Multidrug-resistant Neisseria gonorrhoeae infections have been declared 1 of the top 3 urgent threats to public health.1 N. gonorrhoeae has developed resistance to all antimicrobials currently available,2,3 and there are only a few novel antibiotics in development4–6; therefore, new approaches are urgently needed. A recently described approach calls for targeted therapy with antibiotics previously thought to be ineffective,7,8 which has been made possible by the development of rapid molecular assays that predict in vitro antimicrobial susceptibility.9
In the United States, it is estimated that approximately 80% of N. gonorrhoeae infections are susceptible to ciprofloxacin.10 Previous work has shown that mutation of the gyrase A (gyrA) gene of Neisseria species is associated with ciprofloxacin resistance, particularly mutation at the Ser91 codon9 [30s, http://links.lww.com/OLQ/A163]. Mutations in other loci have been shown to confer resistance to ciprofloxacin, such as parC, however prior studies have demonstrated that mutations in other locations, which confer resistance, usually occur in parallel with the gyrA mutation; prediction of ciprofloxacin susceptibility, therefore, may be done solely by detection of mutation in the gyrA gene [30s–34s, http://links.lww.com/OLQ/A163]. In response, real-time polymerase chain reaction (PCR)–based molecular assays have been developed to predict gonococcal susceptibility to ciprofloxacin [33s, http://links.lww.com/OLQ/A163].
Those rapid molecular genotypic assays have been studied extensively to determine how they correlate with standard agar dilution susceptibility methods [32s–43s, http://links.lww.com/OLQ/A163]. Here we review the previously published reports describing the sensitivity, specificity, negative predictive values (NPV) and positive predictive values (PPVs) of the gyrA genotypic assays for the prediction of N. gonorrhoeae susceptibility to ciprofloxacin.
In September 2016, we conducted a literature search using the following text-word terms: “GYRA” OR “CIPROFLOXACIN” OR “QUINOLONE” AND “GONORRH*” with PubMed as a search platform. Our search resulted in over 900 studies which was further restricted to 831 in the English language. Our only inclusion criterion was the use of gyrA genotypic determination techniques and the comparison of genotype results with conventional methods for determining ciprofloxacin susceptibility. We excluded articles that only evaluated resistant or susceptible strains of N. gonorrhoeae because there would be no comparison of susceptibility results. Similarly, we excluded articles reporting five or fewer isolates with resistant or susceptible phenotypes. Wild type is defined as the most prevalent sequence of the gyrase A gene, which predicts full ciprofloxacin susceptibility. One article was excluded as a cluster analysis was performed and the results provided were insufficient to distinguish which mutant or wild-type strains were associated with which susceptibility results. In total, we identified 30 articles meeting inclusion criteria. We were also aware of 1 article that was not available on PubMed, which fulfilled inclusion criteria. Here, we report sensitivity, specificity, PPV and NPV of the wild-type gyrA gene for predicting susceptibility to ciprofloxacin from those 31 reports.
Sensitivity, specificity and their corresponding 95% confidence intervals (CI) of individual studies were calculated and presented in forest plots (Figs. 1 and 2). For studies with zero event cell count, continuity correction of 0.5 was added to each cell to enable calculations. Heterogeneities of sensitivities and specificities among included studies were assessed by Cochran Q and χ2 test,11 respectively. Given the strong evidence of heterogeneities (P values of χ2 test <0.05) and known correlation between the pair of sensitivity and specificity, we obtained the summary statistics of sensitivity and specificity by using a bivariate random effects modeling approach.12 The summary operating characteristic curves (sROC)13 were constructed and compared between studies using real-time PCR techniques and non–real-time PCR techniques (Fig. 3). All the statistical analyses were conducted using Package MADA (Meta-Analysis of Diagnostic Accuracy) in the R statistical software environment version 184.108.40.206
Of the 31 articles meeting inclusion criteria, 7 different loci for mutations in the gyrA gene were identified from samples collected in 16 countries between the years of 1996 and 2016 [30s–60s, http://links.lww.com/OLQ/A163]. Eleven studies used real-time polymerase chain reaction techniques for determination of the gyrA genotype, whereas 20 were conducted using other methods, such as pyrosequencing, probe hybridization, isothermic chimeric primer-initiated amplification, or gel electrophoresis to determine the gyrA genotype.
