Article Text
PDF
Abstract
Objective To systematically review and appraise published data, to determine the prevalence of Mycoplasma genitalium (MG) in men who have sex with men (MSM) tested at each anatomical site, that is, at the urethra, rectum and/or pharynx.
Design Systematic review and meta-analysis.
Data sources Ovid Medline, PubMed, Embase were searched for articles from 1st January 1981 (the year MG was first identified) to 1st June 2018.
Review methods Studies were eligible for inclusion if they reported MG prevalence in MSM tested at the urethra, rectum and/or pharynx, in at least 50 MSM, using nucleic acid amplification testing. Data were extracted by anatomical site, symptom and HIV status. Summary estimates (95% CIs) were calculated using random-effects meta-analysis. Subgroup analyses were performed to assess heterogeneity between studies.
Results Forty-six studies met inclusion criteria, with 34 reporting estimates of MG prevalence at the urethra (13 753 samples), 25 at the rectum (8629 samples) and 7 at the pharynx (1871 samples). MG prevalence was 5.0% (95% CI 3.5 to 6.8; I2=94.0) at the urethra; 6.2% (95% CI 4.6 to 8.1; I2=88.1) at the rectum and 1.0% (95% CI 0.0 to 5.1; I2=96.0) at the pharynx. The prevalence of MG was significantly higher at urethral and rectal sites in symptomatic versus asymptomatic MSM (7.1% vs 2.2%, p<0.001; and 16.1% vs 7.5%, p=0.039, respectively). MG prevalence at the urethra was significantly higher in HIV-positive compared with HIV-negative MSM (7.0% vs 3.4%, p=0.006).
Conclusion MG was common in MSM, particularly at urethral and rectal sites (5% to 6%). MG was more commonly detected in symptomatic men at both sites, and more common in HIV-positive men at the urethra. MG was uncommonly detected in the pharynx. Site-specific estimates are similar to those for chlamydia and will be helpful in informing testing practices in MSM.
PROSPERO registration number CRD42017058326.
- mycoplasma
- meta-analysis
- gay men
- homosexuality
Statistics from Altmetric.com
Introduction
Mycoplasma genitalium (MG) is a STI, with prevalence estimates in the community ranging from 1.3% to 3.9%.1 The majority of published data provides estimates of urethral infection with less data available on MG infection in gay, bisexual and other men who have sex with men (MSM), particularly for rectal and pharyngeal sites. There has been one prior meta-analysis of community-based studies (n=8) that estimated the overall prevalence of MG in MSM at 3.2% (95% CI 2.1 to 5.1), but clinic studies were excluded, and estimates were predominantly derived from urine samples.1
International guidelines recommend screening for Chlamydia trachomatis (C. trachomatis) and Neisseria gonorrhoeae (N. gonorrhoeae) in MSM,2–4 although there is ongoing debate as to the relative contribution of these STIs at extragenital sites to transmission.5 Guidelines do not recommend screening for MG at any site,2 3 6 as the situation is more complicated than for C. trachomatis or N. gonorrhoeae due to lack of clarity around the natural history of MG, and increasing challenges with treatment due to antimicrobial resistance.7 Testing for MG in men with urethritis is recommended by international guidelines,2 8 as its pathogenic role in this syndrome is well established.9 MG’s association with rectal symptoms and the syndrome of proctitis has been inconsistent across published studies,7 10 with UK and Australian guidelines recommending consideration of testing for MG in men with sexually acquired proctitis.6 8 Site-specific prevalence estimates in MSM are needed to understand the contribution of each anatomical site to MG transmission, and to inform testing practices in MSM.
We undertook a systematic review and meta-analysis of published studies in order to determine the prevalence of MG at the urethra, rectum and pharynx of MSM tested for infection at each site, and to examine the association with symptoms, HIV status and other factors on site-specific prevalence, to inform testing and clinical practice in MSM.
Methods
The study protocol was registered on Prospero (ID CRD42017058326).
