Performance of direct fluorescent antibody tests for routine diagnosis of Chlamydia trachomatis in Russian sexually transmitted disease clinics

Abstract

Testing for Chlamydia trachomatis in Russia is usually done by microscopic examination of genital smears stained with fluorescent antibody provided in locally produced kits. The aim was to assess the sensitivity and specificity of such direct fluorescent antibody (DFA) tests compared with a nucleic acid amplification test (NAAT) (ligase chain reaction) to detect C. trachomatis in 171 cervical smears and 201 urethral smears from men. The patients were at high risk of chlamydial infection and had been recruited at three sexually transmitted disease clinics in Moscow. Among women, DFA test sensitivity was 6% (95% CI 0–14) and the specificity was 92% (95% CI 88–97). Among men, the sensitivity was 9% (95% CI 2–16) and the specificity was 90% (95% CI 83–94). Poor DFA test performance was probably due to poor antibody quality and such tests are not adequate for routine examination of populations with either low or high chlamydial prevalence. As there may remain a place for DFA testing where few patients are seen, the Russian Ministry of Health should enforce registration of diagnostic tests, and Russian manufacturers should seek ways of improving DFA test performance. However, the mainstay of testing should depend on NAATs.

Keywords

Chlamydia trachomatis diagnosis fluorescent antibody test ligase chain reaction Russian clinics

Introduction

The Russian Federation has recently experienced major epidemics of sexually transmitted disease (STD), particularly syphilis and gonorrhoea, as well as declining fertility. Between 1990 and 1997, rates of active syphilis notification increased more than 40-fold.¹ Since 1997 there have been explosive HIV epidemics, concentrated among injecting drug users^2,3 with an estimated 860,000 HIV-infected people by the end of 2003.⁴ There is evidence that around half of female drug users engage in sex-work, and that condom usage among these women is low.⁵ Taken together, these factors suggest that there is significant risk of a generalized epidemic of HIV infection in Russia. Interventions to achieve adequate control of sexually transmitted infections (STIs), including chlamydial, at the population level are recommended as a key element of a comprehensive strategy for HIV prevention by UNAIDS.⁴ The total fertility rate in Russia has declined by 25% since 1992 reaching 1.26 per woman in 2004.⁶

Infections with Chlamydia trachomatis are the most common acquired STIs in developed countries, and often lead to serious sequelae including pelvic inflammatory disease and tubal infertility.^7,8 The majority of recently acquired urogenital infections in both men and women are asymptomatic.⁹

Development of laboratory tests for C. trachomatis that are affordable and have adequate performance for routine diagnostic or screening use has been proved to be difficult. While culture was traditionally considered the gold-standard test, studies conducted during the 1980s suggested that direct microscopic identification of C. trachomatis elementary bodies in genital tract smears using fluorescent antibody in a direct fluorescent antibody (DFA) test¹⁰ was more sensitive in expert hands. However, DFA tests are time-consuming, technically demanding and performance may vary with operator and antibody preparation used. Despite these problems, and because labour supply within STD services in Russia has been abundant, DFA tests using Russian-manufactured antibody preparations have, to some extent, remained the standard for the diagnosis of genital chlamydial infection.

A major advance over recent years has been the commercial introduction and widespread adoption in Europe and North America of nucleic acid amplification tests (NAATs) for C. trachomatis, including the polymerase chain reaction (PCR, Roche Amplicor, Roche Diagnostics), ligase chain reaction (LCR, Abbott LCx, Maidenhead, UK), transcription-mediated amplification (TMA, GenProbe AmpCT and APTIMA [Gen Probe San Diego, CA, USA]) and strand displacement amplification (SDA, Becton Dickinson ProbeTec [NJ, USA]). These tests have shown high sensitivity and specificity when used to examine swabs from urethral and cervical epithelium as well as vaginal and urine specimens, despite relatively low organism load in the latter samples^11,12 and are delivered in batch-processing formats suitable for large-scale diagnostic or screening use.

Given the importance of achieving STD control, both for itself, and for the prevention of generalized HIV epidemics, we assessed the performance in routine practice of DFA testing carried out on genital specimens in three Russian STD clinics compared with that of a NAAT (LCR).