All studies compared genotype results with minimum inhibitory concentration (MIC) assay results using agar dilution techniques, except 1 study, which used available phenotypic susceptibility data without MIC assay results [40s, http://links.lww.com/OLQ/A163]. The cutoff MIC for reduced susceptibility and resistance varied slightly among studies with cutoff values ranging from ≥ 0.06 to 0.125 μg/mL and ≥ 0.05 to 1.0 μg/mL, respectively.
Studies Using Non–Real-Time Polymerase Chain Reaction Methods for Determination of GyrA Genotype
We identified 20 studies (N = 2931 isolates) from 7 countries using non–real-time PCR techniques for the determination of the gyrA genotype. The range of sensitivities and specificities of wild-type gyrA genotype for predicting ciprofloxacin susceptibility reported by those studies was 58.0% to 100% (mean, 91.2%) and 90.5% to 100% (mean, 99.2%), respectively (Figs. 1A and B). Positive predictive values ranged from 76.3% to 100% (mean, 97.9%), whereas NPVs ranged from 72.4% to 100% (mean, 95.4%). The pooled estimate of sensitivity and specificity based on the bivariate mixed-effect modeling approach were 91.7% (95% CI, 85.4–95.5%) and 98.0% (95% CI, 96.7–98.8%), respectively.
Shigemura et al. used denaturing high-performance liquid chromatography and DNA sequencing in 2002 among samples collected in Japan and found 83.3% sensitivity of gyrA mutation for predicting reduced ciprofloxacin susceptibility; however, results improved to 100% when only evaluating mutations at codon 91 of the gyrA gene. Similarly, Dewi et al. found 80.6% sensitivity, which improved to 93.5%; and Tanaka et al. found 75.0% sensitivity, which improved to 100% when analyzing only mutations in codon 91.
Gharizadeh et al. found 10 isolates with mutations in the gyrA gene (2 in region Asp95 and 8 in region Ser91) that showed phenotypic susceptibility to ciprofloxacin; however, the median MIC was higher than for wild-type strains (0.21 and 0.03 μg/mL, respectively). Conversely, Grad et al. report 5 isolates with phenotypic resistance to ciprofloxacin (MIC values > 1 μg/mL) that had no identifiable mutation in gyrA or parC.
Fourteen of the 20 studies included sequence analysis of parC as well as gyrA. Uthman et al. found slightly higher sensitivity of wild-type parC genotype for predicting ciprofloxacin susceptibility compared to wild-type gyrA genotype (100% compared with 97.8%), with similar results reported by Shigemura et al. and Grad et al. Eleven other studies, however, found wild-type parC was not as sensitive or specific for predicting ciprofloxacin susceptibility as wild-type gyrA [30s, 31s, 44s, 45s, 46s, 47s, 51s, 53s, 54s, 56s, 57s, http://links.lww.com/OLQ/A163].
Studies Using Real-Time Polymerase Chain Reaction Assays to Amplify Ser91 Region of the GyrA Gene
In total, 11 studies (N = 4777 isolates) used RT-PCR techniques to amplify the Ser91 region of the gyrA gene. Samples were collected from ten countries between 2007 and 2016. All studies reported high sensitivity and specificity with ranges between 93.8–100% (mean, 98.8%) and 93.2–100% (mean, 99.3%), respectively (Figs. 2A and B). Positive predictive values and NPVs ranged from 94.4% to 100% (mean, 99.4%) and 87.5% to 100% (mean, 98.2%), respectively. The pooled estimate of sensitivity and specificity based on the bivariate mixed-effect modeling approach were 98.2% (95% CI, 96.5–99.1%) and 98.6% (95% CI, 97.0–99.3%), respectively.
Siedner et al. were the first to use only real-time PCR techniques for determination of the gyrA genotype. They reported improvement of assay sensitivity after restricting the assay to amplification of only the Ser91 codon, compared to the entire amplicon of the gyrA gene as was reported previously [34s, http://links.lww.com/OLQ/A163]. Furthermore, 4 studies reported 100% specificity of mutation in the Ser91 region of the gyrA gene for N. gonorrhoeae compared with other subspecies [35s–37s, 41s, http://links.lww.com/OLQ/A163].
Among 252 samples from Canada, Peterson et al. compared ciprofloxacin susceptibility results with the Asp95 region of the gyrA gene and the Asp86 region of the parC gene, both previously associated with ciprofloxacin resistance. They found comparable sensitivity for Asp86 parC in predicting ciprofloxacin resistance. Three other studies using real-time PCR techniques, however, report that amplification of the parC gene was less sensitive and specific than amplification of the gyrA gene, though additional alteration of the parC gene may confer a greater degree of antimicrobial resistance [32 s, 38 s, 39 s, http://links.lww.com/OLQ/A163].
