Clinical Performance of Three Commercial Molecular Diagnostic Assays for the Detection of Fluoroquinolone Resistance-Associated Mutations in Mycoplasma genitalium

ABSTRACT

The high prevalence of macrolide resistance in Mycoplasma genitalium results in an increased reliance on moxifloxacin, the second-line treatment; however, moxifloxacin resistance has also emerged. Because assays that can detect fluoroquinolone resistance-associated mutations will be useful for the management of macrolide-resistant M. genitalium infections, we evaluated the performance of three commercial assays (the Allplex MG & MoxiR Assay [Seegene], LightMix Modular parC kit [TIBMOLBIOL], and MGMO qPCR [NYtor) in comparison with parC gene Sanger sequencing used as the reference. Between January 2018 and December 2020, remnants of M. genitalium-positive clinical specimens received at the French National Reference Center for Bacterial Sexually Transmitted Infections were collected if a Sanger sequencing result was obtained for the parC gene. Overall, 368 M. genitalium-positive specimens were assessed. The clinical sensitivities for the detection of the ParC mutations that are likely of clinical significance were 91.8% (95% CI = 83.2 to 96.2), 98.6% (95% CI = 92.4 to 99.8), and 94.4% (95% CI = 86.6 to 97.8) for the Allplex MG & MoxiR, LightMix Modular parC, and MGMO qPCR kits, respectively, with no significant difference between the three kits. The clinical specificity of the Allplex MG & MoxiR and MGMO qPCR kits was 100% (95% CI = 97.7 to 100 and 98.7 to 100, respectively), which was significantly higher than the specificity of the LightMix Modular parC kit of 95.4% (95%CI = 92.3 to 97.3), for which the interpretation of melting curves may be misleading. These kits should be useful for the selection of antimicrobials in macrolide-resistant M. genitalium infections, although further developments may be necessary because parC mutations involved in fluoroquinolone resistance have not been precisely determined.

INTRODUCTION

Sexually transmitted Mycoplasma genitalium infections are treated with macrolides as part of the first-line treatment according to the most available guidelines, and resistance-guided therapy is recommended in several countries (1–3) (http://www.sti.guidelines.org.au/sexually-transmissible-infections/mycoplasma-genitalium). The high prevalence of macrolide resistance worldwide (4, 5) has resulted in an increased reliance on moxifloxacin, which is the second-line treatment; however, moxifloxacin resistance has emerged (4–7). Thus, assays that are capable of detecting fluoroquinolone resistance-associated mutations would be helpful for the management of macrolide-resistant M. genitalium infections.

Moxifloxacin resistance is mainly associated with a mutation in the parC gene encoding topoisomerase IV. The most common single nucleotide polymorphism (SNP) associated with moxifloxacin MIC increase and treatment failure is G248T (M. genitalium numbering), corresponding to amino acid Ser83Ile substitution (6, 8, 9). Other ParC substitutions, such as Ser83Arg, Asp87Tyr, Asp87Asn, and Gly81Cys, are less frequent (6). They have been associated with MIC increase (9, 10) or with a small number of moxifloxacin treatment failures (6, 8, 11–13). SNPs reported in the parC gene of M. genitalium have not all been associated with moxifloxacin resistance. Many of them remain of unknown significance, mainly because M. genitalium culture and MIC determination are tremendously fastidious. Mutations in the gyrA gene encoding the DNA gyrase have also been reported at low frequency in M. genitalium. However, they have almost never been detected in the absence of ParC mutations and have not been correlated alone with M. genitalium MIC increase or treatment failure (8, 11, 14, 15).

