gyrA Mutations in Mycoplasma genitalium and Their Contribution to Moxifloxacin Failure: Time for the Next Generation of Resistance-Guided Therapy

Abstract

Background

Although single nucleotide polymorphisms (SNPs) in Mycoplasma genitalium parC contribute to fluoroquinolone treatment failure, data are limited for the homologous gene, gyrA. This study investigated the prevalence of gyrA SNPs and their contribution to fluoroquinolone failure.

Methods

Samples from 411 patients (male and female) undergoing treatment for M. genitalium infection (Melbourne Sexual Health Centre, March 2019–February 2020) were analyzed by Sanger sequencing (gyrA and parC). For patients treated with moxifloxacin (n = 194), the association between SNPs and microbiologic treatment outcome was analyzed.

Results

The most common parC SNP was G248T/S83I (21.1% of samples), followed by D87N (2.3%). The most common gyrA SNP was G285A/M95I (7.1%). Dual parC/gyrA SNPs were found in 8.6% of cases. One third of infections harboring parC G248T/S83I SNP had a concurrent SNP in gyrA conferring M95I. SNPs in gyrA cooccurred with parC S83I variations. Treatment failure was higher in patients with parC S83I/gyrA dual SNPs when compared with infections with single S83I SNP alone from analysis of (1) 194 cases in this study (81.2% vs 45.8%, P = .047), and (2) pooled analysis of a larger population of 535 cases (80.6% vs 43.2%; P = .0027), indicating a strong additive effect.

Conclusions

Compared with parC S83I SNP alone, M. genitalium infections with dual mutations affecting parC/gyrA had twice the likelihood of failing moxifloxacin. Although antimicrobial resistance varies by region globally, these data indicate that gyrA should be considered as a target for future resistance assays in Australasia. We propose a strategy for the next generation of resistance-guided therapy incorporating parC and gyrA testing.

Mycoplasma genitalium gyrA sequence variants (M95I and D99N/Y) were present in 8.5% of infections in a clinic population. Moxifloxacin failure occurred in 46% of infections with parC S83I alone, but in 81% with combined parC S83I and a gyrA sequence variations.

Mycoplasma genitalium is a common sexually transmitted infection with limited treatment options that causes urethritis in men and reproductive complications in women [1, 2]. Failure of the first-line therapy (azithromycin) is caused by single nucleotide polymorphisms (SNPs) affecting the target molecule, the 23S rRNA gene [3], which results in an in vitro minimum inhibitory concentration (MIC) increase on the order of 10,000-fold [4].

In contrast to azithromycin, the mechanism of resistance to moxifloxacin, used for macrolide-resistant infections, is not as clear. Fluroquinolones target the DNA gyrase (composed of 2 subunits each of GyrA and GyrB) and topoisomerase IV (comprising 2 subunits each of ParC and ParE), exerting activity by binding to serine at position 83 (Escherichia coli GyrA numbering) and an acidic amino acid 4 positions away (eg, D87 or E87) [5]. This part of the molecule is known as the quinolone resistance determining region (QRDR). Sequence variations in parC impacting the serine at position 83 (eg, S83I, S83R; M. genitalium numbering) and the aspartic acid at position 87 (eg, D87N, D87Y) have been associated with fluoroquinolone failure [6–9] and cause more modest increases in MIC in vitro [4, 10]. The strongest candidate for resistance is the parC S83I variation, with approximately 60% of cases carrying this mutation failing moxifloxacin treatment; moreover, treatment failures are rare in the absence of S83I (approximately 4%) [9]. Although other parC changes may also contribute to failure, rarity and lack of enrichment suggests their contribution at a population level is minor [9]. Of note, for ease of understanding, SNPs will henceforth often be referred to by their amino acid change.

In contrast to the increasingly strong evidence base for parC S83I, there are limited data about the prevalence of gyrA mutations and their contribution to fluoroquinolone failure. Recently, we indicated that combining a gyrA mutation (affecting M95 or D99) with a parC S83I mutation may increase the probability of moxifloxacin failure [7]. However, datasets were too small to draw meaningful conclusions. Because isolating M. genitalium in in vitro culture is challenging and makes routine MIC testing unfeasible, the molecular analysis of samples with known treatment outcomes provides an alternative approach and has the benefit being clinically relevant. This study investigated the prevalence of gyrA mutations in a clinic population and their contribution to fluoroquinolone failure.

