Methods

Participants

Participants were HIV-negative and HIV-positive men, self-identifying as gay or bisexual, who attended the GUM clinic in Brighton for STI screening between February and July 2008. The clinic operates an open access system that sees symptomatic and asymptomatic patients, although some HIV-positive men were recruited through an appointment-only designated HIV–STI clinic at the same site. An information leaflet with study details was offered to all patients, and written informed consent was obtained from all participants. Data on the number of men not asked or who declined to participate and their reasons for declining were not captured.

Data collection

Data were collected on demographics, sexual behaviour over the previous 3 months and clinical symptoms and signs, by retrospective case-note review. Urethral symptoms and signs included: dysuria, urethral discomfort and visible discharge. Rectal symptoms and signs included: discharge, pain, tenesmus, change in bowel habit, bleeding, mucosal erythema and ulceration. Urethral and rectal smear microscopy was performed only in those subjects reporting urethral and rectal symptoms, respectively, reflecting the clinic policy. NGU was defined as polymorphonuclear leucocyte (PMNL) counts of 5 cells/hpf or more on urethral smear. As rectal smears were not obtained from asymptomatic men, the dataset for rectal PMNL counts was incomplete and therefore was not analysed. Data regarding past history of STI and concurrent STI were retrieved from the clinic's computerised results reporting system and case notes. For those participants who attended the clinic for screening more than once during the study period, only their first visit was included for analysis.

Specimen collection

All nurses and doctors who consulted with patients in clinic were involved in recruitment, clinical examination and the obtaining and labelling of specimens. Proctoscopy was performed on all patients to obtain rectal swab specimens unless the patient declined, and in these cases (<1% of participants) a blind swab was taken. In those patients who declined to give either a urine or rectal specimen, the unpaired specimen was processed as usual and included for analysis. MG detection was performed on the routine clinical specimens and no additional specimens were required from participants.

Laboratory testing

Following routine CT nucleic acid amplification testing in the local laboratory by strand displacement amplification using the BD ProbeTec assay (Becton-Dickinson Diagnostic Systems, Sparks, Maryland, USA) residual first-void urine (FVU) and rectal swab specimens were sent to the Sexually Transmitted Bacteria Reference Laboratory of the Health Protection Agency, Colindale, UK, for MG testing. Samples were stored at 4°C before testing and all specimens were processed within 6 weeks of collection. MG DNA was detected using a previously published real-time PCR method that targets the MgPa adhesin gene, with a minor modification to the forward primer.7 8 All MG-positive specimens were also examined using a confirmatory real-time PCR that targets the Mg219 gene.8 Culture methods were used for the detection of urethral and rectal GC.

Follow-up

Patients were not informed if they tested positive for MG unless they had presented or re-presented with symptoms of recurrent urethritis, and in these cases they were treated with 500 mg azithromycin stat followed by 250 mg once a day for 4 days.

Statistical analysis

Statistical analyses were performed using EpiData Analysis and STATA (V. 10). To investigate significant associations between risk factors and MG-positivity, those factors with p values of 0.2 or less and elevated odds ratios (OR) from a single variable analysis were considered for inclusion in a multivariable logistic regression model. The multivariable modelling procedure consisted of a backwards stepwise approach, whereby at each step the least significant value with a p value greater than 0.05 was removed unless it was a substantial confounder (ie, a variable, whose omission leads to a change of 10% or more in the OR of the variables remaining in the model). p Values for comparing the proportion positive between tests were obtained using the McNemar's test for paired observations.

Results

Demographics

Four hundred and thirty-eight MSM were recruited to the study between February and July 2008. Mean age was 36 years (range 16–81 years) and 94.4% of men were of white ethnicity. Ninety out of 438 men (20.5%) were known to be HIV positive. HIV status was recorded for all men, although not all men underwent HIV testing at this visit. The clinic serves a local MSM population of approximately 15 000, and diagnoses between 90 and 110 cases of HIV in MSM annually.

