzyme-linked immunosorbent assays and confirmed by HIV type 1 Western blot, as previously described .
The unit of observation was a genotype-specific HPV annual visit interval. The primary outcome was incident HPV detection after 12 and 24 months of follow-up in men and women who were in HPV genotype–discordant relationships at baseline or the month 12 visit. Incident HPV detection was defined as a newly detected HPV genotype in an individual who was negative for that genotype at the prior annual visit. Incident detection of HPV was assessed separately in men and women. Analyses were also stratified by HR-HPV and LR-HPV genotype status and trial arm.
The primary exposure was the genotype-specific HPV load in the partner at the prior visit and was evaluated as an ordinal variable consisting of 4 categories corresponding to the 4 linear array band intensities. A band intensity of 1 was the reference group. Associations between demographic and behavioral factors and partner HPV load were assessed using Poisson multivariate regression with generalized estimating equations and robust variance estimators. There were 9 instances in which a polygamous man was in an HPV genotype–discordant relationship with two or more female partners in this study. In these cases, the viral load for that genotype-specific HPV was considered as the highest viral load that the man was exposed to during the visit interval, such that a man who had 2 female partners, one with a band intensity equal to 3, and the other with band intensity equal to 1, was assumed to be exposed to a genotype-specific viral load of 3 at that visit. Sensitivity analyses were conducted after excluding these 9 observations.
Relative risks (RRs) and 95% confidence intervals (CIs) were estimated using modified Poisson regression with generalized estimating equations and robust variance estimators. Associations between HPV load and time-fixed covariates, including age and male circumcision status, and time-varying covariates, including HR-HPV and LR-HPV genotype status, detection of other HPV genotypes in the partner at the prior visit, nonmarital relationships, and condom use, were also assessed. Risk factors that were previously identified as potential confounders of the HPV load association in prior studies (eg, male circumcision status ) or that were associated with incident HPV detection with a P value of <.10 in univariate analysis were entered into a Poisson multivariate model to estimate adjusted RRs (aRRs) and 95% CIs of incident HPV detection.
Analyses were performed using Stata software (version 11; StataCorp, College Station, Texas) and R statistical software (version 3.2).
There were 1097 couples assessed for HPV as part of the randomized trial of male circumcision. In this study, 356 couples (32%) were excluded because one or both partners were HIV infected at baseline. Of the 741 couples in which both partners were HIV negative at baseline, 10 were excluded because the man (7 individuals) or woman (3) underwent HIV seroconversion between baseline and the first annual visit. We also excluded 9 couples because either the man (1 individual) or woman (8) had an unknown HIV status after the baseline visit, owing to loss to follow-up or inconclusive results of HIV testing. Of the remaining 722 couples, 632 (88%) had the requisite HPV results available to assess HPV discordance at baseline or the month 12 visit and at the next annual visit by the man (438 couples [69%]) or woman (588 [3%]). These 632 couples consisted of 587 men and 632 women.
Table Table11 shows baseline characteristics for the 632 HIV-negative couples assessed for HPV genotype discordance and incident HPV genotype detection. The median age of women was 25 years (interquartile range [IQR], 22–30 years), and the median age of men was 29 years (IQR, 25–34 years). Only 1 woman (0.2%) and 12 men (2%) reported consistent condom use. Approximately half the men (48% ) were in the intervention arm of the circumcision trial. Most couples (72% ) had detectable HPV infection in either the man (58% ) or woman (56% ) at their initial visit. In 31% of the couples (195), both partners had concordant positive test results for at least 1 HPV genotype, and in 28% (176) no HPV genotypes were detected in either partner. The number of genotypes detected in men with detectable HPV ranged from 1 to 12 (median, 2 genotypes; IQR, 1–3 genotypes), and the number of detectable genotypes in women ranged from 1 to 10 (median, 2 genotypes; IQR, 1–3 genotypes).
HPV infection was assessed at the baseline and month 12 visits in 588 couples with the requisite follow-up to assess incident HPV detection in female partners 1 year later. Among these 588 couples, we identified 270 in which 1 or more genotypes detected in the man (257 individuals) were not also detected in the female partner at the same visit. These couples contributed 611 genotype-specific observations, representing all 37 HPV genotypes assessed by the Roche Linear Array assay.
Of these 611 observations, genotypes in 39% (239) had a band intensity of 1, those in 25% (154) had a band intensity of 2, those in 19% (117) had a band intensity of 3, and those in 17% (101) had a band intensity of 4 (Table (Table2).2). Overall, genotypes in 9.0% of these observations (55) were detected in the female partner 1 year later. Genotypes with band intensity of 4, representing the highest viral load, were more than twice as likely to be detected in female partners at follow-up than were male HPV infections with the lowest band intensity (aRR, 2.52; 95% CI, 1.40–4.54). HPV genotypes with band intensities of 3 (aRR, 1.75; 95% CI, .97–3.17) and 2 (aRR, 1.14; 95% CI, .58–2.27) in males were also more frequently detected in their female partners at the next annual visit than were HPV genotypes with band intensity of 1. This increasing positive relationship between partner HPV load and incident HPV detection was statistically significant (P = .002, by the Cuzick test for trend).