Whereas commercial tests for the detection of macrolide resistance-associated mutations are available and have been evaluated (16–19), research-use only (RUO) or CE-IVD commercial tests for the detection of fluoroquinolone resistance-associated mutations are scarce. A multiplex real-time PCR assay, MG+parC(beta2) assay (SpeeDx Australia), which detects ParC Ser83Ile, Ser83Arg, Ser83Asn, Asp87Asn, Asp87Tyr, and Asp87His mutations, showed high concordance compared with Sanger sequencing (20–22), but the test has not been commercially available to date. A real-time PCR assay, the AmpliSens M. genitalium ML/FQ-Resist-FL (Central Research Institute of Epidemiology, Moscow, Russia), was recently evaluated. This kit may overestimate the prevalence of moxifloxacin resistance because it is designed to detect only the wild-type parC gene and, thus, indicates the presence of any parC mutation in the absence of amplification (23). In this study, we evaluated the performance of three real-time PCR commercial assays for the detection of fluoroquinolone resistance-associated mutations in M. genitalium-positive specimens: the Allplex MG & MoxiR Assay (Seegene, Republic of Korea), LightMix Modular parC kit (TIB MOLBIOL, Germany), and MGMO qPCR (NYtor, the Netherlands). The clinical sensitivity and specificity for the detection of parC mutations that are likely associated with moxifloxacin resistance were determined in comparison with parC Sanger sequencing used as the reference.

MATERIALS AND METHODS

Clinical specimens.

Between January 1, 2018 and December 31, 2020, remnants of M. genitalium-positive clinical specimens received at the French National Reference Center for Bacterial Sexually Transmitted Infections in the Bacteriology Department of Bordeaux University Hospital (France) were systematically collected if a Sanger sequencing result was obtained for the parC gene. The specimens were stored at −80°C until testing. A total of 368 clinical specimens were analyzed, including 188 cervicovaginal swabs, four first-void urine samples, and three rectal swabs from 195 women; and 96 first-void urine samples, 63 rectal, nine urethral, and four throat swabs from 172 men.

Nucleic acid extraction was performed from 200 μL of each clinical sample using the MagNA Pure 96 DNA and viral NA small-volume kit on the MagNA Pure 96 instrument (Roche Diagnostics, France).

Reference method for the detection of fluoroquinolone resistance-associated mutations.

A 214-bp fragment of the parC gene encompassing the quinolone resistance determining region (QRDR) of this gene was amplified from 10 μL of DNA using MG-PARC-A and MG-PARC-B primers, as previously described (7). PCR products were sent to Eurofins Genomics (Germany) for Sanger sequencing. Sequencing data were analyzed using BioEdit version 7.2.5 software (Isis Pharmaceuticals, USA) and were compared with the sequence of the wild-type M. genitalium G37 reference strain (24).

Detection of fluoroquinolone resistance using commercial kits.

The main characteristics of the three evaluated assays are presented in Table 1.

TABLE 1.

Main characteristics of the three commercial assays

Kits	Allplex MG & MoxiR assay (Seegene)	LightMix modular parC (TIBMOLBIOL)	MGMO qPCR (NYtor)
PCR technology	MuDT technology	FRET	Taqman
M. genitalium detection (targeted gene if specified)	Included (unspecified target)	Not included	Included (MgPa)
parC mutations detecteda	A247C (S83R), C248A (S83N), G248T (S83I), G259C (D87H), G259T (D87Y), G259A (D87N)	All mutations within the amplified fragment	G248T (S83I), A247C (S83R), G259A (D87N), G259T (D87Y)
Validated specimen types	Urine, genital swab, liquid-based cytology	Unspecified	Unspecified
Input sample vol	5 μL of DNA extract	5 μL of DNA extract	5 μL of DNA extract
Internal control	Endogenous human and exogenous internal controlb	Not included	Includedc
Positive and negative controls	Included	Not included	Included
No. of tests/kit	50	100	100
No. of reactions/run	Batches; up to 94 samples/PCR run	Batches; up to 94 samples/PCR run	Batches; up to 94 samples/PCR run
Hands-on time (min)d	15 to 45, depending on the no. of samples	15 to 45, depending on the no. of samples	15 to 45, depending on the no. of samples
Test turnaround time	1h15	1h10	1h40
Instrument	CFX96 (Bio-Rad)	LightCycler 480 (Roche)	CFX96 (Bio-Rad)
Automation	Amplification and analysis	Amplification	Amplification and analysis
Data analysis software	Seegene viewer 3.24.000	LightCycler 480 SW 1.5.1	CFX Maestro 1.1
List price/reaction (€), excluding taxes	47.12 €	3.13 €e	7.5 €
CE marking	CE-IVD	RUOf	RUO

^aM. genitalium numbering with M. genitalium amino acid numbering in parentheses.

^bExtraction and inhibition control.

^cInhibition control.

^dNot including DNA extraction.

^eIncluding the LightCycler Multiplex DNA Master, which is not provided in the kit.

