JOURNAL OF CLINICAL MICROBIOLOGY, Nov. 1990, p. 2473-2476 0095-1137/90/112473-04$02.00/0 Copyright © 1990, American Society for Microbiology

Vol. 28, No. 11

Use of Sequential Enzyme Immunoassay and Direct Fluorescent Antibody Tests for Detection of Chlamydia trachomatis Infections in Women JANE R. SCHWEBKE,l* WALTER E. STAMM,' AND H. HUNTER HANDSFIELD12 Department of Medicine, University of Washington and Harborview Medical Center,' and Seattle-King County Department of Public Health,2 Seattle, Washington 98104 Received 7 June 1990/Accepted 15 August 1990

Endocervical infections due to Chlamydia trachomatis remain difficult to diagnose due to the lack of an inexpensive, rapid, and accurate test. We evaluated an alternative strategy for diagnosis in which initial screening was performed with an enzyme immunoassay (Chlamydiazyme) followed by a direct fluorescent antibody (DFA) test on specimens in which the Chlamydiazyme optical density (OD) reading fell in an intermediate zone. Lowering the Chlamydiazyme OD ratio (specimen to control) used to define a positive test from 1.0 (the ratio suggested by the manufacturer) to 0.3 raised the sensitivity of Chlamydiazyme from 73 to 83%. Confirmation of those specimens having OD ratios of 0.3 to 0.99 by DFA testing increased the specificity of Chlamydiazyme from 95 to 100%. This strategy necessitated performance of the DFA test on 5% of the specimens. Lowering the cutoff OD ratio below 0.3 increased the sensitivity even further but required DFA testing on >25% of the specimens. Use of an adjusted positive cutoif value for defining positive enzyme immunoassays followed by DFA confirmation for intermediate-zone readings may be a feasible approach for some laboratories that lack cell culture facilities.

Chlamydia trachomatis continues to be the major cause of endocervical infections in females. These infections are often asymptomatic and if undetected may progress to cause endometritis and pelvic inflammatory disease, with resulting infertility and ectopic pregnancy (10). Thus far, a national screening program for chlamydial infection has not been implemented, in part because of the lack of a diagnostic test that is easy to perform, highly accurate, and inexpensive. In experienced laboratories, cell culture remains the most sensitive and specific test for chlamydial infection but is expensive and not widely available (1). Direct fluorescent antibody (DFA) testing and enzyme immunoassays (EIAs) have been introduced more recently. The sensitivity and specificity of DFA testing can be excellent in experienced hands, but the test performs less well in laboratories without DFA experience (7). Additionally, the DFA test is labor intensive and not well suited to screening large numbers of specimens. In contrast, EIAs are easily automated and thus excellent for large-scale screening. Unfortunately, the sensitivity of EIA has often been significantly less than that of culture, and its specificity has been suboptimal because of cross-reactive antigens (6, 11, 12). An alternative strategy, to which little attention has been given, is the use of one noncultural diagnostic test to confirm another. This approach has several potential advantages. Initial screening could be performed by EIA, which is easily done on large numbers of specimens, while DFA slides could be used to confirm positive tests for those patients whose EIA reading was within an "intermediate" zone. In order to increase sensitivity of the EIA, this intermediate zone could be defined as a range of optical densities that includes all true-positive results but also some false-positive results. Use of the DFA test for confirmation would retain specificity. To assess the feasibility of this approach, we performed cell *

Corresponding author.