METHODS

Study Group and Sample Collection

Sequential M. genitalium–positive samples were collected at Melbourne Sexual Health Centre (MSHC) (March 2019–February 2020) and tested as described previously [8] or as outlined later in this article. Indications for M. genitalium testing at MSHC included urethritis, cervicitis, suspected pelvic inflammatory disease, proctitis, being a sexual contact of M. genitalium infection, and attending for a test of cure (TOC). Patients were prescribed treatment based on treatment guidelines and the outcome of a diagnostic test reporting a macrolide resistance mutation (MRM) [11, 12].

From 411 patients at baseline, 290 (70.6%) infections were macrolide-resistant and 121 (29.4%) were susceptible. A total of 249 patients received a moxifloxacin-containing regimen (Table 1), the majority included 7 days of lead-in with doxycycline; 127 (51.0%) were then treated with 7 days of moxifloxacin and 107 (43.0%) had 7 days of moxifloxacin + doxycycline as combination therapy [8]. One patient had moxifloxacin only for 7 days without doxycycline because of a contraindication to doxycycline and 14 patients (5.6%) had 14 days of moxifloxacin in the setting of pelvic inflammatory disease. Of patients receiving other treatments, 1 received sitafloxacin and the remaining 161 received a nonfluoroquinolone regimen outlined in Table 1. In line with standard of care, patients were asked to return for a TOC 14 to 28 days after completing treatment, but TOC results obtained within 14 to 90 days were eligible for inclusion in the analysis.

Table 1.

Demographic Characteristics and Treatments for 411 Baseline Samples

Demographic	…
Age, median (range)	…
…	27 (16–61)
Gender/orientation, n (%)	…
Men who have sex with women	128 (31.1)
Men who have sex with men	150 (36.5)
Women who have sex with men	89 (21.7)
Women who have sex with women	26 (6.3)
Women who have sex with women and men	4 (1.0)
Transgender	1 (0.2)
Femaleb	13 (3.2)
Treatments at baseline, n (%)	…
MFX (7 d)	1 (0.2)
MFX (14 d)	14 (3.4)
DOX/MFX (sequential)	127 (30.9)
DOX + MFX (combination)	107 (26.0)
STFX	1 (0.2)
Other treatmenta	161 (39.2)
Sample site, n (%)	…
Urine	243 (59.1)
Vagina	91 (22.1)
Cervix	24 (5.8)
Rectum	50 (12.2)
Urethra	3 (0.7)

Abbreviations: DOX, doxycycline; MFX, moxifloxacin; STFX, sitafloxacin.

Includes tetracyclines (doxycycline, minocycline), azithromycin, pristinamycin.

Data unavailable on sexual practices.

Of note, the patient group analyzed in this study overlapped with some samples analyzed in a previous study (August 2019–December 2020) that sequenced parC only [8]. Ethics approval was obtained from Alfred Health ethics committee (Approval:232/16).

Laboratory Analysis of Samples

As part of standard diagnostic procedure, DNA was extracted on the QIASymphony (Microbiological Diagnostic Unit Public Health Laboratory, Melbourne, Australia) and tested using the ResistancePlus MG Assay (SpeeDx Pty Ltd, Sydney Australia), which simultaneously detects M. genitalium and macrolide-resistance mutations. Residual nucleic acid extracts were amplified using M. genitalium–specific primers targeting parC (MG-parC124-F, MG-parC478-R) and gyrA (GyrA_1243F, GyrA_502R), and Sanger sequencing was performed as described previously [13]. SNPs were identified through comparison to M. genitalium strain G37 (GenBank accession NC_000908.2). Sequencing of gyrA was successful for 488/752 (65%) of samples. Analysis was limited to samples for which gyrA sequence was obtained, which included 411 samples collected at baseline; of these, 350 samples also had successful sequencing for parC.

Strain typing was performed through sequencing of mgpB (encoding the adhesin MgPa) for 2 patients that appeared to have mutation selection during treatment. Sanger sequencing was performed with primers MgPa-1 and MgPa-3, as described previously [13]. Subsequently, 1 TOC sample was excluded from the study because evidence indicated it was a new infection rather than a treatment failure (for further information, see Results). The mgpB types were referenced as defined previously [14], and the accession number of equivalent sequences from GenBank is provided.

