Molecular Assay for Detection of Genetic Markers Associated with Decreased Susceptibility to Cephalosporins in Neisseria gonorrhoeae

Abstract

The incidence of antimicrobial-resistant Neisseria gonorrhoeae continues to rise in Canada; however, antimicrobial resistance data are lacking for approximately 70% of gonorrhea infections that are diagnosed directly from clinical specimens by nucleic acid amplification tests (NAATs). We developed a molecular assay for surveillance use to detect mutations in genes associated with decreased susceptibility to cephalosporins that can be applied to both culture isolates and clinical samples. Real-time PCR assays were developed to detect single nucleotide polymorphisms (SNPs) in ponA, mtrR, penA, porB, and one N. gonorrhoeae-specific marker (porA). We tested the real-time PCR assay with 252 gonococcal isolates, 50 nongonococcal isolates, 24 N. gonorrhoeae-negative NAAT specimens, and 34 N. gonorrhoeae-positive NAAT specimens. Twenty-four of the N. gonorrhoeae-positive NAAT specimens had matched culture isolates. Assay results were confirmed by comparison with whole-genome sequencing data. For 252 N. gonorrhoeae strains, the agreement between the DNA sequence and real-time PCR was 100% for porA, ponA, and penA, 99.6% for mtrR, and 95.2% for porB. The presence of ≥2 SNPs correlated with decreased susceptibility to ceftriaxone (sensitivities of >98%) and cefixime (sensitivities of >96%). Of 24 NAAT specimens with matched cultures, the agreement between the DNA sequence and real-time PCR was 100% for porB, 95.8% for ponA and mtrR, and 91.7% for penA. We demonstrated the utility of a real-time PCR assay for sensitive detection of known markers for the decreased susceptibility to cephalosporins in N. gonorrhoeae. Preliminary results with clinical NAAT specimens were also promising, as they correlated well with bacterial culture results.

INTRODUCTION

Neisseria gonorrhoeae, the causative agent of gonorrhea infection, has the second highest reported rate of bacterial sexually transmitted infections in Canada, with >12,000 reported cases in 2012 (36.18 cases per 100,000 population) (1). According to the latest World Health Organization (WHO) reports, worldwide gonococcal infections amount to 106 million cases per year (2). N. gonorrhoeae has acquired resistance to all of the antibiotics commonly used for treatment, including penicillin, tetracycline, spectinomycin, azithromycin, and ciprofloxacin, and reduced susceptibility to the third-generation cephalosporins has been reported (3). In recent years, the MICs to cefixime and ceftriaxone have been increasing, and there have been reports of cephalosporin treatment failures in Canada and around the world (4–8).

Canada has conducted antimicrobial susceptibility testing on N. gonorrhoeae cultures since the mid-1980s to monitor antimicrobial resistance trends and develop an understanding of the molecular subtypes circulating in the population. However, starting in the early 2000s, an increasing number of gonococcal infections have been diagnosed by nucleic acid amplification tests (NAATs) and a decreasing number of laboratories across Canada are culturing N. gonorrhoeae. This is of concern since N. gonorrhoeae cultures are required for antimicrobial susceptibility testing. In fact, >70% of gonococcal infections in Canada are now detected using NAATs, and hence antimicrobial susceptibility data are not available for these isolates (Public Health Agency of Canada, National Microbiology Laboratory, 2014, unpublished data).

Numerous molecular mechanisms for decreased susceptibility (DS) to cephalosporins have been described in N. gonorrhoeae. Two classes of alterations of penA, which encodes penicillin-binding protein 2 (PBP2), have been described: the first is the penA mosaic allele, which contains segments of penA from nongonococcal neisserial species; the second is an alteration in the amino acids (A501, G542, P551) of PBP2 in nonmosaic penA alleles (3, 9, 10). Mutations in the promoter of the repressor gene mtrR, which cause overexpression of the MtrCDE efflux pump system, have been associated with cephalosporin DS (3, 11). Finally, porB1b gene mutations that alter amino acids G120 and A121 in the outer membrane PorB1b porin result in reduced permeability and thus further cephalosporin DS (3, 11). In addition, mutations in PBP1 (ponA) have been observed in N. gonorrhoeae strains with elevated cephalosporin MICs, although this mutation has not been shown to cause resistance in transformation experiments (11).