culture, DFA testing, and EIA simultaneously on endocervical specimens from 1,031 women. MATERIALS AND METHODS Subjects. Sexually active female subjects were recruited from three clinics in Seattle, Wash., from November 1987 to September 1988. Thirty-nine subjects were initially excluded because one or more of the three tests was not performed. Two additional subjects were excluded because of uninterpretable DFA slides. Of the remaining 990 women, 471 were tested at the Seattle-King County Health Department Sexually Transmitted Diseases (STD) Clinic, 245 were tested at the Department of Public Health Family Planning Clinic (FP-A), and 274 were tested at a private family planning clinic (FP-B). Examinations at these facilities routinely include the collection of endocervical swabs for isolation of C. trachomatis by cell culture. During the study period, two additional endocervical swabs were collected for DFA tests (MicroTrak; Syva Co., Palo Alto, Calif.) and EIA (Chlamydiazyme; Abbott Laboratories, North Chicago, 111.). Specimen collection. Clinicians at the three clinics were instructed to use a uniform technique for specimen collection. Cervical secretions were first wiped from the external cervical os with a large cotton-tipped swab. Three separate Dacron-tipped plastic swabs were used to obtain endocervical cells by gentle rotation in the endocervical canal for 10 s. The first swab was placed into transport medium for cell culture, the second was placed in the Chlamydiazyme transport medium per the directions of the manufacturer, and the third swab was gently rolled onto a glass slide which was then fixed with methanol for DFA testing. All specimens were refrigerated and transported to the laboratory within 48 to 72 h of collection. After approximately half of the specimens at each site were collected, the order of specimen collection was reversed so that the EIA specimen was collected before that intended for culture. Diagnostic techniques. (i) Cell culture. Cultures for C. 2473




trachomatis were performed with cycloheximide-treated McCoy cells in 96-well microtiter plates as previously described (13). A fluorescein-conjugated genus-specific monoclonal antibody was used to stain inclusions (13). (ii) Chlamydiazyme. Chlamydiazyme was performed as specified by the manufacturer. A confirmatory blocking assay was performed in all initially positive tests to differentiate false-positive from true-positive results (L. Howard, J. Schachter, W. Stamm, and I. K. Mushahwar, Program Abstr. Int. Soc. Sex. Transm. Dis. Res., abstr. no. 19, 1989). In this assay, the cutoff value suggested by the manufacturer for a positive test was used. Additionally, an optical`density (OD) ratio (the OD of the specimen divided by the OD of the cutoff value for that particular run) was determined to equalize possible differences between runs and to facilitate comparability of the data. All specimens were analyzed by the same technologist in one laboratory. (iii) DFA. Slides were processed and stained according to the instructions of the manufacturer. The number of elementary bodies per slide was determined by use of a Zeiss epifluorescence microscope. Slides were screened at a magnification of x40 and confirmed at x 100 and were considered positive if more than two elementary bodies were identified. Clinical and demographic data. Age, race, and marital status were recorded for all subjects. Also noted were the reason for the clinic visit, sexual history, contraceptive method, and use of antibiotics in the preceding 2 weeks. The examining clinicians noted the presence or absence of cervical ectopy, bleeding induced by the endocervical swabs, and endocervical mucopus (2). Statistical methods. The StatPac Gold Statistical Analysis Package was used to determine probability by chi-square analysis and Student's t test where applicable. RESULTS Study populations. Table 1 summarizes the characteristics of the 990 women for whom all three tests were completed successfully. The three clinic populations were similar in mean age and racial composition. The majority of women at the STD clinic had complaints of genital symptoms or a history of sexual contact with a partner having an STD, whereas routine screening exams were more common at the other clinics (P < 0.001). Significantly fewer STD clinic patients than family planning clinic patients used oral contraceptives (28 versus 60%; P < 0.001). Barrier contraceptives were used with equal frequency by patients at all three sites. Prevalence of and predictors for chlamydial infection. The prevalence of chlamydial infection as determined by cell culture was 10.7% at the STD clinic, 7.3% at FP-B, and 5.8% at FP-A. Among all patients combined, the presence of chlamydial infection was correlated by univariate analysis with younger age (P < 0.001), black race (P 0.05), history of contact with a partner having an STD (P < 0.001), easily induced endocervical bleeding (P < 0.001), and the presence of mucopurulent endocervical discharge (P < 0.001). No correlation was found between chlamydial infection and the method of birth control or the presence of cervical ectopy. Comparison of diagnostic tests. The sensitivity and specificity of Chlamydiazyme and DFA testing compared with cell cultures are shown in Table 2. The overall sensitivity of Chlamydiazyme was 73%, but the sensitivity varied by clinic (FP-B, 90%; FP-A, 86%; and STD clinic, 63%). When examined by reason for visit, the sensitivity was 74% for patients reporting symptoms, 58% for those reporting expo=