Statistical Analysis

To analyze associations between SNPs and treatment outcomes (Fisher exact test), all patients receiving any regimen containing moxifloxacin were grouped. Trends over time were analyzed by ptrend (STATA v17.0, StataCorp LLC).

RESULTS

Patient Demographics and Treatment Outcomes

Patient demographics and treatments are summarized in (Table 1). From 411 cases with baseline samples, 249 received a moxifloxacin-based initial treatment regimen; 55/249 (22.1%) did not attend for TOC within 14 to 90 days of completing treatment (Figure 1A). Of 194 patients attending for TOC, 157/194 (80.9%; 95% confidence interval [CI], 74.7–86.2) were cured and 37/194 (19.1%; 95% CI, 13.8–25.3) failed.

Figure 1.

Overview of sample collection and resistance mutation analysis. A, Sample collection, assignment to moxifloxacin treatment, and treatment outcomes. B, Distribution of significant parC and gyrA single nucleotide polymorphisms (SNPs) stratified by the presence of coincident MRMs. Analysis is limited to samples with available parC sequence. The category of “parC not S83I” includes both WT parC and infrequently detected SNPs. LTFU, lost to follow up; QRDR, quinolone-resistance determining region; TOC, test of cure; WT, wild type.

Prevalence of gyrA and parC Single Nucleotide Polymorphisms in Baseline Samples

For parC, the most common SNP was G248T/S83I (74/350, 21.1%), followed by D87N (2.3%) and S83N (1.7%) (Table 2). For gyrA, the most common SNP was G285A conferring M95I (29/411, 7.1%), followed by D99Y (3/411, 0.7%). No individual sample had multiple SNPs within a single gene (ie, affecting both amino acids S83 and D87 of parC or both M95 and D99 of gyrA). One third of S83I samples also had a gyrA M95I change. Notably, where parC sequence was available, almost all the samples with a gyrA M95I SNP had a corresponding parC S83I change.

Table 2.

Prevalence of SNPs Affecting the Quinolone Resistance–Determining Region of parC and gyrA^a

Gene	DNA Change	Amino Acid Change	No. (%)
parC	…	…	N = 350
	G248T	S83I	74 (21.1)b
	G247C	S83R	2 (0.6)
	G248A	S83N	6 (1.7)
	A247T	S83C	1 (0.3)
	G259A	D87N	8 (2.3)
	G259C	D87H	3 (0.9)
	G259T	D87Y	5 (1.4)
	WT	WT	251 (71.7)
gyrA	…	…	N = 411
	G285T	M95I	1 (0.2)c
	G285A	M95I	29 (7.1)d
	G295A	D99N	1 (0.2)e
	G295T	D99Y	3 (0.7)e
	A283G	M95V	1 (0.2)e
	WT	WT	376 (91.5)

This includes all samples (those treated with moxifloxacin and those not treated with moxifloxacin). Sequence data for parC were unavailable for 61 samples.

Of the 74 cases, 45 had WT gyrA QRDR, 25 had a single nucleotide polymorphism conferring M95 change (24 M95I, 1 M95V), and 4 had a single nucleotide polymorphism conferring a change at D99 (3 D99Y and 1 D99G).

Also had a D87Y change in parC.

24 of these samples also carried a parC single nucleotide polymorphism conferring S83I, the remaining 5 failed parC sequencing.

Also had S83I mutation in parC.

When stratified by macrolide-resistance mutation status, there was a higher proportion of parC S83I changes in the MRM group compared with the non-MRM group (27.5% vs 5.1%, respectively; P < .001) (Figure 1B). Likewise, the proportion of infections with gyrA changes was higher in the MRM group compared with the non-MRM group (11.6% vs 1.0%, respectively; P < .001).

Trends in the Prevalence of gyrA Single Nucleotide Polymorphisms Over Time

Previous studies at MSHC analyzing gyrA and parC SNP prevalence were collectively evaluated (Table 3), with most data available for macrolide-resistant infections. Notably, between 2012 to 2020, there was an increase in parC S83I SNP detection from 13.0% to 27.5% of macrolide-resistant infections (P_trend <.01), as previously reported [9]. The prevalence of gyrA M95I mutations fluctuated over time, but overall increased from 7.4% to 10.0% (P_trend = .8), whereas the proportion of infections carrying parC S83I mutations with a concurrent mutation in gyrA (M95I) increased from 28.6% to 34.8% (P_trend = .4) (Table 3); however, neither reached statistical significance.

