Global Type-Specific Genital Human Papillomavirus Prevalence in Men, by Sexual Orientation: A Systematic Review and Meta-Analysis

Abstract

Background

Knowledge on genital type-specific human papillomavirus (HPV) prevalence among men is important for prevention of HPV-related cancers and other diseases. Men who have sex with men (MSM) have higher anal prevalence than men who have sex with women only (MSW) but for genital HPV this is unclear. We performed a systematic review and meta-analysis of type-specific genital HPV prevalence among men, by sexual orientation.

Methods

MEDLINE and Embase were used for searching publications reporting on male genital HPV prevalence with data from November 2011 onwards. A random-effects meta-analysis was conducted estimating pooled type-specific and grouped external genital and urethral HPV prevalence. Subgroup analyses were conducted for sexual orientation.

Results

Twenty-nine studies were eligible. Of those, 13 studies reported prevalence among MSM, 5 among MSW, and 13 studies did not stratify by sexual orientation. The most common genotypes were HPV-6 and HPV-16 for both anatomical locations, although heterogeneity was high. HPV prevalence was similar among studies reporting on MSW, MSM, and men with unknown sexual orientation.

Conclusions

Genital HPV is common among men, with HPV-6 and HPV-16 being the most common genotypes. Type-specific HPV genital prevalence appears to be similar among MSM and MSW, which contrasts with earlier findings on anal HPV.

Genital HPV is common among men, with HPV-6 and HPV-16 being most common for the external genitals and urethra. No difference is observed for men who have sex with men and men who have sex with women.

Human papillomavirus (HPV) is a nonenveloped DNA virus, which is the most common sexually transmitted infection in the world [1, 2]. There are more than 200 HPV genotypes, labelled as either low-risk HPV (lrHPV) or high-risk HPV (hrHPV), based on their oncogenic potential [2, 3]. The most common clinical manifestation of lrHPV is genital warts, which are highly infectious; 65% of people with sexual partners with genital warts will develop genital warts themselves, and recurrence of warts is common [4]. HrHPV infections can cause several types of cancer. Among women 570 000 and among men 60 000 annual cancer cases are attributable to hrHPV [5]. Annually, an estimated 26 000 penile cancer cases are diagnosed globally, of which approximately 50% are HPV related. Additionally, 80% of the penile intraepithelial neoplasia, a precancer stage, are associated with HPV [5–8].

Most HPV-related research focuses on HPV epidemiology among women, because cervical cancer is the most common HPV-attributable cancer [5]. As a result, HPV prevention strategies are often based on the epidemiology of HPV infection among women, but the role of men in female infection, and the disease burden for men themselves, might be overlooked [9, 10]. Asymptomatic infected men serve as a reservoir for HPV infection and infect women, making male genital HPV infection an important source of cervical HPV infection [11–13]. Moreover, studying genital HPV among men is important for prevention of HPV-related disease in men, such as penile lesions, HPV-related cancer, and genital warts. Therefore, knowing the HPV prevalence among men is important, especially hrHPV-types.

Systematic reviews and meta-analyses have been published on genital HPV prevalence for men. However, they were either published over 10 years ago [14, 15], lacked type-specific information of genital HPV among men [10, 14–16], only focused on national or continental data [16, 17], summarized only findings among men who have sex with men (MSM) [18, 19], did not take sexual orientation into account [10, 14, 16, 20], or focused on anatomical sites other than genital, like oral or anal sites only, or pooled anal and genital infection [19, 21–24]. Therefore, we aimed to summarize the most recent global type-specific genital HPV prevalence data for men, stratified by anatomical genital location and sexual orientation, within a systematic review and meta-analysis.

METHODS

This review has been registered at PROSPERO under protocol number CRD42021260147. The current systematic review and meta-analysis has been executed as outlined in the protocol, with minor deviations in stratifications due to unavailability of required detailed information from publications to conduct planned stratified analyses.

Search Strategy

A systematic review and meta-analysis of the most recent literature was conducted, and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) and Meta-analyses of Observational Studies in Epidemiology (MOOSE) guidelines [25]. The scientific databases Embase and MEDLINE were used, with key search terms being (human papilloma virus OR papillomavirus infection OR papillomaviridae) AND (male OR men) AND (incidence OR prevalence OR epidemiology OR occurrence OR frequency). Full details on the search strategy can be found in the Supplementary Material. All articles were obtained on 25 October 2021 for further evaluation.