In this study, real-time PCR assays were developed to detect single nucleotide polymorphisms (SNPs) in genes associated with DS to extended-spectrum cephalosporins (ESCs) for the purpose of providing surveillance data. Four targets associated with cephalosporin DS (ponA, mtrR, penA, and porB) and one N. gonorrhoeae-specific gene (porA) as an internal positive control were selected and evaluated using N. gonorrhoeae cultures. As proof of principle, all targets were also evaluated using clinical specimens tested by the Aptima CT/NG (Chlamydia trachomatis/Neisseria gonorrhoeae) assay on the Tigris platform (Hologic, Bedford, MA) that also had a matched culture isolate.

MATERIALS AND METHODS

Bacterial isolates and clinical specimens.

Canadian provincial laboratories submit isolates to the National Microbiology Laboratory if they identify resistance to at least one antibiotic or if they do not conduct antimicrobial susceptibility testing (12). From this collection, 241 N. gonorrhoeae strains isolated across Canada between 2001 and 2014 were selected for development of the assay along with N. gonorrhoeae control strains ATCC 49226, F62, FA19, WHO F, WHO G, WHO K, WHO L, WHO M, WHO N, WHO O, and WHO P (13). Isolates were primarily selected to represent a range of cephalosporin MICs, including DS to ceftriaxone (n = 55) and cefixime (n = 32). In addition, 193 isolates were included that were susceptible to both ceftriaxone and cefixime and represented a diverse group of Neisseria gonorrhoeae multiantigen sequence typing (NG-MAST) types and temporal and geographic distribution. Isolates were cultured (from storage at −80°C in brain heart infusion [BHI] containing 20% glycerol) on GC medium base (Difco Laboratories, Detroit, MI) containing 0.2% BioX and incubated for 18 to 24 h at 35°C in a 5% CO₂ atmosphere.

The SNP genotyping assay was then tested with a total of 58 clinical Hologic Aptima CT/NG NAAT specimens. Ten Aptima specimens, consisting of 5 urethral swabs and 5 urine samples, were obtained from Cadham Provincial Laboratory (Winnipeg, MB, Canada). Forty-eight Aptima specimens were obtained from the British Columbia Centers for Disease Control and consisted of 24 N. gonorrhoeae-positive specimens and their corresponding cultured isolates and 24 N. gonorrhoeae-negative control specimens selected in consecutive order of receipt.

Fifty different nongonococcal strains were chosen to assay for cross-reactivity based on similarity to N. gonorrhoeae sequences or the likelihood of their presence in urine or urogenital specimens (Table 1) (14).

TABLE 1.

