550
Value of Confirmation of Chlamydiazyme Enzyme Immunoassay Results in the Detection of Chlamydia trachomatis H. Hallander*, E Jonsson, B. G ~ t r i n
The specificity of an EIA (Chlamydiazyme, Abbott) for detection of Chlamydia trachomatis was evaluated by means ofa monoclonal antibody blocking reagent (Abbott) and a direct fluorescent antibody test (DFA) (Micro Trak, Syva). Of the 12,864 tested specimens from the urethra and cervix, 666 (5.2 %) were positive. When retested with the same technique 625 (4.9 %) were again positive. The result in 514 (4.0 %) specimens was verified by means of the blocking reagent. Of specimens with absorbance values just below the established cut,off limit (0.07-0.114), 31 of 79 (39.2 %) were specifically inhibited, increasing the rate of positive results by 0.25 %. The results confirmed with the blocking reagent were in agreement with those obtained by DFA. The findings show that verification of positive Chlamydiazyme results is strongly advisable, in particular in specimens from the cervix.
The introduction of enzyme immunoassays (EIA) for detection of the chlamydia antigen as an alternative to cell culture has facilitated early recognition of infection. There are however problems concerning the sensitivity and specificity of these tests. Polyclonal antibodies reacting primarily with the common chlamydia lipopolysaccharide antigen are used in the Chlamydiazyme EIA (Abbott Laboratories, USA). False-positive reactions have been reported to a various extent in this EIA. In a review article Stamm (1) summarized the specificity to be between 86 % and 100 % and in another review article Kellogg (2) reported figures between 92.9 % and 98.1%. Recently, Kellogg et al. (3) reported that only 88 % of initially positive Chlamydiazyme results could be confirmed on repeating the test and using a monoclonal blocking antibody. Cross-reactivity National Bacteriological Laboratory, S-105 21 Stockholm, Sweden.
Eur. J. Clin. Microbiol. Infect. Dis.
with other bacteria such as group A streptococci (4), Acinetobacter calcoaceticus (5, 6), Escherichia coli, Gardnerella vaginalis, Neisseria gonorrhoeae and group B streptococci (6) has been shown to occur, particularly if such organisms are present in high numbers. False-positive Chlamydiazyme results have also been reported in assays of urine sediment from patients with bacterial urinary tract infections due to Escherichia coli or Klebsiella pneumoniae (7). Furthermore, faecal specimens are known to give erroneous results (8). An EIA result may however also be recorded as false-positive due to cell culture failure (9). Accurate diagnosis of sexually transmitted diseases is of paramount importance. Reports of false-positive results may have serious legal and social consequences. To improve the specificity of the Chlamydiazyme a monoclonal blocking reagent specific for chlamydia lipopolysaccharide has been developed as a confirmatory assay. In a recent study 11 of 2891 cervical specimens were identified as false-positive by this assay (9). The aim of this study was to evaluate the possibility of reducing false-positive reactions by use of the blocking reagent.
Materials and Methods. The 12,864 specimens tested (approximately 8,000 from the cervix and 4,000 from the urethra) were sent to the Central Microbiological Laboratory of the Stockholm County Council, Sweden, in the period 5 October 1988 to 10 April 1989, for detection of chlamydia antigen. The samples sent from primary health care centres and general practitioners arrived at the laboratory within three days. The material was heterogeneous with regard to patients' ages and symptoms, and different persons were involved in the collection of samples. Specimens were collected using STD swabs and transport kits supplied by Abbott Laboratories. The EIA (Chlamydiazyme, Abbott Laboratories, Chicago, IL, USA) was performed according to the method recommended by the manufacturer. All samples were processed within five days after collection. The absorbance was measured by a Quantumatic spectrophotometer (Abbott). An absorbance value of approximately 0.115 (at 492 nm), which was calculated from the mean of three negative controls + 0.1, was used as the cutoff absorbance value. All specimens with an absorbance value exceeding the cut-off value (666 specimens) or an absorbance value in the range 0.03-0.114 (112 specimens) were retested using the same technique.
