The relatively low rate of macrolide resistance-associated mutations in Russia and Estonia, and the significant difference in macrolide resistance rates in these countries, can be attributed to a number of factors. First, testing for M. genitalium is not introduced everywhere, which may be particularly the case in Russia, and therefore many M. genitalium infections, symptomatic and particularly asymptomatic ones, are not diagnosed and treated. In Russia, in addition, some of the M. genitalium nucleic acid amplification tests have suboptimal sensitivity, which results in missing cases with low M. genitalium loads [30]. Furthermore, screening for Chlamydia trachomatis, with subsequent treatment using azithromycin, is also less frequent in Russia compared to many European Union countries. Finally, in Russia the recommended first-line antimicrobials for M. genitalium infections are josamycin (500 mg three times daily, 10 days) or doxycycline (100 mg twice a day, 10 days) [31]. In Estonia, the recommended first-line antimicrobials are azithromycin (500 mg first day, then 250 mg daily for 4 days) or as alternative treatment moxifloxacin 400 mg daily for 7–10 days. The data on activity of josamycin against M. genitalium and the resistance selection with josamycin are very limited. Its activity in vitro was shown to be slightly lower than that of azithromycin [32]. It was found that macrolide resistance mutations resulting in treatment failures can be rapidly selected during treatment with josamycin [14], with a mutation selection rate of 6.5%, which is comparable with that of the extended course of azithromycin (0–6.5%), yet lower than that of 1 g single dose azithromycin (10%) [7],[33]. It is unknown how frequently doxycycline is administered for M. genitalium infections in Russia (can vary substantially in different regions), but using this drug may contribute to the decreased rate of macrolide resistance as a proportion (~30%) of patients get cured with the recommend doxycycline regimen. Nevertheless, the macrolide resistance selection pressure is still considerable in Russia, particularly because josamycin and azithromycin (1 g single oral dose) are widely used as first-line treatments for C. trachomatis infections.

Data regarding the prevalence of possible fluoroquinolone resistance-associated mutations in Europe remain very limited. In mostly minor recent studies from United Kingdom, France and Germany, a ParC QRDR alteration was found in 5% (1 of 22 samples) [20], 6% (12/200) [17] and 10% (2/19) [25] of samples, respectively. In contrast, in a study from Australia, 15% of specimens collected between 2008 and 2011 carried mutations associated with fluoroquinolone resistance [21]. A report from Japan found parC QRDR mutations in 33% of specimens collected between 2011 and 2013, with a dramatic increase from 20% in 2011 to 47% in 2013 [22]. However, a majority of this increase was due to the S83N mutation that might not result in significantly elevated MICs of moxifloxacin (Jensen et al, unpublished). In our study, the rates of parC QRDR mutations were 6.2% in Russia, ranging from 2.5% in Tula to 7.6% in Moscow, and 5% in Estonia. The predominance of the mutations S83I, S83N, D87N and D87Y in our study is consistent with previous studies [11],[13],[21],[22]. In addition to nonsynonymous mutations at S83 and D87 (80 and 84 in E. coli), which are known to be associated with fluoroquinolone resistance in M. genitalium and other closely related organisms [11],[13],[21],[22],[34], we found several amino acid changes at position 84 (81 in E. coli) (S84G, S84H, S84I, S84R, S84Y) and one mutation at position 90 (87 in E. coli) (I90N), which have not been previously reported or associated with fluoroquinolone resistance, to our knowledge, in mycoplasmas. However, also a ParC S84P mutation was found in one sample from Moscow and this mutation has been previously shown to cause resistance to fluoroquinolones in a clinical M. hominis isolate [35]. The recommended second-line treatment for M. genitalium infections in Russia is ofloxacin (400 mg twice a day, 10 days) [31], although other fluoroquinolones, including moxifloxacin and levofloxacin, can also be used. Given the suboptimal activity of ofloxacin and levofloxacin against M. genitalium and the frequent use of these fluoroquinolones in the treatment of other gynaecological and urological infections (which exerts pressure to select fluoroquinolone resistance in M. genitalium), the relatively low rate of fluoroquinolone resistance mutations in Russia was slightly unexpected. However, the possible resistance to fluoroquinolones was higher than the resistance to macrolides, which is rare internationally.

In the present study, seven samples in Russia and one sample in Estonia harbored both macrolide and possible fluoroquinolone resistance-associated mutations, accounting for about 1% of samples in both countries. Multidrug resistance in M. genitalium, with a prevalence of up to 25%, has recently been reported in studies from Australia and Japan [10],[11],[21],[22],[23]. This is of great concern because there is only one option for third-line treatment, i.e. pristinamycin (1 g four times daily for 10 days) [4], which is not 100% effective. Novel antimicrobials for treatment of M. genitalium infections are crucial and dual antimicrobial therapy has to be considered. Some new antimicrobials such as solithromycin, lefamulin, and zoliflodacin deserve further attention for treatment of M. genitalium infections [5].

The main limitations of the present study included that no information was available whether a sample was submitted before or after exposure to any antimicrobial treatment. In many cases, it was not either possible to reliably identify multiple M. genitalium positive samples from the identical individuals, because no unique patient-specific numbers are used in Russia and, in addition, in some clinics patients can submit their samples anonymously. Consequently, there might have been a few cases of treatment failures in the examined material, creating a limited bias in the prevalence of mutations. It was also impossible to compare resistance rates in different years, because nearly 90% of the samples were obtained in 2014 (n = 560; 64.6%) and 2015 (n = 201; 23.2%), and also because the samples obtained in different years were not representative of all centres.

Conclusions