Cross-reactivity of SNP assays in nongonococcal strains

Species	Strain	Targets showing cross-reactivitya
		ponA	mtrR	porB	penA
Atopobium vaginae	ATCC BAA-55
Bacteroides ureolyticus	ATCC 33387
Candida albicans	ATCC 18804
Corynebacterium glucuronolyticum	ATCC 51860
Corynebacterium urealyticum	ATCC 43043
Corynebacterium xerosis	ATCC 373
Cryptococcus neoformans	ATCC 2517
Enterobacter aerogenes	ATCC 13048
Enterococcus faecalis	ATCC 29212
Enterococcus faecium	ATCC 19434
Escherichia coli	ATCC 35218
Gardnerella vaginalis	ATCC 14018
Klebsiella oxytoca	ATCC 13182
Lactobacillus crispatus	ATCC 33197
Lactobacillus gasseri	ATCC 33323
Lactobacillus iners	ATCC 55195
Lactobacillus jensenii	ATCC 25258
Leptotrichia buccalis	ATCC 14201
Listeria monocytogenes	ATCC 15313
Mobiluncus curtisii	ATCC 35242
Moraxella catarrhalis	ATCC 25238		X
Neisseria animalis	ATCC 49930
Neisseria animaloris	Clinical
Neisseria cinerea	ATCC 14685
Neisseria elongata	ATCC 25295		X
Neisseria flavescens	ATCC 13120				X
Neisseria lactamica	ATCC 23970
Neisseria meningitidis	ATCC 13102	X			X
Neisseria mucosa	Clinical
Neisseria perflava	ATCC 9913
Neisseria polysaccharea	ATCC 43768	X
Neisseria sicca	ATCC 29256
Neisseria subflava	Clinical				X
Neisseria wadsworthii	Clinical
Neisseria weaveri	Clinical
Peptococcus niger	ATCC 27731
Peptostreptococcus anaerobius	ATCC 27337
Prevotella bivia	ATCC 29303
Proteus mirabilis	ATCC 7002
Pseudomonas aeruginosa	ATCC 27853				X
Salmonella typhimurium	ATCC 39183
Staphylococcus aureus	ATCC 29213				X
Staphylococcus epidermidis	ATCC 14990		X
Streptococcus agalactiae	ATCC 12386
Streptococcus gordonii	ATCC 10558
Streptococcus infantis	ATCC 700779
Streptococcus oralis	ATCC 35037
Streptococcus pyogenes	ATCC 19615
Ureaplasma parvum	ATCC 27815
Ureaplasma urealyticum	ATCC 27618

^aIsolates were tested in triplicate.

Antimicrobial susceptibility testing.

MICs were determined using the agar dilution method as previously described (12). Interpretation of the cephalosporin MICs was based on the criteria of the WHO: cefixime DS MIC of ≥0.25 μg/ml and ceftriaxone DS MIC of ≥0.125 μg/ml (2). Since there have been recent reports of cephalosporin treatment failures for infections caused by isolates with MICs as low as 0.032 μg/ml, we also used MIC cutoffs of 0.032 μg/ml and 0.063 μg/ml for sensitivity and specificity calculations (6).

Real-time PCR assay for SNP genotyping.

DNA was extracted from Aptima NAAT specimens using the QIAamp viral RNA minikit according to the manufacturer's instructions (Qiagen, Toronto, ON). Five gene targets were chosen, including 4 associated with cephalosporin resistance, ponA (L421P), mtrR (−35delA), porB (G120/A121), and penA (mosaic), along with the N. gonorrhoeae-specific porA pseudogene as a positive control. SNP targets were selected based on circulating isolates in Canada and previously reported resistance mechanisms. Oligonucleotide primers and probes were chosen for each target region using Primer Express Software version 3.0 (Life Technologies). Gene sequences for porA (GenBank accession no. HE681885.2), ponA (AB727713.1), mtrR (Z25796.1), and porB (M21289.1) were acquired from NCBI. Sequences of penA representing a variety of mosaic and nonmosaic penA types were aligned using Lasergene MegAlign version 11.2.1 (DNAStar, Madison, WI); primers and probes were chosen to detect both the mosaic and nonmosaic sequences. The ponA and mtrR assays contained probes to detect either the wild-type (WT) or the SNP alleles, while the porB assay contained a probe to detect the WT allele, along with an internal positive-control probe that was detected in all isolates (Table 2).

TABLE 2.

