Meta-analysis of Human Papillomavirus Infection Concordance

Anogenital human papillomavirus (HPV) infection is a common sexually transmitted infection that preferentially infects squamous epithelial cells (1, 2). At least 20 million people in the United States are currently infected with HPV, and 5.5 million people become infected annually (3). Although most HPV infections clear spontaneously (4-6), some nononcogenic HPV types (mainly types 6 and 11) cause anogenital warts (7, 8). Persistent infection with oncogenic HPV types, if left untreated, can lead to cancers of the cervix, vagina, vulva, penis, anus, oropharynx, and possibly skin and lung (9-12). Most cervical cancer deaths occur in the developing world (13), primarily because of lack of access to screening and treatment. Although the United States has much lower rates of cervical cancer, it spends about $4 billion annually on treating and managing cervical disease and anogenital warts (14) and an additional $418 million annually on other HPV-related diseases (15).

A key contributor to the incidence of HPV-related disease is the high transmissibility of HPV, with an estimated median value of 40% (range, 5-100%) per unprotected sexual act (16). Persistent HPV infection also increases the risk for disease and sexual partners acquiring the virus. Among women, the median duration of infection is usually less than a year, but oncogenic HPV infections last longer than nononcogenic HPV infections (4, 6, 17, 18). Although fewer data on infection duration have been reported for men, findings suggest that HPV infections clear more quickly for men than for women, and men have similar duration of infection for oncogenic and nononcogenic HPV types (5). Whereas the extent to which sexual partners both harbor anogenital HPV infection has important implications for public health, including helping to understand the risks that partners of infected individuals face and increasing the precision of HPV vaccine cost-effectiveness analyses, previous studies of HPV concordance have yielded mixed findings. To address this question, we conducted a systematic review and meta-analysis of the existing literature on HPV DNA concordance among heterosexual couples.

Two investigators (P.R. and W.P.) independently searched the PubMed and EMBASE databases for studies published through December 2008 whose title, abstract, or keywords referred to HPV infection or concordance among sexual partners. Search terms were ((HPV OR human papillomavirus) OR (papillomavirus infection)) AND (transmission OR concordance). We also manually searched reference sections of identified articles to locate additional studies for inclusion.

Two investigators (P.R. and W.P.) independently screened titles, abstracts, and articles for eligibility using predefined criteria described below. We required that studies reported findings in English on anogenital HPV infection. We excluded studies that reported data only on nonanogenital HPV infections (e.g., oral infections only). We required that studies determined the HPV status of participants using molecular methods (e.g., DNA hybridization, PCR), as opposed to relying solely on clinical examination or histopathology, because some individuals with no clinical evidence of HPV infection are positive for HPV DNA (19) whereas others with acetowhite anogenital lesions are HPV DNA negative (20). We required that studies reported data on heterosexual couples because HPV transmission dynamics in same-sex couples may differ.

Two investigators (P.R. and W.P.) independently completed standardized data extraction forms to code study and participant characteristics that could affect HPV concordance. Study characteristics included study design, specimen collection method, HPV types examined, and HPV detection methods. Participant characteristics included demographics (age and marital status), health (history of HPV-related disease or other sexually transmitted infections and whether males were circumcised), sexual history (age at first intercourse and number of lifetime sexual partners), and relationship characteristics (length of time couples had been together, condom use, and whether couples were monogamous). A third investigator (N.B.) resolved the few coding disagreements. For case-control studies, we treated cases and controls as separate study samples because cases (e.g., women with cervical cancer) may have higher chances of infecting their partners with HPV than would controls (e.g., women with normal Pap smear test results).

