Impact of Patient Characteristics on Performance of Nucleic Acid Amplification Tests and DNA Probe for Detection of Chlamydia trachomatis in Women with Genital Infections

Collection of specimens.Following the history and physical examination, clinicians instructed women to collect a 30-ml first-catch urine specimen for LCx LCR (Abbott Laboratories, Abbott Park, Ill.) and Amplicor PCR (Roche Diagnostic Systems, Branchburg, N.J.) tests. At a subsequent pelvic examination, a urethral swab was collected for tissue culture, and endocervical swabs were obtained in the following order: a swab for Gram-stained smear and culture for Neisseria gonorrhoeae; randomly ordered swabs for PACE2 (Gen-Probe), LCR, and PCR; and a swab for tissue culture. The urethral sample was obtained by inserting the swab 2 to 3 cm into the urethral meatus with a rotary motion, and cervical swabs for each nonculture test were obtained according to the manufacturers' instructions. For collection of the endocervical culture specimen, a cytobrush was used according to each center's usual protocol. Urine was held and transported at 4°C.

Laboratory methods.Research laboratories at each center tested urine specimens using the Amplicor microwell plate PCR test and the LCx LCR test. Urine was processed and tests were performed according to the manufacturers' protocols, except that repeated LCR tests for subjects who had equivocal results on initial LCR testing were performed on specimens that had been frozen at −70°C for up to several weeks. Each center used its own protocol for tissue culture isolation of C. trachomatis. Swabs were transported to the laboratory in sucrose phosphate chlamydia transport medium or Eagle's minimal essential medium in Earle salts, each containing fetal calf serum and antibiotics. Specimens were held at 4°C for a maximum of 18 to 72 h, depending upon the center, or were frozen at −70°C. Culture medium was inoculated onto cycloheximide- and/or DEAE-dextran-treated McCoy cells in 96-well microtiter plates or 1-dram vials and incubated at 35 or 37°C for 48 to 72 h. After incubation, the cells were fixed with methanol or ethanol and stained with locally prepared or commercial major outer membrane protein-specific fluorescein-conjugated antibody reagents. One center also used a lipopolysaccharide-specific antibody stain. Three centers performed a blind passage if no chlamydia inclusions were seen at 48 to 72 h. The centers followed the manufacturers' test kit protocols for commercial tests, except that M4 transport medium was used for the PCR cervical test. For N. gonorrhoeae culture, modified Thayer-Martin medium was inoculated at each site and incubated immediately in a CO₂-enriched environment at 35 to 37°C for up to 48 h. Testing of urine for the presence of blood and leukocyte esterase was performed by dipstick by laboratory technicians upon receipt of urine samples. For both tests, results were scored as negative, trace, or 1+ to 4+; for analysis, ≥1+ was considered positive.

Data analysis.As previously reported for these subjects (2), we compared the use of single test reference standards consisting of a culture or NAAT cervical test result with two-specimen reference standards, which consist of cervical culture or cervical NAAT plus urethral culture or urine NAAT results, respectively. A two-specimen reference standard (patient standard), in which subjects are classified as infected if any reference test at any anatomic site is positive, has been advocated to improve reference standard sensitivity and thereby reduce underestimation of test specificity due to false-negative reference standard results (16). The two-specimen standard can be regarded as a patient standard, while the single reference standard may be regarded as a specimen standard. Because a patient standard may be useful for avoiding bias when comparing performances of cervical and urine tests and to maximize the detection of infected patients, we used this approach to define the presence of chlamydial infection. Since neither LCR nor PCR has been shown to be a superior reference standard, we used LCR as the reference standard NAAT when calculating PCR performance and PCR as the reference standard NAAT when calculating LCR performance (2).

The sensitivity and specificity of each test in the presence or absence of key subject characteristics were calculated. These included age, presence of mucopurulent endocervical discharge, report of concurrent menses, concurrent cervical infection with N. gonorrhoeae, and signs and symptoms of urethral inflammation (dysuria and blood or leukocyte esterase in the urine). Because earlier analyses of these data indicated significant differences in test performance across centers for some tests (2), we used a random-effects model for combining proportions to incorporate between-center differences into variance estimates when combining the results from the five centers. Because our results are essentially descriptive, we did not attempt to control for confounding or to analyze interactions with the research center (18). Finally, because an unamplified DNA probe (PACE2) is widely used in clinical practice, we calculated the relative increase in detection of positives by cervical and urine NAAT using cervical PACE2 as the baseline comparison (see Table 3).

Positivity for C. trachomatis.Of 3,551 women enrolled, the plurality were 20 to 24 years old (median age, 24 years) and black and had been seen in STD clinic settings. Table 1 displays C. trachomatis positivity as measured by a patient standard (any single positive test in a cervical, urethral, or urine specimen), unadjusted for differences in test performance across centers (crude positivity). Chlamydia positivity varied significantly across centers, but at all centers it was highest among adolescents. At all centers, positivity was also high among women with mucopurulent endocervical discharge (30.1% as measured by any positive test) and among women with a positive N. gonorrhoeae culture (37%). The results of gonococcal culture were not available for center C. Women who reported contact with a sex partner with C. trachomatis infection had a chlamydia positivity of 44%, while those who reported contact with a partner with N. gonorrhoeae infection had a chlamydia positivity of 29%.