Eligibility Criteria

Studies were included if: (1) they were peer reviewed with original data from randomized controlled trials, cohort studies, cross-sectional studies, or other studies that provide HPV prevalence; (2) they assessed the genital type-specific HPV prevalence via any well-described genotyping methodology with samples taken from either the external genitals or urethra; (3) the study population consisted of adolescent or adult men; and (4) data collection occurred between 1 November 2011 and 25 October 2021. If a study started its data collection before 1 November 2011, but part of the data was collected after November 2011, the study was included.

Excluded were (1) systematic reviews, commentaries, case reports, abstract-only publications, and unpublished manuscripts; (2) studies analyzing samples obtained only from transgender people; (3) studies with samples only taken from other anatomical locations than the external genitals or urethra, such as anal, oral, urine, or semen samples; (4) studies that specifically recruited symptomatic men; or (5) studies that recruited men based on HPV-related disease or positive HPV status of their sex partner. Language restrictions were not applied.

Data Extraction and Bias Assessment

All data selection and extraction were done by 2 reviewers, J. K. and J. B., independently. If no consensus could be obtained, a third reviewer, M. S. v. d. L., was consulted. Articles were screened on titles and abstracts for eligibility, with help of the online software Rayyan (http://rayyan.qcri.org/). Subsequently, possibly eligible articles were read in full text and a final selection was made. A predefined standardized data extraction form was used to extract information on the following variables: title, first author, year of publication, country, year(s) of data collection, sexual orientation (might also be described as sexual practices, sexual identity, or sexual behaviors; MSM could also be described as gay or bisexual men; men who have sex with only women [MSW] as heterosexual men), anatomical location of the swab, total number of samples taken, age (mean, median, or mode), method of genotyping, information on HIV status, tested HPV types, and number of positive samples per genotype. Number of positive samples were taken for individual genotypes and where possible, grouped prevalence for any HPV and hrHPV. For randomized controlled trials and longitudinal cohort studies, baseline data were extracted, so all observations were cross-sectional and sufficiently homogenous to synthesize study data. If studies lacked information on sexual orientation, type-specific prevalence, time of data collection, method of genotyping, age, or methods of the study, the lacking information was requested from the corresponding author via email. This was done for 37 studies. A maximum of 2 reminders to request additional information was sent. Reminders were sent for 29 articles. In total, 12 authors replied, of whom 6 gave additional information.

Assessment for quality and potential bias was done using of the JBI Critical Appraisal Checklist for Studies Reporting Prevalence Data [26] by J. K. and J. B., independently. This checklist consists of 9 items which can be answered with “yes,” “no,” or “unknown.” If at least 6 items were answered with “yes,” the study was judged to be of sufficient quality (see Supplementary Material).

Statistical Analysis

Meta-analyses were done and forest plots were created using the “Metaprop” function from package “Meta” in RStudio (version 1.3.959). Type-specific HPV prevalence (for lrHPV genotypes 6 and 11, and 14 hrHPV genotypes 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68) [3] and grouped HPV prevalence (any HPV; any hrHPV) summary estimates and 95% confidence intervals (CI) were calculated. In light of potential high heterogeneity in populations sampled, and HPV DNA detection methods, with DerSimonian-Liard random-effects [27] meta-analysis for external genital and urethral HPV infection separately. Freeman-Tukey double arcsine transformations [28, 29] was applied to stabilize the variance and distributions. Heterogeneity between studies was analyzed with the I² index [30, 31]. Values for I² less than 25% were regarded as low, between 25% and 75% as medium, and over 75%, as high heterogeneity.

Subgroup Analyses

Subgroup analyses were done by sexual orientation, that is, MSM, MSW, and men with unknown sexual orientation. A requirement for the subgroup analysis was availability of at least 5 studies per subgroup.

Heterogeneity Assessment

Univariable meta-regression analyses were conducted to assess potential sources of heterogeneity in study-level variables for (1) year of publication; (2) recruitment at sexual health centers (SHCs) or not; (3) number of genotypes assessed in the study; and (4) sample size of the study (<300, ≥ 300). A P value of <.05 was considered to indicate a significant relationship between the predictor and heterogeneity.