Search strategy and selection criteria
This systematic review and meta-analysis was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (online supplementary table 1). Two authors (RLL, HSS) independently searched for published peer-reviewed studies reporting the prevalence of MG in MSM by anatomical site, from 1st January 1981 (the year MG was first identified) to 1st August 2017, with the search updated on the 1st June 2018.
Supplemental material
We included cross-sectional, longitudinal and cohort studies, and randomised controlled trials where baseline MG prevalence was reported, and conference abstracts from major STI conferences between 1st January 2015 and 1st June 2018 for studies that may not have been published yet. The search was performed using Ovid Medline, PubMed and Embase using search terms ((‘mycoplasma genitalium’ or ‘mycoplasma infections’ or ‘mycoplasmosis’) and/or (‘men adj3 sex’ or ‘males adj3 sex’ or ‘homosexual’ or ‘homosexuality’)) (online supplementary table 2). Medical Subject Headings (MeSH) were used where possible. Reference lists of included studies were reviewed to identify other relevant studies. Studies were eligible for inclusion if they were published in English and reported MG prevalence at the urethra, rectum and/or pharynx, in at least 50 MSM, to reduce small study bias. Prevalence at each anatomical site was defined as the prevalence among MSM tested for MG at each anatomical site. Definition of MSM varied between studies. Participants were assumed to be MSM if studies stated that men either had male sexual partners, or if men had a rectal swab collected.11 12
Two reviewers (RLL, HSS) independently screened studies for eligibility using Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia). Abstracts and titles were screened, and then articles reviewed. Differences were resolved by discussion with other reviewers (CSB, LAV and/or DAM). When multiple articles reported on the same study population, we either included the most comprehensive article or the original article published.
Data extraction
Variables extracted included: name of first author, publication year, study design, geographical location (using WHO regions),13 study period, study setting, median or mean age, HIV status, symptom status, symptom location, clinical diagnoses, number of participants, laboratory testing methods and number positive for MG by specimen type (urethral swab, first-void urine, rectal swab, and/or pharyngeal swab). Prevalence estimates with 95% CIs were also extracted. One reviewer (RLL) extracted the data, and a second reviewer (HSS) checked data for transcription errors. Any disagreements were discussed between the two reviewers, with differences resolved by discussion with CSB. Authors were contacted for data where estimates were not stratified by anatomical site, HIV or symptom status, or if the study design suggested that additional information was available. Where data was only presented as percentages and authors did not provide raw data on contact, numerical values were generated from percentages if sufficient data were available. Where CIs were not reported, these were calculated using binomial methods.
Primary outcome
The primary outcome was the prevalence of MG in MSM at the urethra, rectum and/or pharynx. This was calculated as the number of men who tested positive for MG at each site (defined as a positive test using nucleic acid amplification testing) divided by the total number tested for MG at the site.
Secondary outcome
Secondary outcomes included the prevalence of MG either by symptom status (asymptomatic vs symptomatic), HIV status (HIV-negative vs HIV-positive) or recruitment setting or location. Analyses were dependent on sufficient data being available.
Analysis
Summary prevalence estimates were calculated using random-effects meta-analysis with Freeman-Tukey double arcsine transformation, and study-specific 95% CIs computed using score method.14 Included studies were examined using forest plots. I2 statistics were calculated to assess between-study heterogeneity when more than two studies were included, with values of <25%, 25%–75% and >75% representing low, medium and high heterogeneity, respectively. The χ2 statistic was used to assess the strength of the evidence for heterogeneity.
We undertook subgroup analyses and univariable random-effects meta-regression by symptom status, HIV status and broad WHO geographic region,13 to investigate potential sources of heterogeneity. Sensitivity analyses were conducted to determine the effect on summary estimates of removal of either: i) outlier studies, or ii) studies at high risk of being non-representative of MSM. Studies were considered outliers if they fell outside the overall pattern of distribution for prevalence at each site. Studies were considered at risk of not representing the wider MSM community if a) the sample was not a true representation of the broader MSM population, for example select HIV patients, or b) MSM were not clearly defined as men who had penetrative sex in the past year with another man. Data were analysed using Stata (V.14.0, StataCorp, Austin, Texas, USA).