SNP assay primers and probes

Targeta	Primer/probeb	Sequence	Product length (bp)	Probe	Sequencec	SNP detected
porA	F	GTATTTTCAAACGCCACGACG	207	WT	CAGCATTCAATTTGTTC	Indicates isolate is N. gonorrhoeae
	R	GACCGGCATAATACACATCCG
ponA	F	GCGGTCGATAATGAGAAAATGG	131	WT	AGCCGTTGCTGCAGG	WT at L421
	R	ATCCAGCGAAACCAAAGCC		SNP	AGCCGTTGCCGCAGG	L421P
mtrR	F	TCGAACGGGTTGCAAAGC	141	WT	TGCACGGATAAAAAGT	WT at −35 bp
	R	TCGTTTCGGGTCGGTTTG		SNP	GCACGGATAAAAGTC	−35delA
porB	F	GCTTGAAGGGCGGCTTC	154	WT	TTGGC(G/A)CCGGTGTT	WT at G120, A121; if WT probes are absent, implies at least one of G120K, G120R, G120D, A121D, A121G, A121N, A121S
	R	GACAGGTAGCGGTGTTCCC		Control	TGCCGGATTCCCAAGC	Internal positive control detected in all isolates
penA	F1	GGCAATCAAACCGTTCGTG	150	WT1	CCGTGCGCGATAC	Nonmosaic penA allele 1
	F2	TGGATTCCGGCAAAGTGG	118	WT2	TGCGCGACGATAC	Nonmosaic penA allele 2
	R	ATAATGCCGCGCACATCC		Mosaic	ACCGTACAAGATACCCA	Mosaic penA allele

^aThe porA assay contains only one probe. The ponA and mtrR assays contain one FAM (WT)- and one VIC (SNP)-labeled probe. The porB assay contains two FAM (WT)-labeled probes to account for a silent mutation in the WT allele, along with an internal positive control probe. The penA assay contains three probes, labeled with FAM, VIC, and NED in a single reaction.

^bF, forward; R, reverse.

^cBold underlined bases denote SNP locations.

Real-time PCR was performed in a reaction volume of 25 μl, consisting of 12.5 μl of 2× TaqMan genotyping master mix (Life Technologies), 900 nM each primer (final concentration), 250 nM each probe, 5 μl of 1 ng/μl template DNA, and Ambion nuclease-free H₂O (Life Technologies). PCR amplification and detection of amplification products were performed on a ViiA 7 instrument (Life Technologies). Thermal cycling conditions were as follows: initial preheating at 60°C for 30 s, denaturation at 95°C for 10 min; 45 cycles (40 cycles for N. gonorrhoeae culture isolates) of 95°C for 15 s and 60°C for 1 min, and a final elongation step of 60°C for 30 s. Each real-time PCR was performed in triplicate. Results were considered to be positive if they had a quantification cycle (C_q) value of <40 (15) and a relative fluorescence of probe minus the baseline (ΔRn) value of >0.5 for porA, ponA, mtrR, and penA and of >0.7 for porB.

SNPs detected by the assay were validated by comparison with aligned gene sequences obtained through whole-genome sequencing (WGS) (16).

To determine the limits of detection (LODs) of the real-time PCR assay, DNA from N. gonorrhoeae control strains F62, WHO K, and WHO L was extracted using the QIAamp viral RNA minikit and quantified using a Qubit fluorometer (Life Technologies). Ten-fold serial dilutions were performed (0.1 fg/μl to 1 ng/μl). The LOD was determined for each real-time SNP assay by testing each isolate in duplicate and recording the lowest dilution that produced a positive result as defined above.

Calculation of sensitivity and specificity.

Sensitivity measures the percentage of isolates with DS containing the SNP of interest, while specificity represents the percentage of susceptible isolates containing a WT allele. Since treatment failures have been observed with a cefixime MIC of 0.032 μg/ml (6), sensitivities and specificities were calculated with cutoffs of 0.032 μg/ml, 0.063 μg/ml, or 0.125 μg/ml for each antibiotic. To calculate sensitivity and specificity, isolates were characterized into four categories: true positive (TP), having a high MIC and SNP detected by real-time PCR; false-positive (FP), having a low MIC and SNP detected by real-time PCR; false-negative (FN), having a high MIC and WT result; and true negative (TN), having a low MIC and WT result. Calculations were performed as follows: sensitivity = TP/(FN + TP) · 100; specificity = TN/(FP + TN) · 100 (17).

NG-MAST.

Isolates were characterized by NG-MAST based on the sequence of the porB and tbpB genes as previously described (18). Aptima specimen NG-MAST types were confirmed using WGS data (16), when available. The sequence type (ST) was determined using the NG-MAST website (www.ng-mast.net).

