Detection of Chlamydia trachomatis and Neisseria gonorrhoeae Infections in North American Women by Testing SurePath Liquid-Based Pap Specimens in APTIMA Assays

MATERIALS AND METHODS

A total of 1,615 female subjects signed consent forms while being enrolled during 2004 at six clinical sites in North America: University of North Carolina (UNC), Chapel Hill, NC; University of Alabama (UAB), Birmingham, AL; St. Joseph's Healthcare (SJH), Hamilton, ON, Canada; University of Oklahoma—Tulsa (UOT), Tulsa, OK; University of Illinois—Urbana (UIU), Urbana, IL; and Planned Parenthood of Minnesota/South Dakota (MNSD), St. Paul, MN. Each patient had a cervical broom collection into a SurePath preservative-fluid tube, which was processed for cytology. The residual specimen was transferred to an APTIMA specimen transfer tube and tested with each APTIMA assay. A CS sample was also collected by using the APTIMA collection kit. For each assay, the results on the SurePath specimen were compared to a subject-infected status where the patient's CS sample was positive by two APTIMA assays (APTIMA COMBO 2 and APTIMA CT and APTIMA COMBO 2 and APTIMA GC). Sensitivity, specificity, and predictive values were computed based on binomial distribution. Fisher's exact test was used to calculate P values, and 95% confidence intervals were calculated. The patient's age, ethnic origin, and reason for clinic visit were recorded. To determine the performance of the APTIMA COMBO 2, APTIMA CT, and APTIMA GC assays on LPT specimens, positives, negatives, false positives, and false negatives were calculated.

RESULTS

Table 1 is a summation of the information gathered on the questionnaire and shows that the overall mean age of the women enrolled in the study was 29.6 years. Patients from the SJH site were younger (19.9 years) and from the UIU site older (37.8 years). Ethnicity or race varied among sites. The UIU and MNSD patients were predominantly white non-Hispanic women (93.1% and 87.1%, respectively), whereas those from UAB were predominantly black non-Hispanic women (91.9%). Women enrolled at the UNC site were 75% black non-Hispanic and 25% white non-Hispanic. At SJH, these percentages were 23.6% and 60%, respectively. The UOT enrollment had a more even distribution of the three predominant combinations. The UAB and UNC women were attending for sexually transmitted disease (STD) screening, whereas the other four study sites reported that most were attending the clinic for an annual examination. Overall, 71.1% (1,148/1,615) of the women were without symptoms. The majority of patients attending for their annual examination were asymptomatic compared to those attending for STD screening.

Table 2 shows the numbers of patients enrolled at each study site, varying from 20 from UNC to 465 at UOT. Of the 1,615 women enrolled, 113 were infected only with C. trachomatis, 15 with both C. trachomatis and N. gonorrhoeae, and 25 with only N. gonorrhoeae. A total of 20 patients were positive in the APTIMA CT assay but negative by APTIMA COMBO 2, and 3 were positive by APTIMA COMBO 2 but negative in the APTIMA CT test. One woman's CS sample was positive in the APTIMA GC test but negative in the APTIMA COMBO 2 assay. Overall, the prevalence of infection was 7.9% (128/1,615) for C. trachomatis (confidence interval [CI], 6.7 to 9.45) and 2.5% (40/1,615) for N. gonorrhoeae (CI, 1.8 to 3.45). The prevalence of C. trachomatis infections ranged from 0.9% at UIU to 17.0% at SJH. N. gonorrhoeae infections were most prevalent at UAB (7.4%) and not prevalent at UIU and UNC.

Table 3 summarizes the results of testing the CS samples for C. trachomatis and N. gonorrhoeae using the APTIMA COMBO 2, APTIMA CT, and APTIMA GC assays, according to the presence or absence of symptoms. Infection with C. trachomatis was diagnosed in 52 (11.1%) of 467 symptomatic patients, compared to 76 (6.6%) of 1,148 patients without STD symptoms (P = 0.003); N. gonorrhoeae was diagnosed in 17 (3.6%) and 23 (2.0%) of symptomatic and asymptomatic patients, respectively (P = 0.075).

The calculation of the agreement of results of testing CS and LPT samples using the APTIMA COMBO 2 assay was performed for C. trachomatis detection and showed positive agreement (PA) of 78.2% (65.0 to 88.2%), negative agreement (NA) of 99.5% (98.3 to 99.9%), and overall agreement (OA) of 97.0% (95.0 to 98.4%) in symptomatic patients, compared to PA of 86.8% (77.1 to 93.5%), NA of 99.5% (98.9 to 99.8%), and OA of 98.6% (97.7 to 99.2%) in asymptomatic patients. The respective results for N. gonorrhoeae detection for the two specimen types with the APTIMA COMBO 2 assay were PA of 88.2% (63.6 to 98.5%), NA of 100% (99.2 to 100%), and OA of 99.6% (98.5 to 99.9%) in patients with symptoms, compared to PA of 95.7% (78.1 to 99.9%), NA of 100% (99.7 to 100%), and OA of 99.9% (99.5 to 100%) in those that were asymptomatic.