Vol. 11, 1992
In all samples with absorbance values exceeding the cut-off value or in the range 0.03-0.114, a murine monoclonal blocking antibody to chlamydia lipopolysaccharide (Abbott) was added as specified by the manufacturer to determine reduction of absorbance. Samples to which specimen dilution buffer had been added were stored at +4 °C for no longer than 24 hours; otherwise samples were stored at -20 °C for a maximum of three days before retesting and addition of the blocking reagent. Inhibition of > 50 % was considered significant. A direct fluorescent antibody test (DFA) (Micro Trak, Syva, USA) was also performed in a subsample of specimens with an absorbance value above 0.03 collected in the period March to April 1989 (200 specimens). An amount of 0.3 ml of the specimen in dilution buffer was centrifuged at 2500 x g for 30 min. The pelleted material was placed on a microscopic slide, air-dried, fixed with methanol and stained according to the manufacturer's directions. The number of elementary bodies was counted using a Leitz epifluorescence microscope (magnification x 1250). Samples for microscopy were given a code number and read by an experienced technician who was unaware of the E I A results. Results andDiscusslon. During the seven-month study period 666 of the 12,864 specimens tested were positive when first tested for Chlamydia trachomatis antigen with the Chlamydiazyme EIA. On retesting only 625 of these specimens were positive, indicating that 6.2 % of the original results might have been false-positive. A similar figure (5.7 % ) has been reported by Kellogg et al. (10). The prevalence of Chlamydia trachomatis positive specimens in the study material was accordingly reduced from 5.2 % to 4.9 %. The majority of initially positive specimens not confirmed as positive on retesting (26 of 41) had low absorbance values (0.115-0.5) but 12 of the 41 specimens had absorbance values above 0.5, seven of them above 2.0. The decrease in absorbance in the repeated test might have been due to technical factors, such as incomplete removal by washing of unbound conjugate in the first round of testing. On confirmatory testing using the blocking reagent, 514 of the 625 positive specimens were verified as being positive. However, 111 of the specimens did not demonstrate a blocking reaction, constituting 17.8 % of the positive specimens and a further 0.9 % of false-positive reactions for all 12,864 samples. The predictive value
551
of a positive test increased with increasing absorbance values (Figure 1) but there were still 10.2 % false-positive results in samples with absorbance values > 0.5. The somewhat lower rate of falsepositive results of 12 % obtained by Kellogg et al. (3) is probably due to the fact that they used more homogenous clinical material in their study. To further analyse the unexpectedly high number of false-positive results a subsample was examined in more detail. Over a period of two months 200 specimens tested with the EIA which had absorbance values > 0.03 were also analysed by the DFA. In 194 cases (97 %) the results agreed with those obtained with the blocking test: 123 specimens were positive in both tests and 71 negative in both tests. Elementary bodies in low numbers (3 to 9) were also found in 3 of 74 negative specimens with 16-35 % inhibition. Another three specimens were negative in the DFA although fulfilling the criteria for specific inhibition. In general there was good correlation between the number of elementary bodies and the absorbance value. The results suggest that absor-
8locked C'/.~ 100
32~Y'?z
90
80 70 60
5o 2~47
40 30 20
10 0
0,03-0,069 0.07-0.085 O.OB~O.114 0.115-0,23 0,231-0,5
l E~ False negative
Cut-off
>0.5
Absorbance (492 nm) ~
False poslllve
Figure 1: Percent of specificallyblocked specimens in different absorbance value groups of 737 positive specimens, 625 with absorbance values > 0.115, and 112 negative specimens with absorbance values of 0.03-0.114. A total of 12,864 specimens was tested.