^fRUO, research use only; CE-IVD, approved Communauté Européenne marking for in vitro diagnostic medical devices.

(i) Allplex MG & MoxiR assay. The Allplex MG & MoxiR Assay (Seegene, Republic of Korea) is a CE-IVD-marked multiplex real-time PCR assay that detects M. genitalium (undisclosed target) and six mutations in the parC gene (A247C -M. genitalium numbering-, G248A, G248T, G259C, G259T, and G259A). An endogenous human gene is detected as an internal control, except in urine specimens, to which the exogenous ASTI IC(2) internal control is added prior to DNA extraction. The Allplex MG & AziR kit was validated for urine samples, genital swabs, and liquid-based cytology. Assays were performed using 5 μL of DNA extract and a CFX96 real-time PCR system (Bio-Rad) according to the manufacturer’s instructions. Data were analyzed using Seegene viewer software version 3.24.000.

(ii) LightMix Modular parC kit. The LightMix Modular parC kit (TIB MOLBIOL, Germany) is a RUO-marked FRET PCR assay using melting curve analysis. The validated specimen types are not specified. This kit was designed to detect mutations in the parC gene but was not designed to detect M. genitalium. In the manufacturer’s instructions, mutations are identified according to the melting temperature (Tm): Tm ≈ 62.8°C corresponds to specimens harboring the C234T mutation (silent mutation which is not associated with fluoroquinolone resistance), Tm ≈ 61.8°C corresponds to specimens harboring the G248T mutation (fluoroquinolone resistance-associated mutation), Tm ≈ 58.8°C corresponds to the wild-type parC gene, and Tm < 58.0°C corresponds to specimens harboring other fluoroquinolone resistance-associated mutations. The kit does not provide any internal control. Assays were performed with 5 μL of DNA extract and 4 μL of LightCycler Multiplex DNA master (Roche Diagnostics) according to the manufacturer’s instructions using a LightCycler 480 real-time PCR system (Roche). Data were analyzed using LightCycler 480 software version 1.5.1.

(iii) MGMO qPCR kit, version 1.0. MGMO qPCR (NYtor, the Netherlands) is a RUO-marked multiplex real-time PCR assay that detects M. genitalium by targeting the mgpB gene and four fluoroquinolone resistance-associated mutations in the parC gene (G248T, A247C, G259A, and G259T). An inhibition control is included in the kit. Assays were performed using 5 μL of DNA extract and a CFX96 real-time PCR system (Bio-Rad) according to the manufacturer’s instructions. Data were analyzed using CFX Maestro software version 1.1.

Data analysis.

Sanger sequencing of parC PCR products was considered the reference standard for detecting fluoroquinolone resistance-associated mutations. The Ser83Ile, Ser83Arg, Asp87Tyr, and Asp87Asn changes in the ParC protein were considered likely of clinical significance because increased MICs have been observed in mutated isolates (9, 10) or because moxifloxacin treatment failures were reported in the presence of these mutations (6, 8, 11–13). The Gly81Cys mutation was also considered of clinical significance because increased moxifloxacin MIC was reported (25). In contrast, the ParC Ser83Asn alteration, which did not significantly increase fluoroquinolone MICs (9), was not included in the ParC alterations that were likely to have clinical significance. The parC C234T mutation is a silent mutation which is not associated with fluoroquinolone resistance.

The clinical sensitivity and specificity for fluoroquinolone resistance-associated mutation detection were calculated together with the corresponding 95% confidence intervals (CIs). The kappa (κ) value was determined as a measure of the overall agreement.

Ethics statement.

Remnants of specimens were preserved at the Centre de Ressource Biologique-Bordeaux Biothèque Santé (CRB-BBS) of Bordeaux University Hospital under collection number BB-0033-00094 and authorization AC-2014-2166 from the French Ministry of Higher Education and Research with no information regarding patient identity.

