High Prevalence of Antibiotic-Resistant Mycoplasma genitalium in Nongonococcal Urethritis: The Need for Routine Testing and the Inadequacy of Current Treatment Options

1Mycoplasma genitalium and Chlamydia trachomatis are recognized causes of common genital tract morbidities, such as pelvic inflammatory disease and nongonococcal cervicitis in women and nongonococcal urethritis (NGU) in men. UK- and US-based antimicrobial treatment guidelines for these syndromes are aimed at Chlamydia infection, which is more frequently detected and has a strong evidence base for causing chronic disease, whereas M. genitalium is not usually recommended among organisms for routine testing in routine sexually transmitted infection screening. NGU, diagnosed using both clinical presentation with Gram-stained urethral smear microscopy, is empirically treated with a 1-week oral course of doxycycline or a single oral dose of 1 g of azithromycin. Both are effective against C. trachomatis, although clinical failure to eradicate Chlamydia with single-dose azithromycin has been reported. For M. genitalium, doxycycline has poor efficacy, evidenced by high rates of clinical failure with tetracyclines; treatment failure with single-dose azithromycin is increasingly evident and associated worldwide with strains possessing single-nucleotide polymorphisms (SNPs) in its ribosomal RNA gene (domain V, 23S rRNA). Moxifloxacin, a fluoroquinolone demonstrating efficacy against M. genitalium, is recommended following treatment failure with macrolides and tetracyclines. Fluoroquinolone resistance–associated SNPs in gyrase (gyrA) or topoisomerase IV (parC) genes have been detected in M. genitalium from patients exhibiting fluoroquinolone treatment failure and phenotypic resistance in vitro.

2In this study of men symptomatic of urethritis, we investigated the prevalence of M. genitalium in comparison to established causes of urethritis and determined frequencies of known genetic resistance markers to macrolides and fluoroquinolones within M. genitalium. We sought to establish whether prevalence of M. genitalium and resistance was local only to our population or could be applicable more generally, using a previously validated dual-locus typing system.

3This was an observational study of all men prospectively presenting with symptoms of urethritis to the Genitourinary Medicine Clinic at St George's Hospital, Tooting, London, between 28 September and 15 December 2011. This study was conducted with the approval of the South West London Research Ethics Committee (number Q0803 71). Patients were designated as having urethritis, NGU, nonchlamydial NGU, or no urethritis. “Urethritis” was defined as any patient receiving treatment specifically for urethritis in that clinical episode and at least 1 of the following: (1) recent history of dysuria, urethral discomfort, or urethral discharge, and visible urethral discharge on examination; (2) ≥5 neutrophils per high-power field on urethral Gram stain. NGU was defined as urethritis where gonorrhea was excluded by at least 2 of Gram stain, nucleic acid amplification test (NAAT), and routine gonococcal culture. Nonchlamydial NGU was defined as NGU where chlamydia had been excluded by NAAT. Only patients consenting to routine urethral smears were included in the study. Patients had a urethral smear prepared with a cotton-tipped swab or plastic loop and were asked to provide a maximum of 50 mL first-catch urine specimen. Clinical notes of patients with urethritis were reviewed at 3 months after presentation to assess instances of treatment failure.

4First-void urine samples were collected routinely for chlamydia and gonorrhea NAATs using the Viper system (Becton Dickinson, Oxford, UK). The residual of these samples were retrieved and used for detection of M. genitalium and T. vaginalis using primers as previously described. DNA extracts positive for M. genitalium were investigated for SNPs in the quinolone resistance–determining regions (QRDRs) of gyrA, gyrB, and parC and macrolide resistance SNPs in 23S rRNA and also typed using a multiple locus variable number tandem repeat marker in MG309 (MG-309-STR) and mgpB SNP analysis. Further details of the protocol are provided in Supplementary Data 1.

