OBJECTIVE: To estimate the clinical performance characteristics of a real-time polymerase chain reaction (PCR) assay using vaginal/rectal swabs from antepartum (35–37 weeks of gestation) and intrapartum women.
METHODS: The assay evaluated is a qualitative, automated, real-time PCR test for the detection of group B streptococci, with results available in approximately 75 minutes. Enrollment in this multicenter clinical study occurred between October 2005 and January 2006. Vaginal/rectal swabs were analyzed by nursing personnel (intrapartum tests) or by laboratory technologists (all others). Polymerase chain reaction assay results were compared with culture using standard methods, including selective broth medium, and to a predicate nucleic acid amplification test.
RESULTS: Of 1,028 enrolled women, 234 were deemed ineligible, and 10 had unresolved test results. Of the 784 remaining women, valid PCR assay results were obtained on the first test attempt for 93.0%. Performance characteristics relative to culture were sensitivity 91.1%, specificity 96.0%, positive predictive value 87.8%, negative predictive value 97.1%, and accuracy 94.8%. These results exceeded those obtained using the predicate nucleic acid amplification test.
CONCLUSION: Performance characteristics of the PCR assay exceed the threshold recommended by the Centers for Disease Control and Prevention when compared with culture. The test is sufficiently robust to be performed for intrapartum patients in a point-of-care setting by medical professionals.
LEVEL OF EVIDENCE: II
A rapid, real-time polymerase chain reaction assay demonstrates excellent sensitivity and specificity relative to a predicate nucleic acid amplification test.
From the 1Phoenix Perinatal Associates/Obstetrix Medical Group of Phoenix, Phoenix, Arizona; 2Administration/Microbiology, Kaiser Permanente, North Hollywood, California; 3Department of Pediatrics, Section on Medical Genetics, Wake Forest University School of Medicine, Winston-Salem, North Carolina; 4Department of Pathology, Wishard Memorial Hospital, Indianapolis, Indiana; 5Obstetrical & Gynecological Associates, Houston, Texas; and 6Department of Pathology, Magee Women’s Hospital and Research Institute, University of Pittsburgh, Pittsburgh, Pennsylvania.
Supported by Cepheid (Sunnyvale, CA). Cepheid was responsible for the design of the study, recruitment of centers, training of individuals to perform tests, periodic site visits, and performance of the statistical analyses.
Presented at the 33rd annual scientific meeting of the Infectious Diseases Society for Obstetrics and Gynecology in Monterey, California, August 2–5, 2006.
Corresponding author: Rodney K. Edwards, MD, MS, Phoenix Perinatal Associates, 3877 North 7th Street, #400 Siete Square IV, Phoenix, AZ 85014; e-mail: email@example.com.
Financial Disclosure Dr. Jordan has been a paid consultant for Cepheid (Sunnyvale, CA). The other authors have no potential conflicts of interest to disclose.
Streptococcus agalactiae, or group B streptococci (GBS), are gram positive, aerobic coccoid bacteria that can be isolated from the vagina and rectum of 15–35% of pregnant women.1,2 In the absence of intervention, the organism is the most common cause of early-onset (within the first 7 days of life) neonatal infection, with an incidence of 1 to 2 per 1,000 live births.3 Early-onset infections are due to vertical transmission of the organism from mother to infant during the birth process and can result in mortality or serious morbidity, such as cognitive or visual impairment.4
Administration of intravenous intrapartum antibiotic prophylaxis is effective in reducing the incidence of early-onset neonatal GBS infections.5–7 Guidelines from the Centers for Disease Control and Prevention (CDC) recommend that all women should be screened at 35–37 weeks of gestation and that those women found to be colonized with group B streptococci should receive intrapartum antibiotic prophylaxis either with penicillin G or ampicillin.8 However, serial cultures from pregnant women have suggested that GBS colonization may be intermittent9; therefore, the desired information is the colonization status at the time of labor. An ideal screening test for GBS colonization would be performed at the time of labor and would yield results in time to administer intrapartum antibiotic prophylaxis to those women found to be colonized.
The CDC-recommended method for detecting GBS colonization involves collecting a combined vaginal and rectal specimen and culturing the organism in a selective broth medium, followed by subculturing onto blood agar plates. This process requires 2–3 days; therefore, screening must occur at some point during the antepartum course. Additionally, the positive predictive value of antepartum cultures done within 6 weeks of delivery in predicting culture results at the time of labor has been reported to be only 67–87%.10,11 This situation leads to antibiotics being administered to women who are at very low risk of delivering an infant who will develop an early-onset GBS infection. In addition, the negative predictive value is imperfect—approximately 90–95%.10,12 Therefore, antepartum screening fails to identify some women who are colonized at the time of labor.