Bias and Sensitivity Analysis

For estimates based on more than 10 studies, publication bias was tested using Egger test, with P < .05 indicating significant publication bias [32, 33]. A first sensitivity analysis consisted of performing meta-analyses with only studies without substantial bias as judged in the bias assessment. A second sensitivity analysis was conducted excluding studies that did not report prevalence stratified by sexual orientation.

Only 11 studies provided some information on human immunodeficiency virus (HIV) status. As the participants of those studies differed in sexual orientation and anatomical location, this led to subgroups that were too small. Therefore, no sensitivity analysis could be performed for HIV status.

RESULTS

Literature Search

The search strategy identified a total of 2906 records, resulting in 1928 articles after deduplication (Figure 1). After screening titles and abstracts, 187 articles were read in full text for eligibility, resulting in 29 articles with data from a total of 17 802 men for the systematic review and meta-analysis (Table 1). Of these studies, 22 analyzed the HPV prevalence of the external genitals (15 648 men), 5 studies of the urethra (2014 men), and 2 studies analyzed samples of both the external genitals and urethra (140 men). Thirteen studies reported prevalences for MSM (3320 men), 5 for MSW (2063 men), 7 had a combination of MSM and MSW without stratifying results by sexual orientation (4140 men), and 6 studies did not mention sexual orientation (8326 men) (Table 1). Studies from all continents were included in the study (Africa n = 1, Americas n = 5, Asia n = 12, Europe n = 8, Australia n = 3). Of the 29 studies, 22 were considered to have sufficient quality (Figure 1). Details of the quality assessment can be found in the Supplementary Material.

Figure 1.

Flow chart presenting the steps for study selection from the databases. Abbreviation: HPV, human papillomavirus.

Table 1.

Overview of all Included Articles in the Main Analysis

Country of Data Collection	First Author (Year of Publication)	Years of Data Collection	No. of Men With Genital Specimen	Age, y, Mean, Median, or Mode (Range, IQR, or SD)	Anatomical Site	No. of Genotypes Tested	Assay Used for Genotyping
MSM only
Australia	Chow (2021)	2010–2018	356	Mean 18.8 (range 16–20)	EG	28	Anyplex II HPV28 Detection Assay
France	Cotte (2021)	2012–2014	115	Median 34 (IQR 27.3–41.7)	EG	28	Anyplex II HPV28 Detection Assay
Italy	Sammarco (2016)	2013	50	Mean 38 (IQR 20–53)	EG and UR	37	Promega Restriction Fragment Length Polymorphism
Italy	Uccifferi (2018)	2013	90	Mean 35 (IQR 22–53)	EG and UR	37	PrimmBiotech Sanger sequencing
Netherlands	Woestenberg (2020)	2009–2017	514	Median 22 (range 16–24)	EG	25	DDL Diagnostics Laboratory SPF10 DEIA-LIPA
Taiwan	Lin (2018)	2013–2016	279	Median 26 (IQR 23–31)	EG	37	Roche Linear Array Assay
Taiwan	Strong (2020)	2015–2016	250	Mean 28.89 (range 20–50)	EG	37	Roche Linear Array Assay
Thailand	Chuerduangphui (2018)	2014–2015	346	Mode 21–30 (range 18–60)	UR	37	Reverse line blot hybridization
UK	King (2015)	2010–2012	446	Median 30 (range 18–40)	EG	21	Bio-Plex Luminex xMAP
USA	Kahn (2019)	2012–2015	133	Median 23 (range 13–26)	EG	33	Not mentioned
USA	Winer (2021)	2016–2018	665	Mean 23 (range 18–26)	EG	37	Roche Linear Array Assay
MSW only
Australia	Chow (2019)	2014–2017	298	Mean 18.4 (range 17–19)	EG	37	Roche Linear Array Assay
Australia	Machalek (2017)	2014–2016	511	Median 23 (range 16–23)	EG	37	Roche Linear Array Assay
Iran	Davarmanesh (2020)	2017–2018	415	Mean 33 (SD 8.14)	EG	14	Roche Cobas 4800 HPV test
MSM and MSW, reported separately
Brazil	Wendland (2021)	2016–2017	819	Median 21.13 (range 16–25)	EG	37	Roche Linear Array Assay
Thailand	Wittawatmongkol (2019)	2013–2014	49	Median 18 (IQR 17–20)	EG	37	Roche Linear Array Assay
MSM and MSW, not reported separately
China	Xin (2017)	2015–2016	198	Median 33 (range 18–66)	EG	37	Hybribio 37 HPV GenoArray Diagnostic Kit
Japan	Matsuzawa (2020)	2011–2015	798	Mean 55.4 (range 20–94)	EG	31	GENOSEARCH HPV31 kit
Malaysia	Khoo (2021)	2014–2016	389	Median 40 (range 18–60)	EG	16	Life Technologies BGISEQ-100
Netherlands	Luttmer (2015)	2011–2012	170	Median 22 (range 18–64)	EG	25	DDL Diagnostics Laboratory SPF10 DEIA-LIPA
Russia	Popova (2021)	2018–2019	256	Median 35 (range 18–67)	UR	14	Amplisens HPV HCR genotype
South Africa	Chikandiwa (2017)	2011–2012	283	Mean 38 (SD 8)	EG	37	Roche Linear Array Assay
USA	Gargano (2017)	2013–2014	2046	Range 14–59	EG	37	Roche Linear Array Assay
Sexual orientation not reported
China	Wang (2021)	2015–2020	3690	Mean 41.6 (range 20–85)	EG	37	HybriBio HPV Genotyping Kit
China	Wei (2018)	2014	1513	Range 18–55	EG	16	Melting curve analysis
China	Xiang (2021)	2017–2019	1711	Mean 34.73 (range 17–89)	EG	23	Yaneng BIOscience reverse line blot hybridization
China	Zhang (2017)	2013–2015	233	Range 19–83	UR	15	Multiplex PCR SLAN-96P
Mexico	Gallegos-Bolaños (2017)	2010–2012	166	Mean 36.07 (range 21–68)	UR	18	Digene hybrid capture assay
Russia	Donnikov (2019)	2010–2016	1013	Mean 30 (SD 8.6)	UR	21	HPV Quant-21