RESULTS

A total of 368 M. genitalium-positive specimens were studied. Among them, 254 (69.0%) harbored a wild-type parC, 74 (20.1%) harbored mutations that were likely to be associated with fluoroquinolone resistance as defined above, and 40 (10.9%) harbored mutations that were unlikely of clinical significance (Table 2). The Ser83Ile mutation was the most frequently encountered fluoroquinolone resistance-associated mutation, being present in 83.8% (62/74) of specimens harboring clinically significant mutations and was once associated with a silent mutation due to a C234T nucleotide substitution. The Asp87Asn mutation was present in 8.1% (6/74) of cases and was associated with the C234T silent mutation in one case. The Asp87Tyr and Gly81Cys substitutions were each present in 4.1% (3/74) of specimens, with the latter mutation being associated once with the C234T silent transition. Among the 40 mutations that were unlikely of clinical significance, Ser83Asn was present in 20% (8/40 cases), whereas the C234T silent transition was found in 57.5% (23/40) of specimens, alone or associated once with the Ser83Asn ParC substitution. In three specimens, nucleotide substitutions also led to mutations in the QRDR of ParC at positions 84 (Ser84Pro) or 87 (Asp87Gly). These mutations were considered unlikely to be of clinical significance because no data have been available to date.

TABLE 2.

Mutations in the parC gene detected by Sanger sequencing in the 368 M. genitalium-positive specimens studied

Mutation status	Mutation in parC genea	Mutation in the ParC proteinb	No. of specimens	Total
No mutation	WTc	WT	254	254
Likely associated with fluoroquinolone resistance	G248T	S83(80)I	61	74
	G259A	D87(84)N	5
	G259T	D87(84)Y	3
	G241T	G81(78)C	2
	G248T & C234T	S83(80)I & SM	1
	G259A & C234T	D87(84)N & SM	1
	G241T & C234T	G81(78)C & SM	1
Not likely associated with fluoroquinolone resistance	C234T	SM	22	40
	G248A	S83(80)N	7
	G355T	A119(117)S	3
	T318G	H106(103)Q	1
	C332T	S111(108)L	1
	T250C	S84(81)P	1
	A260G	D87(84)G	1
	G248A & C234T	S83(80)N & SM	1
	T250C & G272A	S84(81)P & R91(88)K	1
	G291A	SM	1
	T306G	SM	1
Total				368

^aM. genitalium numbering.

^bM. genitalium numbering with E. coli numbering in parentheses.

^cWT, wild-type; SM, silent mutation.

Among the 368 M. genitalium-positive specimens evaluated, some specimens did not yield resistance results (Tables 3–4). Using the Allplex MG & MoxiR Assay, one (0.27%) cervicovaginal specimen was invalid because of the absence of amplification of the internal control, and one (0.27%) urine specimen did not yield amplification of the M. genitalium undisclosed target gene. Using the MGMO qPCR kit, the amplification of the M. genitalium MgPa target gene was not obtained for seven (1.9%) specimens, including four cervicovaginal swabs and three urine samples. Regarding the LightMix Modular parC kit, 12 (3.3%) specimens (eight cervicovaginal, two rectal swabs, and two urine samples) did not yield melting curves, and the results of three specimens (0.8%, one urine and two rectal swabs), among which two harbored the Ser83Ile mutation, could not be determined because the Tm values obtained were not interpretable according to the Tm values reported in the manufacturer’s instructions.

TABLE 3.

Performance of the three commercial kits for the detection of fluoroquinolone resistance-associated mutations that are likely of clinical significance^a

		parC sequencing results
Commercial assay (manufacturer)	parC resistance-associated mutation detection result	Presence of mutation that is likely of clinical significance	No mutation that is likely of clinical significance	Total	Overall % agreement (95% CI), κ value	Sensitivity (%) (95% CI)	Specificity (%) (95% CI)
Allplex MG & MoxiR (Seegene)	Detected	67	0	67	98.4 (96.5 to 99.3) κ = 0.95	91.8 (83.2 to 96.2)	100 (97.7 to 100)
	Not detected	6	293	299
	Total	73	293	366
	Invalid or NA	1	1	2
LightMix Modular parC (TIB MOLBIOL)	Detected	70	13	83	96.0 (93.5 to 97.6) κ = 0.88	98.6 (92.4 to 99.8)	95.4 (92.3 to 97.3)
	Not detected	1	269	270
	Total	71	282	353
	NA	0	12	12
	ND	3	0	3
MGMO qPCR (NYtor)	Detected	68	0	68	98.9 (97.2 to 99.6) κ = 0.96	94.4 (86.6 to 97.8)	100 (98.7 to 100)
	Not detected	4	289	293
	Total	72	289	361
	NA	2	5	7

^aCI, confidence interval; NA, not amplified; ND, not determined because the obtained Tm values were not interpretable according to the Tm values reported in the manufacturer’s instructions.