RESULTS

Bacterial isolates and specimens tested.

Overall, 241 clinical N. gonorrhoeae isolates from 2001 to 2013, along with 11 N. gonorrhoeae reference isolates, representing 77 different STs were tested with our assay. Of the isolates, 103 (40.9%) had ceftriaxone MICs of ≤0.016 μg/ml, 94 (37.3%) had MICs of 0.032 to 0.063 μg/ml, and 55 (21.8%) had DS (MICs of ≥0.125 μg/ml). Cefixime susceptibilities were as follows: 108 isolates (42.9%) had MICs of ≤0.016 μg/ml, 112 isolates (44.4%) had MICs of 0.032 to 0.125 μg/ml, and 32 isolates (12.7%) had DS (MICs of ≥0.25 μg/ml).

The 10 N. gonorrhoeae-positive specimens with no bacterial cultures were obtained from 3 males and 7 females and included 5 urine samples and 5 swabs (3 cervical, 1 vaginal, and 1 penis/urethral). The 24 Aptima specimens with matched bacterial cultures and 24 N. gonorrhoeae-negative Aptima controls were obtained from 41 males and 7 females (Table 3). The N. gonorrhoeae-negative NAAT specimens were obtained from the penis/urethra (n = 1), urine (n = 13), rectum (n = 4), throat (n = 4), and cervix/vagina (n = 2). Of the 24 matched culture isolates, 16 (66.7%) isolates had ceftriaxone MICs of ≤0.016 μg/ml, 7 isolates (29.2%) had MICs of 0.032 to 0.063 μg/ml, and 1 isolate (4.2%) had DS (MIC of ≥0.125 μg/ml), while 15 isolates (62.5%) had cefixime MICs of ≤0.016 μg/ml, 8 isolates (33.3%) had MICs of 0.032 to 0.125 μg/ml, and 1 isolate (4.2%) had DS (MIC of ≥0.25 μg/ml).

TABLE 3.

SNP assay and NG-MAST results from Aptima NAAT specimens compared with MICs from matched culture isolates^a

Sample no.	Source	Matched isolate MICb		SNP assay results from Aptima NAAT specimens				NG-MAST
		Ceftriaxone	Cefixime	ponA	mtrR	porB	penA
37200A	Penis/urethra	0.25	0.25	SNP	SNP	SNP	Mosaic	ST-3158
37201A	Rectum	0.004	0.008	WT	WT	SNP	WT	ST-5985
37202A	Rectum	0.032	0.063	UNDc	UND	SNP	UND	UND
37203A	Rectum	0.032	0.016	SNP	SNP	SNP	WT	ST-9665
37204A	Throat	0.008	0.008	WT	WT	SNP	WT	ST-5985
37205A	Urine	0.008	0.008	WT	WT	SNP	WT	ST-5985
37206A	Urine	0.008	0.016	WT	WT	SNP	WT	ST-5985
37207A	Rectum	0.016	0.016	WT	WT	SNP	WT	ST-5985
37208A	Throat	0.016	0.016	SNP	SNP	SNP	UND	UND
37209A	Rectum	0.032	0.032	SNP	SNP	SNP	WT	ST-8502
37210A	Throat	0.032	0.016	SNP	SNP	SNP	WT	ST-9665
37211A	Urine	0.032	0.016	SNP	SNP	SNP	WT	ST-9665
37212A	Vagina	0.004	0.008	WT	WT	SNP	WT	ST-10169
37213A	Cervix	0.016	0.032	WT	WT	SNP	WT	ST-8890
37214A	Urine	0.008	0.032	SNP	WT	WT	WT	ST-5444
37215A	Urine	0.032	0.063	WT	WT	SNP	WT	ST-6339
37216A	Urine	0.016	0.032	WT	WT	SNP	WT	ST-8890
37217A	Urine	0.008	0.032	WT	WT	WT	WT	ST-2992
37218A	Urine	0.008	0.016	WT	WT	SNP	WT	ST-8890
37219A	Urine	0.002	0.004	WT	WT	WT	WT	ST-10170
37220A	Urine	0.063	0.063	SNP	SNP	SNP	WT	ST-8502
37221A	Urine	0.008	0.008	WT	WT	SNP	WT	ST-8890
37222A	Urine	0.008	0.016	WT	WT	WT	WT	ST-10171
37223A	Urine	0.008	0.016	WT	WT	WT	WT	ST-4643

^aMICs indicating reduced susceptibility are shown in bold.