Methods

Overview and patients

We recruited men and women over the age of 18 years attending three STD clinics, designated A, B and C, in the Moscow Oblast between March 2000 and March 2001 who were assessed to be at high risk for chlamydial and gonococcal infections. They were tested for C. trachomatis by both a DFA test and a NAAT (LCR). The protocol, instruments and all subsequent amendments were reviewed and ratified by a District Health Authority Ethics Review Board in London, UK, and by the Ethics Review Board of the Central Institute of Skin and Venereal Diseases, Moscow, Russian Federation.

Procedure

A detailed operations manual for the study was prepared and provided to each participating clinic, and staff were given training. Subjects were included who tested negative for gonorrhoea on an initial Gram-stained genital smear and who either had an abnormal clinical examination (e.g. vaginal discharge), were the sexual partners of patients with gonorrhoea, or, in a smear, had evidence of polymorphonuclear leucocytosis, or Gram-negative organisms which were extracellular, or intracellular but of atypical morphology. Subjects were recruited as part of a wider trial of the performance of gonococcal diagnostic tests.

Patients meeting the inclusion criteria were invited to participate by the study doctor, provided with a written information sheet, and asked to sign a consent form indicating their willingness to participate. Consenting patients were then allocated a study number. All subjects were interviewed by the study doctor using a standard proforma to obtain their characteristics and medical and sexual history, and these and the results of the previous and a further clinical examination were recorded on the proforma. Then cervical swabs were obtained from women and urethral swabs from men for a C. trachomatis DFA test, culture for Neisseria gonorrhoeae and LCR assay. All subjects who were regarded as C. trachomatis-positive by the local DFA test were advised to abstain from sexual activity and were treated with doxycycline 100 mg b.i.d. for seven days and further managed according to standard clinical guidelines. Subjects who were found to be Gram-stain or culture-positive for N. gonorrhoeae after collection of specimens for the study were given ciprofloxacin 500 mg orally and doxy-cycline 200 mg b.i.d. for one week.

Specimen collection, processing and microbiological methods

Cervical or urethral specimens were obtained from each subject at the recruitment visit. A disposable loop specimen for a DFA test was applied to a slide and the smear stained and read in Moscow according to the manufacturer's protocol. A specimen for the LCR assay, to be carried out in London, was obtained using a thin LCR swab, placed in transport medium, and refrigerated immediately at +4°C. Such specimens in transport medium were taken on ice, on the same day, to the microbiology laboratory of the Central Institute of Skin and Venereal Diseases. On arrival, these specimens were placed immediately at –80°C. Before transportation to London all specimens were thawed, mixed thoroughly and then the swab squeezed against the side of the tube and discarded. The sample was divided into two equal volumes in 2 mL screw-capped vials and the aliquots refrozen at –80°C. One aliquot was left frozen at –80°C and the other was transported to London. Aliquots were packed in boxes of 100. The vials were kept cold but they thawed during transport. They were, however, refrozen at –80°C within six hours of leaving Russia. All specimens obtained were tested for C. trachomatis using the LCR assay (Abbott Laboratories) according to the manufacturer's instructions. Specimens were categorized as positive, negative or equivocal according to the manufacturer's cut-off value. Specimens providing positive results were retested at a dilution of 1:4 as were those for which there were equivocal results. After this, any equivocal results were considered negative. Box records were kept for all specimens to check for evidence of cross-contamination in transport, but none was found.

Some specimens taken from patients attending the STD clinic of St Mary's Hospital, London, were also examined in London using C. trachomatis antibody that was part of a Russian DFA test.

Statistical methods

Data were analysed using SAS version 8.2. Concordant and discordant positive and negative results were cross-tabulated and sensitivities and specificities calculated. Exact 95% CI on sensitivities and specificities were calculated from exact binomial distributions.

Results

Patient characteristics and chlamydial prevalence

Two hundred and four women and 232 men were recruited to the study. Of the women, 108 (53%) met a single inclusion criterion (6% complaining of an abnormal vaginal discharge, 21% being contacts of men with gonorrhoea and 26% having an abnormal cervical smear); 96 (47%) met two or more inclusion criteria. Of these, 21% were aged 18–19 years; 28% aged 20–24 years and 51% were 25 years or more. Ten women had taken antibiotics in the previous month and four had performed vaginal cleaning in the morning prior to clinic attendance. Seventy-four percent of women complained of genital symptoms and 79% had evidence of an abnormal vaginal discharge on examination. Forty-three percent of women reported their last vaginal sexual intercourse as being within the previous week, and 24% between eight and 14 days prior to recruitment. Twenty percent of women reported more than one sexual partner in the previous week and 77% more than one partner in the previous month. The influence of these factors on the odds of chlamydia-positivity by the LCR among 176 women for whom LCR results were available (prevalence by LCR = 19%; n = 34) was examined by both univariate and multivariate (logistic regression) analysis. Having had more than one partner in the previous week was the only factor significantly associated (adjusted OR = 5.2; 95% CI = 1.5–18.7; P = 0.012) on multivariate analysis.