552
bance values above the cut-off value and not changed by the addition of blocking reagent represent real false-positive reactions. Using the confirmed results as reference the predictive value for an initially positive test would be only 77 % (514/666), or 82 % after retesting, prior to the addition of blocking reagent (514/625) in low prevalence material like this. With respect to the false-positive results, it is of interest that 105 of 109 specimens still positive after the blocking reaction which could be evaluated were collected from the cervix and only four from the urethra, corresponding to 1.3 % and 0 . 1 % of all cervix and urethral samples respectively. The genital tract is often heavily colonized with organisms such as enterobacteria, anaerobes and other species known to cross-react in the Chlamydiazyme assay. Kellogg et al. (3) attributed their false-positive results to vaginal contamination and recommended more careful specimen collection. The blocking reagent also offers the possibility of increasing the sensitivity while maintaining the specificity. Therefore the monoclonal antibody was used in 112 specimens with absorbance values below the given cut-off value (0.115) but above 0.03. Specific inhibition was recorded in 32 specimens (Figure 1). All but one Was found in the interval between 0.07 and the cut-off value, giving 39.2 % (31/79) positive specimens in that range of measurement. The remaining specimen had an absorbance value of 0.069. Thus a cut-off value of 0.07 instead of 0.115 would have increased the overall detection rate by 0.25 % to 4.25 % after it was reduced from 5.2 % to 4.0 % using blocking reagent for verification. Such confirmatory tests are strongly recommended, particularly in low prevalence clinical material such as the present in which a few falsepositive results can have an unacceptable effect on the predictive value for a positive test, 0.77 (or 0.82 after retesting) in our material. Confirmation tests are also recommended by Kellogg et al. (2,
3). Nonculture methods have been advocated for use in screening programmes in view of their favourable sensitivity and specificity compared with culture and their lower costs (11). However, the problems arising from false-positive tests must also be considered when new rapid assays are introduced for decentralized laboratory testing (12).
Eur. J. Clin. Microbiol. Infect. Dis.
References
1. Slamm WE: Diagnosis of Chlamydia trachomatis genitourinary infections. Annals of Internal Medicine 1988, 108: 710-717. 2. Kellogg JA" Clinical and laboratory consideration of culture versus antigen assays for detection of Chlamydia trachomatis from gen!tal specimens. Archives of Pathology and Laboratory Medicine 1989, 113: 453-.460. 3. Kellogg JA, Selple JW, Murray CL, Levisky JS: Effect of endocervical specimen quality on detection of Chlamydia trachomatis and on the incidence of falsepositive results with the Chlamydiazyme method. Journal of Clinical Microbiology 1990, 28: 1108-1113. 4. Po~er K, Sanehez N, Roblin P, McHugh M, Hammerschlag MR: Lack of specificity of Chlamydiazyme for detection of vaginal cblamydiat infection in prepubertal girls. Pediatric Infectious Disease 1989, 8: 358--360. 5. Saikku P, Puolakkainen M, Leinonen M, Nurminen M, Nissinen A: Cross-reactivity between Chlamydiazyme and Acinetobacter strains. New England Journal o[ Medicine 1986, 314: 922. 6. Taylor-Robinson D, Thomas B J, Osborn MF: Evaluation'of an enzyme immunoassay (Chlamydiazyme) for detecting Chlamydia trachomatis in genital tract specimens. Journal of Clinical Pathology 1987, 40: 194199. 7. Demaio J, Boyd RS, Rensi R, Clark A: False-positive Chlamydiazyme results during urine sediment analysis due to bacterial urinary tract infection. Journal of Clinical Microbiology 1991, 29: 1436-1438. 8. Riordan T, Ellis DA, Matlhews PI, Ratcliffe SF: Falsepositive results with an ELISA for detection of chlamydia antigen. Journal of Clinical Pathology 1986, 39: 1276-1277. 9. Moncada J, Schachter J, Bolan G, Engelman J, Howard L, Mushawar I, Ridgway G, Mumtaz G, Stature W, Clark A: Confirmatory assay increases specificity of the Chlamydiazyme test for Chlamydia trachomatis infection of the cervix. Journal of Clinical Microbiology 1990, 28: 1770--1773. 10. Kellogg JA, Seiple JW, Levisky JS: Efficacy of duplicate genital specimens and repeated testing for confirming positive results for Chlamydiazyme detection of Chlamydia trachomatis antigen. Journal of Clinical Microbiology 1989, 27: t218-122I. 11. Estany A, Todd M, Vasquenz M, McLaren R: Early detection of genital chlamydial infection in women: an economic evaluation. Sexually Transmitted Diseases 1989, 16: 21-27. 12. Coleman P, Varilek V, Mushahwar IK, Marchlewicz B, Stafford J, Hansen J, Kurpiewski G, Grier T: TestPack Chlamydia; a new rapid assay for the direct detection of Chlamydia trachomatis. Journal of Clinical Microbiology 1989, 27: 2811-2814.