Cepheid (Sunnyvale, CA) has developed a qualitative in vitro diagnostic test to detect GBS DNA in vaginal/rectal swab specimens. This test, the Xpert GBS Assay, uses automated real-time polymerase chain reaction (PCR) technology. It is performed on the GeneXpert Dx System, a platform in which sample preparation, amplification, and real-time detection all are fully automated and completely integrated. The objective of this study was to estimate the test performance characteristics of the Cepheid Xpert GBS Assay using vaginal/rectal swabs from women at 35–37 weeks gestation (antepartum) or during labor (intrapartum). We compared the performance of this test to culture and to a predicate nucleic acid amplification test, the IDI-Strep B Assay (BD GeneOhm, San Diego, CA) run on the Cepheid SmartCycler platform (Cepheid, Sunnyvale, CA).
MATERIALS AND METHODS
All subjects gave written informed consent before participation. Women were eligible for the intrapartum arm of the study if they were in labor, had no contraindication to vaginal examination (eg, placenta previa), had not received any systemic or vaginal antibiotic treatment in the preceding week, and did not have an urgent indication to proceed to delivery. Women were eligible for the antepartum arm of the study if they were at 35–37 weeks of gestation, had no contraindication to vaginal examination, and had not received any systemic or vaginal antibiotic treatment in the preceding week. Subjects were enrolled between October 2005 and January 2006. The study was approved by the institutional review board of each participating center.
Two double swab sets (four swabs) of vaginal/rectal samples were collected from each woman. The collection technique included brushing the two swabs together with a twirling motion after the sample had been collected, to minimize swab-to-swab variation in a given double swab set. All clinicians collecting samples participated in a training session to review the desired collection technique.
One swab was used for culture only, while two other swabs were used for the Xpert GBS Assay and the IDI-Strep B Assay. The discarded swabs from both molecular tests also were cultured for group B streptococci. For women in the intrapartum arm of the study, the Xpert GBS Assay was run in the labor and delivery unit by nurses. Assays for the antepartum arm of the study were conducted by technologists in the laboratory.
Our sample collection protocol specified both vaginal and rectal sites of collection, in keeping with guidelines recommended by the Centers for Disease Control and Prevention.8 The Cepheid clinical trial protocol approved by the U.S. Food and Drug Administration closely followed these recommended guidelines. Accordingly, samples that were taken only from vaginal or rectal sites, or samples taken from perivaginal or perirectal sites, were excluded from the analysis.
The Xpert GBS tests were conducted according to the instructions in the clinical protocol. The Xpert GBS Assay includes reagents for the simultaneous detection of the target GBS DNA, a sample-processing control to monitor processing conditions, and an internal control to monitor PCR conditions and the presence of molecules that can inhibit a PCR reaction. An additional mechanism of control within the Xpert GBS Assay is a feature called probe check. Probe check verifies reagent rehydration, PCR-tube filling in the cartridge, hydrolysis probe integrity, and dye stability. The sample-processing control and internal control do not affect sensitivity of the target sequence because the concentration of the internal control is adjusted so that it does not compete with the target. The sample-processing control and internal control were tested extensively during the study by using substances that can potentially inhibit the PCR process. In the presence of inhibitors or when there is GBS DNA present in a sample, the sample-processing control and the internal control will be repressed.
For those tests not yielding results on the first assay attempt, the remaining elution reagent from the first cartridge was placed in a second cartridge for testing according to the package insert. If a result was not obtained after two attempts, the result was considered to be unresolved. After testing, the remaining swab in the elution reagent was transferred into Lim broth (Todd-Hewitt broth with 15 microgram/mL nalidixic acid and 10 microgram/mL colistin; PML Microbiologicals, Wilsonville, OR or BBL, Kansas City, MO) for culture, and cartridges were kept at 2–8°C until the remaining swabs were cultured.
The IDI-Strep B tests were conducted according to the instructions in the package insert.12 Specimens not yielding results on the first assay attempt were retested following the instructions in the package insert. If a result was not obtained after 2 attempts, the result was considered to be unresolved. After testing, the remaining swab in the elution reagent was transferred into Lim broth for culture.
Culturing for GBS was carried out by incubating in Lim broth for 18–24 hours at 35±2°C in a 5% CO2 environment, followed by subculturing onto a 5% sheep blood agar plate and an NEL plate (Northeastern Laboratory Services; Winslow, ME). The blood agar plate was incubated for 18–24 hours at 35±2°C in a 5% CO2 environment. The NEL plate was incubated anaerobically for 18–24 hours at 35±2°C. Colonies on the blood agar plates suspicious for group B streptococci were confirmed with latex agglutination testing. Data from the NEL plates were included in the analysis only when blood agar plates could not be read (eg, overgrowth of Proteus sp.), with the presence of orange-colored colonies being confirmatory for β hemolytic GBS.