Male partners' HPV genotypes were 58% less likely to be detected in women at follow-up if the man was in the intervention arm (immediate circumcision) of the male circumcision trial (aRR, 0.42; 95% CI, .23–.76; Table Table2).2). There was no difference in the association between HPV load and incident HPV detection by trial arm (Supplementary Table 1). There also were no statistically significant associations between incident HPV detection in women and high-risk HPV status, genital ulcer disease in the woman, condom use, or sex with partners outside of the relationship. In a sensitivity analysis, a similar relationship between higher HPV load in male partners and incident HPV detection in women was observed when the analysis was stratified by HR-HPV or LR-HPV genotypes (Table (Table33).
HPV infection was assessed at baseline and month 12 visits in 438 couples with the requisite follow-up to assess incident HPV detection in male partners 1 year later. Among these 438 couples, we identified 194 in which 1 or more genotypes detected in the woman were not also detected in the male partner at the same visit. Only 6 of these men had 2 female partners who had detectable infection with the same HPV genotype (9 observations). Overall, these couples contributed 508 observations, representing 36 of 37 HPV genotypes assessed by the Roche HPV Linear Array assay.
Of these 508 observations in female partners, genotypes in 31% (157) had a band intensity of 1, those in 17% (86) had band intensity of 2, those in 22% (112) had band intensity of 3, and those in 30% (153) had band intensity of 4 (Table (Table4).4). Incident HPV genotype detection in men was similar to that in women: genotypes in 10.6% of observations (54 of 508) were detected in men 1 year later. Genotypes with a band intensity of 4 in female partners were associated with an aRR of male HPV detection at follow-up of 2.83 (95% CI, 1.50–5.33), compared with genotypes with the lowest band intensity. This association did not differ substantially by HR-HPV or LR-HPV status (Table (Table5)5) or if men who had >1 partner with detectable infection with the same genotype were excluded from the analysis. Female genotypes with band intensities of 2 or 3 were not associated with incident HPV detection in men.
The HPV infections observed in women were 48% less likely to be detected in men at follow-up if they were in the intervention arm of the circumcision trial (aRR, 0.53; 95% CI, .30–.95; Table Table4).4). Similar to women, we also observed no difference in the association between HPV load and incident HPV detection in men when stratified by trial arm (Supplementary Table 1). Incident HPV detection in men also was not statistically significantly associated with HR-HPV status, genital ulcer disease in the man, inconsistent condom use, or sex with partners outside of the relationship (Table (Table44).
Higher HPV load in male partners was also associated with persistent HPV detection in these men at follow-up. Among male partners who had HPV results 1 year later (83% [216 of 529]), the rates of persistence were 28% (25 of 91; RR, 2.00; 95% CI, 1.11–3.60) for HPV genotypes with a band intensity of 4, 20% (20 of 96; RR, 1.65; 95% CI, 1.06–2.59) with a band intensity of 3, and 13% (17 of 135; RR, 1.12; 95% CI, .67–1.89) with a band intensity of 2, compared with 9% (19 of 207) for genotypes with a band intensity of 1.
HPV load in females was associated with an increased frequency of persistent HPV detection in these women at follow-up. After exclusion of 6 polygamous men, 98% of female partners (185 of 188) had HPV results available 1 year later for assessment of persistent infection. At follow-up, the rates of persistent detection in women were 57% (82 of 145; RR, 2.86; 95% CI, 2.02–4.07) for HPV genotypes with a band intensity of 4, 55% (48 of 108; RR, 2.18; 95% CI, 1.50–3.18) for those with a band intensity of 3, and 33% (27 of 83; RR, 1.75; 95% CI, 1.14–2.69) for those with band intensity of 2, compared with 18% (19 of 207) for genotypes with band intensity of 1.
The risk of HPV persistence was significantly less in circumcised men relative to that in men in the control arm (RR, 0.39; 95% CI, 0.20–0.73); however, male circumcision status did not impact HPV persistence in women (RR, 0.84; 95% CI, .63–1.11). High-risk HPV genotypes were no more likely to persist than LR-HPV genotypes in men (RR, 0.74; 95% CI, .48–1.14) or women (RR, 0.95; 95% CI, .74–1.20).