TABLE 4.

parC mutation detection by commercial kits compared with parC Sanger sequencing^a

			Allplex MG & MoxiR (Seegene)								LightMix modular parC (TIB MOLBIOL)						MGMO qPCR (NYtor)
Mutation status	Mutations by Sanger sequencing	Corresponding protein mutationb	G259C	A247C	G259T	G248A	G259A	G248T	WT	Invalid or NA	Tm~62.8°C SM	Tm~61.8°C G248T	Tm~58.8°C WT	Tm < 58.0°C Other ressiance-associated mutaions	ND	NA	Mutation detected	Mutation not detected	NA	No. of specimens
No mutation	WT	WT	0	0	0	0	0	0	253	1	5	0	235	2	0	12	0	249	5	254
Likely associated with fluoroquinolone resistance	G248T	S83I	0	0	0	0	0	57	3	1	0	59	1	0	1	0	59	0	2	61
	G259A	D87N	0	0	0	0	5	0	0	0	0	0	0	5	0	0	4	1	0	5
	G259T	D87Y	0	0	3	0	0	0	0	0	0	0	0	3	0	0	3	0	0	3
	G241T	G81C	0	0	0	0	0	0	2	0	0	0	0	2	0	0	0	2	0	2
	G248T & C234T	S83I & SM	0	0	0	0	0	1	0	0	0	0	0	0	1	0	1	0	0	1
	G259A & C234T	D87N & SM	0	0	0	0	1	0	0	0	0	0	0	0	1	0	1	0	0	1
	G241T & C234T	G81C & SM	0	0	0	0	0	0	1	0	0	0	0	1	0	0	0	1	0	1
Mutation that is unlikely of clinical significance	C234T	SM	0	0	0	0	0	0	22	0	22	0	0	0	0	0	0	22	0	22
	G248A	S83N	0	0	0	7	0	0	0	0	0	0	0	7	0	0	0	7	0	7
	G355T	A119S	0	0	0	0	0	0	3	0	0	0	3	0	0	0	0	3	0	3
	T318G	H106Q	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	1	0	1
	C332T	S111L	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	1	0	1
	T250C	S84P	0	0	0	0	0	0	1	0	0	0	0	1	0	0	0	1	0	1
	A260G	D87G	0	0	0	0	0	0	1	0	0	0	0	1	0	0	0	1	0	1
	G248A & C234T	S83N & SM	0	0	0	1	0	0	0	0	0	1	0	0	0	0	0	1	0	1
	T250C & G272A	S84P & R91K	0	0	0	0	0	0	1	0	0	0	0	1	0	0	0	1	0	1
	G291A	SM	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	1	0	1
	T306G	SM	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	1	0	1
Total			0	0	3	8	6	58	291	2	27	60	243	23	3	12	68	293	7	368

^aWT, wild type; SM, silent mutation; NA, not amplified; ND, not determined because the Tm values obtained were not specified in the manufacturer’s instructions.

^bM. genitalium numbering.

The overall agreements with parC sequencing for the detection of the mutations that were likely associated with fluoroquinolone resistance were 98.4% (95% CI = 96.5% to 99.3%) for the Allplex MG & MoxiR assay kit, 96.0% (95% CI = 93.5% to 97.6%) for the LightMix Modular parC kit, and 98.9% (95% CI = 97.2% to 99.6%) for the MGMO qPCR kit, with no significant difference between the three kits (Table 3). The k values were 0.95, 0.88, and 0.96, respectively.

The clinical sensitivities for the detection of the ParC mutations that were likely of clinical significance were 91.8% (95% CI = 83.2% to 96.21), 98.6% (95% CI = 92.4% to 99.8), and 94.4% (95% CI = 86.6% to 97.8) for the Allplex MG & MoxiR, LightMix Modular parC, and MGMO qPCR kits, respectively. Although the sensitivity of the Allplex MG & MoxiR Assay appeared lower than the sensitivity of the two other kits, there was no significant difference, as indicated by overlapping CIs. Among the specimens yielding an accurate parC amplification, the Allplex MG & MoxiR kit missed three specimens harboring the Ser83Ile mutation, whereas the LightMix Modular parC kit missed one and the MGMO qPCR kit did not (Table 4). In addition, the Allplex MG & MoxiR assay and the MGMO qPCR kit did not detect the three specimens harboring the Gly81Cys mutation, as expected, because these two assays were not designed to detect it. In contrast, these three specimens were accurately detected as mutated using the LightMix Modular parC kit.