For women in the antepartum arm of the study, the routine test results (culture using selective broth medium and/or IDI Strep-B) performed as part of the study were reported for patient management if ordered. Decisions regarding whether to provide intrapartum antibiotic prophylaxis were based on these results.
Women in the intrapartum arm of the study underwent antenatal screening cultures as part of their routine prenatal care. Decisions regarding whether to provide antibiotic prophylaxis were based on these antenatal screening cultures. However, Xpert GBS results for women in the intrapartum arm of the study were made available for clinical use. If the Xpert GBS result was positive in a patient who otherwise would not receive antibiotics, the choice to provide intrapartum antibiotic prophylaxis could be made. These results were not used to withhold antibiotics from women who had positive antenatal screening cultures but negative intrapartum Xpert GBS results.
The following rules were used in the comparison of culture results to Xpert GBS and to IDI-Strep B: A sample was considered culture positive if either culture from its respective swab pair (Xpert GBS or IDI-Strep B) was positive. A sample was considered culture negative if both cultures from its respective swab pair were negative. For each molecular test, a true-positive result was defined as positive by the molecular test and by culture. A false-positive result was defined as positive by the molecular test and negative by culture. A true-negative result was defined as negative by the molecular test and negative by culture. A false-negative result was defined as negative by the molecular test and positive by culture. Using these definitions, sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated for each molecular test.
Sample size was based on the proportion of concordant responses to a reference standard to form a one-sample P test for Xpert GBS and IDI-Strep B separately and a two-sample P test, with each P value obtained separately relative to the reference standard. The alpha value was set at 0.05 and the β value between 0.05 and 0.10 to balance the sample sizes. Fisher exact test was used to assure that distinct categories, such as race, gender, lot, and site, were homogeneous before combining results for an overall analysis.
The Xpert GBS homogeneity for various measures was analyzed with n×2 tables using the Fisher exact test. These measures included the six sites, four test lots, two patient types (intrapartum and antepartum), and five race categories. If homogeneity was confirmed, data would be pooled and compared with the IDI-Strep B pooled results. Exact binomial confidence intervals were calculated about proportions. The sample size was calculated to have a 90% power (alpha=0.05) to determine at least a 0.1 difference in test performance characteristics between the IDI-Strep B and Xpert GBS assays if one existed. A sample size of at least 160 culture-positive samples was needed to fulfill these criteria. All statistical tests were performed using SAS 9.1 (SAS Institute Inc., Cary, NC).
During the study period, a total of 1,028 women were enrolled and had samples collected, including 548 in the intrapartum arm and 480 in the antepartum arm of the study. Figure 1 shows the accountability of patients for the Xpert GBS analysis. In the intrapartum arm of the study, 126 patients were considered ineligible because samples were collected with vaginal/perianal swab instead of vaginal/rectal swab and one patient was enrolled twice—in this latter case, the second sample was excluded. The inappropriately collected samples did not follow the recommended CDC sampling protocol, deviating from the clinical protocol that was accepted by the U.S. Food and Drug Administration. Of the remaining subjects, three had unresolved test results after two attempts, and they were not included in the analysis. Therefore, there were 418 patients analyzed from the intrapartum arm of the study. In the antepartum arm of the study, 96 patients were considered ineligible because samples were not collected from both vaginal and rectal sites per protocol, and one patient was enrolled twice (the second sample was excluded). In addition, 10 subjects in this arm of the study also were enrolled in the intrapartum arm of the study; only their intrapartum results were analyzed. For these 10 subjects, there were no changes in the classification (Xpert GBS or culture) for any subject at the intrapartum sampling (relative to antepartum). Of the remaining subjects, seven had unresolved test results after two attempts, and they were not included in the analysis. Therefore, there were 366 patients analyzed from the antepartum arm of the study.
The number of subjects analyzed from each study site ranged from 50 to 237. Study sites were distributed throughout the United States. The study population consisted of 40.8% white, 30.7% Hispanic, 20.7% African American, 5.5% Asian, and 2.3% other races. Average age at time of specimen collection was 27±6 years, with a range of 15.9 to 47.7 years. Based on results of analysis with n×2 tables using Fisher exact test, homogeneity was established across the six sites, four test lots, two patient types (intrapartum and antepartum), and five racial categories (data not shown). Therefore, data across these groups were pooled for calculation of test performance characteristics.