Assessment of bias and quality
The potential presence of publication and small study bias was assessed using funnel plots of proportions against study samples size and the Egger test.15 To evaluate within study bias, we adapted an instrument designed by Hoy et al, which examines both the internal and external validity of selected studies (online supplementary table 3).16 The tool consisted of eight questions and two reviewers (RLL, HSS) independently assessed each study as being low, medium or high risk of bias for each item, with differences resolved by discussion with CSB.
Results
Study selection
The search process identified 4518 records and 2595 duplicates were removed. Titles and abstracts of 1923 records were screened and 1586 excluded, leaving 337 records for full-text screening. Forty-six records were included in the final analysis (figure 1, included studies described in online supplementary table 4).
Flow diagram of studies included in meta-analysis of the prevalence of Mycoplasma genitalium among men who have sex with men (MSM). epub, electronic publication; n, number; NAAT, nucleic acid amplification testing.
Study characteristics
Of the 46 included studies, 34 reported MG prevalence at the urethra (n=13 753 samples), 25 at the rectum (n=8629 samples) and 7 at the pharynx (n=1871 samples) (table 1). Sample sizes ranged from 51 to 6293 (online supplementary table 4). Nineteen studies provided information on the presence of symptoms,10 17–34 and 29 on HIV status.10 18–22 24–30 32 33 35–48 Most studies (n=20 (42%)) were from Europe,18 23 26 29 33 34 37 38 40 42 44–46 49–55 13 (28%) were from the Western Pacific, although all were from Australia and China,10 17 20 28 30–32 35 41 47 48 56 57 12 (26%) were from the Americas, predominately the USA,19 21 22 24 25 27 36 39 58–60 W1 and the WHO Africa region contained only one South African study (2%),43 (table 1, online supplementary table 4). Overall, 36 (78%) studies were cross-sectional,10 17–26 28 33 35–37 39–42 44–49 51–59 W1 6 (13%) were retrospective cohorts29–31 34 50 60 and 4 (9%) were prospective cohort studies27 32 38 43 (table 1, online supplementary table 4).
Characteristics of included studies (n=46)
Mycoplasma genitalium prevalence at the urethra
Prevalence of MG at the urethra (n=34 studies) ranged from 0% to 25.5%, with a summary estimate of 4.6% (95% CI 3.0 to 6.4; I2=94.4%) (table 2).
Analyses assessing prevalence of Mycoplasma genitalium by anatomical site
Of the 15 studies reporting symptom status at the urethra, 12 included symptomatic men and 9 asymptomatic men. Prevalence of MG was higher among men with urethral symptoms (7.1% (95% CI 4.7 to 9.7; I2=67.2)) than among asymptomatic men (2.2% (95% CI 1.0 to 3.7; I2=81.8)) (p<0.001) (table 2, online supplementary figure 1).
Twenty studies examining prevalence of MG at the urethra reported HIV status; 18 studies included men living with HIV and 14 included HIV-negative men. Urethral prevalence of MG was higher among men living with HIV (7.0% (95% CI 3.0 to 12.2; I2=86.3)) than among HIV-negative men (3.4% (95% CI 1.8 to 5.5; I2=82.6)) (p=0.006) (table 2, online supplementary figure 2).
Prevalence of urethral MG varied across geographical regions (table 2), although assessment of geographical regions was impacted by the limited number of countries providing data in each region. Prevalence of urethral MG was 1.3% (95% CI 0.2 to 6.9) in the single African study, 2.2% (95% CI 1.3 to 3.3; I2=75.3) across European studies (n=14 studies), 7.4% (95% CI 4.8 to 10.6; I2=86.5) in studies from the Americas (n=8 studies) and 8.2% (95% CI 4.0 to 13.6; I2=97.1) across studies from the Western Pacific (n=11 studies). Prevalence of MG did not differ by recruitment setting (table 2).