Value of Confirmation of Chlamydiazyme Enzyme Immunoassay Results in the Detection of Chlamydia trachomatis H. Hallander*, E Jonsson, B. G ~ t r i n
The specificity of an EIA (Chlamydiazyme, Abbott) for detection of Chlamydia trachomatis was evaluated by means ofa monoclonal antibody blocking reagent (Abbott) and a direct fluorescent antibody test (DFA) (Micro Trak, Syva). Of the 12,864 tested specimens from the urethra and cervix, 666 (5.2 %) were positive. When retested with the same technique 625 (4.9 %) were again positive. The result in 514 (4.0 %) specimens was verified by means of the blocking reagent. Of specimens with absorbance values just below the established cut,off limit (0.07-0.114), 31 of 79 (39.2 %) were specifically inhibited, increasing the rate of positive results by 0.25 %. The results confirmed with the blocking reagent were in agreement with those obtained by DFA. The findings show that verification of positive Chlamydiazyme results is strongly advisable, in particular in specimens from the cervix.
The introduction of enzyme immunoassays (EIA) for detection of the chlamydia antigen as an alternative to cell culture has facilitated early recognition of infection. There are however problems concerning the sensitivity and specificity of these tests. Polyclonal antibodies reacting primarily with the common chlamydia lipopolysaccharide antigen are used in the Chlamydiazyme EIA (Abbott Laboratories, USA). False-positive reactions have been reported to a various extent in this EIA. In a review article Stamm (1) summarized the specificity to be between 86 % and 100 % and in another review article Kellogg (2) reported figures between 92.9 % and 98.1%. Recently, Kellogg et al. (3) reported that only 88 % of initially positive Chlamydiazyme results could be confirmed on repeating the test and using a monoclonal blocking antibody. Cross-reactivity National Bacteriological Laboratory, S-105 21 Stockholm, Sweden.
Eur. J. Clin. Microbiol. Infect. Dis.
with other bacteria such as group A streptococci (4), Acinetobacter calcoaceticus (5, 6), Escherichia coli, Gardnerella vaginalis, Neisseria gonorrhoeae and group B streptococci (6) has been shown to occur, particularly if such organisms are present in high numbers. False-positive Chlamydiazyme results have also been reported in assays of urine sediment from patients with bacterial urinary tract infections due to Escherichia coli or Klebsiella pneumoniae (7). Furthermore, faecal specimens are known to give erroneous results (8). An EIA result may however also be recorded as false-positive due to cell culture failure (9). Accurate diagnosis of sexually transmitted diseases is of paramount importance. Reports of false-positive results may have serious legal and social consequences. To improve the specificity of the Chlamydiazyme a monoclonal blocking reagent specific for chlamydia lipopolysaccharide has been developed as a confirmatory assay. In a recent study 11 of 2891 cervical specimens were identified as false-positive by this assay (9). The aim of this study was to evaluate the possibility of reducing false-positive reactions by use of the blocking reagent.
Materials and Methods. The 12,864 specimens tested (approximately 8,000 from the cervix and 4,000 from the urethra) were sent to the Central Microbiological Laboratory of the Stockholm County Council, Sweden, in the period 5 October 1988 to 10 April 1989, for detection of chlamydia antigen. The samples sent from primary health care centres and general practitioners arrived at the laboratory within three days. The material was heterogeneous with regard to patients' ages and symptoms, and different persons were involved in the collection of samples. Specimens were collected using STD swabs and transport kits supplied by Abbott Laboratories. The EIA (Chlamydiazyme, Abbott Laboratories, Chicago, IL, USA) was performed according to the method recommended by the manufacturer. All samples were processed within five days after collection. The absorbance was measured by a Quantumatic spectrophotometer (Abbott). An absorbance value of approximately 0.115 (at 492 nm), which was calculated from the mean of three negative controls + 0.1, was used as the cutoff absorbance value. All specimens with an absorbance value exceeding the cut-off value (666 specimens) or an absorbance value in the range 0.03-0.114 (112 specimens) were retested using the same technique.
