Article Text
PDF
Abstract
Objectives Resistance to both macrolides and fluoroquinolones has been reported in Mycoplasma genitalium; however, due to limited diagnostics, studies are often small and confined to specific geographical areas. This study sought to determine the rate of predicted resistance in M. genitalium-positive specimens referred for diagnostic testing.
Methods Seventy-four M. genitalium-positive specimens, referred to the national reference laboratory (2010-2013) from 19 centres across England, were blinded and anonymised. Specimens were examined for markers predictive of resistance to macrolides and fluoroquinolones using PCR followed by sequence analysis of 23S rRNA gene, or gyrA and parC, respectively.
Results 23S rRNA gene PCR sequencing revealed that 82.4% (61/74) of specimens harboured a single nucleotide polymorphism (SNP) associated with macrolide resistance. Differences were observed between the rates of predicted macrolide resistance in male (95.1% (58/61)) and female (23.1% (3/13)) patients (P = <0.001). By contrast, all specimens for which sequencing data were available (73/74) yielded wild-type gyrA sequences; and 58/61 (95.1%) had wild-type parC genes. Three specimens (3/61 4.9%) had SNPs in the parC gene associated with fluoroquinolone treatment failure, and all three also had predicted resistance to macrolides.
Conclusions Eighty-two per cent and 4.9% of M. genitalium specimens had SNPs associated with macrolide and fluoroquinolone resistance, respectively. Due to lack of widespread availability of testing for M. genitalium in the UK, this study sample was likely to be sourced from patients who may have already failed first-line macrolide therapy. Nevertheless, this study highlights the need for both greater access to M. genitalium diagnostics and genetic antimicrobial resistance testing.
- m genitalium
- antibiotic resistance
- treatment
Statistics from Altmetric.com
Introduction
Mycoplasma genitalium is a cause of urethritis in men and cervicitis in women,1 and is also increasingly associated with upper genital tract infection and pelvic inflammatory disease (PID).2–4 Prevalence of infection in sexual health attenders in England is unknown because M. genitalium detection is not part of the routine STI health screen; therefore, published data regarding this tend to be from small-scale studies.5–7 Opportunistic screening of the general population (both in England and in other countries) suggests M. genitalium prevalence rates of approximately 2% in women and less in men,8–11 while higher rates (3%–17%) have been reported in patients attending sexual health clinics.12–21
Tests for M. genitalium are not widely available in primary diagnostic laboratories in the UK. To aid patient management where M. genitalium infection was suspected, the national reference laboratory at Public Health England (PHE), London implemented an assay for the molecular detection of the organism. In the absence of M. genitalium testing, men with non-gonococcal urethritis (NGU) are usually treated syndromically, with a focus on treating presumptive Chlamydia trachomatis, with either one dose of azithromycin 1 g or doxycycline 100 mg twice daily for 7 days.22 Specimens sent to the reference service are most likely received from those patients who have failed treatment.
M. genitalium is typically susceptible to macrolide, tetracycline and fluoroquinolone antimicrobials in vitro.23 Treatment with tetracyclines is not recommended due to high rates of clinical failure,24 and so macrolides and fluoroquinolones are currently recommended as first-line and second-line therapy, respectively.22 However, as with other bacterial STIs, resistance to these antimicrobials is of growing concern. Treatment failures associated with resistance to azithromycin have been reported.21 25–31 Resistance in M. genitalium is rarely detected phenotypically, but rather is inferred by detecting single nucleotide polymorphisms (SNPs) in the 23S rRNA gene (for macrolide resistance), or gyrA and parC genes (for fluoroquinolone resistance). SNPs in region V of the 23S rRNA gene that are known markers of macrolide resistance have been observed in between 14.2%26 and 58%20 of cohorts tested. Of increasing concern are reports by Jensen et al 29, Twin et al 27 and Couldwell et al 30 of paired pre- and post-treatment patient specimens, where resistance-associated SNPs were present only in post-treatment samples, indicating that the treatment itself was selecting for minority resistant variants. In addition, there have been reports of treatment failures with moxifloxacin, and detection of fluoroquinolone resistance-associated SNPs in the parC and gyrA genes21 28 30 32 in between 5%21 and 15%33 of the patient specimens tested. Detection of these SNPs did not appear to be associated with prior antimicrobial therapy.30 Of significant public health concern is the emergence of M. genitalium strains that appear to carry resistance-associated mutations to both azithromycin and moxifloxacin28 30 as there are very limited alternative treatment options.