The prevalences, sensitivities, specificities, positive predictive values (PPV), and negative predictive values (NPV) for C. trachomatis and N. gonorrhoeae determined by testing the LPT samples in the assays are summarized in Table 4. The APTIMA COMBO 2 assay performed on the LPT samples detected 109 of the 128 CS sample-positive patients (85.2%). The APTIMA CT assay performed on the LPT sample detected 114 of the positives (89.1%). The APTIMA CT assay recorded 19 (1.1%) extra positives and 14 (0.8%) extra negatives. The APTIMA COMBO 2 test detected 8 (0.5%) extra positives and 19 (1.2%) extra negatives. The percent sensitivities of the APTIMA COMBO 2 and APTIMA CT assays for the detection of C. trachomatis-infected women by testing the liquid-based Pap samples ranged from 75% at UOT to 100% at three of the six study sites (data not shown). Table 4 also shows that the specificities of the APTIMA COMBO 2, APTIMA CT, and APTIMA GC assays were very high (98.7 to 100%). The specificities did not vary significantly from site to site (P = 0.003). In testing LPT specimens, the percent sensitivities and specificities, respectively, by site for the detection of C. trachomatis were 100 and 100 at UIU, 100 and 100 at MNSD, 82.3 and 96.9 at SJH, 75.0 and 99.8 at UOT, 86.7 and 99.7 at UAB, and 100 and 100 at UNC using the APTIMA COMBO 2 test. The percent values with the APTIMA CT test were 100 and 99.7, 100 and 99.5, 87.9 and 93.8, 83.3 and 99.5, 90.0 and 98.6, and 100 and 100 at each respective site. Overall, the PPV and NPV were 93.2% and 98.7% using the APTIMA COMBO 2 test and 85.7% and 99.1% when the APTIMA CT assay was used to diagnose C. trachomatis infections. From site to site, the NPV for both tests showed a tight range between 96.9 and 100%, with three of the sites at 100% (data not shown). The three sites with the NPV at 100% also demonstrated their PPV at 100%. UAB and UOT demonstrated strong PPV and NPV, in the mid- to high 90% range. The SJH site recorded the lowest PPV, at 84.8% with the APTIMA COMBO 2 test and 76.3% for the APTIMA CT assay.

Calculations of the sensitivities, specificities, and predictive values for N. gonorrhoeae are also summarized in Table 4. By APTIMA COMBO 2 testing of the LPT sample, the overall prevalence of N. gonorrhoeae infections was 2.3%, ranging from 0% at UIU and UNC to 3.1% at SJH (data not shown), and 92.5% of the CS samples positive for N. gonorrhoeae were detected by both the APTIMA COMBO 2 and APTIMA GC assays. PPV and specificity were 100% overall and at each of the study sites. Sensitivity was 100% at three of the four sites where N. gonorrhoeae was found, with only three extra negatives by APTIMA COMBO 2 and APTIMA GC reported at UAB (sensitivity, 90%). The values reported for the APTIMA GC assay were equally impressive for sensitivity, specificity, PPV, and NPV, with consistent reporting from site to site.

DISCUSSION

To our knowledge, this is the first publication on the use of the APTIMA assays (APTIMA COMBO 2, APTIMA CT, and APTIMA GC) to detect women infected with C. trachomatis and/or N. gonorrhoeae by testing the residuum from a SurePath liquid-based Pap collection. Our study involved a substantial number (n = 1,615) of women from six different geographical regions in North America, with a range of races and ethnicities, seeking either an annual pelvic examination or screening for a sexually transmitted infection. Although the majority of women screened were asymptomatic, 41.1% (69/168) of the infected women had symptoms (Table 3). Performance of the APTIMA assays for C. trachomatis demonstrated very high specificities and NPV (Table 4). The PPV were 93.2% for the APTIMA COMBO 2 assay and 85.7% for the APTIMA CT assay. The NPV for C. trachomatis were 98.7% for the APTIMA COMBO 2 assay and 99.1% for the APTIMA CT assay. One of the study sites accounted for seven of the false positives, suggesting that the results may have been influenced by technical difficulties in processing the samples or collection of the LPT sample. Examination of these hypotheses is under way. These observations may also be a reflection of the exquisite analytical sensitivity of the APTIMA assays (3, 4), which allows for the detection in samples of small amounts of the target nucleic acid, near the assay cutoffs (12). This observation with C. trachomatis has not been observed for N. gonorrhoeae and may be a reflection of the APTIMA COMBO 2 and APTIMA GC assays having lower analytical sensitivities for the gonococcus than for C. trachomatis or of a consistently higher level of N. gonorrhoeae than C. trachomatis in clinical specimens.