Table 3.

Prevalence of parC S83I and gyrA M95I in Macrolide-Resistant Infections Over Several Study Periods From Melbourne Sexual Health Centre

Sample Collection Date Range	parC S83I (%)	gyrA M95I (%)	Proportion of parC S83I With a Coincident gyrA M95I (%)a	Study
July 2012–June 2013b	7/54 (13.0)	4/54 (7.4)	4/14 (28.6)	[6]
June 2016–June 2018	49/321 (15.3)	12/213 (5.6)	11/39 (28.2)c	[7]
March 2019–February 2020d	69/251 (27.5)	25/251 (10.0)	24/69 (34.8)	Current study

Analysis of trends: parC P_trend = 0.0011, gyrA P_trend = 0.8, parC/gyrA combination P_trend = 0.4.

For all infections (ie, macrolide-susceptible and macrolide-resistant) for the 2012–2013 study period, parC S83I, 14/140 (10.0%); gyrA M95I 4/140 (2.9%); proportion of parC S83I with gyrA M95I, 4/14 (28.6%).

From 49 baseline samples with S83I, 10 had no gyrA sequence available.

For all infections (ie, macrolide-susceptible and macrolide-resistant) for 2019–2020 study period, parC S83I, 74/350 (21.0%); gyrA M95I 30/411 (7.3%); proportion of parC S83I with gyrA M95I, 24/74 (32.4%).

Association Between parC and gyrA Sequences and Microbiologic Outcomes for Moxifloxacin Treatment

From 169 infections with known treatment outcomes following moxifloxacin, parC SNPs conferring S83I and S83R were both significantly associated with microbiologic treatment failure (Table 4), with 24/40 (60.0%; 95% CI, 43.3–75.1) of infections harboring the S83I change failing moxifloxacin (P = .0001). The S83R change was rare; however, both cases failed treatment. Other SNPs in the parC QRDR conferring S83N/C (3 in total) and D87N/H/Y (9 in total) were only found in infections that were successfully treated with moxifloxacin. From 194 infections treated with moxifloxacin, gyrA M95I was significantly associated with treatment failure, with 11/15 (73.3%; 95% CI, 44.9–92.2) of cases carrying this change failing moxifloxacin (P = .0001). Other gyrA changes conferring D99N (n = 1) and D99Y (n = 2) were uncommon but were also only found in moxifloxacin failures.

Table 4.

Association Between SNPs Affecting parC and gyrA in the Quinolone Resistance–Determining Region^a and Moxifloxacin Outcomes

Gene	DNA Change	Amino Acid Change	Treatment Outcome
			Success (%)	Failure (%)	P value
parC	G248T	S83I	16/40 (40)	24/40 (60)	.0001
	G247C	S83R	…	2/2 (100)	.0022
	G248A	S83N	2/2 (100)	…
	A247T	S83C	1/1	…
	G259A	D87N	5/5 (100)	…
	G259C	D87H	1/1	…
	G259T	D87Y	3/3 (100)	…
	WT	WT	111/115 (96.5)	4/115 (3.5)	Reference
	…	Total	139/169(82.2)	30/169 (17.8)
gyrA	G285A	M95I	4/15 (26.7)	11/15 (73.3)	.0001
	G295A	D99N	…	1/1	.195
	G295T	D99Y	…	2/2	.039
	WT	WT	153/176 (86.9)	23/176 (13.1)	Reference
	…	Total	157/194	37/194
parC; gyrA	…	S83I; gyrA WT	13/24 (54.2)	11/24 (45.8)
parC; gyrA	…	S83I; M95/D99	3/16 (18.8)	13/16 (81.2)
parC; gyrA	…	parC WT; gyrA WT	111/115 (96.5)	4/115 (3.5)

Only changes affecting S83 and D87 of parC and M95 and D99 of gyrA are shown.