Figure 3 shows the sROC for studies using non-real-time PCR and real-time PCR techniques.
We systematically reviewed studies that have compared Neisseria gonorrhoeae gyrA genotype results with conventional methods of ciprofloxacin susceptibility determination. Among studies using real-time PCR techniques there appears to be strong evidence that gyrA genotype results are both highly sensitive and specific for the prediction of N. gonorrhoeae susceptibility to ciprofloxacin. Furthermore, studies have shown that gyrA genotype determined by real-time PCR is 100% specific to N. gonorrhoeae compared to other Neisseria subspecies [35s–37s, 41s, http://links.lww.com/OLQ/A163].
Studies using non–real-time PCR techniques, such as Pyrosequencing and probe hybridization, reported wider ranges of sensitivities. That finding is likely in part due to the incorporation of multiple different mutations in the gyrA gene that may not be associated with resistance; the range of sensitivities improved when only analyzing known resistance mutations. Notably, the study by Gharizadeh et al. compared gyrA mutation with phenotypic susceptibility results; when looking at MIC assay results, it was noted that samples with mutant gyrA genotypes had higher median MICs than wild-type samples.
We also compared the sROC between studies using real-time PCR and non–real-time PCR techniques. The summary estimates of sensitivity and specificity were well separated, though the confidence regions overlapped slightly. Based on that, we therefore conclude that the real-time PCR technique may be a more accurate technique in gyrase A genotype testing than other techniques.
Rapid molecular assays for determination of resistance are enabling targeted antimicrobial therapy that could curb the emergence of antimicrobial resistance. The FDA-approved molecular assay for Mycobacterium tuberculosis uses real-time PCR for amplification and molecular probes for the detection of mutations within the rifampin-resistance determining region. That assay has an overall sensitivity of 97.6% and specificity between 98.1% and 99.2%.15 Real-time PCR amplification of known resistance genes has also been shown to be clinically effective in screening for Methicillin-resistant Staphylococcus aureus, with low false-negative and false-positive rates between 0.0–7.3% and 0.0–5.4% respectively.16 Additionally, rapid molecular assays are available for carbapenem-resistant Enterobacteriaceae and have been shown to detect most resistance genes without yielding false positives.17
Therefore, given the robust performance of the real-time PCR gyrA assays, N. gonorrhoeae genotyping of gyrA has the potential to enable reliable targeted therapy with ciprofloxacin. Targeted therapy with ciprofloxacin might reduce the ongoing selection pressure caused by the widespread empiric use of extended-spectrum cephalosporins and have implications for the enhanced control of ceftriaxone resistant N. gonorrhoeae.8 A recent study suggested that treatment might be a major driver of ceftriaxone resistance among N. gonorrhoeae.18
In November 2015, University California Los Angeles implemented routine gyrA testing on all N. gonorrhoeae positive clinical specimens. Results from testing are used to promote targeted therapy with ciprofloxacin based on genotype susceptibility results.19,20 Further studies are underway to characterize the impact of the assay on patient clinical outcomes.21
There were a few limitations to our study. Primarily this was an analysis of previous studies and therefore subject to the same limitations as the studies being reviewed. Additionally the prevalence of ciprofloxacin susceptible N. gonorrhoeae differs in various regions of the world,10,22 and gyrA results may not be of substantial benefit in regions where the prevalence of ciprofloxacin resistance is very high. Furthermore, our study spanned 20 years, during which time technological advances likely impacted the sensitivity and specificity of test results. However, as most studies, including the earliest studies, report high sensitivity and specificity of gyrA testing for predicting ciprofloxacin susceptibility, the changes in technology likely did not impact the overall findings, but furthered the utility of gyrA testing, and therefore support, rather than negate our findings.
Gyrase A genotype testing is a novel approach to combating the emergence of multi drug-resistant N. gonorrhoeae. Among studies comparing gyrA genotype results with conventional methods of susceptibility testing, there appeared to be strong evidence that gyrA genotype results are both sensitive and specific in predicting ciprofloxacin in vitro susceptibility, with real-time PCR techniques being a more accurate approach compared with non–real-time PCR techniques.