Reviewers extracted data on the status of each partner for infection with any HPV type and HPV types 6, 11, 16, and 18. We examined these four HPV types because they are the primary causes of HPV-related diseases (e.g., genital warts, cervical cancer, and anal cancer; refs. 7-9) and recently licensed HPV vaccines protect against two or all four of these viral types (21-23). Reviewers created measures of positive concordance, defined as both partners having the HPV outcome of interest. Thus, both partners being negative did not increase concordance. We usually refer to positive concordance simply as concordance for the remainder of this report. We examined any-type concordance (both partners had HPV), same-type concordance (both partners had 1 or more HPV types in common), and type-specific concordance for HPV types 6, 11, 16, and 18.

We required that studies reported HPV infection data for both members of couples in a manner that allowed us to assess at least 1 of these HPV concordance measures. We included type-specific concordance only for studies that reported data that did not combine multiple HPV types (e.g., the HPV DNA probe indicated the presence of either HPV type 16 or 18). During data extraction, we excluded articles that inconsistently reported concordance data that queries to the authors did not resolve.

We pooled data across studies using random-effects meta-analysis to examine any-type concordance, same-type concordance, and type-specific HPV concordance among couples. Using methods similar to those from previous studies (24, 25), we determined if concordance levels for HPV types 6, 11, 16, an 18 were higher than chance would predict by calculating expected concordance levels and using random-effects meta-analysis to calculate odds ratios and 95% confidence intervals (95% CIs). Analyses also examined sex-dependent same-type concordance, defined as the proportion of men who had the same HPV types (1 or more types) as their HPV-positive female partners and the proportion of women who had the same HPV types (1 or more types) as their HPV-positive male partners. We examined potential correlates of sex-dependent concordance using random-effects meta-regression, using two-tailed tests and 0.05 as the critical α level. Analyses excluded a study that reported data on only 1 couple, because the absence of study-level variance precluded its inclusion in random-effects meta-analyses. We report I² values as an indication of heterogeneity among studies. We conducted analyses using the Comprehensive Meta-Analysis Version 2 software.

Of the 2,070 titles and abstracts and 130 articles that we reviewed, 30 articles met inclusion criteria (Fig. 1; refs. 24-53). These articles reported data from 33 study populations that included HPV concordance data for 2,972 couples (median, 45; range, 4-499; Table 1). Studies provided too few data to report meaningful summary statistics on participant characteristics. Year of study publication ranged from 1985 to 2008 (median, 1994). Study locations included Europe (52%), Asia (36%), Latin America (15%), and the United States (9%). Most studies used cohort (76%) and cross-sectional (88%) study designs. Researchers collected specimens from a wide range of anatomic sites using various methods. To detect HPV, most studies used DNA hybridization alone (42%) or in combination with PCR (42%). Whereas most studies (61%) tested for 5 or more HPV types, the most common were HPV type 16 (100%), type 18 (94%), type 6 (70%), and type 11 (70%).

Both partners in 37.7% of couples were infected with any type of HPV (Table 2). In 25.5% of couples, both partners were infected with 1 or more of the same HPV types. More couples had both members infected with HPV type 16 (9.0%) or type 6 (8.6%) than with type 11 (4.6%) or type 18 (2.2%). Concordance was higher than chance would predict for HPV types 11, 16, and 18 (all P < 0.05), whereas concordance for HPV type 6 reached borderline statistical significance (P = 0.06; Table 3). In couples where both members were HPV-positive, 63.2% were infected with 1 or more of the same HPV types (Table 2).

Same-type concordance for people with HPV-infected partners was lower for men than for women (Table 4; Figs. 2 and 3). Among male partners of HPV-positive women, 36.1% (95% CI, 22.7-52.0%) were infected with 1 or more of the same HPV types. Conversely, among female partners of HPV-positive men, 55.1% (95% CI, 40.3-69.1%) were infected with 1 or more of the same HPV types.

