Article Text
PDF
Abstract
Objectives Performance of the new Becton Dickinson ProbeTec GC Qx assay on the BD VIPER platform was evaluated to ascertain whether confirmatory testing is required in our clinical setting.
Methods Positive predictive value (PPV) was determined by comparison with culture and a confirmatory nucleic acid amplification test (NAAT)-based Neisseria gonorrhoeae assay from genital and extragenital samples (rectal and pharyngeal) collected from a genitourinary medicine (GUM) clinic.
Results Among 14 223 clinical genital samples, 149 (1.0%) specimens were positive using the ProbeTec GC Qx assay, automated on the VIPER platform; 141 of these were confirmed by either culture or a real-time PCR targeting two gonococcal-specific targets (PPV 94.6%; 95% CI 90% to 98%). Among 840 pharyngeal samples, 26 (3.1%) were positive by the ProbeTec GC Qx assay; 13 were confirmed (PPV 50%; 95% CI 30% to 70%). Among 593 rectal samples, 17 tested positive by the ProbeTec GC Qx assay; all were confirmed (PPV 100%; 95% CI 80% to 100%).
Conclusions The lower 95% CI of the PPV for the ProbeTec GC Qx assay for genital specimens was >90% in this GUM clinic population, and therefore confirmatory testing for genital specimens is not required. Confirmatory testing of pharyngeal and rectal samples should continue in line with national guidelines.
- Neisseria gonorrhoeae
- Laboratory diagnosis
- NAAT
Statistics from Altmetric.com
Introduction
Testing of asymptomatic, sexually active people for Chlamydia trachomatis and Neisseria gonorrhoeae infection is widespread in Europe and North America. In most settings, gonorrhoea is less common than chlamydia infection: in 2008 there were 209 000 new diagnoses of chlamydia in the UK compared with 16 629 new diagnoses of uncomplicated gonorrhoea.1
The BD VIPER system (Becton Dickinson, Franklin Lakes, New Jersey, USA) is an automated sample processor and fluorescent reader using strand displacement amplification for qualitative detection of N gonorrhoeae and C trachomatis DNA. The previous Becton Dickinson assay was the ProbeTec ET GC assay, utilising similar technology but comprising manual extraction via heating and lysis and a semi-automated strand displacement amplification procedure. The newer methodology has a fully automated extracted mode, which uses alkaline lysis of cells and binding of DNA to ferric oxide to extract N gonorrhoeae DNA and includes an extraction control to monitor failed extraction and PCR inhibition.
The ProbeTec GC Qx and ET GC assays target different regions of the N gonorrhoeae pilin inverting protein homologue gene, so there may be differences in the performance characteristics of these assays. The manufacturer claims sensitivity of 99.3% and specificity of 99.4% for genital samples.
Alternative commercial assays for the detection of N gonorrhoeae are the Gen-Probe APTIMA Combo 2 assay on the Tigris platform (Gen-Probe, Inc, San Diego, California, USA), the Roche AMPLICOR CT/NG assay (Roche, Branchburg, New Jersey, USA) and the Abbott RealTime CT/NG assay on the m2000 platform (Abbott Molecular Diagnostics, Des Plaines, Illinois, USA). Each of these assays detects different targets. The APTIMA Combo 2 assay detects N gonorrhoeae 16S rRNA using transcription-mediated amplification, and the m2000 CT/NG assay detects the N gonorrhoeae Opa gene using real-time PCR.
For chlamydia nucleic acid amplification tests (NAATs), repeat testing using the same assay is considered adequate, as this reduces inaccurate results due to processing errors but also reduces sensitivity in samples with a low pathogen load.2 There have been reports of a relatively low specificity of the older ProbeTec ET GC assay for N gonorrhoeae and the Roche COBAS AMPLICOR leading to unacceptably low positive predictive values (PPVs) in a low prevalence population; therefore repeat testing on the same platform is not considered adequate for GC because of cross-reaction with commensal Neisseria spp.3 4 The Health Protection Agency (HPA) and the Centers for Disease Control and Prevention recommend that, for N gonorrhoeae confirmatory testing, a different target should be considered if the PPV of an assay is <90% to improve PPV.5–7 Our aim was to ascertain the PPV of the BD ProbeTec GC Qx assay in our genitourinary medicine (GUM) clinic attendee population for genital and extragenital sites.
