Performance of a commercial polymerase chain reaction test for endocervical Chlamydia trachomatis infection in a university hospital population.

OBJECTIVES: To examine the accuracy of a commercial polymerase chain reaction (PCR) test (Amplicor CTR, Roche Diagnostic Systems, Branchburg NJ) for identification of endocervical chlamydial infections through both laboratory evaluation and among a diverse teaching hospital patient population. METHODS: Testing of reliable threshold inocula and reproducibility were carried out using laboratory stock organisms. Paired endocervical samples from patients with a wide range of indications were tested by PCR and an established culture procedure, and discrepant pairs were further analyzed to determine true results. RESULTS: Laboratory evaluation suggested that one copy of target DNA from a viable organism consistently yielded a positive result, and test reproducibility was very good, with an overall coefficient of variation of 15%. Compared to true results in 1,588 paired clinical samples from 1,489 women with a 10% prevalence of infection, the PCR test and culture yielded respective sensitivities of 87.4% and 78.0%, and negative predictive values of 98.6% and 97.6%. Specificity and positive predictive value for both tests were 100%. Cost per specimen was nearly identical at $18.84 and $18.88 respectively. Polymerase inhibitors and organisms lacking target DNA were not found in false-negative PCR samples. CONCLUSION: This commercial PCR test is accurate, cost-competitive, and much faster than culture for diagnosis of endocervical chlamydia infections in our population of intermediate prevalence of chlamydial infection.

ally active adolescent and young adult women, which is highly effective. 1,11 These serovars are poor antigens, so serodiagnosis has not been helpful. Identification of the organism by more affordable tests such as enzyme immunoassay, lz direct fluorescent antibody staining, 13 Pap test, 4 and DNA probes 15 is relatively insensitive. Culture of the organism is slow, labor intensive, and expensive but has traditionally been taken as the gold standard for laboratory diagnosis of genital chlamydial infection. Recently, DNA amplification tests have emerged as competitors for the preferred method for diagnosis. Given the value of broadly based screening for prevention of genital chlamydial infection and its complications as recommended by the Centers for Disease Control and Prevention (CDC), 6 extensive information on the performance of various laboratory tests for C. trachomatis in geographically and culturally diverse settings is important. We therefore undertook this study to determine the relative accuracy of culture and polymerase chain reaction (PCR)-based testing among our southeastern U.S., hospital-based patient population where chlamydial infections are of intermediate prevalence.