We further refined these sex-dependent concordance estimates by stratifying on whether studies recruited individuals with HPV or HPV-related disease (e.g., invasive cervical cancer) and their partners. Men with HPV-positive female partners had 1 or more of the same HPV types more often in studies that recruited men with HPV-related disease (65.8%; 95% CI, 48.5-79.7%) compared with studies without this inclusion criterion for men (27.2%; 95% CI, 15.0-44.2%; P = 0.002; Table 4). We did not find this difference for women. Same-type concordance was equally high in studies that recruited couples with women having HPV-related disease (51.9%; 95% CI, 34.3-69.1%) compared with studies that did not recruit women with HPV-related disease (63.0%; 95% CI, 39.2-81.9%; P = 0.463).

We further examined these sex-dependent concordance estimates for potential correlates. For men, we examined only studies that did not recruit men with HPV-related disease (k = 12 studies) because this was shown to strongly affect concordance levels. Because we did not find this difference for women, we included all studies (k = 15 studies). Among men, studies that used PCR to detect HPV had higher same-type concordance (k = 10 studies; 35.0%; 95% CI, 20.5-52.9%) compared with studies that did not use PCR (k = 2 studies; 2.7%; 95% CI, 0.8-8.9%; P < 0.001), where concordance was defined as the proportion of men positive for 1 or more of the same HPV types as their HPV-infected female partners. None of the other variables examined (specimen collection site for men, number of HPV types tested for, year of article publication, study location) were associated with sex-dependent concordance. Among women, the number of HPV types tested for, the year of article publication, whether the study used PCR, and study location were not associated with sex-dependent concordance.

Our comprehensive review of data from several thousand heterosexual couples from four continents found moderate to high positive concordance, defined as both partners having the HPV outcome of interest. Concordance was greater than one would expect by chance for HPV types 11, 16, and 18, and results were suggestive for HPV type 6. Whereas two small primary studies have reported similar results (24, 25), our meta-analysis confirms these findings in a substantially larger and more diverse sample. Of couples in which both members were HPV-positive, about two thirds (63.2%) were infected with 1 or more of the same HPV types. This level of concordance is consistent with the high transmissibility of HPV (16).

Our finding that female coital partners of HPV-positive men were more likely to be infected with the same types of HPV (1 or more types) compared with male partners of HPV-positive women makes sense, given that women may be more susceptible to HPV infection and take longer to clear HPV infections than men (4-6, 17, 18, 54). Exposure to HPV in men often involves the keratinized epithelium of the penis that may be less susceptible to HPV infection than the mucosal epithelium of the cervix (54). HPV infections may also persist longer in women compared with men. Estimates of the median duration of HPV infections among women have ranged from 4.3 to 11.1 months for nononcogenic HPV types and from 6.5 to 14.8 months for oncogenic HPV types (4, 6, 17, 18). Among men, the median duration of HPV infections is about 6 months for both oncogenic and nononcogenic HPV types (5), with circumcision possibly reducing HPV acquisition and speeding clearance (55-57). The gender difference was not observed in studies where recruitment involved individuals with HPV-related disease. However, in studies where recruitment did not involve individuals with HPV-related disease, the difference was apparent and may be partly due to sampling from suboptimal anatomic sites and difficulty in obtaining adequate specimens, problems that occur more often when testing men for HPV (58, 59).

HPV concordance in heterosexual couples has important clinical and public health implications. HPV infection and subsequent HPV-related disease pose a substantial burden worldwide (10). Female partners of men with HPV-related disease should be encouraged to get screened for HPV-related disease, given that they have a high likelihood of concomitant infection and that most HPV infections in couples are of the same viral types. Screening may also benefit male partners of HPV-infected women, although an HPV DNA test has not yet been approved for clinical use in men. Partners of HPV-infected individuals may receive additional benefits from educational counseling and screening for other sexually transmitted infections if encouraged to see a health care provider (60). However, only 62% of health care providers encourage women with either abnormal Pap smears or positive HPV tests to tell their sexual partners to see a clinician (61). We are not aware of data about referral of female partners of men with HPV-related disease, but high concordance levels suggest that clinicians may be missing many opportunities to encourage HPV-infected patients to notify their partners and encourage them to seek care. It is likely that many of these partners are unknowingly infected with HPV and may be at risk for HPV-related disease.