Among couples initially discordant for 1 or more HPV genotypes, we found that a high HPV load, as measured by linear array band intensity, was associated with an increased risk of new detection of the specific genotype(s) in both male and female heterosexual partners 1 year later. The association between HPV load in partners and incident HPV detection may be mediated by a higher infectious dose at the time of transmission; longer exposure to HPV, owing to an increased frequency of HPV persistence in partners with high viral load; or some combination of these mechanisms. Indeed, previous studies found that a higher viral load was associated with HPV persistence [31, 32]. It is also plausible that higher-level HPV genotypes represent more-recent infection, in which case incident detections of HPV genotypes in partners more likely represent first-time infections that have not been immunologically controlled.
The rate of new HPV detection among men and women was almost identical. While one study reported similar detection rates between men and women in heterosexual couples , other studies have found that female-to-male transmission may be 2-fold higher [13–15, 33]. Detection rates can vary by follow-up interval, specimen collection method, coital frequency, autoinoculation rates, contaminating DNA from recent sexual intercourse, participant age, and partnership length [13–16, 33]. Although we found no significant differences in HPV detection by gender in this study, it is possible that men or women may be at higher risk for infection, owing to sampling limitations.
A high HPV load has been associated with increased morbidity. Among both men and women, higher viral load has been associated with persistence of HR-HPV [31, 34], which was also observed in this study. Higher viral load among men is associated with detection of HR-HPV at multiple penile sites  and with flat penile lesions . In addition, higher viral load in women has been associated with the incidence of cervical lesions [37–39]. Combined with the current study's findings that HR-HPV load correlates with newly detected HPV in heterosexual partners, interventions that reduce the HPV load may have public health benefit.
Potential interventions, such as a therapeutic HPV vaccine or medical male circumcision, may avert new HPV infections. While prophylactic HPV vaccines do not have substantial impact on established infection, peptide-based therapeutic HPV vaccines have been designed to treat these individuals [40, 41]. These vaccines appear to have cross-protection against nonvaccine genotypes. If these vaccines could also be successful in lowering the HPV load, they may also assist in lowering transmission. In addition, since male circumcision decreases the HR-HPV load among men and their female partners [28, 29], the procedure may reduce HR-HPV transmission within couples. The extent to which a population-level reduction in HPV load would reduce overall HPV transmission and HPV-associated diseases should be evaluated in future studies.
Previous studies found that male circumcision in HIV-negative men decreased the incidence of HR-HPV detection by approximately 33% in men and by 23% in their female partners [19, 21, 42–46]. The increased efficacy rates in this study of male circumcision to protect both men (47%) and women (58%) are likely more accurate than our previous randomized trial findings since this study only evaluated HPV-discordant couples and did not include concordant HPV-negative couples, which would reduce observed efficacy.
There are limitations with this study. The interpretation of HPV epidemiology is difficult, particularly when the interval between sample collections is long. This may have led to underestimation of the number of new detections, since individuals may have both acquired and cleared HPV genotypes between visits. Second, noninvasive methods of HPV detection have been historically difficult, owing to inadequate cell recovery [47, 48]. Consequently, we may also have misclassified persistent infection as new events if we did not detect HPV infection at the previous study visit. Moreover, we have previously demonstrated that β-globin detection is significantly lower among circumcised men, compared with uncircumcised men , which could also affect our classification of discordant relationships or new HPV detection. However, we were conservative in our estimates of new detection, which were restricted to sequential samples with amplifiable cellular or viral DNA to ensure the adequacy of sample collection. In addition, the hybridization signal strength is based on intensity relative to that of a high and low β-globin control, which provides some level of internal sampling and amplification efficiency control. Third, newly detected virus may also reflect reactivation of latent HPV infection after a loss of local immune response to viral infection, which is likely more common with increasing age [49, 50]. Last, we assessed penile HPV at the coronal sulcus only, not at other genital sites. However, our results were robust to adjustment for behavioral and clinical variables that may have confounded the association between viral load and new detection among heterosexual partners, including male circumcision status.
In conclusion, we found that a partner's HPV load is associated with new detection of HPV infections in heterosexual couples. Our findings highlight the potential role of viral load in the natural history of HPV transmission and the potential public health benefit of interventions, such as male circumcision, that reduce HPV load.
Supplementary materials are available at http://jid.oxfordjournals.org. Consisting of data provided by the author to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the author, so questions or comments should be addressed to the author.
Acknowledgments.We thank the study participants and the Rakai Community Advisory Board, whose commitment and cooperation made this study possible.
Disclaimer.The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Financial support.This work was supported by the National Institutes of Health (NIH; grants U1AI51171, 1K23AI093152-01A1 [to A. A. R. T.], and T32AI102 [to M. K. G.]); the Bill and Melinda Gates Foundation (grant 22006.02); the National Institute of Allergy and Infectious Diseases (NIAID), NIH (grants 1K23AI093152-01A1, U01-AI-068613, and 3U01-AI075115-03S1); the Division of Intramural Research, NIAID, NIH; and the Doris Duke Charitable Foundation (clinician scientist development awards to A. A. R. T. [no. 22006.02] and M. K. G.).
Potential conflicts of interest.All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.