Mycoplasma genitalium prevalence at the rectum
Prevalence of MG at the rectum (n=25 studies) ranged from 0% to 29.9%, with a summary estimate of 6.1% (95% CI 4.5 to 7.9; I2=89.0) (table 2). One outlier study by Ong et al reported a prevalence of 29.9%.30 A sensitivity analysis excluding this study resulted in a summary estimate of 5.4% (95% CI 4.2 to 6.8; I2=81.9) (online supplementary table 6).
Of the seven studies reporting symptom status at the rectum, five included symptomatic men and four asymptomatic men. Prevalence of MG was higher among men with rectal symptoms (16.1% (95% CI 7.2 to 27.5; I2=82.9)) than asymptomatic men (7.5% (95% CI 5.4 to 10.0; I2=72.5)) (p=0.039) (Table 2, online supplementary figure 3).
Eleven studies examining prevalence of MG at the rectum reported HIV status; all included men living with HIV, with seven also including HIV-negative men. Rectal prevalence of MG was not different among men living with HIV (10.6% (95% CI 5.5 to 17.0; I2=89.4)) compared with HIV-negative men (6.8% (95% CI 1.2 to 15.8; I2=94.4)) (p=0.456) (Table 2, online supplementary figure 4).
The prevalence of rectal MG varied across geographical regions (table 2). Prevalence of rectal MG was 0.0% (95% CI 0.0 to 4.7) in the single African study, 4.3% (95% CI 3.1 to 5.7; I2=56.0) across European studies (n=10 studies), 6.3% (95% CI 4.1 to 8.9; I2=65.0) in studies from the Americas (n=7 studies) and 9.5% (95% CI 5.1 to 15.1; I2=95.5) across studies from the Western Pacific (n=7 studies). Prevalence of MG did not differ by recruitment setting (table 2).
Mycoplasma genitalium prevalence at the pharynx
Prevalence of MG at the pharynx (n=7 studies) ranged from 0% to 13.4%, with a summary estimate of 1.0% (95% CI 0.0 to 5.1; I2=96.0) (table 2, online supplementary figure 5). Six of the seven studies reported pharyngeal prevalence between 0% and 2%. One outlier study by Jiang et al reported a prevalence of 13.4%.57 A sensitivity analysis excluding this study resulted in a summary estimate of 0.0% (95% CI 0.0 to 0.3; I2=36.6) (online supplementary table 6). As symptom status was only reported in one study,32 subgroup analysis by symptoms was not performed.
Two studies examining prevalence of MG at the pharynx reported HIV status. One study included MSM living with HIV only, and the other included both men living with and without HIV, with no difference in the prevalence of pharyngeal MG by HIV status (p=0.891) (table 2).
The prevalence of pharyngeal MG did not differ by geographical region, ranging from 0.0% (95% CI 0.0 to 5.2) across European studies (n=2 studies), to 1.8% (95% CI 0.3 to 9.3) in studies from the Americas (n=1 study) and 1.8% (95% CI 0.0 to 10.3) across studies from the Western Pacific (n=4 studies) (table 2). Prevalence of MG did not differ by recruitment setting (table 2).
Between-study bias
Assessment of funnel plots indicated no small study effects (online supplementary table 6-8), although inclusion criteria required studies to contain ≥50 MSM. There was no evidence of publication bias by the Egger test, with a coefficient of 0.23 (95% CI −0.65 to 1.11; p=0.573).
Within-study bias
Online supplementary table 5 provides a summary of risk of bias for included studies, using the adapted tool from Hoy et al (online supplementary table 3). Twenty-one studies (46%) were deemed to be at low risk of bias, 23 (50%) at medium risk and 2 studies (4%) at high risk of bias.60 W1 Only one study was considered as low risk of bias across all domains.40 Thirteen studies randomly or consecutively selected participants or specimens.10 18 22–25 32 40 44 49 52 55 Seven studies did not state the study population to be MSM but used rectal samples from men,49 50 52 54 55 60 W1 which for the purpose of this meta-analysis were assumed to be from MSM.11 12 W2 Four papers did not provide a numerator or denominator, instead presenting percentage data.41 53 57 60 Sensitivity analyses which removed studies deemed to be at high risk of representative bias or outlier studies did not show any major shift in summary estimates (online supplementary table 6,7).