The overall results for the Xpert GBS Assay and the IDI–Strep B Assay are shown in Tables 1 and 2. Because there were 10 subjects excluded from the Xpert analysis and three excluded from the IDI-Strep B analysis due to unresolved samples after two assay attempts, the total number of subjects in each of these two tables differs slightly. Of the 784 analyzed patients, the Xpert GBS Assay yielded results on the first test attempt for 729 (93.0%) women.
Table 3 reports the overall test performance characteristics for the Xpert GBS Assay and the IDI-Strep B Assay compared with culture. Compared with the IDI-Strep B Assay, the Xpert GBS Assay demonstrated statistically significantly better sensitivity and negative predictive value (The 95% confidence intervals do not overlap.) and statistically similar specificity, positive predictive value, and accuracy. The two tests are compared with one another with the McNemar test in Table 4. The Xpert GBS Assay was statistically significantly more likely to be positive than was the IDI-Strep B test. Of the 24 subjects that were Xpert GBS positive and culture negative (Table 1), 15 of those samples were also IDI-Strep B Assay positive (data not shown). It is possible that samples positive for both PCR assays were actually true positive for GBS and false negative by culture, because the genetic targets for the two assays are distinct. The detection of two or more distinct genetic targets by using nucleic acid amplification methods has been used as a “gold standard” for defining true-positive results in other disease indications.13
Possible interfering substances were considered, including blood, feces, mucous, lubricant, meconium, and amniotic fluid. Of the analyzed patients, 238 (30.4%) had samples that were reported by the individual collecting the sample to be contaminated with at least one of these substances. However, the sensitivity and specificity of the Xpert GBS test were not significantly altered by the presence of these substances (data not shown).
Intrapartum antibiotic prophylaxis against early-onset neonatal infections with group B streptococci has had a significant effect on reducing the incidence of these infections. Within 3 years of the publication by the CDC of the first guideline for intrapartum antibiotic prophylaxis, the incidence of this infection had decreased by more than 60%.14 Since that time, it has been demonstrated that a screening-based approach to selecting women to receive prophylaxis is superior to a risk-factor-based approach.15 Because of the time required to obtain results, culture-based screening strategies are limited by the need to perform these screening cultures during the antepartum period. The value of these antepartum screening cultures for predicting colonization status at the time of labor is less than optimal.11
We report in this article the results of a clinical evaluation of the Xpert GBS Assay. When compared with culture, the test exhibited excellent sensitivity and specificity. The CDC has specified that an adequate rapid intrapartum test must be at least 85% sensitive compared with culture.8 The lower limit of the 95% confidence interval for the sensitivity of the Xpert GBS Assay in this study exceeded that threshold. Furthermore, the test was shown to be superior to the IDI-Strep B rapid test, which is also clinically available. The IDI-Strep B test requires a number of manual manipulation steps and a significant amount of hands-on time to attain results, and it does not include a sample processing control. The Cepheid Xpert GBS test includes a sample processing control to control for adequate processing of the target bacteria and can be performed by personnel in a labor and delivery setting who do not have formal laboratory training. In addition to demonstrating superior analytical performance, the Xpert GBS test does not have to be run in batches, a feature that is critical to providing actionable results in the intrapartum setting.
A question raised by the possibility of intrapartum screening is whether a significant proportion of women will have a meaningful delay in the administration of intrapartum antibiotic prophylaxis when screening is done intrapartum. Of the clinical sites that participated in the intrapartum testing, all opted to follow their own standard patient risk management as advised in the clinical study protocol; Xpert GBS results generated in the study were not used solely for patient management. Therefore, not all subjects were explicitly managed to generate the fastest possible turnaround time for the Xpert GBS Assay. Nonetheless, meaningful time analysis data were collected, which support the use of this test for effective intrapartum testing. The median time from sample collection to reporting results was 2.1 hours, and from Xpert initial run start to reporting results it was 1.5 hours, indicating that there were delays in starting the Xpert GBS Assay, because the priority was patient care and not the study protocol.
We believe that the turnaround time for this test will be less than 1.5 hours when the Xpert GBS Assay is used for intrapartum screening, because obtaining the result will be a patient care priority. In addition, in this study, the intrapartum tests were performed by nurses in the Labor and Delivery Unit. Omitting the need to send samples to the laboratory further decreases the likelihood that intrapartum screening would delay the administration of antibiotic prophylaxis to a clinically meaningful degree.