Abbreviations: EG, external genital; HPV, human papillomavirus; IQR, interquartile range; MSM, men who have sex with men; MSW, men who have sex with women only; PCR, polymerase chain reaction; UR, urethral.

Type-Specific HPV Prevalence

For the 2 HPV genotypes with highest oncogenic potential, HPV-16 and HPV-18, stratified forest plots of the pooled prevalence, including subgroup analyses, are displayed in Figure 2. For external genital, the pooled HPV-16 prevalence based on 24 studies was 3.6% (95% CI, 2.4%–5.1%), with high heterogeneity across the studies (I² = 98%) (Figure 2A). The pooled HPV-18 prevalence was 1.8% (95% CI, 1.2%–2.5%) (Figure 2B). Heterogeneity across the studies was also high for HPV-18 (I² = 83%).

Figure 2.

Forest plots for the estimations of global HPV-16 (A) and HPV-18 (B) infection at the external genital and urethral sites among men. Subgroup test is the test for differences by sexual orientation. Abbreviations: CI, confidence interval; HPV, human papillomavirus; I², heterogeneity; n, number of HPV-positive samples; MSM, men who have sex with men; MSW, men who have sex with women only; N, total number of samples; P, proportion positive for HPV.

The pooled prevalence of urethral HPV-16 was 5.3% (95% CI, 1.7%–10.5%), with high heterogeneity across the 7 studies (I² = 94%) (Figure 2A). The pooled urethral HPV-18 prevalence was 1.6% (95% CI, .1%–4.6%) (Figure 2B). Again, heterogeneity across studies was high (I² = 92%).

The genotype with the highest pooled prevalence was HPV-6; the pooled external genital prevalence was 5.6% (95% CI, 3.1%–8.8%) and the pooled urethral prevalence was 5.5% (95% CI, 1.3%–12.1%) (Table 2). The second most prevalent genotype was HPV-16 for both anatomical locations. For most genotypes, the pooled urethral prevalence was slightly lower than the external genital prevalence, with the exception of HPV-16, HPV-33, and HPV-58. The heterogeneity was high for all genotypes at both anatomical locations, except for external genital infections of HPV-31, HPV-33, HPV-39, and HPV-56, all having medium heterogeneity (Table 2).

Table 2.