Vol. 11, 1992
In all samples with absorbance values exceeding the cut-off value or in the range 0.03-0.114, a murine monoclonal blocking antibody to chlamydia lipopolysaccharide (Abbott) was added as specified by the manufacturer to determine reduction of absorbance. Samples to which specimen dilution buffer had been added were stored at +4 °C for no longer than 24 hours; otherwise samples were stored at -20 °C for a maximum of three days before retesting and addition of the blocking reagent. Inhibition of > 50 % was considered significant. A direct fluorescent antibody test (DFA) (Micro Trak, Syva, USA) was also performed in a subsample of specimens with an absorbance value above 0.03 collected in the period March to April 1989 (200 specimens). An amount of 0.3 ml of the specimen in dilution buffer was centrifuged at 2500 x g for 30 min. The pelleted material was placed on a microscopic slide, air-dried, fixed with methanol and stained according to the manufacturer's directions. The number of elementary bodies was counted using a Leitz epifluorescence microscope (magnification x 1250). Samples for microscopy were given a code number and read by an experienced technician who was unaware of the E I A results. Results andDiscusslon. During the seven-month study period 666 of the 12,864 specimens tested were positive when first tested for Chlamydia trachomatis antigen with the Chlamydiazyme EIA. On retesting only 625 of these specimens were positive, indicating that 6.2 % of the original results might have been false-positive. A similar figure (5.7 % ) has been reported by Kellogg et al. (10). The prevalence of Chlamydia trachomatis positive specimens in the study material was accordingly reduced from 5.2 % to 4.9 %. The majority of initially positive specimens not confirmed as positive on retesting (26 of 41) had low absorbance values (0.115-0.5) but 12 of the 41 specimens had absorbance values above 0.5, seven of them above 2.0. The decrease in absorbance in the repeated test might have been due to technical factors, such as incomplete removal by washing of unbound conjugate in the first round of testing. On confirmatory testing using the blocking reagent, 514 of the 625 positive specimens were verified as being positive. However, 111 of the specimens did not demonstrate a blocking reaction, constituting 17.8 % of the positive specimens and a further 0.9 % of false-positive reactions for all 12,864 samples. The predictive value
551
of a positive test increased with increasing absorbance values (Figure 1) but there were still 10.2 % false-positive results in samples with absorbance values > 0.5. The somewhat lower rate of falsepositive results of 12 % obtained by Kellogg et al. (3) is probably due to the fact that they used more homogenous clinical material in their study. To further analyse the unexpectedly high number of false-positive results a subsample was examined in more detail. Over a period of two months 200 specimens tested with the EIA which had absorbance values > 0.03 were also analysed by the DFA. In 194 cases (97 %) the results agreed with those obtained with the blocking test: 123 specimens were positive in both tests and 71 negative in both tests. Elementary bodies in low numbers (3 to 9) were also found in 3 of 74 negative specimens with 16-35 % inhibition. Another three specimens were negative in the DFA although fulfilling the criteria for specific inhibition. In general there was good correlation between the number of elementary bodies and the absorbance value. The results suggest that absor-
8locked C'/.~ 100
32~Y'?z
90
80 70 60
5o 2~47
40 30 20
10 0
0,03-0,069 0.07-0.085 O.OB~O.114 0.115-0,23 0,231-0,5
l E~ False negative
Cut-off
>0.5
Absorbance (492 nm) ~
False poslllve
Figure 1: Percent of specificallyblocked specimens in different absorbance value groups of 737 positive specimens, 625 with absorbance values > 0.115, and 112 negative specimens with absorbance values of 0.03-0.114. A total of 12,864 specimens was tested.