As noted, all samples with a gyrA change at M95 or D99 with available parC sequence also had a parC change of S83I, meaning any independent assessment of the contribution of gyrA SNPs to treatment failure was not possible. However, the additive effect of combining a gyrA change with parC S83I on treatment outcomes could be assessed. Patients with a combined parC S83I/gyrA mutation (either M95 or D99 change) were significantly more likely to fail moxifloxacin treatment than those with a parC S83I mutation alone (13/16 [81.2%] vs 11/24 [45.8%]; P = .047) (Table 5). To encompass a larger dataset, this analysis was expanded to incorporate data from 3 prior studies at MSHC that included similar clinic populations [6, 7]. Using this pooled dataset (N = 535; n = 68 with parC S83I ± gyrA mutations), infections with dual mutations were twice as likely to fail moxifloxacin compared with those with a parC S83I change alone (25/31 [80.6%] failure vs 16/37 [43.2%]; P = .0027) (Table 5).

Table 5.

Comparison of Moxifloxacin Outcomes for Infections With a Single Mutation in parC and Dual Mutations Affecting Both parC and gyrA From 3 Studies at Melbourne Sexual Health Centre

Period of Sample Collection	Treatment Failures/Total Number of Cases With Specified Mutations (%)			Study
	Dual parC S83I/gyrA M95orD99 Mutation	parC S83I Mutation Only	Fisher Exact Test
July 2012–June 2013	4/6 (67)	0/1 (0)	0.43	[6]
June 2016–June 2018	8/9 (88.9)	5/12 (41.7)	0.067	[7]
March 2019–February 2020	13/16 (81.2)	11/24 (45.8)	0.047	This study
Pooled analysis	25/31 (80.6)	16/37 (43.2)	0.0027	…

Bold indicates a statistically significant finding of P < .05.

Analysis of the Acquisition of gyrA and parC Mutations During Treatment With Moxifloxacin

Of the moxifloxacin failures, only 2 patients had infections that had a different sequence at TOC compared with baseline (Table 6). Patient 282 received (1) sequential doxycycline followed by moxifloxacin, which failed (TOC-1), then (2) doxycycline and pristinamycin in combination, which failed (TOC-2), then (3) combination doxycycline with sitafloxacin (lost to follow-up so no cure data available). Samples had gyrA sequences of D99N (baseline), WT (TOC-1), then M95I (TOC-2). Notably, the chromatogram for the middle sample had a double peak (G/A) at base G295, indicating a mixture of WT sequence and D99N (data not shown). Each sample was strain type 30 (equivalent to GenBank accession GU226233), indicating that a new infection between timepoints was unlikely.

Table 6.

Analysis of Samples With Apparent Acquisition of Resistance Mutations During Treatment

Casea	Baseline			TOC1				TOC2				TOC3
	parC/gyrA	mgpB typec	Treatment	Outcome	parC/gyrA	mgpB type	Treatment	Outcome	parC/gyrA	mgpB type	Treatment	Outcome
282	S83I/D99N	30	DOX/MXF	Positive	S83I/WT,D99Nb	30	DOX + PRIS	Positive	S83I/M95I	30	DOX/STFX	LTFUa
411	NA/WT	108	DOX/MXF	Positive	S83I/M95I	127	MIN	Negative	…	…	…	…

Abbreviations: DOX, doxycycline; LTFU, lost to follow up; MIN, minocycline; MXF, moxifloxacin; PRIS, pristinamycin; STFX, sitafloxacin; TOC, test of cure.

The timings for testing of case 282 were day 0 (baseline), day 75 (TOC1), day 188 (TOC2). The patient was lost to follow-up after TOC2 visit.

The sequence chromatogram showed a mixture of 2 genotypes, whereas the background signal was very low. WT was the dominant sequence.

As defined previously [14].

Patient 411 was treated with (1) doxycycline followed by moxifloxacin, then (2) minocycline. Although sequencing revealed gyrA WT (baseline) changing to gyrA M95I (TOC), the strain type of these samples differed by 6 SNPs (type 108 [MK673423], and type 127 [KU856544]), indicating a new infection was likely. As mentioned in the Methods, the TOC sample for this patient was excluded from the analysis.