Uninfected sexual partners may be an important target population for HPV vaccination, provided they are in the recommended age range for HPV vaccine. The Advisory Committee on Immunization Practices currently recommends routine vaccination of females ages 11 or 12 years, with catch-up vaccination for females through age 26 (21, 22), while providing a permissive recommendation for HPV vaccination of males ages 9 through 26 years (23). The benefits of vaccination to individuals seronegative to HPV types included in the vaccine are clear, and emerging evidence suggests that HPV vaccine may also help people who previously had and cleared an infection (62, 63), although additional research among such individuals is needed. It is unlikely that people have been infected with all four HPV types in the quadrivalent HPV vaccine (64), making the vaccine a potentially beneficial prophylaxis for people already exposed to some HPV types. Women's higher rates of infection when their partners are HPV-positive lend further support to the recommendations of the Advisory Committee on Immunization Practices, which made a stronger recommendation for HPV vaccination among age-appropriate females than males (21-23). Our findings may provide HPV vaccine cost-effectiveness analyses with increased precision when estimating the potential consequences of having a sexual partner who is HPV-positive and how it may differ for males and females.

Strengths of our meta-analysis include a comprehensive search strategy, careful data extraction methods, and use of random-effects modeling techniques. The few additional studies identified after we searched the PubMed database suggest that we identified the most relevant published studies. We identified some important sources of variability in study concordance estimates, but because of poor reporting by primary studies, we were unable to identify additional sources of variation that we believe are likely to exist but are presently unknown. For this reason, we believe that these concordance levels should be viewed as tentative estimates that may differ for some populations.

As the field matures, new concordance studies should use more rigorous research designs. All but a few studies were cross-sectional, meaning many of the studies were not able to establish the temporality or direction of HPV transmission between sexual partners. Longitudinal designs for studies of HPV discordant couples would more effectively address the dynamics of HPV transmission, clearance, and persistent infection (37). More rigorous research designs will have the added benefit of allowing researchers to move beyond description to more robust hypothesis testing. More complex study designs would also allow researchers to better understand first transmission, reinfection, and back-and-forth passage within couples; concordance in couples in which 1 partner has received HPV vaccine; and concordance after treatment for HPV-related lesions.

Concordance studies should also characterize their participants in greater detail and study more meaningful populations. Fewer studies recruited men with HPV-related disease than recruited women with such disease. Many studies did not report basic information on participant demographics, health, sexual history, and relationship characteristics. Whereas studies often used state-of-the-art methods for determining HPV DNA prevalence, some studies also did not report basic information on study design and methods. We recommend future HPV concordance and transmission studies consider collecting and reporting the basic participant, couple, and study characteristics shown in Table 5. The studies in our meta-analysis sampled geographically diverse populations, including many in developing countries where the majority of cervical cancer deaths occur (13). However, no studies were conducted in Africa, parts of which have the highest cervical cancer mortality rates in the world (13).

Our findings can help inform cost-effectiveness analyses, which have helped guide regulatory and funding policies about HPV vaccine. High-quality longitudinal studies are needed to better understand HPV concordance and transmission among heterosexual couples. In future efforts, researchers also should be more diligent in reporting characteristics of participants, couples, and their studies. Reporting such detailed information about participants and couples may help identify high-risk subsets of individuals that warrant focused interventions to prevent HPV-related disease. Our findings suggest the need for greater attention to sexual partners of HPV-infected individuals.

N. Brewer and P. Reiter, research grant, Merck & Co.; N. Brewer, unrestricted research grant, GlaxoSmithKline.

We thank Evan Myers for his feedback on an earlier draft of the manuscript.

Grant Support: Research grants from the American Cancer Society (MSRG-06-259-01-CPPB) and the Cancer Control Education Program at Lineberger Comprehensive Cancer Center (R25 CA57726).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.