Currently, the only resistance data for M. genitalium available for England were reported by Pond et al.21 The study was limited to one inner-London sexual health clinic and the specimens were from male patients only. Pond et al described macrolide resistance-associated mutations in 41% of M. genitalium-positive clinical specimens, while a fluoroquinolone resistance-associated mutation was detected in the parC gene in a single clinical specimen. This specimen had wild-type sequence in the 23S rRNA gene.21 The following study was performed to determine the prevalence of SNPs associated with resistance to macrolides and fluoroquinolones in M. genitalium-positive specimens referred to the national reference laboratory and collected from both male and female patients across England between the start of 2010 and the end of 2013.
Materials and methods
Cohort tested
Eight hundred and fifty-eight clinical specimens were referred to the national reference laboratory for the detection of M. genitalium in the period 2010–2013 from 23 geographically diverse laboratories in England. Of these, 109 (12.7%) (80 males, 14 females, 15 unknown) were positive for M. genitalium-specific DNA using a modified MgPa adhesion gene real-time PCR34 (forward primer MgPa355F modified to 5′GAG AA(A/G) TAC CTT GAT GGT CAG CAA3′) on initial testing. Specimens were stored at −20°C. Patient identifiers were removed from records of positive specimens before performing resistance testing in this study. Due to anonymisation it was not possible to identify if all specimens were from unique patients or if some patients gave mutliple specimens for testing.
Twenty M. genitalium-positive clinical specimens (20/109) were unavailable for further testing, that is, insufficient specimen remained. Total DNA was extracted from 89 (89/109, 81.7%) clinical specimens (referred from 20 different laboratories) using the MagNA Pure Compact Nucleic Acid Isolation Kit I (Roche). Eighty-four specimens (84/89, 94.4%) (71 males, 13 females) from 19 different laboratories were reconfirmed as M. genitalium-positive (using the same assay as previously). The remaining five specimens (5/89, 5.6%) were M. genitalium-negative on retesting, and the DNA was assumed to have degraded on storage; these samples were not included in the study.
Amplification of genes of interest for macrolide and fluoroquinolone resistance detection
Nested PCR was used to amplify genes of interest, namely 23S rRNA gene, gyrA and parC, to allow sequencing of regions containing potential resistance-associated SNPs. Specimens were amplified retrospectively, nested PCR was used to increase the potential for amplification of specimens with low concentrations of M. genitalium DNA. Primer sequences are given in table 1.
Primer sequences used to amplify genes of interest for sequencing
For 23S rRNA gene amplification, PCR round-one master mix contained: 0.4 µM of each primer, 2 mM MgCl2, 2.5 µL 10X Taq polymerase buffer (Qiagen, Manchester, UK), 200 µM each of dATP, dTTP, dCTP and dGTP, 5 U Taq DNA polymerase (Qiagen) and 5 µL template DNA in a 25 µL reaction. Reactions were run on a thermocycler using the following programme: initial denaturation at 94°C for 3 min followed by 35 cycles of 94°C for 1 min, 55°C for 30 s and 72°C for 1 min with a final extension of 72°C for 10 min. Round-two PCR master mix contained 25 µL MyTaq Red Mix (BioLine, London, UK), 0.5 µM of each primer and 2.5 µL round-one amplicon as template in a 50 µL reaction. Reactions were run on a thermocycler using the following programme: initial denaturation at 95°C for 3 min followed by 40 cycles of 95°C for 15 s, 60°C for 15 s and 72°C for 10 s.
For gyrA and parC amplification, PCR round-one master mix contained 0.4 µM each primer, 2.5 µL 10X JumpStart Taq buffer (Sigma-Aldrich, Gillingham, UK), 2 mM MgCl2, 200 µM each of dATP, dTTP, dCTP and dGTP, 1.25 U JumpStart Taq polymerase (Sigma-Aldrich) and 2.5 µL template DNA in a 25 µL reaction. Reactions were run on a thermocycler using the following programme: initial denaturation of 94°C for 1 min followed by 40 cycles of 94°C for 30 s, 57°C for 30 s and 72°C for 1 min with a final extension of 72°C for 1 min. The round-two PCRs were performed as for round-one but with 2 µL of the round-one amplicon used as template and a final reaction volume of 50 µL (double volumes of each reagent were used, except for the JumpStart Taq polymerase, of which 1.25 U were used). For gyrA, the round-one amplicon was diluted 1:10 with nuclease-free water (Promega, Southampton, UK) prior to use as template.
Sequencing and sequence analysis
Amplicons derived from the three gene targets (23S rRNA, gyrA and parC) were sequenced and the data analysed using BioNumerics (V.6.1; Applied Maths, Austin, Texas, USA). Multiple alignments of the sequenced clinical specimen fragments were performed with reference sequences for M. genitalium strains M6320 (GenBank: NC_018497.1), G37 (GenBank: L43967.2), M2288 (GenBank: NC_018498.1), M6282 (GenBank: NC_018496.1), M2321 (GenBank: NC_018495.1) and ATCC M. genitalium strain 49895 DNA sequence (sequenced in-house). The 23S rRNA gene sequences with macrolide resistance-associated mutations reported by Twin et al 27 and the parC sequence containing the fluoroquinolone resistance-associated mutation reported by Pond et al 21 (GenBank accession number HF947096) were included in alignments for these genes.