1. Centers for Disease Control and Prevention. Sexually Transmitted Disease Surveillance 2013. Atlanta: U.S. Department of Health and Human Services, 2014.
2. Unemo M, Nicholas RA. Emergence of multidrug-resistant, extensively drug-resistant and untreatable gonorrhea. Future Microbiol 2012; 7:1401–1422.
3. Bolan GA, Sparling PF, Wasserheit JN. The emerging threat of untreatable gonococcal infection. N Engl J Med 2012; 366:485–487.
4. Foerster S, Golparian D, Jacobsson S, et al. Genetic Resistance determinants, in vitro time-kill curve analysis and pharmacodynamic functions for the novel topoisomerase II inhibitor ETX0914 (AZD0914) in Neisseria gonorrhoeae
. Front Microbiol 2015; 6:1377.
5. Butler MS, Blaskovich MA, Cooper MA. Antibiotics in the clinical pipeline at the end of 2015. J Antibiot (Tokyo) 2016.
6. GlaxoSmithKline, A Dose-Ranging Study Evaluating the Efficacy, Safety, and Tolerability of GSK2140944 in the Treatment of Uncomplicated Urogenital Gonorrhea Caused by Neisseria gonorrhoeae
. In: ClinicalTrials.Gov., 2014- [cited 2016 Aug 12] (Bethesda(MD): National Library of Medicine (US)): p. Available from: https://clinicaltrials.gov/ct2/show/NCT02294682
NLM Identifier: NCT02294682.
7. Buono SA, Watson TD, Borenstein LA, et al. Stemming the tide of drug-resistant Neisseria gonorrhoeae
: the need for an individualized approach to treatment. J Antimicrob Chemother 2015; 70:374–381.
8. Klausner JD, Kerndt P. Cephalosporin resistance in Neisseria gonorrhoeae
infections. JAMA 2013; 309:1989–1991.
9. Low N, Unemo M. Molecular tests for the detection of antimicrobial resistant Neisseria gonorrhoeae
: when, where, and how to use? Curr Opin Infect Dis 2016; 29:45–51.
10. Kirkcaldy RD, Harvey A, Papp JR, et al. Neisseria gonorrhoeae
antimicrobial susceptibility surveillance—The Gonococcal Isolate Surveillance Project, 27 Sites, United States, 2014. MMWR Surveill Summ 2016; 65:1–19.
11. Cochran WG. The comparison of percentages in matched samples. Biometrika 1950; 37:256–266.
12. Reitsma JB, Glas AS, Rutjes AW, et al. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005; 58:982–990.
13. Irwig L, Macaskill P, Glasziou P, et al. Meta-analytic methods for diagnostic test accuracy. J Clin Epidemiol 1995; 48:119–130; discussion 131–2.
14. R Development Core Team: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, 2016.
15. Boehme CC, Nabeta P, Hillemann D, et al. Rapid molecular detection of tuberculosis and rifampin resistance. N Engl J Med 2010; 363:1005–1015.
16. Kim JU, Cha CH, An HK, et al. Multiplex real-time PCR assay for detection of methicillin-resistant Staphylococcus aureus
(MRSA) strains suitable in regions of high MRSA endemicity. J Clin Microbiol 2013; 51:1008–1013.
17. Miller S, Humphries RM. Clinical laboratory detection of carbapenem-resistant and carbapenemase-producing Enterobacteriaceae. Expert Rev Anti Infect Ther 2016; 14:705–717.
18. Fingerhuth SM, Bonhoeffer S, Low N, et al. Antibiotic-resistant Neisseria gonorrhoeae
spread faster with more treatment, not more sexual partners. PLoS Pathog 2016; 12:e1005611.
19. Allan-Blitz LT, Humphries RM, Hemarajata P, et al. Implementation of a rapid molecular assay for determination of Neisseria gonorrhoeae
susceptibility in a large health system. Poster Presented at: ID Week 2016; October 26–30; New Orleans.
20. Allan-Blitz LT, Humphries RM, Hemarajata P, et al. Implementation of a rapid genotypic assay to promote targeted ciprofloxacin therapy of Neisseria gonorrhoeae
in a large health system. Clin Infect Dis, 2016.
22. Młynarczyk-Bonikowska B, Kujawa M, Młynarczyk G, et al. Resistance to ciprofloxacin of Neisseria gonorrhoeae
strains isolated in Poland in 2012–2013. Med Dosw Mikrobiol 2014; 66:99–104.