^bMICs were calculated from matched N. gonorrhoeae culture isolates.

^cUND, undetermined result.

Genetic marker performance.

All of the 252 N. gonorrhoeae isolates tested positive for the N. gonorrhoeae-specific porA pseudogene. The assay concordance (percentage of isolates called correctly as WT or SNP) with sequencing for the 252 N. gonorrhoeae isolates was 100% for ponA (78 WT and 174 SNP) and penA (160 WT and 92 mosaic). For the mtrR gene, the assay correctly predicted 99.6% of the isolates (90/91 WT and 161/161 SNPs). One isolate gave a false-negative result for mtrR instead of a WT result due to the replacement of an A in the 5-A homopolymer at −35 with a C, preventing the probe from binding. For the porB gene, 95.2% of the isolates (69/70 WT and 171/182 SNPs) were correctly identified. Twelve isolates gave false-negative results for porB due to the presence of the porB1a allele rather than the porB1b allele.

The SNP assays were tested with 50 non-N. gonorrhoeae control species and 24 N. gonorrhoeae-negative NAAT specimens. The N. gonorrhoeae-specific porA pseudogene was negative for all non-N. gonorrhoeae species. Cross-reacting species for each SNP assay are listed in Table 1. For the negative Aptima samples, one porB and two penA assays showed cross-reactivity. These three cross-reacting negative Aptima specimens were all throat swabs.

The LODs were 50 fg/reaction for porA, ponA, mtrR, and porB, and 500 fg/reaction for penA. Sensitivities and specificities were measured for each SNP assay (Table 4) and for combinations of SNPs (Table 5).

TABLE 4.

Sensitivities and specificities of the SNP real-time PCR assays when interpreted individually following testing of N. gonorrhoeae culture isolates (n = 252)^a

SNP assay	Results for ceftriaxone at an MIC of:						Results for cefixime at an MIC of:
	≥0.032 μg/ml (n = 149)		≥0.063 μg/ml (n = 116)		≥0.125 μg/ml (n = 54)		≥0.032 μg/ml (n = 143)		≥0.063 μg/ml (n = 112)		≥0.125 μg/ml (n = 90)
	% sensitivity (TPb)	% specificity (TNc)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)
ponA	98 (146)	72.8 (75)	100 (116)	57.4 (78)	100 (54)	39.4 (78)	95.8 (138)	66.7 (72)	99.1 (111)	55 (77)	98.9 (89)	47.5 (77)
mtrR	94.6 (141)	80.4 (82)	98.3 (114)	65.2 (88)	96.3 (52)	44.7 (88)	92.4 (133)	73.8 (79)	98.2 (110)	62.6 (87)	97.8 (88)	54.7 (88)
porB	93.3 (139)	64.8 (59)	94.8 (110)	50.8 (63)	96.3 (52)	36.6 (68)	90.9 (130)	57.7 (56)	92.9 (104)	47.7 (61)	93.3 (84)	42 (63)
penA Mosaic	61.1 (91)	99 (102)	69 (80)	91.2 (124)	75.9 (41)	74.2 (147)	63.2 (91)	99.1 (107)	78.6 (88)	97.1 (136)	91.1 (82)	93.8 (152)

^aSensitivity = TP/(FN + TP) · 100; specificity = TN/(FP + TN) · 100.

^bTP, true positive (isolates contain the SNP and have high MICs).

^cTN, true negative (isolates do not contain the SNP and have low MICs).

TABLE 5.