TABLE 1. Characteristics of study population by clinical Characteristic

Results from: STD clinic FP-A FP-B


(n = 473) (n = 245) (n = 274)

Mean age (yr)





Race (%) White Black

65 27

68 17

75 16

NS 0.001

Reason for visit (%) Symptoms Contact with STD Screening

68 21 10

7 2 88

41 4 53

'0.001 s0.001 '0.001

Mean no. of sexual partners in past 2 mo Birth control method (%) None Oral contraceptives Condom

Diaphragm Prevalence of C. trachomatisc


38 28 13 4



19 60 10 2



il 62 13 8 7.3


'0.001 '0.001 NS NS


Population characteristics limited to patients for whom all three tests were performed. b STD clinic versus FP-A and FP-B. NS, Not significant. '


By culture.

sure to STD, and 87% for those who underwent routine screening. There was no significant difference in the sensitivity of Chlamydiazyme among specimens from women who reported using oral contraceptives versus those from women who reported using any other method of birth control. Collection of the EIA swab after the initial swab for culture did appear to increase the number of false-negative Chlamydiazyme results (3.4% in the first half of the study versus 1.6% in the second half). The initial specificity of EIA was 99.8%; it rose to 100%, however, when the blocking assay was used to confirm all initially positive results. The relationship between the number of inclusion-forming units in culture and the sensitivity of Chlamydiazyme is shown in Fig. 1. Greater numbers of inclusion-forming units were correlated with higher sensitivity of Chlamydiazyme (P < 0.0001 by x2 test for linear trend). The DFA test had an overall sensitivity of 90% compared with culture (Table 2) and did not vary significantly by clinic (STD clinic, 92%; FP-B, 89%; and FP-A, 86%). Approximately 2% of the DFA slides collected at each clinic were uninterpretable. When examined by reason for visit, the sensitivity was 92% for patients reporting symptoms, 82% for those reporting exposure to STD, and 88% for those who underwent routine screening. There were two false-positive DFA tests (specificity = 99.8%). Use of DFA as confirmation for EIA intermediate-zone results. In an attempt to improve the sensitivity of the EIA, the OD reading defining a positive result was lowered. Table 3 demonstrates the range of sensitivity and specificity for the assay when various OD readings were chosen to define an intermediate zone. Although the sensitivity was markedly improved by lowering the OD ratio value to 0.1 or 0.2, the specificity at these levels was such that .27.5% of specimens required DFA confirmation. At an OD ratio of 0.3, confirmatory DFA tests were required for 50 specimens (5%) and seven additional chlamydial infections were diagnosed.


VOL. 28, 1990


TABLE 2. Sensitivity, specificity, and predictive value of Chlamydiazyme and MicroTrak with culture as the standard (n = 990) Result [% (no. correct/no. tested)]


Chlamydiazyme Chlamydiazyme with blocking assay MicroTrak DFA



Positive predictive value

Negative predictive value

73 (61/83) 73 (61/83) 90 (75/83)

99.8 (905/907) 100 (907/907) 99.8 (905/907)

96.8 (61/63) 100 (61/61) 97.4 (75/77)

97.6 (905/927)

Direct effect on patient treatment by using EIA intermediate-zone results with DFA confirmation. A retrospective chart

review was performed for the STD clinic patients with positive Chlamydia cultures to determine whether a positive Chlamydiazyme screening result would have altered patient management. Of the 51 patients with chlamydial infection, 45 charts were available for review. Of these 45 women, 28 (62%) received doxycycline at the initial visit before their Chlamydia diagnostic test results were available, including 13 women who were given the drug for a history of contact with a partner with chlamydial infection or gonorrhea, 9 for a clinical diagnosis of mucopurulent cervicitis based upon physical examination and endocervical Gram stain, 4 for a clinical diagnosis of pelvic inflammatory disease, and 2 for a diagnosis of gonorrhea. Of the 17 patients who did not receive appropriate therapy for Chlamydia at the initial visit, 9 (53%) had a positive Chlamydiazyme result with the cutoff value suggested by the manufacturer. If an intermediatezone OD ratio value of 0.3 had been used, two additional patients would have been correctly diagnosed within 24 to 48 h. DISCUSSION The diagnosis of chlamydial infections in women continues to be problematic. First, infected women are often asymptomatic and thus can be diagnosed only by laboratory screening tests or by contact tracing following identification of infected partners (3). Second, a uniformly accepted definition of cervicitis has yet to be established (2, 5, 9). Except for the most obvious cases in which there is overt endocervical mucopus, an accurate diagnosis requires the aid of a Chlamydia detection method. Third, no ideal detection method currently is available. 1001~
10,000 n=7 n=1 5