The clinical specificity of the Allplex MG & MoxiR and the MGMO qPCR kits was 100% (95% CI = 97.7% to 100% and 98.7% to 100%, respectively), which was significantly higher than the specificity of the LightMix Modular parC kit, which was 95.4% (95% CI = 92.3% to 97.3%) (Table 3). Among the 13 specimens that were falsely considered fluoroquinolone-resistant by the kit, eight (61.5%) harbored the Ser83Asn mutation, two were wild-type, and the three others harbored mutations of unknown significance at positions 84, 87, and 91 (Table 4).

Regarding the handling characteristics of the kits, only one kit, the Allplex MG & MoxiR Assay, has been CE-IVD marked. The two others are intended for RUO. The Allplex MG & MoxiR and the MGMO qPCR kits both included an internal control, which was not the case of the LightMix Modular parC kit (Table 1). However, the manufacturer’s instructions did not provide any Ct threshold for invalid results in cases of weak amplification of the internal control. Analysis of the MGMO qPCR result curves was easy, but the nature of the mutation detected remained unknown to the operator. The analysis of the Allplex MG & MoxiR Assay was entirely performed by the dedicated software, and the nature of the mutation was displayed, which is not the case of the two other kits. The analysis of the melting curves generated with the LightMix Modular parC kit results was sometimes misleading. The Tm values provided in the manufacturer’s instructions were inadequate to determine the resistance status of all strains. Indeed, the comparison with previously characterized wild-type and mutated specimens (not included in the kit) was necessary to accurately determine the wild-type or resistance status of the M. genitalium strain. In addition, in cases of association of the C234T silent mutation with another mutation, a shift of the expected Tm value was observed, leading either to a noninterpretable result (one specimen harboring Ser83Ile and one specimen harboring Asp87Asn) or to a false-resistant result (one specimen harboring Ser83Asn) (Table 4).

DISCUSSION

This study aimed to evaluate the performance of three commercial kits for the detection of fluoroquinolone resistance-associated mutations in M. genitalium. The use of such kits should improve the turnaround time of results compared with parC amplification followed by Sanger sequencing, and may also be less expensive than sequencing according to the kit.

The MGMO qPCR and the LightMix Modular parC kits did not yield amplification or melting curves in seven (1.9%) and 12 (3.3%) specimens, respectively. These failures may be random events or may be associated with low M. genitalium DNA load. To address this question, the M. genitalium detection Ct-values obtained with the Allplex MG & MoxiR Assay were examined in the nonamplified specimens in comparison with the M. genitalium detection Ct-values in accurately amplified samples. The mean of Ct-values of nonamplified specimens were 35.8 for the MGMO qPCR kit and 36.6 for the LightMix Modular parC kit, which was significantly higher than the mean of Ct values obtained for the accurately amplified samples (31.7, P < 0.05 for the MGMO qPCR kit, and 31.6, P < 0.001 for the LightMix Modular parC kit) (data not shown). This finding suggests that amplification failures were associated with lower M. genitalium DNA load.

In this study, the Gly81Cys mutation was considered of likely clinical significance because moxifloxacin MICs were shown to be a 4-fold increase compared with those of the wild-type strains (25). However, more studies are needed to determine whether this mutation is involved in moxifloxacin treatment failure. If the Gly81Cys mutation had not been considered of clinical significance here, the sensitivities of the two kits that were not designed to detect it would have been higher: 95.7% (95% CI = 88.1 to 98.5) and 98.6% (95% CI = 92.2 to 99.7) for the Allplex MG & MoxiR and the MGMO qPCR kits, respectively, whereas their specificity would have been unchanged (data not shown).