Methods
Traditionally, diagnostic assessments are made by comparing the test with a reference assay for all samples. This would represent an expensive and time-consuming approach. As this was a service evaluation, we used a methodology similar to that reported by Golden et al,8 who determined the PPV of a NAAT for N gonorrhoeae by confirming only the samples that tested as positive in a larger number of specimens without performing reference tests in all samples.
There is currently no clear consensus on standards for evaluations of NAATs. The choice of ‘gold standard’ becomes complicated because NAATs are likely to be more sensitive than culture-based methods, and genuine positives may not be recognised. We calculated PPV by comparing screening NAAT results with a composite ‘gold standard’ consisting of bacterial culture and two confirmatory real-time PCR results.
Specimen collection
Between 23 June 2009 and 15 January 2010, patients attending the Courtyard (GUM) Clinic at St George's NHS Healthcare Trust were tested for infection with N gonorrhoeae and C trachomatis using the ProbeTec CT/GC Qx assay, automated on the VIPER platform as per the manufacturer's instructions.
Consecutive routine specimens of urine samples, endocervical, urethral, rectal and pharyngeal wet swabs were included in this service evaluation as outlined below. All male patients were tested with a first-void urine for chlamydia/gonorrhoea NAAT. In addition, clinically indicated swabs for gonococcal culture and the ProbeTec GC Qx assay were taken from urethral, rectal and pharyngeal sites. Male patients giving a history of dysuria or urethral discharge had a urethral swab collected for Gram stain and culture for N gonorrhoeae. Men who have sex with men had culture and NAAT performed on rectal and pharyngeal swabs based on a history of anal and/or oral sex. Three swabs were taken simultaneously from the throat by sticking them together with tape and rotating them. One swab was for N gonorrhoeae culture, one for BD ProbeTec GC Qx and one for BD ProbeTec ET GC (used for confirmatory testing at the Sexually Transmitted Bacteria Reference Laboratory (STBRL)). Rectal swabs were collected without a speculum unless rectal symptoms were present. The culture swab was taken first, followed by the GC Qx and the ET GC swab. Urethral swabs were taken from female patients for gonorrhoea culture. After insertion of a speculum, endocervical swabs were collected for culture and BD ProbeTec GC Qx in that order.
Testing by ProbeTec GC Qx
Samples positive for N gonorrhoeae in the initial ProbeTec GC Qx assay were retested using the same sample and a fresh extraction using the same assay. A NAAT was considered positive if positive on the repeat run. Samples that did not repeat as positive were considered equivocal, as this may have resulted from a processing error, and a repeat sample was requested; these samples were not used in the analysis of PPV. The confirmatory testing method is shown in figure 1.
Outline of confirmatory testing protocol.
Confirmation by real-time PCR
All samples that tested positive for N gonorrhoeae by the ProbeTec GC Qx assay were forwarded to the STBRL, where they were examined blind using two N gonorrhoeae-specific real-time PCRs. Some negative samples were also sent.
Upon receipt, 200 μl ProbeTec ET GC transport buffer was extracted using the MagNAPure Compact (Roche Diagnostics, Indianapolis, Indiana, USA) and examined using a modified version of a previously published N gonorrhoeae real-time PCR assay which targets the porA pseudogene.9 Briefly, PCR was performed using 25 μl reactions containing 5 μl extracted DNA, 4 mM MgCl2, 1 U Taq DNA polymerase (Applied Biosystems, Foster City, California, USA), 250 nM forward and reverse primers, and 500 nM probe (table 1). Also included in this PCR were 80 mM primers and probe of a previously described internal control that targets the human RNase gene, to monitor both inhibition and specimen quality.11 PCR was performed using a Rotor-Gene 6000 (Qiagen, Hilden, Germany) using the following cycling conditions: 10 min at 95°C, followed by 40 cycles of 95°C for 30 s and 60°C for 60 s.