Number Inclusion Forming Units

FIG. 1. Number of inclusion-forming units per positive Chlamydia culture specimen correlated with the sensitivity of Chlamydiazyme (confidence interval [CI] = 95%; P < 0.0001 for linear trend).

0.10 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0a

93 89 83 81 81 79 79 78 76 73


Specificity 60.4 77.6 95.3 98.4 98.8 99.3 99.6 99.8 100 100

aCutoff value suggested by manufacturer.

No. of DFA slides to be interpreted

of total) ~~~~~~~~~~(%

613 (61.9) 272 (27.5) 50 (5.0) 22 (2.2) 18 (1.8) il (1.1) 8 (0.8) S (0.5) 3 (0.3) 0 (0)



a smaller laboratory might be able to interpret up to 25% of the slides, thus bringing the sensitivity of Chlamydiazyme close to 90%. In our study, the sensitivity of Chlamydiazyme varied significantly with the clinic site. The reason for this difference is unclear. The intensity of infection could be a factor, since Chlamydiazyme appears to gain sensitivity with increasing amounts of antigen, and increased amounts of antigen are typically found in acutely symptomatic patients (12). The latter were overrepresented in the STD clinic group, yet the distributions of IFUs in culture were equivalent at each clinic. Variation in collection technique among clinics is one possible explanation but seems unlikely since all clinics achieved comparable sensitivity with the DFA test and the latter is an excellent measure of collection technique. A more important variable in our particular study may have been the differences in transport time from the clinics to the laboratory. Ninety-six percent of the specimens from the STD clinic were processed within 24 h of collection, compared with 80% of the specimens from FP-A and 34% from FP-B. Longer transport times may have diminished the sensitivity of culture at the last two sites. Specimens with low inclusion counts would have been falsely negative in culture, thereby artificially elevating the sensitivity of Chlamydiazyme, which is often negative in specimens with few inclusion-forming units. Given this hypothesis, one might also expect to find false-positive DFA tests. Two were found; we suspect that more were not identified because the DFA test and culture are close to one another in sensitivity. Thus, when the culture was falsely negative because of a small number of inclusion-forming units which were at that time nonviable, these same specimens were those that had few elementary bodies and may not have been identified as

positive. False-positive results were previously a major concern when EIA was used for screening low-prevalence populations, even when the cutoff suggested by the manufacturer was used. The addition of a blocking assay to confirm positive results for Chlamydiazyme eliminated false-positives in our study. Others have noted similar improvement in specificity by using the blocking assay (Howard et al., Int. Soc. Sex. Transm. Dis. Res. 1989). False-positive results may be partly the result of improper specimen collection. Excessive amounts of mucus appear to cause false-positive readings, which can be verified as true-negatives by use of the blocking assay (personal observation). It is not known if the blocking assay could serve to differentiate true-positives from false-positives within an intermediate-zone reading. The use of cytobrushes to collect endocervical specimens might also help to increase the sensitivity of EIA by increasing the amount of antigen present (4, 8). In summary, we have demonstrated the sensitivity of the Chlamydiazyme EIA for endocervical specimens to be 73% in a large group of symptomatic and asymptomatic women when it was compared with a sensitive culture system. The addition of the Chlamydiazyme blocking assay eliminated false-positive results. Using the strategy of screening all patients by EIA at lower OD cutoff values resulted in increased EIA sensitivity with decreased specificity. Addition of DFA to the intermediate-zone specimens returned the