5To detect the difference of 25% and 4% of M. genitalium in patients with and without urethritis, respectively, with 80% power and α =.05, 111 urethritis and nonurethritis cases each were required. Prevalence of infections was compared between urethritis and no urethritis and also compared within urethritis cases for the same infection, using χ2 and McNemar test, respectively.

6MG191 sequence type (ST) type was assigned on the basis of previously described types numbered 1–80. Maximum likelihood phylogeny was reconstructed from ClustalW alignments using RAxML v7.3.2 using a generalized time reversible (GTR) model of nucleotide substitution with gamma model of rate heterogeneity. Branch support values were generated from 1000 bootstrap replicates.

7217 men were recruited, 18 of whom were men who have sex with men (MSM). One hundred ten of 217 had any urethritis by study definition and 102 of 217 had NGU. The prevalence of C. trachomatis and M. genitalium were similar in NGU cases and both significantly higher than in those with no urethritis (Table 1). The prevalence of both M. genitalium and C. trachomatis among MSM and heterosexual men with NGU was similar (data not shown). The clinic database was checked for all male attendees during the exact period of the study. Of 184 men diagnosed with NGU, 14 were diagnosed with C. trachomatis, giving a prevalence of 7.6% (95% confidence interval [CI], 4.4%–12.7%), lower than in our study. This may have reflected a different threshold for diagnosing NGU in the general clinic compared to our study criteria.

8Genotypic resistance data were available for both the macrolide and fluoroquinolone antibiotics for all 22 M. genitalium–positive clinical samples (Table 2). Macrolide resistance was detected in 9 samples (41%; [95% CI, 20%–62%]). A single sample, that is, 4.5% (95% CI, 1%–21%), possessed a mutation in the QRDR of parC associated with fluoroquinolone resistance, and this genotype possessed no genotypic resistance markers of macrolide antibiotics. The parC sequence of the isolate containing a SNP within the QRDR and the sequences of domain V of the 23S rRNA gene have been deposited within GenBank under the nucleotide accession numbers HF947096 and HF572928–HF572950, respectively.

9There were no obvious clinical associations with antibiotic resistance. Macrolide-associated resistance markers were carried in 6 of 17 and 3 of 5 of those with NGU and no urethritis, respectively.

10In 1 of these patients, carrying A2058G, the clinical notes indicated reception of single-dose azithromycin for a prior episode of NGU some months previously, raising the possibility of either treatment failure or selection of macrolide resistance from the previous episode. However, none of the 6 cases with NGU and macrolide resistance, 2 of whom were treated with doxycycline and 4 with single-dose azithromycin, returned to the clinic in the following 6 months with episodes of persistent or recurrent urethritis.

11All 22 M. genitalium DNAs were successfully sequence typed by the MG191 SNP typing system. Eighteen of 22 were successfully typed using MG-309-STR. Table 2 displays MG191 (mgpB) and MG-309-STR types together with 23S rRNA type. Eleven of 22 M. genitalium infections were caused by 8 STs identical to MG191 alleles in the literature, designated ST 2, 3, 4, 8, 21, 23, 24, and 44. The remaining 11 sequence types exhibited novel MG191 alleles which we have designated as Novel STs A–J (Table 2). Sequences used for typing were deposited in GenBank under accession numbers KC445139 to KC445161 for the mgpB locus and KC445162 to KC445182 for the MG309 locus.

12Successfully typed MG-309-STR sequences were combined end-to-end with their respective MG191 sequences to form a dual locus sequence type (DLST). Phylogenetic analysis revealed a tree of 18 DLSTs, divided clearly into 2 major clusters, which we designated Major Cluster A (MCA) and B (MCB) (Figure 1). A phylogenetic tree, constructed using only MG191 sequences on study sequence types, produced identical assignments into these major clusters with poorer bootstrap values (data not shown). We therefore assigned the 4 sequence types, in which MG-309-STR had failed, to these major clusters accordingly (Table 2). Seven of 12 MCA types exhibited genotypic resistant mutations (4 A2058G, 2 A2059G, and 1 A2059C), compared with 2 of 10 in MCB (A2058G and A2059G) (Table 2 and Figure 1). We determined the phylogenetic relationship of study M. genitalium STs with 80 other previously typed strains from Southern Europe and North Africa, which were typed using MG191-ST alone (Supplementary Figure). The tree suggests that our study M. genitalium STs, including those carrying 23S rRNA mutants, are distributed widely in this group of STs.