SUBJECTS AND METHODS
Laboratory evaluation of PCR testing for C. trachomarls was undertaken prior to initiation of clinical testing. Using dilutions of low-titer positive control stock employed in our culture system (C. trachomatis serovar E), we calculated the threshold number of target DNA copies needed to consistently yield a positive PCR result. Further, we assessed the reproducibility of the test in our hands among different assays by repetitive testing of a sample that gave results in the quantifiable range of the test, absorbance 0 to 3, in accord with the NCCLS EP5T guideline.
From October 1996 to April 1997, unselected women attending obstetric, gynecologic, and emergency clinics at our institution for whom testing for genital chlamydia was felt by their clinicians to be indicated were candidates for this study. Indications included pelvic pain, cervicitis, presence of other STD, risk factors for STD, symptomatic partner, new obstetric patient screening, asymptomatic sexually active adolescents and young adults, and others. No characteristic excluded patients from participation, and samples were excluded from analysis only if they were not endocervical swabbings or if they were rejected for reasons of improper transport or identification. Chlamydia specimens were obtained after all other endocervical samples were taken. The culture specimen was taken first, rotating a Dacron-tipped, plastic stick swab in the endocervical canal and transporting it in 2M sucrose-phosphate medium at 4C to the laboratory within 24 hr. The PCR specimen was then taken using the same technique; the swab was rotated and squeezed out in the manufacturer's transport medium and discarded. The transport medium was held at room temperature or refrigerated before delivery to the laboratory within 24 hr for processing. Verbal consent was obtained from each patient for the additional endocervical swabbing, as approved by our institutional review board. Results of PCR testing were in no case made available outside the laboratory.
The PCR test used, Amplicor CT (R) (Roche Diagnostic Systems, Branchburg, NJ) was carried out according to the manufacturer's procedure by personnel who had successfully completed the standard three-day training course offered by the manufacturer. Briefly, the test employs biotinylated primers targeting a highly conserved 207 base pair segment in the genome of a cryptic plasmid found exclusively in all serotypes of C. trachomatis. In the presence of thermostable polymerase and deoxyribonucleotide triphosphates, thermocycling reproduces each copy of target DNA to (in theory) 10 9 copies after 30 cycles. Copies are then bound, excess reagent washed out, and avidin-peroxidase added. Biotinylated target DNA in the sample binds the avidin, and colorimetric changes induced by peroxidase, measured spectrophotometrically at 450A, signify the presence of target DNA. A nuance of this system is the addition of uracil Nglycosylase, which destroys amplified DNA under ambient conditions, thus minimizing the risk of contamination of other specimens in the laboratory. All specimens were assayed in duplicate. If absorbance results in both aliquots was ->0.5, the specimen was positive for C. trachomatis; if <0.2, negative. If either aliquot fell in the range ->0.2 to < 0.5, the specimen was assayed again in triplicate and reported as positive if three or more of the five results was ->0.25, negative if three or more were < 0.25. Our culture technique has been previously reported. 17,18 Briefly, shell vials containing a coy-erslip with a monolayer of McCoy cells were inoculated in duplicate with clinical specimen by centrifugation at 1,700g at 35 C for 1 hr, then incubated at 35 C in 5% CO z for 64-68 hours in medium containing cycloheximide to suppress cellular activity. One of the aliquots was then blindly subcultured to another monolayer, which was then treated identically to the primary culture sequence. The other coverslip was stained with iodine and examined at 400 magnification for cellular inclusions of C. trachomatis. A positive resul consisted of one or more typical inclusions in the primary or passage culture. Equivocal specimens and those with few inclusions were independently evaluated by two experienced observers, and both must have interpreted it as positive for a final positive result to be assigned. If either observer interpreted a specimen as negative or if no typical inclusions were seen, a negative result was issued.
Discrepancy analysis followed predetermined rules. For all discrepant pairs, culture and the plasmid-target PCR test were repeated. In addition, among the first 600 samples, PCR using primers targeting DNA coding for the major outer membrane protein (MOMP)of C. trachomatis was conducted. Results of the plasmid-target and MOMPtarget PCR tests agreed in every case, so MOMPtarget PCR testing was discontinued for the balance of the study. True-positive samples were defined as those that were: 1) culture-positive and PCR-positive on initial testing, 2) culture-positive and PCR-negative on initial testing, and 3) culturenegative and PCR-positive on initial testing and positive by repeat culture or PCR testing. Thus, all samples with a positive culture were taken as truepositive regardless of PCR result. True-negative samples were defined as those that were: 1) culture-negative and PCR-negative on initial testing and 2) culture-negative and PCR-positive on initial testing and negative by repeat culture and PCR testing.
Available PCR specimens from culture-positive, PCR-negative pairs were tested for the presence of polymerase inhibitors by repeat PCR testing of the specimen after "spiking" it with a low-titer positive control standard--a negative result from this mixture indicates the presence of polymerase inhibitors in the specimen. Test cost analysis was undertaken by evaluation of direct expenses over a six-month interval for reagents, supplies, and ma-

RESULTS
Positive threshold evaluation of the PCR test employed the low-titer control used in our culture system, which over the years has reliably produced a mean of 3.6 inclusion-forming units/100 pL sample. Through trial and error we found that 1:24 was the greatest dilution of this stock which consistently produced a positive result in the PCR test, which employs a 50-pL sample. It is known that each viable C. trachomatis organism contains about 10 copies of target DNA. Thus, we calculate that approximately one copy of target DNA is needed to consistently yield a positive PCR result (3.6 10 0.5/24). Reproducibility testing employed that 1:24 dilution of stock positive control, and those results are shown in Table 1.
Results from culture and PCR testing were available from 1,588 paired endocervical samples from 1,489 women. Table 2 shows the demograph- ics of the group. Table 3 shows the correlation of initial culture and PCR testing. Results of discrepancy resolution testing of all culture-negative/PCRpositive pairs is shown in Table 4, along with the retesting done for culture-positive/PCR-negative pairs, which was limited because our protocol predetermined that all culture-positive samples were true-positive. Finally, initial results of each test were compared with true results determined by discrepancy resolution. (