specificity for the combined tests to 100%, indicating that the DFA test can be a reliable confirmatory method for intermediate-zone Chlamydiazyme readings. The usefulness of this strategy in a given setting will depend upon available laboratory facilities and the prevalence of infection in the population tested. Further experience with this approach at another testing site would be of interest to assess its feasibility. ACKNOWLEDGMENTS This work was supported by Public Health Service research training grant AI-07140 and program project grant AI-12192 from the National Institutes for Health. We thank Vivien Hanson and the Seattle-King County Family Planning Clinic, Amy Pollack and the Seattle-King County Planned Parenthood Clinics, Victory Murphy for data analysis, Aggie Clark for technical assistance, and Oran Walker for preparing the manuscript. LITERATURE CITED 1. Barnes, R. C. 1989. Laboratory diagnosis of human chlamydial infections. Clin. Microbiol. Rev. 2:119-136. 2. Brunham, R. C., J. Paavonen, C. E. Stevens, N. Kiviat, C. Kuo, C. W. Critchlow, and K. K. Holmes. 1984. Mucopurulent cervicitis-the ignored counterpart in women of urethritis in men. N. Engl. J. Med. 311:1-6. 3. Centers for Disease Control. 1985. Chlamydia trachomatis infection: policy guidelines for prevention and control. Morbid. Mortal. Weekly Rep. 34(Suppl. 3S):53S-74S. 4. Ciotti, R. A., S. J. Sondheimer, and I. Nachamkin. 1988. Detecting Chlamydia trachomatis by direct immunofluorescence using a cytobrush sampling technique. Genitourin. Med. 64:245-246. 5. Katz, B. P., V. A. Caine, and R. B. Jones. 1989. Diagnosis of mucopurulent cervicitis among women at risk for Chlamydia trachomatis infection. Sex. Transm. Dis. 16:103-106. 6. Lefebvre, J., H. Laperrière, H. Rousseau, and R. Massé. 1988. Comparison of three techniques for detection of Chlamydia trachomatis in endocervical specimens from asymptomatic women. J. Clin. Microbiol. 26:726-731. 7. Lipkin, E. S., J. V. Moncada, M.-A. Shafer, T. E. Wilson, and J. Schachter. 1986. Comparison of monoclonal antibody staining and culture in diagnosing cervical chlamydial infection. J. Clin. Microbiol. 23:114-117. 8. Moncada, J., J. Schachter, M. Shipp, G. Bolan, and J. Wilber. 1989. Cytobrush in collection of cervical specimens for detection of Chlamydia trachomatis. J. Clin. Microbiol. 27:18631866. 9. Moscicki, B., M. Shafer, S. G. Millstein, C. E. Irwin, and J. Schachter. 1987. The use and limitations of endocervical Gram stains and mucopurulent cervicitis as predictions for Chlamydia trachomatis in female adolescents. Am. J. Obstet. Gynecol. 157:65-71. 10. Schachter, J. 1978. Chlamydia infections. N. Engl. J. Med. 298:428-435. 11. Smith, J. W., R. E. Rogers, B. P. Katz, J. F. Brickler, P. L. Lineback, B. Van Der Pol, and R. B. Jones. 1987. Diagnosis of chlamydial infection in women attending antenatal and gynecologic clinics. J. Clin. Microbiol. 25:868-872. 12. Stamm, W. E. 1988. Diagnosis of Chlamydia trachomatis genitourinary infections. Ann. Intern. Med. 108:710-717. 13. Stamm, W. E., M. Tam, M. Koester, and L. Cles. 1983. Detection of Chlamydia trachomatis inclusions in McCoy cell cultures with fluorescein-conjugated monoclonal antibodies. J. Clin. Microbiol. 17:666-668.

Use of sequential enzyme immunoassay and direct fluorescent antibody tests for detection of Chlamydia trachomatis infections in women.

Endocervical infections due to Chlamydia trachomatis remain difficult to diagnose due to the lack of an inexpensive, rapid, and accurate test. We eval...
841KB Sizes 0 Downloads 0 Views