DISCUSSION

There has been an incomplete understanding of the molecular basis for M. genitalium fluoroquinolone resistance, and this impacts the development of resistance assays with strong predictive values for microbial cure/failure with fluoroquinolones. This study of patients attending the largest public sexual health clinic in Australia captured all treatment histories and tests of cure, enabling careful evaluation of the impact of specific parC and gyrA mutations on moxifloxacin outcomes. It also allowed us to pool data from prior studies to provide more confidence around estimates. This study found that mutations in gyrA were common (8.6% prevalence) and associated with a corresponding parC mutation (usually S83I). Furthermore, although the parC S83I mutation was associated with moxifloxacin failure, the degree of treatment failure was substantially increased (nearly doubled) when a gyrA mutation was combined with parC S83I, suggesting a synergistic effect. Predictably, both gyrA and parC mutations were concentrated in the patients with macrolide-resistant infections, consistent with the causal strains having been subjected to multiple rounds of past treatment for M. genitalium, similar to previous observations [15].

Data have recently emerged on prevalence of fluoroquinolone resistance SNPs, with the richest information available from East Asia and Australia [9, 16–18]. In Japan, analysis of men with urethritis found levels of gyrA mutations have trended upward from around 2% in 2013 to 10% in 2017, with the most reported combination being either gyrA M95I or D99N paired with parC S83I [17]. Elevated levels of gyrA changes have also been reported in Korea [15]. Few gyrA mutations have been reported in other geographic regions. A study from the United States reported a single gyrA M95I mutation out of 26 M. genitalium samples from pregnant women collected from 2018–2019 [19], whereas a study in France detected a single case with parC S83I/gyrA M95I genotype from 283 samples collected from 2018 to 2019 [20]. Additional studies from Europe and Africa have not detected significant gyrA mutations at position M95 or D99 [21–25]. Overall, this indicates that levels of gyrA mutations are higher in some parts of the Western Pacific.

Current evidence for the contribution of gyrA mutations in quinolone treatment failure is limited. Some studies have described an association between specific mutations and moxifloxacin failure, with gyrA M95I alone [26], dual parC S83I/gyrA M95I [6, 7], and other combinations including parC S83I/gyrA D99N/Y/G [6, 7, 27]; however, in most of these studies, sample sizes were too small to reach statistical significance and draw meaningful conclusions about their contribution to moxifloxacin failure. The current study found a clear association between the presence of a parC/gyrA mutation combination and moxifloxacin failure, with the most common mutation combination being parC S83I/gyrA M95I. MIC data can augment our understanding of treatment failure; MICs for moxifloxacin were reported to be ≤ 0.25 mg/L for isolates that were WT at both parC/gyrA [4], 2 to 4 mg/L for isolates with parC S83I/gyrA WT [4, 28], and 2 to 16 mg/L for isolates with both parC S83I and gyrA changes [4, 28]. However, these data come from different laboratories and are very limited in sample size; they therefore do not provide a conclusive picture on how each of these genes contributes to resistance.

The elevated treatment failure after combining different mutations in M. genitalium parallels quinolone-resistance mechanisms for Neisseria gonorrhoeae. In N gonorrhoeae, mutations in gyrA are primarily responsible for resistance, with common amino acid changes S91F/Y and changes at D95. Similar to the findings of this study, mutations in the alternative fluoroquinolone target of N gonorrhoeae (in this case, gonococcal parC) may occur concurrently and enhance resistance but are unlikely to occur independently [29–31]. Notably, in contrast to M. genitalium, in which typically a single mutation is observed in parC, double mutations are common in N gonorrhoeae gyrA (affecting both the key serine and aspartic acid) and further elevate resistance.

The development of azithromycin resistance during treatment has been well documented [3, 14]. In contrast, published evidence for parC or gyrA mutation acquisition during quinolone treatment is limited. In this study, 1 infection changed from gyrA D99N to a mixture of gyrA D99N/WT (after moxifloxacin treatment) to gyrA M95I (after nonfluoroquinolone treatment). It appears counterintuitive that a resistance mutation would revert to WT after fluoroquinolone treatment, then change to a different resistance mutation while not under fluoroquinolone selective pressure. mgpB-based strain typing suggested that this is a single continuous infection. However, it is possible the patient was infected with a new strain carrying the same mgpB type, although the sequence type for this infection appears uncommon, having not previously been described in Australian datasets [32, 33]. It is also possible that the combination of mutations in this infection led to a fitness disadvantage, and the reversion to gyrA wild type. In a study from a similar clinic population, there was evidence of mutation acquisition during a case with sitafloxacin treatment [7], although strain typing was not performed to identify reinfection.

