Detection of parC gene mutations associated with quinolone resistance in Mycoplasma genitalium: evaluation of a multiplex real-time PCR assay

Introduction Increasing levels of antibiotic resistance are complicating treatment for the sexually transmitted pathogen Mycoplasma genitalium . Resistance to fluoroquinolones is associated with mutations in the parC gene. Although the precise mutations conferring resistance are not fully understood, the single nucleotide polymorphism (SNP) G248T/S83I is most implicated. Aim To evaluate the performance of the MG+parC(beta2) assay (SpeeDx, Australia), which detects single nucleotide polymorphisms (SNPs) in the parC gene at amino acid position S83 (A247C/S83R, G248T/S83I, G248A/S83N) and D87 (G259A/D87N, G259T/D87Y, G259C/D87H). Methods Clinical samples were analysed by MG+parC(beta2) assay and results compared to Sanger sequencing. Sensitivity, specificity, and predictive value for treatment failure were calculated. Results From analysis of 205 samples, the MG+parC(beta2) assay performed with a high sensitivity 98.2% (95% CI:90.3–100) and specificity 99.3% (95% CI:96.3–100) for parC SNP detection with a kappa of 0.97 (95% CI:0.94–1.00). The predictive value of G248T/S83I detection (the most common SNP, prevalence of 13% in the study population) was analysed with respect to treatment failure (patients received sequential doxycycline-moxifloxacin). The positive-predictive-value for moxifloxacin failure after detection of S83I was only 44% (95% CI:24.4–65.1), while negative-predictive-value was high at 96.9% (95% CI:92.7–99.0), suggesting that other SNPs are contributing to resistance. Conclusion MG+parC(beta2) performed with high concordance compared to Sanger sequencing. Such qPCR assays can assist in understanding causes of treatment failure, inform the development of diagnostic assays, and can be applied to surveillance of mutations in populations. Due to an incomplete understanding of the basis for fluoroquinolone resistance, such tests do not appear to be ready for clinical application.


INTRODUCTION
Mycoplasma genitalium is a sexually transmitted pathogen associated with a number of syndromes in males and females [1,2]. Treatment of infection is complicated by increasing levels of resistance to the recommended first line treatment, the macrolide azithromycin, and second line treatment, the fluoroquinolone moxifloxacin [3].
Research tools that detect SNPs can help to develop an understanding of causes of treatment failure and inform the development of diagnostic assays. Quantitative PCR-based assays are generally quicker, more convenient, and cheaper than Sanger sequencing. We previously evaluated the test MG+parC(beta) (SpeeDx, Sydney, Australia) that detects M. genitalium plus five parC SNPs (excluding G248A/S83N) [15]. In this study we evaluate the performance of the revised assay, MG+parC(beta2), that detects the six above mentioned SNPs in parC (Table 1).

Samples used for this study
M. genitalium-positive samples (n=242), were previously collected from patients undergoing sequential doxycyclinemoxifloxacin treatment for macrolide-resistant M. genitalium [6]. The samples included 223 pre-treatment samples (a diagnostic sample taken prior to any antibiotic therapy) and 19 test of cure samples. Sample types included 166 urine, and 45 anal, 14 cervical, 14 vaginal/high vaginal, and two urethral swabs, plus one swab of unknown site.

Analysis of results in the context of treatment outcomes
The treatment outcomes for the patients in this study group are known, with 92% of patients cured by sequential doxycyclinemoxifloxacin treatment, and 8% failing treatment [6]. This allowed calculation of the positive predictive value of the assay for treatment failure. The most common SNP in the study The parC SNP G248T/S83I is the most strongly implicated mutation in M. genitalium fluoroquinolone resistance [7][8][9]. The results of this study indicate that detection of S83I has a low predictive value for moxifloxacin failure (particularly when used following doxycycline) and suggests that fluoroquinolone resistance may have a more complex basis than macrolide resistance, as outlined in the Introduction. Notably, in this study the pre-treatment of patients with doxycycline prior to moxifloxacin may have impacted cure, so the findings here do not necessarily translate to the situation where preceding doxycycline is not used. However, it is also important to note that quinolone resistance assays are likely to be used in the context of sequential doxycyclinemoxifloxacin treatment as doxycycline is recommended within sequential treatment for M. genitalium in Australia and the UK [16,17], and doxycycline is recommended for initial treatment of NGU (Europe, UK, Australia) [16][17][18], and as one of two recommended regimens for NGU in the USA [19].
As mentioned above, resistance to fluoroquinolones appears to be more complex than the situation for macrolides. There are other SNPs that are not captured by current parC assays that may contribute to resistance. These include SNPs affecting amino acids M95 and D99 in gyrA, which may have an additive effect when combined with parC SNPs [9]. Undefined variations elsewhere in the genome may also contribute. The results of this study demonstrate current limitations in using parC-based assays to triage patients away from moxifloxacin treatment while we still don't have a clear understanding of mechanisms of resistance [20]. This is particularly important in an environment where there are limited alternative treatments for M. genitalium.
This MG+parC(beta2) assay was used in an earlier study using a smaller study group [21]. Sanger sequencing wasn't used to confirm mutations, which were few in number (n=7), and treatment data were only available for three patients, so it is difficult to draw comparisons with the current study.

Study limitations
Lower load samples are less likely to yield a successful sequencing result, so it is possible that they were not included in the analysis. Additionally, results for Australian samples may not be generalizable to other locations.

CONCLUSION
MG+parC(beta2) showed very high specificity and sensitivity for detection of SNPs potentially associated with fluoroquinolone resistance. Such quantitative PCR-based assays are generally quicker and cheaper than Sanger sequencing, making this assay a useful tool for detecting parC mutations to investigate treatment failure and for the surveillance of mutations in populations.