DISCUSSION
Our calculation that one copy of target DNA in the specimen will consistently produce a positive result in the Amplicor CT test is in keeping with the mechanisms of PCR technology. However, the calculation assumes the presence of DNA from viable organisms only. The proportion of target DNA in our specimen derived from viable organisms is not known but should be high based on the procedures used to prepare our stock, which exclude cell debris. Because each viable organism in a clinical specimen carries 10 copies of target DNA, we are comfortable that a single organism will yield a positive PCR test under proper conditions. Our reproducibility testing shows excellent consistency of the test and is comparable to results provided by the manufacturer. It should be noted, however, that reproducibiity of this test among a subset of samples with equivocal results has recently been called into question. 19   The population of women we studied is a mixture of high-risk patients and those at lower risk who were subject to screening protocols. This mixture is typical of hospital-based practices. Additional data on patient characteristics were not prospectively collected, as they apparently have little significance with regard to risk for chlamydia genital infection, 9 and no data suggest that they impact test accuracy. The prevalence of cervical infection by C. trachomatis in our population was 10%, intermediate between private office practices and STD clinics. The PCR-based test performed very well in this setting, with results of initial testing in agreement with our established culture technique in 96.5% of specimen pairs. Further, after identification of true results for each specimen by discrepancy analysis, the PCR test yielded sensitivity of 87.4%, specificity of 100%, and negative-and positive-predictive values of 98.6% and 100% ,respectively. The corresponding values for culture were 78.0%, 100%, 97.6%, and 100%. These values, including the trend toward better sensitivity with PCR than culture, are comparable to those found in other studies that also employed high-quality culture techniques, z-z3 We feel that our decision to count all culturepositive pairs as true-positive is well founded. This result. We use iodine stain which enhances test specificitynprobably at the cost of sensitivity zs--when compared to fluorescent antibody stain, because it provides for evaluation of inclusion morphology. Using criteria recommended by the CDC, 18 two experienced observers were required to conclude independently that inclusion location, color, granularity, peripheral clearing, and nuclear indentation were all appropriate for a positive result.
Moreover, as discussed by Green et al., z6 the bias associated with discrepancy analysis-based estimates of PCR test sensitivity is minimized by assigning high specificity to culture results. We were not surprised that reculture of specimens from culture-positive discrepant pairs was often negative, as all of them showed low-titer infectivity initially, and the detrimental effect of an additional freeze-thaw cycle on inclusion count is well known, z7 We believe the insensitivity of direct fluorescent antibody stain and enzyme immunoassay tests make them inappropriate for use in resolving disagreement between culture and PCRbased tests. Ligase chain reaction testing was not available to us at the time of this study.
Inhibitors of the PCR process have been described in cervical specimens at a frequency of 19% by Verkooyen et al., z8 using the method of detection we used. Using more sophisticated methods, Bass et al. z found inhibitors in 4% of their specimens. We found no evidence of PCR inhibitors in the nine culture-positive/PCR-negative samples we tested. C. trachomatis lacking the cryptic plasmid which contains the DNA targeted by Amplicor CT has been described, z9 All eight culturepositive/plasmid PCR-negative samples we tested were also negative by PCR testing with primers for MOMP DNA, indicating that chlamydia lacking plasmids was not the explanation for false-negative PCR results in these specimens. Our culturepositive/PCR-negative specimens may have resulted from improper specimen procurement, transport, or processing. We did not monitor the adherence of clinicians to the specimen procurement procedure. Specimen adequacy has been shown to be an important determinant of the sensitivity of this test. 3 We were unable to assess this factor because of cytolytic agents included in the transport medium.
Our study suggests that the Amplicor CT PCRbased test for endocervical C. trachomatis infection offers accuracy at least equivalent to culture at the same cost, and provides results much more rapidly, in a large population with intermediate prevalence of infection. Where resources provide for it, this PCR test may be the preferred method for laboratory diagnosis of endocervical chlamydial infections.