Intrapartum screening would pose an additional challenge for those women who report a history of a life-threatening allergy to β-lactam antibiotics. Because of the inability of intrapartum screening to comment on whether an isolate of group B streptococci is susceptible to alternative antibiotics, these women would need either to 1) undergo skin testing during the antepartum course to validate their allergy, 2) have an antepartum screening culture and if necessary, antibiotic sensitivities performed, or 3) be given an alternative antibiotic for prophylaxis to which group B streptococci are universally susceptible (eg, vancomycin).
This study has shown that the Xpert GBS Assay has the potential to offer the alternative of screening for colonization with group B streptococci in the intrapartum phase, rather than at the antepartum phase of gestation. Such intrapartum screening would target for prophylaxis only those women who are colonized with group B streptococci at the time of labor, avoiding the risks of antibiotic exposure for those women who are not at significant risk of delivering an infant with an early-onset infection with group B streptococci and avoiding the omission of prophylaxis for those women who are colonized at the time of labor but were not colonized at 35–37 weeks of labor. Additional benefits of intrapartum screening would be improved targeting of antibiotic prophylaxis in those women with scant or no prenatal care, those who tested negative when screened earlier in the pregnancy but who develop risk factors during labor, and those who present in preterm labor before the gestational age at which antepartum screening would have been performed. Perhaps establishing rapid and reliable intrapartum screening performed in the labor and delivery area eventually will result in additional decreases in the incidence of early-onset neonatal GBS infections and in morbidity and mortality associated with those infections.
1. Dillon HC Jr, Gray E, Pass MA, Gray BM. Anorectal and vaginal carriage of group B streptococci during pregnancy. J Infect Dis 1982;145:794–9.
2. Regan JA, Klebanoff MA, Nugent RP. The epidemiology of group B streptococcal colonization in pregnancy. Vaginal Infections and Prematurity Study Group. Obstet Gynecol 1991;77:604–10.
3. McKenna DS, Iams JD. Group B streptococcal infections. Semin Perinatol 1998;22:267–76.
4. Schuchat A. Epidemiology of group B streptococcal disease in the United States: shifting paradigms. Clin Microbiol Rev 1998;11:497–513.
5. Yow MD, Mason EO, Leeds LJ, Thompson PK, Clark DJ, Gardner SE. Ampicillin prevents intrapartum transmission of group B streptococcus. JAMA 1979;241:1245–7.
6. Boyer KM, Gadzala CA, Kelly PD, Gotoff SP. Selective intrapartum chemoprophylaxis of neonatal group B streptococcal early-onset disease. III. Interruption of mother-to-infant transmission. J Infect Dis 1983;148:810–6.
7. Boyer KM, Gotoff SP. Prevention of early-onset neonatal group B streptococcal disease with selective intrapartum chemoprophylaxis. N Engl J Med 1986;314:1665–9.
8. Schrag S, Gorwitz R, Fultz-Butts K, Schuchat A. Prevention of perinatal group B streptococcal disease. Revised guidelines from CDC. MMWR Recomm Rep 2002;51:1–22.
9. Boyer KM, Gadzala CA, Kelly PD, Burd LI, Gotoff SP. Selective intrapartum chemoprophylaxis of neonatal group B streptococcal early-onset disease. II. Predictive value of prenatal cultures. J Infect Dis 1983;148:802–9.
10. Yancey MK, Schuchat A, Brown LK, Ventura VL, Markenson GR. The accuracy of late antenatal screening cultures in predicting genital group B streptococcal colonization at delivery. Obstet Gynecol 1996;88:811–5.
11. Edwards RK, Clark P, Duff P. Intrapartum antibiotic prophylaxis 2: positive predictive value of antenatal group B streptococci cultures and antibiotic susceptibility of clinical isolates. Obstet Gynecol 2002;100:540–4.
12. Davies HD, Miller MA, Faro S, Gregson D, Kehl SC, Jordan JA. Multicenter study of a rapid molecular-based assay for the diagnosis of group B Streptococcus colonization in pregnant women. Clin Infect Dis 2004;39:1129–35.
13. Schachter J, Hook EW, Martin DH, Willis D, Fine P, Fuller D, et al. Confirming positive results of nucleic acid amplification tests (NAATs) for Chlamydia trachomatis: all NAATs are not created equal. J Clin Microbiol 2005;43:1372–3.
14. Schrag SJ, Zywicki S, Farley MM, Reingold AL, Harrison LH, Lefkowitz LB, et al. Group B streptococcal disease in the era of intrapartum antibiotic prophylaxis. N Engl J Med 2000;342:15–20.
15. Schrag SJ, Zell ER, Lynfield R, Roome A, Arnold KE, Craig AS, et al. A population-based comparison of strategies to prevent early-onset group B streptococcal disease in neonates. N Engl J Med 2002;347:233–9.
© 2008 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
Figure. No caption available.