A total of 232 men were recruited to the study. Of these, 146 (63%) met a single inclusion criterion (9% complaining of an abnormal urethral discharge, 14% being contacts of women with gonorrhoea and 40% having an abnormal Gram-stained urethral smear); 86 (37%) met two or more inclusion criteria. There were 15% aged 18–19 years, 29% aged 20–24 years and 66% were 25 years or more. Seven had taken antibiotics in the previous month. Seventy-eight percent of men complained of genital symptoms and 78% had evidence of an abnormal ure-thral discharge on examination. Thirty-four percent reported last vaginal sexual intercourse within the previous week, and 19% between eight and 14 days prior to recruitment. None reported having more than one new sexual partner in the previous month. The influence of these factors on the odds of chlamydia-positivity by the LCR among 201 men for whom LCR results were available (prevalence by LCR = 33%; n = 66) was examined by both univariate and multivariate (logistic regression) analysis. Recruitment at clinic C was the only factor significantly associated (adjusted OR = 3.1; 95% CI = 1.2–8.1; P = 0.0282) on multivariate analysis.

Performance of DFA tests compared with LCR: patient-based analysis

Specimens for both DFA tests and the LCR assay were available from 171 women and 201 men. Among women the prevalence of C. trachomatis as judged by the LCR assay was 19% (n = 33), and among men 33% (n = 66). All LCR-positive results were ones confirmed at a 1:4 dilution of the original specimen. Numbers of concordant and discordant tests and sensitivities and specificities for DFA tests compared with the LCR (gold standard) are shown in Table 1, separately for each clinic and for all clinics taken together.

Table 1

Sensitivity and specificity of DFA test compared with LCR among subjects by clinic and for all subjects

		Clinic A		Clinic B		Clinic C		All subjects
		DFA–	DFA+	DFA–	DFA+	DFA–	DFA+	DFA–	DFA+
Females	LCR-negative	46	4	54	5	27	2	127	11
	LCR-positive	11	0	10	2	10	0	31	2
		%	95% CI	%	95% CI	%	95% CI	%	95% CI
	Sensitivity (95%CI)	0	(0–29)	16.67	(0–38)	0	(0–31)	6.06	(0–14.2)
	Specificity (95%CI)	92	(85–99)	91.53	(85–99)	93	(84–100)	92.03	(88–97)
	PVP							15.38	(2–45)
	PVN							80.38	(73–86)
		DFA–	DFA+	DFA–	DFA+	DFA–	DFA+	DFA–	DFA+
Males	LCR-negative	31	3	50	7	40	4	121	14
	LCR-positive	7	1	23	2	30	3	60	6
		%	95% CI	%	95% CI	%	95% CI	%	95% CI
	Sensitivity (95%CI)	12.5	(0–35)	8	(0-19)	9.09	(0-19)	9.09	(2.2–16)
	Specificity (95%CI)	91.18	(82–100)	87.72	(79-96)	90.91	(82–99)	89.63	(83–94)
	PVP							30.00	(11–54)
	PVN							66.85	(59–74)

DFA = direct fluorescent antibody; LCR = ligase chain reaction; PVN = Predictive value negative; PVP = Predictive value positive

The sensitivity and specificity of DFA tests among women were: 0% (95% CI: 0–29) and 92% (95% CI: 85–99), respectively, at clinic A; 17% (95% CI: 0–38) and 92% (95% CI: 85–99) at clinic B; and 0% (95% CI: 0–31) and 93% (95% CI: 88–97) at clinic C. Among women recruited from all clinics the overall sensitivity of DFA tests was 6% (95% CI: 0–14) and the specificity was 92% (95% CI: 88–97).