Clinical specimens

Four hundred and twelve paired sets of rectal and FVU specimens were obtained and 26 men gave FVU specimens alone. Of the total 412 rectal and 438 urine samples processed, 20 rectal and 11 FVU specimens were MgPa real-time PCR positive, respectively. When examining these specimens using the confirmatory Mg219 PCR assay, 26 (84%) specimens were confirmed as positive; however, due to the reproducibility of the MgPa-positive results and the extensive validation of the MgPa gene as an appropriate target for the laboratory detection of MG, all 31 MgPa-positive results were regarded as true positives.

Overall prevalence and symptom/sign associations

The overall prevalence of MG at either site in this patient group was 6.6% (95% CI 4.5% to 9.4%) compared with CT 7.8% (95% CI 5.4% to 10.7%) and GC 5.0% (95% CI 3.2% to 7.5%). The prevalence of urethral MG infection was 2.7% (12/438) and rectal MG infection was 4.4% (19/438) (see table 1).

At the urethral site, urethral symptoms (11/12), particularly dysuria (8/12), were significantly associated with MG positivity (p<0.001), and all smears for urethral microscopy in symptomatic men showed PMNL counts greater than 20 cells/hpf. There was no association between rectal symptoms and rectal MG (p=0.3). The dataset for rectal smear microscopy was incomplete and therefore was not analysed. MG was isolated simultaneously at both urethral and rectal sites in two subjects, both of whom were HIV positive but had no other concurrent STI. Urethral and rectal MG were significantly associated with the presence of non-chlamydial NGU (p<0.001), but were not associated with concurrent CT or GC infection (p>0.999 for both pathogens). There was some evidence of an increasing risk of MG positivity with increasing numbers of sexual partners (see table 2).

Prevalence in HIV-positive men

When analysed according to HIV status, stark differences in the prevalence of MG were noted between HIV-positive and HIV-negative MSM. MG positivity rates in HIV-positive MSM in FVU and rectal specimens were 7/90 (7.8%) and 12/83 (14.1%), respectively, compared with 5/348 (1.4%) and 6/329 (1.8%) in HIV-negative MSM (see table 3).

MG was significantly associated with HIV positivity; a finding not observed with either CT or GC (see table 4). Furthermore, MG was more prevalent than CT (p=0.15) and GC (p=0.02) in this group but not in MSM overall. The symptom association of dysuria with the presence of urethral MG was significant in this group of MSM (p=0.04) although no association between rectal MG and anorectal symptoms was observed.

Discussion

Urethral MG in MSM

The observed prevalence of urethral MG was low but similar to that of CT and GC. Nearly all men with urethral MG had symptomatic urethritis and smears for urethral microscopy indicated evidence of marked inflammation, supporting findings from previous work.9 10

Interestingly, only nine of 57 (15.8%) non-chlamydial NGU diagnoses were attributable to MG, leaving other sexually transmitted pathogens such as Ureaplasma urealyticum, Trichomonas vaginalis and herpes simplex virus as possible causes. More commonly in MSM, organisms including coliforms and oropharyngeal flora as well as trauma may account for other cases, although in clinical practice, these would all have been assumed to be bacterial infections and treated with 1 g oral azithromycin in this clinic.

Rectal MG in MSM

The prevalence of rectal MG was 4.4%, comparable to rates of rectal CT and GC in MSM, although MG was not associated with symptoms of proctitis. Francis et al6 reported a prevalence of 5.4% among 500 MSM in San Francisco, but more recently, a large study of MSM at sex-on-premises venues in Australia found a lower rectal MG prevalence (1.6%)11 and in both studies there was no clear association with symptoms of proctitis.

At present it is not known if MG is a rectal pathogen or is simply carried at this mucosal site. Previous studies examining rectal prevalence have detected the organism in healthy individuals, although those studies have had small participant numbers or used culture rather than molecular techniques for detection.5 12 We were unable to compare rectal mucosal inflammation between MG-negative and MG-positive participants, as routine rectal microscopy is not undertaken in asymptomatic individuals at our clinic. Previous work has found no association of PMNL counts with rectal MG infection (D Taylor-Robinson, unpublished data) but further research is warranted to investigate this association. Following rectal inoculation with MG it is possible that MG persists and colonises the rectum, although this may still provide a reservoir for onward urethral infection in those MSM practising insertive anal intercourse.