Type-Specific Genital Pooled Prevalence Estimates Among Men Globally for External Genitals and Urethra

Genotype	External Genital							Urethral
	No. of Reports	n	Pooled %	95% CI	I 2, %	Subgroup Test P Value	Egger Test P Value	No. of Reports	n	Pooled %	95% CI	I 2, %
Low risk
HPV-6	22	15 056	5.6	3.1–8.8	98	.73	.25	4	1319	5.5	1.3–12.1	89
HPV-11	22	15 056	3.0	1.4–5.2	97	.83	.42	4	1319	2.2	.2–5.7	79
High risk
HPV-16	24	15 641	3.6	2.4–5.1	94	.42	.20	7	2154	5.3	1.7–10.5	92
HPV-18	24	15 641	1.8	1.2–2.5	83	.93	.95	7	2154	1.6	.1–4.6	94
HPV-31	23	15 226	0.7	.4–1.1	73	.91	.52	6	1808	0.6	.0–3.4	92
HPV-33	22	15 056	0.6	.3–1.0	69	.02	.42	6	1808	1.5	.1–3.8	85
HPV-35	20	14 451	0.8	.4–1.3	83	.19	.93	6	1808	0.6	.0–2.8	88
HPV-39	21	14 840	1.9	1.4–2.4	66	.97	.80	6	1808	1.0	.0–3.6	90
HPV-45	22	15 056	1.1	.7–9.6	85	.55	.15	6	1808	0.9	.0–3.1	87
HPV-51	22	15 010	2.8	1.9–3.9	88	.94	.73	6	1808	2.2	.3–5.7	90
HPV-52	22	15 056	2.3	1.6–3.1	84	.06	.53	6	1808	1.7	.4–3.7	77
HPV-56	23	15 226	1.0	.6–1.4	73	.24	.63	6	1808	0.9	.0–4.2	93
HPV-58	22	15 056	1.8	1.1–2.8	90	.29	.31	7	2154	2.1	.8–4.0	79
HPV-59	22	15 010	2.2	1.4–3.2	88	.69	.35	6	1808	1.3	.0–3.4	85
HPV-66	22	15 010	2.2	1.3–3.3	91	.07	.61	4	1409	1.6	.0–6.3	92
HPV-68	21	14 840	1.3	.7–2.1	86	.61	.49	6	1808	0.6	.0–2.7	88

Subgroup test is the test for differences between sexual orientation subgroups. P value in bold indicates statistical significance (P < .05).

Abbreviations: CI, confidence interval; HPV, human papillomavirus; I², heterogeneity.

Grouped HPV Prevalence

The pooled grouped prevalence for any HPV at the external genitals, was 35.8% (95% CI, 29.3%–42.6%), and at the urethra 25.9% (95% CI, 6.4%–52.6%) (Supplementary Figure 1). The pooled grouped prevalence for any hrHPV at the external genitals was 21.0% (95% CI, 16.6%–25.8%); at the urethra this was 12.7% (95% CI, 8.9%–17.0%) (Supplementary Figure 2). All pooled grouped estimates had high heterogeneity across the studies.

Subgroup Analyses

No difference by sexual orientation in external genital HPV-16 prevalence was observed (P = .42), nor for HPV-18 (P = .93)(Figure 2). For the other type-specific external genital prevalences, no significant differences by sexual orientation were found, except for HPV-33 (Table 2); HPV-33 prevalence differed significantly by sexual orientation (MSM = 0.8%, 95% CI, .03%–1.6%; MSW = 0.3%, 95% CI, .0%–.6%; unknown sexual orientation = 0.7%, 95% CI, .2%–1.4%). No significant differences by sexual orientation were found for either of the pooled grouped estimates (P = .89 for any HPV, P = .66 for any hrHPV). For urethral prevalence, subgroup analyses for sexual orientation could not be conducted, as the requirement of having at least 5 studies in each subgroup was not met.

Heterogeneity Assessment

Univariable metaregressions were conducted for 4 variables (ie, year of publication, recruitment at SHCs, number of genotypes in assay, study sample size). None of the variables could explain the heterogeneity across studies, both for type-specific and grouped prevalence estimates (Supplementary Table 1).

Bias and Sensitivity Analyses

No significant publication bias was found using the Egger test for any of the type-specific or grouped external genital estimates. For urethral estimates, publication bias could not be estimated due to the low number of included studies.