Discussion
To our knowledge, this is the first meta-analysis to determine the prevalence of MG by anatomical site in MSM tested for infection at each site, and to explore the association between MG and specific factors such as symptom and HIV status. We combined data from 46 studies, representing nearly 24 000 samples. Prevalence of MG in MSM ranged from 0% to 29.9%, with summary estimates of 6.2% (95% CI 4.6 to 8.1; I2=88.1) at the rectum, 5.0% (95% CI 3.5 to 6.8; I2=94.0) in the urethra and 1.0% (95% CI 0.0 to 5.1; I2=96.0) in the pharynx, but estimates were limited by high heterogeneity. MG was more common in MSM with symptoms at the urethra or rectum and more common at the urethra in MSM living with HIV.
The prevalence of MG in this study was similar to C. trachomatis prevalence estimates in MSM, which have been reported in the order of 3.5%–3.7% at the urethra, 5.6% (95% CI 4.8 to 6.3) at the rectum and 0.5% (95% CI 0.2 to 0.9) at the pharynx.W3-W5 Screening for C. trachomatis and N. gonorrhoeae at urethral, rectal and pharyngeal sites in MSM is recommended in many countries.2–4 However, WHO states a number of key principles must be satisfied when screening for an infection,W6 including: 1) the natural history of the condition should be understood and 2) the cost-benefit of finding and treating cases should be balanced.W6 MG meets neither criteria, with limited knowledge of the natural history of asymptomatic MG infection, particularly in MSM, and increasing antimicrobial resistance resulting in greater use of costly antimicrobials that can be associated with serious side effects.7
Our meta-analysis confirmed MG to be uncommonly detected in the pharynx of MSM (1%), although estimates were limited by small numbers of included studies. When the outlier study by Jiang et al was excluded, the overall estimate declined to 0.0% (95% CI 0.0 to 0.3; I2=36.6).57 In contrast to N. gonorrhoeae, where pharyngeal prevalence is 5.5%–8.3%,W7 W8 pooled estimates of MG appear similar to those for chlamydia at 0.5% .w5 These data suggest the pharynx does not play a significant role in transmission of MG.
While MG is an established cause of urethritis, its contribution to proctitis has been less clear. Several studies have reported no association between rectal MG and proctitis.7 22 46 However in this meta-analysis MG was associated with symptoms at both the urethra and rectum. These data support recommendations that MG testing should be undertaken in men with urethral symptoms,2 8 and suggest that MG is associated with rectal symptoms in MSM, supporting UK and Australian guidelines that recommend consideration of MG testing in men with proctitis.6 8
Our meta-analysis found the prevalence of MG to be higher at the urethra among men living with HIV compared with HIV-negative men. The pattern was similar at the rectum, but there were fewer studies, limiting this comparison. A prior meta-analysis reported an association between MG and HIV (OR=2.01, 95% CI 1.44 to 2.79),W9 but only two studies with data specific to MSM were included.W8 While our data suggests MG may be more common at the urethra and rectum in HIV-positive men, specific recommendations around testing for MG in men living with HIV cannot be inferred from these results due to the high heterogeneity and limited number of studies. Overall however, these data add plausibility to a relationship between MG and HIV, as has been previously reported.W9
There were several important limitations to this study. Random-effects meta-analysis takes into consideration real or larger heterogeneity between studies, but despite this we recorded high heterogeneity in most of the summary estimates. Symptoms and HIV status were only known for a subset of studies, reducing precision around these estimates. Assessment of the influence of geographical region on MG prevalence was greatly impacted by the limited number of countries providing regional data. While the prevalence of MG appeared highest in the Western Pacific, studies from this region only originated from Australia and China. Similarly, only one South African study contributed to African estimates, and there were no studies from South East Asian or Eastern Mediterranean regions. High heterogeneity may have resulted from a combination of all three of the risk factors explored (ie, symptoms, HIV status and geographical location), but the small number of studies available for each end point limited our ability to perform multivariable meta-regression analyses.14 W10 As information on potentially important factors such as age and risk behaviour were not consistently reported, we could not assess the impact of these on heterogeneity. There was limited power to look at the influence of time (ie, year of study) within different subgroups. Few of the included studies were from systematically sampled populations, as screening is not recommended for MG, and testing has generally been limited to symptomatic individuals and specialist services with access to research assays, resulting in over-representation of STI clinics compared with other sites. Thus, estimates may represent a select population, contributing to the significant heterogeneity and limited generalisability of the findings. Finally, this review was limited to studies published in English only, which also limits the generalisability of our findings; however, only four studies were excluded on this basis.