13Mycoplasma genitalium was a frequent but undiagnosed cause of NGU, as prevalent as C. trachomatis in our symptomatic, mainly heterosexual male cohort. In addition, >40% of M. genitalium infections had genotypic resistance to macrolide antibiotics, one of the recommended first-line options for NGU. A single patient, approximately 5% of the cohort, was infected with a strain demonstrating genotypic resistance to fluoroquinolones. Patients with M. genitalium infection, with or without genotypic resistance, separated into distinct genotypic clusters and were also related to a diverse population of international control sequence types. This indicated that our findings, rather than being a local clonal phenomenon that could have biased the data, were more likely to represent established infection and resistance rates in this patient population as a whole.

14This also complements evidence that selection of macrolide-resistant M. genitalium during suboptimal treatment contributes to cases of treatment failure rather than ongoing horizontal transmission of resistant bacterial strains. Overall, our findings indicate that there is a need to provide routine testing for M. genitalium in those presenting with symptoms of urethritis and perhaps cervicitis and that first-line recommended antimicrobial options for treating these syndromes need reappraisal.

15These treatment options will need to take into account the need to have effective regimens against both C. trachomatis and M. genitalium and should question the value of recommending either doxycycline or single-dose azithromycin alone. Whereas extended-dose azithromycin for NGU may be more effective in preventing the development of macrolide resistance and treatment failure, the in vitro and clinical data suggest that it will not work for those carrying 23s rRNA resistance-associated SNPs. Future effective NGU regimens are likely to include fluoroquinolones. Solitary parC mutations, like the one documented in this study, have been reported in cases of fluoroquinolone treatment failure of M. genitalium, and worryingly, there already exist reports of moxifloxacin treatment failure in M. genitalium–associated persistent NGU. Collectively, these findings point to the need to evaluate effectiveness of new treatment strategies, including combination therapies and novel antimicrobials.

16Effective management can be further guided by providing routine testing for M. genitalium infection, and tests of cure following therapy. The prospect of rapid and accurate tests for infections that include validated genetic markers of resistance will allow for appropriately targeted treatment and reduce need for complex regimens. Current rapid test platforms in development, including those recently approved for C. trachomatis and Neisseria gonorrhoeae, offer potential for such resistance tests to be realized.

17The prevalence of M. genitalium in NGU in our study was perhaps at the lower end of the range of that described in previous work of 15%–25%. Recent reports in South Africa mirror our findings of higher rates of M. genitalium infection relative to C. trachomatis, and importantly, comparable rates of macrolide resistance and emerging evidence of resistance to resistance to fluoroquinolones were reported in Sydney, Australia, suggesting that the problem may be globally widespread.

18Although this study is limited by being based in a single clinic as well as conducted over a relatively short period of time, we addressed this by assessing the genetic diversity of M. genitalium within our cohort, utilizing a typing system with the discriminatory power to be useful in the study of sexual networks. This robustly revealed 2 major clusters of M. genitalium, both of which contained drug-resistant mutations. Each cluster itself contained different MG191 STs, many of which had been previously identified. The MG191 typing system alone is less discriminatory and more useful to describe broad epidemiological STs. Using this system, we showed that most STs within each cluster were not closely related as they were distributed broadly among international genotypes also typed using MG191.

19In conclusion, the prevalence of macrolide-resistant M. genitalium in this population is much higher than previously thought and M. genitalium is detected as frequently as C. trachomatis in cases of NGU. Current testing and treatment guidelines should be modified to incorporate these findings.

20Supplementary materials are available at Clinical Infectious Diseases online (http://cid.oxfordjournals.org). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.