Primer and probe nucleic acid sequences for PorA and Opa real-time PCRs
Specimens that tested negative using this assay were re-examined using a further assay that targets the Opa gene.10 PCR was performed using 20 μl reactions containing 5 μl extracted DNA, 5 mM MgCl2, 1 U Taq DNA polymerase, 250 nM forward and reverse primers, and 500 nM probe (table 1). PCR samples were run on a Rotor-Gene 6000 using the following cycling conditions: 10 min at 95°C, followed by 50 cycles of 95°C for 15 s and 60°C for 60 s.
Culture
VCAT (containing vancomycin, colistin, amphotericin and trimethoprim)-selective culture plates (Biomerieux, Marcy l'Etoile, France) were inoculated in the GUM clinic and incubated (48 h, CO2). Presumptive colonies were Gram stained, oxidase tested and subcultured on to a purity plate. N gonorrhoeae was confirmed by MicroTrak N gonorrhoeae culture confirmation fluorescent antibody test (Trinity Biotech Plc, Co Wicklow, Ireland) and API NH (BioMerieux). Corresponding culture plates for all ProbeTec GC Qx extragenital samples that tested as positive were incubated for a total of 72 h to maximise growth of Neisseria spp.
Analysis of test performance
A sample was treated as a true positive if positive by either culture or confirmatory PorA or Opa real-time PCR. Stata 8.0 was used to determine PPV and 95% CIs. A limitation of this methodology is that sensitivity and specificity were not determined because negative samples were not repeated by confirmatory NAAT and were not compared with the culture result. However, we can estimate the specificity if we assume a value for the sensitivity of the assay, based on the following relationship12:
Results
PPV and prevalence for genital samples
In total, 14 223 genital specimens (5851 male urine samples, 8075 cervical and 297 male urethral swabs) were tested. Two samples that tested as positive in the initial ProbeTec GC Qx assay did not repeat as positive on the repeat ProbeTec GC Qx run. These two samples were reported as equivocal and excluded from the analysis. Of these 14 221 samples, 150 (1.1%) from 149 patients tested positive for N gonorrhoeae using the ProbeTec GC Qx assay. Of these positive samples, 141/149 (94.7%) were confirmed by either culture or HPA confirmatory PCR,9–11 giving a PPV of 94.6% (95% CI 90% to 98%) (table 2). Of the remaining eight patients with unconfirmed positive genital specimens, three had a confirmed N gonorrhoeae infection at another site, suggesting that these are genuine positive samples, raising the PPV of genital samples to 96.7% (95% CI 93% to 100%). Of the confirmed positive genital samples, four (2.7%) were culture positive but not detected by confirmatory PCR.
Numbers of ProbeTec GC Qx positive samples confirmed by supplementary NAAT and culture for different sample types
On the basis of the relationship described in equation 1, if we assume a sensitivity of 99.3%, then the estimated specificity will be 99.96%. Even if we assume a much lower sensitivity—for example, 50%—the estimated specificity remains at 99.96%.