Ethics
Patients were referred for the purposes of primary diagnostics as part of the reference service remit therefore ethical approval was not sought or required. Public Health England has permission to handle these data under the Health Service (Control of Patient Information) regulation 2002, overseen by the Confidentiality Advisory Group. After primary diagnostic testing and prior to resistance testing, all specimens were blinded and anonymised.
Analysis of clinical data
Limited clinical data were available for these specimens and, where provided, were analysed in respect of patient age group, sex, specimen collection site, presence or absence of symptoms.
Statistical analysis
Statistical analysis was performed using the χ2 test where appropriate.
Results
Patient demographics
A total of 84 clinical specimens from individual patients were reconfirmed as positive for M. genitalium after recovery from storage. Seventy-nine per cent of these patients were under 35 years of age with 37/71 (52%) of males falling within the age group of 25–34 years and 7/13 (54%) of females falling within the age group of 16–24 years. Patients over the age of 35 years made up only 20% of the patients tested, with no females (of known age) in this age range.
Urine was the most commonly referred specimen type for male patients (59/71 (83%)) and swab specimens (eg, cervical, high-vaginal and introital swabs) were the most commonly collected samples types from women (7/13 (62%)). Symptoms such as recurrent urethritis, discharge and dysuria were recorded for the majority (87%) of male patients, but no clinical details were provided for the majority (77%) of the female patients. No treatment history was available for any of the patients.
Detection of macrolide resistance-associated mutations
Sequencing of the 23S rRNA gene was successful for 74 specimens (61 males, 13 females). Macrolide resistance-associated SNPs (A2058G, A2508C or A2059G; Escherichia coli numbering system) were detected in 61/74 (82.4%) of the specimens tested (table 2). Strains predicted to be macrolide-resistant were detected in 95.1% (58/61) of the specimens tested from males, and male patients were more likely to carry a ‘resistant’ strain than female patients (p≤0.001); wild-type 23S rRNA sequences, indicating susceptible strains, were detected in 76.9% (10/13) of the specimens from female patients (table 2). The most common SNP detected in specimens from both male and female patients was A2059G, which was found in 57.3% (35/61) of ‘resistant specimens’. An A2058G SNP was found in 36.1% (22/61) of the specimens tested, and A2059C was found in 4/61 (6.6%) (table 2). Two specimens, both from male patients, had heterogeneous (A2058G/wild-type or A2059G/wild-type) sequences in the 23S rRNA gene.
Summary of the sequence detected in the 23S rRNA gene at positions 2058 and 2059 (Escherichia coli numbering) for Mycoplasma genitalium isolates from male and female patients
Detection of fluoroquinolone resistance-associated mutations
Sequence data for gyrA were available for 73/74 (98.7%) specimens (60 males, 13 females) for which 23S rRNA sequencing had been successful, and all specimens contained wild-type gyrA sequences.
Sequence data for parC were available for only 61/74 specimens (49 males, 12 females) for which 23S rRNA sequencing was successful (table 3). Most (58/61, 95.1%) contained wild-type parC genes; however, two different parC SNPs were detected in specimens from three male patients; G248T (which causes a Ser(83)>Ile amino acid substitution) or G259T (which causes an Asp(87)>Tyr substitution). Both of these mutations have been previously identified in M. genitalium-positive specimens with associated fluoroquinolone treatment failure20 28 30 32 33 (table 3). Of note, a macrolide resistance-associated SNP was also detected in all of these specimens (parC G248T/23S rRNA A2059G and parC G259T/23S rRNA A2058G).
Summary of the sequence detected in the parC gene of Mycoplasma genitalium in clinical specimens for which 23S rRNA sequence data were also available
Discussion
The results reported in this study are of concern; prevalence of M. genitalium infection was 12.7% in our cohort with mutations previously associated with macrolide resistance detected in 95.1% of the specimens sourced from male patients. Male patients were statistically more likely to have a ‘macrolide-resistant’ specimen than female patients, where 76.9% of the specimens tested contained wild-type 23S rRNA gene sequences and therefore a predicted macrolide-susceptible phenotype. Two specimens from male patients contained M. genitalium 23S rRNA sequences that were heterogeneous (wild-type/A2058G or wild-type/A2059G). The 23S rRNA gene is present as a single copy gene in M. genitalium and therefore these data are indicative either of variants within a single strain or of two strains within the same specimen with different macrolide susceptibility genotypes.
