From July 1, 2012, through June 30, 2013, samples were collected from consecutive M. genitalium–infected participants at the Melbourne Sexual Health Centre in Australia (5). Detection of M. genitalium, load quantitation, and sequence analyses were performed as described previously (5,8). Overall, 155 patients (112 men, 43 women) with PCR-confirmed M. genitalium infection were recruited, representing 90% of patients with infections diagnosed at the Centre over the study period. We obtained adequate samples from 140 of the 155 patients to generate baseline parC and gyrA gene sequences; these 140 formed the study group.

Patients were initially given a single dose of 1 g azithromycin. The 54 for whom this treatment failed (positive by PCR test-of-cure at day 28 or persistent symptoms before day 28, with no identified reinfection risk) were given moxifloxacin (400 mg/d for 10 d). The 6 for whom moxifloxacin treatment failed were given pristinamycin (1 g 4×/d for 10 d). This study was approved by The Alfred Hospital Ethics Committee (no. 150/12), and informed consent was obtained from patients.

In pretreatment specimens, various single-nucleotide polymorphisms (SNPs) were observed in the parC and gyrA QRDR (Table 1; Technical Appendix). Of the 19 (13.6%) of 140 samples with ParC substitutions, 16 had S83 mutations (14 S83I, 2 S83R) and 3 had D87N substitutions.

We found a significant association between detection of ParC S83 mutations and treatment failure. M. genitalium from all 6 patients for whom moxifloxacin failed but from only 3 of the 48 patients for whom moxifloxacin was effective had the ParC S83 mutation (p<0.0001 by Fisher exact test) (Table 2, patients 1–9). The 3 infections successfully treated despite a change at ParC S83 are of interest. For these patients, low bacterial load may have contributed to therapeutic success (9), led to spontaneous clearance, or resulted in false-negative follow-up PCR (Table 2). However, in contrast, treatment failed for 1 patient with a low anal load of M. genitalium and S83 change. Similar to the findings for this study, in the parent cohort of 155 patients, organism load influenced apparent azithromycin cure; 7% of infections carrying markers of azithromycin resistance were cured by azithromycin, and organism load was significantly lower than that among those with resistant infections for whom azithromycin treatment failed (5).

The prevalence of S83 changes in this study is higher than that detected in a study at Sydney Sexual Health Centre (Sydney, New South Wales, Australia) (8.4%, n = 143) (10). Studies in Japan reported prevalence ranging from 3.6% (n = 28) to 29.4% (n = 51) and 36.8% (n = 19) (7,8,11), although 1 study involved a cohort at higher risk (female sex workers). A low prevalence of S83 mutation has been observed in Europe (1.5% in France, 5% in England and Germany) (12–14). This mutation has been associated with moxifloxacin failure in 3/3 cases in the Sydney-based study (6).

SNPs that changed the ParC acidic residue (D87) were rare (2.1%) and because of low numbers could not be associated with treatment outcomes. Other studies found higher frequency of this change (3.5%–7.1%) (7,8,10,15); authors of 1 study reported an association with levofloxacin failure (15).

M. genitalium GyrA lacks the S83 residue common to GyrA of other bacteria, having instead a methionine at the equivalent position (M95). This enzyme is therefore probably partially resistant to quinolones. GyrA changes (at M95 or D99) occurred at a frequency of 5.0% (7/140) but could not be correlated with treatment outcome because they occurred concurrently with S83 changes in ParC. Previously, a GyrA M95I change was associated with M. genitalium treatment failure in 1 patient (6).

Patients who received moxifloxacin were followed up with a PCR test-of-cure at 14 and 28 days. For the 6 for whom treatment failed, the mutation profiles in follow-up specimens were unchanged from the initial premoxifloxacin sequence, suggesting lack of resistance selection in vivo after moxifloxacin.

A total of 60 (42.9%) of the 140 pretreatment samples had macrolide-resistance mutations (5). Both macrolide and parC fluoroquinolone mutations at S83 or D87 were present in 12 (8.6%) of the 140 samples. Prevalence of fluoroquinolone resistance markers was higher in samples with (20%, 12/60) than without (8.8%, 7/80) macrolide-resistance mutations, although this difference did not reach statistical significance (p = 0.08). This finding suggests that successive treatment failures with first-line, then second-line, antimicrobial drugs are generating strains resistant to 2 classes of drugs. Previous studies found lower levels of combined macrolide and fluoroquinolone mutations in men attending a urology clinic (3/51, 5.9%) (7) and higher levels in a high-risk population (female sex workers; 4/16, 25%) (11).

This study has limitations. The resistance profiles for the infecting strains of M. genitalium were not tested in in vitro culture. There may be other unknown changes in the genome that confer resistance to the drugs of interest. In addition, the resistance levels reported are probably underestimates because samples were collected in 2012–2013 and levels have probably risen since then (7).

We found high frequency of ParC S83 changes associated with fluoroquinolone resistance in a sexually transmitted infection clinic in urban Australia; these changes were associated with moxifloxacin failure. The high level of dual markers for macrolide/fluoroquinolone resistance suggests successive treatment failure after sequential monotherapy leading to the serious outcome that ≈10% of M. genitalium infections are not treatable with recommended or readily available antimicrobial drugs. In the absence of alternatives, treatment with pristinamycin cured all 6 patients with dual-class resistance infections (G.L. Murray et al., unpub data).

This study highlights the urgent need for antimicrobial drug resistance surveillance and the value of diagnostic assays that report the presence of resistance markers to optimize treatment. Our results suggest that it is time to reconsider the indications for azithromycin and invest in trials of different available as well as novel classes of antimicrobial drugs for M. genitalium treatment. They also raise serious concerns about sequential use of monotherapy and the need to evaluate combination therapies as we enter a new era of untreatable sexually transmitted infections.