PPV and prevalence for extragenital samples
In total, 593 rectal and 840 pharyngeal swab samples were tested. Of these specimens, 17/593 (2.87%) rectal swabs and 26/840 (3.10%) pharyngeal swabs tested positive for N gonorrhoeae, with the ProbeTec GC Qx assay. Of these positive samples, all 17 rectal swabs were confirmed positive, giving a PPV of 100% (95% CI 80% to 100%). Of the 26 pharyngeal samples, 13 were confirmed by either GC culture or in-house HPA confirmatory PCR, with a PPV of 50% (95% CI 30% to 70%) (table 2). The number of samples tested as positive by the ProbeTec GC Qx assay, confirmed by NAAT and culture, are shown in table 2. For positive results, which were unconfirmed, two of the 13 pharyngeal samples had one or more samples confirmed positive for N gonorrhoeae from another site, and these were therefore considered to be likely genuine positive samples. If these samples were considered true positives, the PPV would be 57.7% (95% CI 39% to 77%) for pharyngeal samples. Of the confirmed positive samples, one (3.8%) pharyngeal swab and one (5.9%) rectal swab were culture positive but not detected by confirmatory NAAT. Nine samples were confirmed by NAAT but were not confirmed by culture. Four of the unconfirmed pharyngeal samples grew Neisseria meningitidis upon culture. All extragenital samples were positive upon retesting by ProbeTec GC Qx.
Discussion
The ProbeTec GC Qx assay is licensed by the Food and Drug Administration and CE marked for use with urethral, endocervical and self-collected vulvovaginal swabs plus male and female urine samples. This assay is not licensed for use on pharyngeal, rectal or eye swabs. For genital samples, we found a PPV of 94.6% (95% CI 90% to 98%) in our clinic population with a confirmed N gonorrhoeae rate of 1.0%. As the lower CI is 90%, it is reasonable to use the ProbeTec GC Qx assay without routine confirmatory testing in this setting. Current national guidelines state that confirmatory GC NAATs are required if the PPV is <90%, and confirmatory testing is unnecessary for genital specimens in our GUM population.5 6 However, in populations where the prevalence of N gonorrhoeae is low—for example, primary healthcare settings—the PPV may be lower, and confirmatory testing by a second NAAT, with a second genetic target, may be required.
Our routine practice is to retest all NAAT-positive samples on the same platform, although this is not explicitly recommended by the manufacturer. This is to reduce inaccurate results due to sample processing errors. This practice of repeat testing of positive samples on the same platform is not universally practiced, but there were only two specimens, neither of them extragenital, that did not retest as positive. The inclusion of this small number of specimens would not alter these findings.
During the evaluation period, no culture positive samples were ProbeTec GC Qx negative from the same site. This supports the manufacturer's claim that the assay has a high sensitivity of >99.5%.
Akduman et al13 found that the ProbeTec ET GC assay had a PPV of 84.9% for testing of urine samples. In a large multicentre study, the PPV of the assay for female urine samples varied from 54.8% to 100%. The site reporting the lowest PPV of 54.8% had a N gonorrhoeae prevalence of 1.2%, which is similar to our reported prevalence.14 The PPV reported here shows that the ProbeTec GC assay, in extracted mode, has a higher PPV than the ProbeTec ET GC assay and is likely to have a higher specificity than that reported of the ProbeTec ET GC assay.
There are concerns regarding the use of NAATs for the detection of N gonorrhoeae from rectal and pharyngeal samples, because of lower specificity. Although NAATs are the most sensitive assays available,15–17 the risk of false-positive results is high because of the genetic similarity between N gonorrhoeae and commensal Neisseria spp., leading to cross-reactions. Optimal methods for detection of N gonorrhoeae infection at extragenital sites are uncertain, and no commercial NAAT is licensed for detection of N gonorrhoeae from extragenital sites. Here, we found that our NAAT identified more pharyngeal GC infections than culture, in keeping with previous studies.18–20 Indeed NAATs are redefining the epidemiology of gonococcal infection. Out of 26 pharyngeal samples initially testing as positive, 13 (50%) were not confirmed by either culture or confirmatory NAAT. There are two possible explanations: (a) these samples had a low inoculum of N gonorrhoeae; (b) cross-reactivity resulted in an increased number of false-positive ProbeTec GC Qx results. The ProbeTec ET GC and Roche COBAS AMPLICOR CT/NG assays are known to cross-react with strains of Neisseria subflava and Neisseria cinerea resulting in false-positive results,21 22 and these species may be present in the oropharynx as commensal bacteria in up to a third of individuals.23 Such cross-reactivity has not yet been reported with the APTIMA Combo 2 test.24 Specificities of the APTIMA Combo 2 and ProbeTec ET GC for oropharyngeal swabs have been found to be similar, and have been reported to be >99.4%.16 However, Ota et al15 determined the PPV of the ProbeTec GC ET assay to be only 82.6% for pharyngeal swabs. It is likely that the low PPV determined here reflects lower specificity of this assay.