Our meta-analysis found the prevalence of MG to be similar to that of C. trachomatis in MSM tested across all three anatomical sites, with associations seen between symptoms and HIV-positivity in subgroup analyses. This meta-analysis provides site-specific estimates for MG in MSM that, despite high heterogeneity, represent biologically plausible patterns of infection. These data provide an evidence base to inform testing and clinical practice, and highlight the need for further research in this population to understand the pathogenic role and natural history of MG in MSM.
Key messages
Prevalence of Mycoplasma genitalium (MG) at the urethra, rectum and pharynx was estimated in men having sex with men (MSM) tested for infection at each site.
MG prevalence was 5.0% (95% CI 3.5 to 6.8; I2=94.0) at the urethra; 6.2% (95% CI 4.6 to 8.1; I2=88.1) at the rectum and 1.0% (95% CI 0.0 to 5.1; I2=96.0) at the pharynx.
The prevalence of MG was higher at urethral and rectal sites in symptomatic versus asymptomatic MSM.
MG prevalence at the urethra was higher in HIV-positive compared with HIV-negative MSM.
Acknowledgments
The authors would like to thank Lorena Romero for assistance with creating the search strategy for this project. The authors would also like to thank the following researchers taking the time to respond to emails inquiring about the availability of data, and where requested sourcing stratified data from their databases from prior published research projects: Britain’s third National Survey of Sexual Attitudes and Lifestyles (Natsal-3), V. Bremer, S. Bruisten, L. Chen, C. Cox, R. Cunha, I. De Baetselier, A. de Jong, J. Dionne-Odom, R. Douglas, A. Ebel, C. Gillespie, J. Gratrix, D. Gesink, D. Getman, S. Hakre, D. Ham, S. Hillier, H. Jalal, K. Jansen, J. Jensen, M. Kamb, A. Kreuter, J. Kriesel, M. Lefebvre, A. Libios, L. Manhart, V. Massari, S. Mehta, C. Mercer, L. Meyn, T. Mezzini, A. Moghaddam, S. Mukherjee, E. Munson, V. Padovese, S. Pereyre, R. Peters, E. Picot, T. Read, J. Reynolds-Wright, P. Rice, D. Richardson, N. Shah, K. Shigehara, A. Singh, K. Templeton, A. Upton, M. Unemo, C. van der Veer, B. Wang, L. Xiao.
References
Footnotes
Handling editor Nicola Low
Twitter @lenkavod, @EricPFChow
Presented at This work was previously presented as a poster at the STI & HIV 2019 World Congress, Vancouver, Canada. Poster P525: Prevalence of Mycoplasma genitalium by anatomical site in men who have sex with men: a systematic review and meta-analysis.
Contributors All authors contributed significantly to the work.
Funding RLL is supported by an Australian Government Research Training Program (RTP) Scholarship. TRHR was supported by NHMRC early career fellowship no. 1091536.
Competing interests EPFC reports grants from Merck & Co., grants from Seqirus Australia, outside the submitted work; CB reports that Melbourne Sexual Health Centre has received funding from Speedx outside the submitted work.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.