In the first sensitivity analysis, analysis was limited to articles meeting at least 6 of 9 quality criteria; this was the case for 22 of 29 articles (Figure 1). Details of the quality assessment can be found in the Supplementary Material. Of those 22 studies, only 1 study assessed urethral HPV prevalence, making a sensitivity analysis for urethral HPV not possible.

After exclusion of studies with insufficient quality, pooled external genital prevalences for HPV-16 and HPV-18 were comparable to the prevalences of the main analysis (HPV-16, 4.0% vs 3.6%; HPV-18, 1.9% vs 1.8%) (Supplementary Table 2). The pooled grouped estimates were similar to the main analyses as well (any HPV, 36.8% vs 35.8%; any hrHPV, 21.3% vs 21.0). Heterogeneity of all pooled estimates was similarly high compared to the main analysis, and no differences in the subgroup analyses were found compared to the main analysis (Supplementary Table 2).

The second sensitivity analysis, with only studies that reported prevalences of men with known sexual orientation, was conducted for external genital HPV only, as the number of studies on urethral HPV became too low. Pooled type-specific and grouped HPV prevalences were comparable to the main analysis. The heterogeneity for many of the type-specific estimates was lowered compared with the main analysis; of the 16 types, 9 had medium heterogeneity in this sensitivity analysis, ranging from I² = 41% to I² = 72%. The remaining genotypes remained high in heterogeneity ranging from I² = 77% to I² = 89% (Supplementary Table 3).

DISCUSSION

This systematic review and meta-analysis provides an up-to-date summary of global genital HPV prevalence among men. The pooled external genital and urethral prevalences were comparable for the type-specific and grouped estimates, and HPV-6 and HPV-16 were most common. Hardly any differences in type-specific and grouped prevalences were found between MSW, MSM, and men with unknown sexual orientation. This is in contrast to earlier findings in anal HPV, where MSM had a consistently higher HPV prevalence than MSW.

To our knowledge, this is the first study summarizing and reporting urethral and external genital HPV prevalence for men at a type-specific level. Additionally, our study included publications from all continents. This is the first systematic review making a comparison for genital HPV prevalence between MSM, MSW, and men with unknown sexual orientation. As homosexuality is illegal or socially unaccepted in some countries, some studies did not provide information on sexual orientation. In a comparison between MSM and MSW only, studies without information on sexual orientation would be excluded, potentially leading to selection bias. By adding the third category of unknown sexual orientation, our study gives a better summary of global male genital HPV prevalence. Finally, a methodological strength is the use of the Freeman-Tukey double arcsine transformation, which minimizes the risk of underestimation of the size of the confidence interval, and an overestimation of the degree of heterogeneity across the observed proportions when having observed proportions close to 0 or close to 1, as for the type-specific estimates in our study [34].

Limitations of our study should be mentioned. Firstly, studies used different HPV assays, varying in sensitivity and specificity; this could lead to an increased heterogeneity of observed prevalences. Additionally, grouped prevalences were taken directly from the studies. As not all studies genotyped the same genotypes, and different definitions were used for hrHPV between studies, grouped prevalence estimates are hard to interpret. It should be mentioned that the most oncogenic hrHPV genotypes, that is HPV-16 and HPV-18, were included in all hrHPV definitions and there was large overlap in used hrHPV genotypes. Unfortunately, not all articles reported essential information needed for inclusion in our systematic review. Many of our requests for additional information were unanswered, especially from studies conducted in low- and middle-income countries. Therefore, these studies were excluded from the meta-analysis, potentially leading to selection bias. A final limitation was that it was not possible to conduct subgroup or meta-regression analyses for important determinants of HPV prevalence (ie, HIV status, prevalence by age, age of sexual debut, number of recent sex partners, or continents) as this information was often missing or numbers were too low.

We found that HPV-6 was the most prevalent HPV genotype at external genitals and urethra, followed by HPV-16. Only 1 other meta-analysis has reported genital type-specific HPV prevalence for men for external genital and urethral HPV, which was among MSM only [18]. That study found slightly higher estimates than in our study (eg, external genital HPV-16, 4.6% vs 3.6%). This might be due to the fact that only studies with information on HIV status were included in that meta-analysis. Possibly only participants with higher risk profiles to acquire HPV knew their HIV status. The same meta-analysis among MSM is the only meta-analysis reporting grouped HPV prevalences for both the external genitals and urethra, with any HPV prevalences similar to our estimates (external genital, 36.4% vs 36.2%; urethral, 14.8% vs 15.4%) [18]. Other systematic reviews did not include meta-analyses, and described very broad ranges of prevalence for any HPV among men, making a comparison with our study results not possible [14, 15].