A further limitation may be the order in which swabs were taken. For rectal and endocervical swabs, culture swabs were taken before swabs for the NAAT, and therefore the NAAT may have been disadvantaged and resulted in an underestimate of PPV. The pharyngeal swabs were collected for all three samples simultaneously, by taping them together, as we considered it unpleasant for patients to undergo three successive samplings. Although practitioners rotated the triple swab, there may well have been sampling differences.
We found a PPV of 100% (95% CI 80% to 100%) for rectal swabs; of 17 samples that tested as positive, all were confirmed by either culture or the second NAAT. However, this is too small a number to give sufficient confidence in this estimate, and larger numbers are required to confirm that the lower CI is >90%. Others have also shown the high PPV of rectal swabs for detection of N gonorrhoeae, but sample sizes were also small.15 16 In compliance with national guidelines, we recommend that extragenital samples that test positive by NAAT should be confirmed. Samples that test as positive but are not confirmed should be reported as equivocal, and a repeat sample requested.5–7
Discrepant analysis of samples that are reference standard negative will increase the PPV. As this was a service evaluation, partner status was not included: the result is that the PPV reported here may be an underestimate of the true value. However, for unconfirmed ProbeTec GC Qx positive samples, a partial discrepant analysis was performed; samples that tested as positive in the ProbeTec GC Qx assay were considered a true positive if confirmed positive results were available from one or more alternative sites.
We used N gonorrhoeae-specific real-time PCRs to act as confirmatory assays. Interestingly, six specimens that were both ProbeTec GC Qx and culture positive were not confirmed by the STBRL real-time PCR assays. There may be numerous reasons for this, including DNA degradation, as the confirmatory testing was performed off site and therefore specimens incurred a period of transportation delay before testing.
It is possible that some culture-negative samples that tested as positive in the ProbeTec GC Qx assay may not have been confirmed even if genuinely positive. Choice of confirmatory assay is important, as this should detect a different target from the screening NAAT and have good sensitivity and specificity. These could include real-time PCR assays available from reference laboratories or developed in-house—for example, PorA/Opa PCR—or reciprocal arrangements with a laboratory with a different platform.25
In summary, we have demonstrated that the ProbeTec GC Qx assay can achieve a high PPV when used to test genital specimens in our GUM clinic population, and therefore, in accordance with national guidelines, confirmatory testing is unnecessary. However, testing with a second supplementary NAAT may still be desirable in lower prevalence populations.
Key messages
Testing of genital specimens by the ProbeTec GC Qx assay does not require confirmatory testing in our genitourinary medicine clinic population.
Confirmatory testing of pharyngeal swabs that have tested as positive should continue as recommended by national guidelines.
The positive predictive value for rectal swabs looks promising, but larger numbers are required to confirm that the lower CI is >90%.
Acknowledgments
We thank the staff at the Departments of Medical Microbiology and GUM at St George's Hospital, London, and at STBRL, CFI, Colindale. In particular, we thank Danuta Dubeck, Pauline Greene, John Hadfield, Helen Liddy and Sima Patel for assistance.
References
Footnotes
Competing interests This work was partly funded by Becton Dickinson.
Provenance and peer review Not commissioned; externally peer reviewed.