Overall, the LightMix Modular parC kit showed a significantly lower specificity than the two other kits. Indeed, this FRET-based kit detects all mutations occurring in the studied parC gene region, including mutations that were unlikely to be associated with fluoroquinolone resistance. The precise boundaries of the targeted DNA fragment are undisclosed, but amino acids at positions 106, 111, and 119 seem not to be included, as mutations at these positions were not detected by the kit in the five concerned specimens (Table 4). It should be noted that if the detection of any mutation found by Sanger sequencing only was considered, the specificity of the kit would have been 99.3% (95% CI = 97.4 to 99.8), with a k value of 0.98 (data not shown). However, detecting some parC mutations as fluoroquinolone-resistant that are unlikely to be associated with fluoroquinolone resistance is a major drawback for a kit intended to guide therapy.

The implementation of fluoroquinolone resistance-guided therapy is more complex than the macrolide resistance-guided therapy already recommended in several guidelines (1, 2) (http://www.sti.guidelines.org.au/sexually-transmissible-infections/mycoplasma-genitalium) because many various SNPs can occur in the parC and gyrA genes. In addition, their relative contribution to moxifloxacin treatment failure and moxifloxacin MIC increase remains to be precisely determined. In a recent study based on Australian data, it was shown that in the absence of Ser83Ile, macrolide-resistant M. genitalium infections had a 96.4% probability of moxifloxacin cure (6). The absence of mutations at positions 83 and 87 had a 97.8% probability of cure using moxifloxacin. Thus, including the detection of substitutions at both positions 83 and 87 appears to be of interest in commercial kits, although the Ser83Asn substitution, which is not associated with moxifloxacin MIC increase, should be either not included in the assay (which is the case of the MGMO qPCR kit) or individually displayed by the assay (which is the case of the Allplex MG & MoxiR Assay). Indeed, because mutations at positions 83 and 87 other than the Ser83Ile substitution are less commonly associated with treatment failure (6), kits displaying the precise nature of the detected mutation present an advantage. Moxifloxacin treatment could still be considered in their presence if any alternative antimicrobial treatment is available. Considering the difficulties in determining which ParC mutations are truly associated with treatment failure, another strategy focusing only on the parC wild-type sequence to predict microbial cure following moxifloxacin treatment rather than inferring susceptibility in the absence of resistance mutation has recently been suggested (26). Accordingly, two proof-of-concept molecular tests that focus on the wild-type ParC positions 83 and 87 and based on post-PCR melting curve analysis have recently been developed (27). Overall, waiting for the determination of the precise significance of each parC SNP on moxifloxacin treatment outcomes and the subsequent development of accurate commercial kits, the kits evaluated in this study should be of help to decide the possible use of a last resort treatment, such as pristinamycin or minocycline (1, 3).

One limitation of this study is that only specimens yielding a Sanger sequencing result for the parC gene were included. Thus, specimens with an M. genitalium DNA load that was too low for Sanger sequencing were not evaluated in this study because of the lack of a reference standard method in these cases. This study did not evaluate the performance of the kits for the detection of M. genitalium. Notably, the TIBMOLBIOL kit was not designed to detect M. genitalium. However, the use of these kits is intended after the previous use of a macrolide resistance detection kit; thus, the M. genitalium positivity of the specimen would have already been demonstrated. In addition, this study did not intend to evaluate the prevalence of fluoroquinolone resistance among the specimens studied; it was only designed to study the clinical performance of the three commercial kits. It should be noted that among the 368 M. genitalium-positive specimens collected, the percentage of mutations that were likely associated with moxifloxacin resistance (20.1%, 74/368) and the highest frequency of the Ser83Ile mutation (82.4%) followed by the Asp87Asn (8.1%) and Asp87Tyr (4.1%) substitutions is in accordance with proportions reported in patients attending STI centers (6, 8). Thus, the studied sample group was representative of specimens on which these kits were intended to be applied in routine.

In conclusion, the Allplex MG & MoxiR and MGMO qPCR kits showed good performance for the detection of parC mutations that were likely associated with moxifloxacin resistance, whereas the LightMix Modular parC kit showed a significantly lower specificity, with the analysis of melting temperatures potentially prone to errors. In the context of high macrolide resistance prevalence and the increased reliance on moxifloxacin worldwide, the use of such assays will be useful to optimize the choice of antimicrobials for the treatment of M. genitalium infections. However, further developments may be necessary because parC mutations involved in fluoroquinolone resistance remain to be precisely determined.