High heterogeneity was found in our meta-analysis, despite our efforts to minimize or explain observed heterogeneity. For example, to minimize methodological heterogeneity (ie, differences due to study design), only baseline data of cohort studies and randomized controlled trials were used, making these studies more comparable to cross-sectional studies. Additionally, we assessed if the sample size could explain methodological heterogeneity, which was not the case. A variation in prevalences was expected due to different sampling and HPV DNA detection methods [4], leading to potential clinical heterogeneity (ie, differences in composition of the study population, and differences in measurements). Meta-regression by assay was not possible, as the variety in DNA detection assays was large. As a proxy, we used the number of genotypes tested for, which did not explain heterogeneity. Recruitment at SHCs might result in a study population with a higher risk of genital HPV infection than recruitment elsewhere leading to clinical heterogeneity. Nevertheless, in our meta-regression this also did not explain heterogeneity. It is therefore possible that male genital HPV prevalence differs across the world and across populations, leading to high heterogeneity. This is likely as high heterogeneity is also found in meta-analyses for anal HPV among men and genital HPV among women [18, 23, 35], due to different prevalences around the world.

Hardly any differences in pooled genital prevalence between MSW and MSM were observed, which is in line with a study among SHC clients [36]. This is a contrast to estimates for anal HPV, where MSM have significantly higher prevalences than MSW [23]. An explanation for higher anal HPV prevalence among MSM versus MSW is that most MSM have receptive anal intercourse and MSW, by definition, do not. Additionally, MSM in general have more sexual partners compared with MSW, which are both risk factors for anal HPV [37]. A higher genital HPV prevalence among MSM than among MSW would be expected, assuming that most MSW acquire genital infections from women with vaginocervical infections and most MSM acquire genital infections from MSM with anal infections, combined with the much lower genital HPV prevalence in women [38] than anal HPV prevalence in MSM [23]. Therefore, our finding of similar genital prevalences between MSW and MSM is unexpected. A possible explanation may be that the majority of the study populations in our meta-analyses included men with a young mean or median age, and in young adult women the genital HPV prevalence is much higher than among older women, even comparable to anal HPV prevalence of MSM [39]. Young MSW may therefore be exposed to women with a high genital HPV prevalence, and young MSM may be exposed to MSM with a high anal prevalence.

Male genital HPV prevalence is suggested to stay stable over age [15, 40], in contrast to HPV prevalence in women which decreases sharply after the age of 25 [35]. Why this is not reflected in genital prevalence among MSW is a conundrum, as is the question of whether the age-specific genital HPV prevalence is different for MSW and MSM. Therefore, HPV transmission dynamics of penile-vaginal intercourse and penile-anal intercourse need further elucidation. Additionally, studies included in our review did not report whether MSM had predominantly insertive or receptive sex. Future research should explore differences in genital HPV prevalence between those groups. Another important research gap we found is that few studies estimated genital HPV stratified by HIV status. HIV increases the risk for anal and cervical HPV infection and related cancers [41, 42]; it is likely to be an important risk factor for male genital HPV infection too. Therefore, more research on the dynamics of HIV and genital HPV for both MSW and MSM is recommended.

No clear differences in genital HPV prevalence between MSM and MSW were found. Over the last 15 years many countries have introduced HPV vaccination. This probably reduced genital HPV prevalence among (young) men (both MSM and MSW) directly in countries with gender-neutral vaccination; herd immunity may have reduced male genital HPV prevalence among MSW in countries with girls-only vaccination [43]. We encourage future researchers to study genital HPV prevalence for both MSM and MSW, and to study whether trends in genital HPV prevalence vary between these groups.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank the libraries of the National Institute of Public Health and the Environment and of Amsterdam UMC location AMC for their help in building a search strategy. We would like to thank Jenny Chen and Liangzi Zhang for their help in translating articles written in Chinese. Finally, we thank the authors who responded to our request for additional information.

Disclaimer . The funder had no role in study design, data collection and analysis, interpretation of data, decision to publish, or preparation of the manuscript.

Financial support. This work was supported by the Ministry of Public Health, Welfare and Sport, the Netherlands.