552
bance values above the cut-off value and not changed by the addition of blocking reagent represent real false-positive reactions. Using the confirmed results as reference the predictive value for an initially positive test would be only 77 % (514/666), or 82 % after retesting, prior to the addition of blocking reagent (514/625) in low prevalence material like this. With respect to the false-positive results, it is of interest that 105 of 109 specimens still positive after the blocking reaction which could be evaluated were collected from the cervix and only four from the urethra, corresponding to 1.3 % and 0 . 1 % of all cervix and urethral samples respectively. The genital tract is often heavily colonized with organisms such as enterobacteria, anaerobes and other species known to cross-react in the Chlamydiazyme assay. Kellogg et al. (3) attributed their false-positive results to vaginal contamination and recommended more careful specimen collection. The blocking reagent also offers the possibility of increasing the sensitivity while maintaining the specificity. Therefore the monoclonal antibody was used in 112 specimens with absorbance values below the given cut-off value (0.115) but above 0.03. Specific inhibition was recorded in 32 specimens (Figure 1). All but one Was found in the interval between 0.07 and the cut-off value, giving 39.2 % (31/79) positive specimens in that range of measurement. The remaining specimen had an absorbance value of 0.069. Thus a cut-off value of 0.07 instead of 0.115 would have increased the overall detection rate by 0.25 % to 4.25 % after it was reduced from 5.2 % to 4.0 % using blocking reagent for verification. Such confirmatory tests are strongly recommended, particularly in low prevalence clinical material such as the present in which a few falsepositive results can have an unacceptable effect on the predictive value for a positive test, 0.77 (or 0.82 after retesting) in our material. Confirmation tests are also recommended by Kellogg et al. (2,
3). Nonculture methods have been advocated for use in screening programmes in view of their favourable sensitivity and specificity compared with culture and their lower costs (11). However, the problems arising from false-positive tests must also be considered when new rapid assays are introduced for decentralized laboratory testing (12).
Eur. J. Clin. Microbiol. Infect. Dis.
References
1. Slamm WE: Diagnosis of Chlamydia trachomatis genitourinary infections. Annals of Internal Medicine 1988, 108: 710-717. 2. Kellogg JA" Clinical and laboratory consideration of culture versus antigen assays for detection of Chlamydia trachomatis from gen!tal specimens. Archives of Pathology and Laboratory Medicine 1989, 113: 453-.460. 3. Kellogg JA, Selple JW, Murray CL, Levisky JS: Effect of endocervical specimen quality on detection of Chlamydia trachomatis and on the incidence of falsepositive results with the Chlamydiazyme method. Journal of Clinical Microbiology 1990, 28: 1108-1113. 4. Po~er K, Sanehez N, Roblin P, McHugh M, Hammerschlag MR: Lack of specificity of Chlamydiazyme for detection of vaginal cblamydiat infection in prepubertal girls. Pediatric Infectious Disease 1989, 8: 358--360. 5. Saikku P, Puolakkainen M, Leinonen M, Nurminen M, Nissinen A: Cross-reactivity between Chlamydiazyme and Acinetobacter strains. New England Journal o[ Medicine 1986, 314: 922. 6. Taylor-Robinson D, Thomas B J, Osborn MF: Evaluation'of an enzyme immunoassay (Chlamydiazyme) for detecting Chlamydia trachomatis in genital tract specimens. Journal of Clinical Pathology 1987, 40: 194199. 7. Demaio J, Boyd RS, Rensi R, Clark A: False-positive Chlamydiazyme results during urine sediment analysis due to bacterial urinary tract infection. Journal of Clinical Microbiology 1991, 29: 1436-1438. 8. Riordan T, Ellis DA, Matlhews PI, Ratcliffe SF: Falsepositive results with an ELISA for detection of chlamydia antigen. Journal of Clinical Pathology 1986, 39: 1276-1277. 9. Moncada J, Schachter J, Bolan G, Engelman J, Howard L, Mushawar I, Ridgway G, Mumtaz G, Stature W, Clark A: Confirmatory assay increases specificity of the Chlamydiazyme test for Chlamydia trachomatis infection of the cervix. Journal of Clinical Microbiology 1990, 28: 1770--1773. 10. Kellogg JA, Seiple JW, Levisky JS: Efficacy of duplicate genital specimens and repeated testing for confirming positive results for Chlamydiazyme detection of Chlamydia trachomatis antigen. Journal of Clinical Microbiology 1989, 27: t218-122I. 11. Estany A, Todd M, Vasquenz M, McLaren R: Early detection of genital chlamydial infection in women: an economic evaluation. Sexually Transmitted Diseases 1989, 16: 21-27. 12. Coleman P, Varilek V, Mushahwar IK, Marchlewicz B, Stafford J, Hansen J, Kurpiewski G, Grier T: TestPack Chlamydia; a new rapid assay for the direct detection of Chlamydia trachomatis. Journal of Clinical Microbiology 1989, 27: 2811-2814.
![[Comparison of polymerase chain reaction and enzyme immunoassay (Chlamydiazyme) in detection of Chlamydia trachomatis from male urethritis].](/assets/img/hold.png)