MG in HIV-positive MSM

HIV-positive MSM are disproportionately affected by STI compared with their HIV-negative counterparts,13 and the difference in prevalences of MG observed between these two groups in our study reflects this. Francis et al6 reported similar observations between HIV-positive and HIV-negative MSM in 2005 with respect to rectal MG, and Martinelli et al14 detected MG in urethral specimens from 52 out of 187 (27.8%) HIV-infected heterosexual male injecting drug users compared with only eight out of 114 healthy volunteers. On the other hand, Hartmann et al15 found no association with HIV status in a study in which only three out of 61 (5%) male and female HIV-infected patients tested positive for MG, although that study did not specify the gender or sexuality of the participants.

Sexual behaviour could account for our findings; HIV-positive men may be more likely to change sexual partners more frequently and may have more unprotected sex with other HIV-infected men. Nevertheless, our study shows that despite a correlation of higher numbers of partners with MG infection, unprotected anal sex (self-reported) was not associated. Alternatively, mucosal T-cell immunodeficiency in the rectum may make HIV-positive men more susceptible to persistent infection with MG.

The interplay between STI and rising incident HIV infection has gained much attention over recent years, with several studies linking the acquisition of HIV to co-infection with bacterial STI and genital ulcer disease.16–18 It follows that an organism such as MG, with the potential to cause mucosal inflammation, may also facilitate HIV transmission. Interestingly, an increased risk of HIV transmission was suggested in one study comparing HIV-concordant and HIV-discordant couples19; HIV-positive concordance was associated with the presence of antibodies to MG but not to syphilis, CT or hepatitis B. Although no studies so far have looked specifically at HIV shedding in seminal fluid in the presence of MG infection, a recent study showed increased HIV-DNA shedding from the cervices of women infected with MG who had high mycoplasmal DNA loads.20 Sadiq et al21 showed increased HIV-RNA shedding in semen samples from men with gonococcal or chlamydial urethritis, which fell following treatment of the urethritis. Further work is needed in this area, with specific regard to MG, to investigate its role in HIV transmission.

Treatment of MG urethritis with doxycycline has low clearance rates,22 and whereas earlier studies reported high success rates with a single dose or extended courses of azithromycin,23 more recently, significant failure rates with azithromycin have been noted.24 Given that both these antibiotics are recommended for the treatment of NGU and non-specific proctitis in MSM, the potential for undiagnosed and persistent MG infection exists. Clinicians should remain alert to the possibility of MG as a cause of persistent urethritis and consider testing specifically for MG in men presenting with ongoing urethral symptoms.

There are some limitations of this study that should be considered. First, data were collected retrospectively, thereby relying on thorough history taking and careful case-note documentation. Self-reported sexual behaviour is subject to inaccuracies in recall and only behaviour in the preceding 3-month period was collected. Second, the actual numbers of men who declined to participate or were not offered the chance to participate in the study were not recorded. Although we acknowledge this as a potential source of bias, we observed that most men attending the clinic were offered the study and very few declined to participate. Third, the study was conducted from a single centre, limiting the sample population to Brighton; more studies are needed in other UK cities to determine whether or not this population is representative.

In this study, we have shown high rates of MG infection among HIV-positive MSM at both urethral and rectal sites, and that MG is more prevalent in HIV-positive MSM than other bacterial STI. While national guidelines advise screening for urethral and rectal GC and urethral CT in all MSM attending GUM clinics,25 there is currently insufficient evidence to support similar recommendations for MG in this group. The introduction of a commercial assay for MG detection should allow further work in large and more widespread MSM populations to clarify the role of MG in the rectum and to inform whether screening for and empirical treatment of MG should be considered in MSM, particularly in those who are HIV infected and report high-risk sexual behaviour.