Sensitivities and specificities of the SNP real-time PCR assays when interpreted in combination following testing of N. gonorrhoeae culture isolates (n = 252)^a

No. of SNPsb	Results for ceftriaxone at an MIC of:						Results for cefixime at an MIC of:
	≥0.032 μg/ml (n = 149)		≥0.063 μg/ml (n = 116)		≥0.125 μg/ml (n = 54)		≥0.032 μg/ml (n = 143)		≥0.063 μg/ml (n = 112)		≥0.125 μg/ml (n = 90)
	% sensitivity (TPc)	% specificity (TNd)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)	% sensitivity (TP)	% specificity (TN)
4 SNPs	56.4 (84)	98.8 (89)	63.8 (74)	91.1 (112)	72.7 (40)	75.5 (139)	58.7 (84)	99 (95)	72.9 (81)	96.9 (123)	84.4 (76)	94 (140)
≥3 SNPs	93.3 (139)	82.2 (74)	98.3 (114)	66.7 (82)	98.2 (54)	45.1 (83)	91.6 (131)	75 (72)	97.3 (109)	63.8 (81)	98.9 (89)	55.7 (83)
≥2 SNPs	98.7 (147)	77.8 (70)	100 (116)	58.5 (72)	100 (55)	39.1 (72)	96.5 (138)	69.8 (67)	99.1 (111)	55.9 (71)	98.9 (89)	47.7 (71)
≥1 SNP	98.7 (147)	54.4 (49)	100 (116)	41.5 (51)	100 (55)	27.7 (51)	97.2 (139)	49 (47)	99.1 (111)	39.4 (50)	98.9 (89)	33.6 (50)

^aSensitivity = TP/(FN + TP) · 100; specificity = TN/(FP + TN) · 100.

^bMinimum number of SNPs present among ponA, mtrR, porB, and penA.

^cTP, true positive (isolates contain the SNP and have high MICs).

^dTN, true negative (isolates do not contain the SNP and have low MICs).

Aptima NAAT specimens.

Of the 24 N. gonorrhoeae-positive Aptima specimens with matched cultures, assay concordance was 100% for porB, 95.8% for ponA and mtrR, and 91.7% for penA compared to the WGS results (Table 3). One rectal swab specimen gave indeterminate results for ponA (neither SNP nor WT was positive), mtrR (neither SNP nor WT was positive), and penA (both WT2 and mosaic probes produced a positive result). One pharyngeal specimen also had an indeterminate penA result due to positive results from both the WT and the mosaic probes. One hundred percent identity was observed between NG-MAST sequences obtained from matched Aptima and culture specimens for specimens that could be typed (Table 3).

DISCUSSION

As NAATs have become the primary method of laboratory diagnosis of gonorrhea, fewer cultures are isolated from patients, providing less information about antimicrobial susceptibilities (19). In this study, we developed real-time PCR assays that permitted us to test N. gonorrhoeae cultures and Aptima specimens for SNPs associated with DS to ESCs in gonococcal infections. While this assay cannot replace culture-based MIC determination, it can aid surveillance by providing insight into the prevalence of genes associated with DS to ESCs in N. gonorrhoeae NAAT specimens for which no culture is available.

We applied our test to 252 N. gonorrhoeae isolates along with 50 non-N. gonorrhoeae isolates. The porA assay gave positive results for all N. gonorrhoeae specimens and isolates and negative results for all non-N. gonorrhoeae specimens and isolates, indicating that this is an appropriate N. gonorrhoeae-specific genetic marker. The assays were effective in identifying SNPs in both culture and NAAT specimens. The LOD of each real-time PCR assay was approximately 50 fg of DNA, corresponding to approximately 25 N. gonorrhoeae genomes (20). The low LOD and high sequence identity found between Aptima specimens and their matched culture isolates provide proof of principle that direct molecular characterization can be performed on Aptima specimens. In this assay, two of the N. gonorrhoeae-negative pharyngeal specimens and one N. gonorrhoeae-positive pharyngeal specimen showed cross-reaction with the penA mosaic probe, and one sample showed cross-reaction with the porB probe. Although this is a possible limitation in the use of DNA from a NAAT specimen, it should be noted that the non-N. gonorrhoeae organisms that gave positive results are not normally found in urine but can be commonly found in the respiratory tract. Balashov et al. (14) also observed that molecular assays may not be applicable for extragenital specimens due to the prevalence of non-N. gonorrhoeae species at extragenital sites.