Positivity for C. trachomatis, stratified by test type and key clinical characteristics and adjusted for variations in test performance across centers, is shown in Table 2. Regardless of age, both LCR and PCR detected significantly more positive cervical tests than PACE2 (P < 0.001 to P < 0.03 for separate comparisons), and LCR detected significantly more positive tests at the cervix than did PCR (P ≤ 0.001). Regardless of the presence of mucopurulent endocervical discharge, both LCR and PCR detected significantly more positives than PACE2. When mucopurulent endocervical discharge was present, detection of positives by cervical LCR did not differ significantly from that by cervical PCR. However, among women without mucopurulent discharge, while cervical PCR still detected more positives than PACE2, cervical LCR detected more positives than cervical PCR (P < 0.001). LCR and PCR detected similar numbers of positives in the urine of infected women, regardless of age or the presence of mucopurulent endocervical discharge.

Among women who reported concurrent menses, LCR detected more positives at the cervix than either PACE2 or PCR (P < 0.05) (Table 2). Detection of positives was not significantly different, however, for PCR and PACE2 in this group. Of note, for all cervical tests, including NAAT, positivity did not differ significantly in women who reported concurrent menses relative to those who did not. Positivity of NAAT cervical tests exceeded that of NAAT urine tests in women who reported menses; however, this increase was not statistically significant. Among women who did not report concurrent menses, cervical LCR and PCR detected more positives than PACE2 (P < 0.001 and P < 0.02, respectively), while LCR detected more positives than PCR (P < 0.001). Among women with N. gonorrhoeae detected at the cervix, cervical LCR detected significantly more positives than PACE2 (P < 0.001); cervical PCR did not, and LCR detected significantly more positives at the cervix than PCR (P = 0.01). Positivity by either cervical NAAT was higher than by PACE2 among women without cervical gonorrhea (for LCR, P = 0.002; for PCR, P = 0.05).

The relative increase in detection of positives by either NAAT performed at the cervix and in urine over cervical PACE2 is shown in Table 3. In general, use of either cervical or urine LCR was associated with the highest relative increases in estimated positivity, particularly among women who might be expected to have lower organism burdens or a lower expected likelihood of chlamydial infection based on the absence of mucopurulent endocervical discharge or age of >25 years. For example, among women without mucopurulent endocervical discharge, the relative increase in positivity of cervical LCR over PACE2 was high (46%) compared to a 35% relative increase among those with such discharge. Similarly, the relative increase in detection by cervical PCR over PACE2 was higher among women ≥20 years old than among younger women (31 versus 19%). The highest relative increases in estimated positivity over PACE2 using urine NAAT were seen with urine LCR, particularly among women reporting dysuria (73%). On average, use of the cervical or urine NAAT resulted in an estimated increase of positivity among all subjects of 33% over PACE2.

Calculated performance of tests by cervical and urine findings.Table 4 presents sensitivities, with 95% confidence intervals, of tests performed at the cervix adjusted for variability among test centers as described previously (2, 18). While sensitivities were generally higher in younger women, these differences did not reach statistical significance. Importantly, concurrent menses had no apparent effect on the calculated sensitivity of any individual test. However, PCR sensitivity was significantly higher among women with mucopurulent endocervical discharge than among those without such discharge (P = 0.004). LCR sensitivity was significantly higher among women infected with N. gonorrhoeae than among those not infected (P = 0.05); indeed, in this subgroup of women, we observed the highest sensitivity of any test studied. Interestingly, the presence of hematuria or a positive urine leukocyte esterase test was associated with somewhat higher estimated sensitivities of cervical PCR, but not of any other cervical test.

The estimated specificities of the different cervical tests are summarized in Table 5. Neither concurrent menses nor coinfection with N. gonorrhoeae at the cervix had a significant effect on the calculated specificity of any individual test. However, while the differences did not reach statistical significance, LCR specificity was lower among women with mucopurulent endocervical discharge than among those without such discharge (P = 0.05) and among women infected with N. gonorrhoeae (P = 0.22). PACE2 was less specific among women ≥20 years old (P = 0.04). The presence of hematuria was associated with higher estimated specificity of cervical LCR, but not of any other cervical test.

The performance of NAAT on urine is summarized in Tables 6 and 7. While both PCR and LCR demonstrated higher sensitivities in the presence of dysuria, mucopurulent endocervical discharge, or hematuria, these differences were not statistically significant. However, the sensitivities of both NAAT were significantly higher among women with a positive urine leukocyte esterase test. The specificities of both tests did not differ by any patient characteristic studied. The sensitivities of NAAT performed on urine differed from those of cervical NAAT only for LCR; cervical LCR was more sensitive than urine LCR among women <20 years old (89 versus 80%, respectively; P = 0.02) and also in women with a negative urine leukocyte esterase test (82 versus 70%; P = 0.01). Urine PCR was less specific than cervical PCR among women who were not menstruating or had no dysuria (99.4 versus 99.7% for both groups; P = 0.04) or who had hematuria (99.3 versus 100%; P = 0.051) or a positive urine leukocyte esterase test (99.2 versus 99.9%; P = 0.03).