Current M. genitalium treatment guidelines recommend resistance-guided therapy incorporating a macrolide-resistance test to individualize treatment based on azithromycin resistance [34–36]. We recently proposed refinement of this algorithm based on the strong contribution of the parC S83I SNP to fluoroquinolone failure, and recommended detection of the parC S83I mutation/parC S83 WT to direct to the use of moxifloxacin [37]. The results of our current study allow us to further stratify infections and individualize therapy. The next generation of resistance-guided therapy incorporates dual-class detection of macrolide-resistance and parC and gyrA resistance mutations/susceptibility, to optimize first-line cure and antibiotic stewardship (Figure 2). The proposed pathway predicts >96% first-line cure for the majority of patients in 2022, who have either macrolide-susceptible infection and can receive azithromycin or have macrolide-resistant infection without the parC S83I change and can receive moxifloxacin. In the presence of macrolide-resistant infections with parC S83I but gyrA WT, there are 3 options that can be selected based on antimicrobial availability, cost, contraindications, and cure. Cure is graded ranging from 60% with moxifloxacin, to 70% to 75% with minocycline or pristinamycin, and 80% to 90% for sitafloxacin (least available globally). For macrolide-resistant cases with parC S83I with a gyrA M95 or D99 mutation, both moxifloxacin and sitafloxacin have very low cure rates and should be avoided, leaving minocycline or pristinamycin with 70% to 75% cure as the only options. Of note, there are currently few commercial tests to detect mutations in parC, and these often detect mutations that have not been strongly associated with fluoroquinolone failure; none are available that target gyrA.

Figure 2.

Proposed clinical management algorithm for patients with Mycoplasma genitalium infection. Patients with a sexually transmitted infection syndrome would initially be tested for established causes. The M. genitalium test would incorporate detection of macrolide resistance mutations in the 23S rRNA gene (MRM), the presence of parC S83I or parC WT, and single nucleotide polymorphisms in gyrA affecting M95 and D99 (eg, M95I). Patients with M. genitalium would be recalled while on doxycycline and an antimicrobial selected based on the resistance profile. This flow chart provides clinicians with recommended treatment options that had the highest predicted microbial cure as indicated in green. Less effective treatments have been shaded orange. Clearly, there is considerable regional and national variability in the availability of antimicrobials such as pristinamycin and sitafloxacin and cost considerations.

Although still at prevalence estimates of <10%, higher levels of gyrA-resistance mutations are found in Australia and Japan than elsewhere, which may reflect increased use of fluoroquinolones in these countries and imported resistance from regional neighbors. It is reasonable, however, to expect dual parC/gyrA mutation resistance to spread over time to other regions as the use of fluoroquinolones increases in direct response to escalating macrolide resistance and limited alternatives to fluoroquinolones.

Study limitations include the use of samples from a single clinic, although this is the only public sexual health clinic servicing a population of 5 million people. This study was performed in a geographic location with high levels of fluoroquinolone resistance; it is therefore of interest to see how outcomes compare at other locations. The proposed treatment algorithm is based on data collected in Australia, which may limit generalizability.

CONCLUSIONS

Although the main mechanism of fluoroquinolone resistance for M. genitalium is the S83I mutation affecting parC, this study found that gyrA mutations enhance treatment failure and are likely to become more common. Given the increased risk of fluoroquinolone treatment failure in the presence of dual parC/gyrA mutations, consideration should be given toward incorporating gyrA mutations into future diagnostic assays, generating a new resistance-guided strategy. This would further improve antibiotic selection, first-line cure, and stewardship for M. genitalium.

Notes

Author contributions . G. L. M., E. L. P., and C. S. B conceived the study. G. L. M., K. B, T. P. C., and J. L. H. generated and analyzed the sequence data. C. S. B., L. A. V., M. D., and E. L .P. compiled the clinical data. G. L. M., C. S. B., and E.L.P. performed the data analysis and wrote the first manuscript draft. All authors contributed to drafting and editing of the final manuscript.

Financial support . This work was supported by an Australian Research Council (ARC) Industrial Transformation Research Hub Grant (IH190100021, G. L. M., C. S. B., D. M. W.) and Victorian Medical Research Acceleration Fund grant (G. L. M., C. S. B.). C. S. B. and S. M. G. are supported by Australian National Health and Medical Research Council Investigator Grants (GNT1173361 and 1197951, respectively).