Although the number of N. gonorrhoeae infections was lower, both the APTIMA COMBO 2 and APTIMA GC assays were highly sensitive and specific for it. The PPV and NPV ranged from 97.4 to 100% (Table 4). Previous studies to determine whether liquid-based Pap specimens might be suitable for the detection of C. trachomatis or N. gonorrhoeae have used direct fluorescent antibody staining, AMPLICOR PCR, or the ligase chain reaction (LCR) test on PreservCyt ThinPrep samples. A study by Inhorn et al. (8) compared the results of C. trachomatis direct fluorescent antibody staining of ThinPrep LPT samples to the results for CS smears from 636 women and reported PA in 43 (6.8%) and discrepant results in 11 (1.7%). Adjudication of the discrepant results by LCR and PACE 2 nucleic acid hybridization favored the ThinPrep sample (45%) and the CS smear (55%). The sensitivities at the three study sites varied from 62.5% to 100%, but the specificities were consistent across sites. This type of variation from site to site was also observed in our study and emphasizes the need to train both microbiology and cytology technologists in handling the dual-testing protocols. Hopwood et al. (7), testing CS and ThinPrep samples, showed 100% concordance of 19 LCR positives and 562 negatives, which were stable for 5 months. Koumans et al. (9) tested ThinPrep samples from 255 sexually active adolescent women, using LCR for C. trachomatis and N. gonorrhoeae, and also performed other forms of testing on urine and CS samples (LCR for C. trachomatis and N. gonorrhoeae, PCR for C. trachomatis and N. gonorrhoeae, transcription-mediated amplification for C. trachomatis, and culture), which broadened the reference standard for comparison of tests and samples to diagnose infected patients. They reported strong agreement (0.97) for C. trachomatis between LCR on ThinPrep and LCR on CS samples collected into the kit transport medium (kappa, 0.92, where kappa is a statistical measurement of interobserver agreement which compensates and corrects for the proportion of agreement that might be due to chance); for N. gonorrhoeae the agreement was 0.99 (kappa, 0.96). The sensitivity of LCR on the liquid-based Pap sample was higher for detection of C. trachomatis (93%) than for N. gonorrhoeae (81%). In contrast to the ThinPrep studies of Koumans, we found that with SurePath specimens, the N. gonorrhoeae assays were more sensitive (APTIMA COMBO 2, 92.5%, and APTIMA GC, 92.5%) than the C. trachomatis tests (APTIMA COMBO 2, 85.2%, and APTIMA CT, 89.1%). The lower values for C. trachomatis testing of the SurePath liquid-based Pap samples may have been related to low levels of C. trachomatis rRNA being present, which may lead to variable results using multiple tests (13), and to the volume of residuum available for testing. We did not record volumes routinely, but Hawthorne et al. (6) showed that the volume of liquid-based Pap material from a ThinPrep transport tube may impact the number of reactive samples reported. Using COBAS AMPLICOR PCR, they showed that 840 samples with a volume of >5 ml were positive for C. trachomatis in 33 (3.9%) and for N. gonorrhoeae in 8 (0.9%), compared to 80 samples with a volume of <5 ml which were negative for both C. trachomatis and N. gonorrhoeae. Some of these studies have led to Cytyc Corporation seeking FDA approval for AMPLICOR testing of ThinPrep samples, and Gen-Probe, Inc., recently received clearance from the FDA for APTIMA COMBO 2, APTIMA CT, and APTIMA GC testing of ThinPrep samples transferred to an LPT. In a multicenter trial using the APTIMA assays to test ThinPrep liquid-based Pap specimens (Fuller et al., unpublished data), sensitivities and specificities for C. trachomatis (96.7 and 99.3%) and N. gonorrhoeae (92.3 and 99.8%) which were not influenced by cervical symptoms in 1,626 women were found. The present study, using SurePath liquid-based Pap samples for 1,615 women from six centers in North America, showed some variations in sensitivities and specificities among study sites for C. trachomatis testing, but the high predictive values indicate that SurePath liquid-based Pap test residuum could be used to screen women for C. trachomatis and N. gonorrhoeae in the APTIMA assays.