In a study by Allen et al. (6), isolates with cefixime MICs of ≥0.125 μg/ml had 25% treatment failures, while failures were 1.9% for MICs of ≤0.125 μg/ml. One treatment failure isolate had an MIC of 0.032 μg/ml. For this reason, we determined the sensitivity and specificity values using MIC cutoffs of 0.032 μg/ml, 0.063 μg/ml, and 0.125 μg/ml. The ponA, mtrR, and porB markers all had high sensitivities for both cefixime and ceftriaxone at the three MICs selected, indicating that organisms with elevated MICs were more likely to contain these SNPs. Specificities for ponA and porB were lower than those for mtrR for both antibiotics. A low specificity for ponA is not unexpected, as, although ponA L421P is often found in organisms with elevated ESC MICs, particularly those that also contain the penA mosaic allele, the L421P variant itself did not produce elevated ESC MICs in transformation experiments (11, 21). This assay tested two adjacent SNPs (PorB 120/121), giving a WT result if there was a WT sequence in both positions or a SNP result if there was variation in at least one position. It is possible that only a specific amino acid change contributes to increased MICs to ESCs at this locus (22–24). In addition, previous studies have shown that porB mutations do not affect ESC susceptibility in the absence of the mtrR −35del SNP (11, 25). When porB mutations are calculated only in the presence of the mtrR −35del SNP, sensitivity decreases by ∼2% and specificity increases by ∼4 to 6%. As expected, mtrR had the highest specificity of the ponA, mtrR, and porB markers, as a mutation in the mtrR promoter sequence of the regulator of the MtrCDE efflux pump has been shown to contribute to an increase in ESC MICs in transformation experiments (11).

The penA assay had a high specificity for ceftriaxone at an MIC of ≤0.063 μg/ml, and at all three cefixime MIC breakpoints tested, implying that the mosaic allele is absent in the majority of low MIC isolates. The presence of the mosaic allele indicates the DS of an isolate. In fact, 98.9% of isolates with the mosaic penA allele exhibited ceftriaxone and cefixime MICs of ≥0.032 μg/ml, with only one isolate exhibiting an MIC of ≤0.032 μg/ml for both antibiotics.

There are many limitations to using molecular detection techniques to predict antimicrobial resistance, especially in an organism such as N. gonorrhoeae. The mechanisms causing resistance are complex and multifactorial (25–27). N. gonorrhoeae is highly recombinogenic and naturally competent, allowing for transformation of DNA from other commensal species (28). Target sequences may cross-react with other similar species, thus causing decreased assay specificities. False-negative results could occur if Aptima specimens do not contain enough N. gonorrhoeae DNA template for SNP detection. In addition, a porA mutant N. gonorrhoeae strain that may result in a false-negative result due to sequence variations was recently discovered (29, 30). Care was taken to limit the selection bias of N. gonorrhoeae isolates for the validation of the assay by choosing isolates with a range of cephalosporin MICs, NG-MAST types, and temporal and geographic distributions; however, the Aptima specimens are not representative of the population, as they were collected from one region within a limited time frame.

The results of this study highlight the utility of a molecular method of surveillance of antimicrobial susceptibilities in the absence of N. gonorrhoeae culture isolates. Upon testing this assay with a wide range of gonococcal isolates, representing a variety of STs and ESC MICs, we found good agreement between the results generated from the SNP assay and the sequence data. In addition, we found porA to be a suitable N. gonorrhoeae-specific marker. Through detection of resistance determinants on the molecular level, this assay could be used to evaluate trends in the presence of molecular markers associated with ESC DS using the full range of specimens, both NAAT and culture.