The prevalence of fluoroquinolone resistance-associated mutations in the cohort tested was markedly lower than that seen for the macrolides. Previously reported mutations20 28 30 32 33 were detected in specimens from only three patients (6.1%), all male, and in all three cases a macrolide resistance-associated mutation in the 23S rRNA gene (A2058G or A2059G) was also detected. Without the use of further techniques such as whole genome sequencing, it is not possible to say definitively whether the mutations detected in the 23S rRNA and parC genes in these specimens were sourced from the same strain; however, there was no evidence of heterogeneous sequence data for the parC gene in the specimens analysed.
The prevalence of M. genitalium infection was slightly lower than the 16.7% prevalence reported previously by Pond et al. 21 However, the study by Pond et al tested samples from male patients in inner-London only, whereas the data presented here were generated from specimens collected across England and from both sexes. Differences in the cohorts tested could account for this difference.
In the absence of widely available M. genitalium tests, the UK treatment guidelines22 recommend that clinics treat male patients with non-gonococcal urethritis with a single dose of 1 g azithromycin or 1 week of doxycycline as first-line therapy. Some patients may receive a second course of therapy, for example, a further course of azithromycin, doxycycline or moxifloxacin, before specimens are referred to the national reference laboratory for M. genitalium testing and, in practice, the majority of specimens referred to the reference service are sourced from those patients who remain symptomatic post-treatment. Therefore, the samples in this study clearly have a significant selection bias towards M. genitalium-positive specimens sourced from symptomatic patients who have already received, and failed, at least one course of antibiotic therapy. Indeed, where clinical data were recorded, 67% of the patients had urogenital symptoms. It is therefore not surprising that a high prevalence of macrolide resistance was found in this study, although in the absence of prescribing data we cannot be sure of this. One possible explanation for the difference in macrolide resistance between the male and female patients is that the samples from females may have come from patients tested as contacts of men diagnosed with M. genitalium. Under these circumstances, they would therefore have been less likely to have received treatment prior to testing. At the time of specimen collection for this study, the role of M. genitalium infection in PID was unclear, therefore referral of patients with signs and symptoms of PID for M. genitalium testing was not routine. Dean et al found macrolide resistance in 56% of M. genitalium-positive clinical specimens from women with signs and symptoms of PID.35 More studies are required to ascertain whether macrolide resistance is more strongly associated with symptomatic females. It may be the case that the prevalence of macrolide resistance in the women in our study (23.1%) is more representative of the general population.
Further work to perform sequence typing on these samples would be valuable to determine the relative extent to which ‘macrolide-resistant’ specimens represent clonal transmission of ‘resistant’ strains, as opposed to resistance induced de novo in individuals by the use of 1 g azithromycin therapy.
It is of some reassurance that SNPs associated with fluoroquinolone resistance were detected in only three specimens; the study was small, but this may reflect less selection pressure for the development of fluoroquinolone resistance during this time period. While the data presented in this study may not be representative of all the isolates of M. genitalium circulating within the UK sexual health clinic population, it does clearly highlight the intrinsic problems associated with trying to manage this infection syndromically. There is clearly a need for M. genitalium testing to be more widely available, and this should ideally include testing for antimicrobial resistance, although this is currently precluded by the lack of suitable susceptibility testing methods. Molecular assays for the detection of macrolide resistance-associated mutations are becoming more commonplace however and implementation of these as part of testing algorithms should be considered.36–38
In conclusion, we have reported a high prevalence of predicted resistance to macrolide antibiotics in M. genitalium in specimens referred to the national reference laboratory. The work performed in this study has highlighted the need for urgent discussion regarding the appropriateness of current testing and treatment guidelines, particularly as commercial tests for M. genitalium become more widely available. A larger study, which is also more representative of the sexual health clinic population, would be very helpful to guide these decisions.
Key messages
Very high rates of macrolide resistance (82.4%) were predicted in Mycoplasma genitalium specimens referred for diagnostic testing.
Significant differences were observed between the rates of predicted macrolide resistance in specimens sourced from male (95.1% (58/61)) and female (23.1% (3/13)) patients (p≤0.001).
Three clinical specimens contained M. genitalium with predicted resistance to both macrolides and fluoroquinolones.
Current treatment regimens and testing guidelines should be reviewed.
Acknowledgments
The authors would like to acknowledge Prof. Catherine Ison for her contribution to this study.
References
Footnotes
Handling editor Jackie A Cassell
Contributors RP carried out all the laboratory work and prepared the first draft of the manuscript for this study. SA supervised this work and contributed to the manuscript. HF and NW contributed to the manuscript and the analysis of results.
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests Since the laboratory work described in this report was completed, AMRHAI has received funding from SpeeDx for a kit evaluation. HF has participated in the Consensus Steering Group for Mycoplasma genitalium diagnosis in the UK, arranged and funded by Hologic.
Provenance and peer review Not commissioned; externally peer reviewed.