The sensitivity and specificity of DFA tests among men were: 13% (95% CI: 0–35) and 91% (95% CI: 81–100), respectively, at clinic A; 8% (95% CI: 0–19) and 88% (95% CI: 79–96) at clinic B, and 9% (95% CI: 2–16) and 90% (95% CI: 83–94) at clinic C. Among men recruited from all clinics, the overall sensitivity of DFA tests was 9% (95% CI: 2–16) and the specificity was 90% (95% CI: 83–94).

We compared the distribution of LCR optical density (OD) readings between LCR-positive results that were DFA-positive (concordant positives) and LCR-positive results that were DFA-negative (discordant positives) separately for men and women. For both sexes, concordant-positive results had a higher median OD reading than discordant-positives, and concordant-negatives had a lower median OD reading than discordant-negatives, but the differences were not statistically significant by Kruskall-Wallis χ² analysis. We also examined differences in subject characteristics according to test concordance separately for men and women. No statistically significant differences were found by either univariate or multivariate analysis between concordant- and discordant-positives, or between concordant- and discordant-negatives for either sex. Among 31 women who had positive LCR results but negative DFA results, LCR results for a second specimen were available in 30 cases and, of these, 22 (78%) were positive and eight were negative. Among 60 men who had positive LCR results but negative DFA results, LCR results for a second specimen were available in 45 cases and, of these, 41 (91%) were positive and four were negative.

We calculated predictive values of positive and negative tests for different levels of true chlamydial prevalence among groups of women to be tested and 95% CI for these (Figure 1 – top panel). We also calculated the expected numbers of true- and false-positives and -negatives per 100 women tested, at different group prevalences (Figure 1 – bottom panel). The prevalence of chlamydial infection may be expected to be around 10% in sexually active groups of young women. At a prevalence of 11%, the DFA test would yield one true-positive, seven false-positive, 82 true-negative and 10 false-negative results per 100 tested. At a prevalence of 41%, that might be expected in high-risk groups such as commercial sex workers, the DFA test would yield two true-positive, five false-positive, 54 true-negative and 43 false-negative results per 100 tested.

Figure 1

Predictive values of positive and negative tests and expected number of true- and false-positives and -negatives per 100 tested at different prevalences among females

Examination of Russian C. trachomatis fluorescent antibody in London

Genital tract specimens taken from patients in London were used. Thirty specimens that were shown to be C. trachomatis-positive by both the LCR assay and the MicroTrak DFA test were examined by a DFA test using a Russian propriety C. trachomatis antibody. Because of intense nonspecific background staining, it was not possible to detect typical fluorescing elementary bodies with any certainty.

Discussion

The findings suggest that the sensitivity of DFA tests for C. trachomatis in routine use in the Russian Federation, using locally produced antibody preparations, and in the hands of personnel from three clinics in the Moscow Region, was <10% for both men and women, with a specificity for both sexes of about 90%. There are at least two possible explanations for these findings. First, the LCR assay may be responsible for false-positive results as these were known to occur, particularly towards the end of its commercial existence.¹³ However, of positive specimens retested at a 1:4 dilution, all were confirmed as positive. Furthermore, specimens that had not been used completely for testing, from 172 men, were refrozen at –80°C and tested again four years later in a PCR assay, by courtesy of Dr Jorgen S Jensen (Copenhagen). There was excellent concordance between the results of the first and second examinations. In addition, 72% of LCR-positive/DFA–negative results for women and 91% of similar results for men were positive by the LCR assay on a specimen taken on a different day and transported separately. Even if those subjects whose second specimen tested negative were reclassified as being in the LCR-negative group, the sensitivity would only reach 12.5% for women and 9.8% for men. Moreover, there was no evidence of contamination by analysis for clustering of positive results among specimens transported in the same boxes. We believe, therefore, that errors in conducting the LCR assay or cross-contamination among specimens are unlikely to have influenced the results significantly.

Secondly, and probably the main reason for the discrepancy between the LCR assay and DFA test results is the poor specificity of the locally produced antibody preparation. This was amply demonstrated in London where smears stained with a Russian proprietary antibody preparation were very difficult to read because of high levels of background staining. Undoubtedly, this must have hampered staff in the Russian Federation, particularly where specimens contained small numbers of chlamydial elementary bodies. Suffice to say, the performance of the DFA test was consistently poor in the three clinics studied. Estimates of sensitivity and specificity make this test inadequate for screening either low prevalence groups, where the vast majority of positive results will be false and 90% of true-positive results will be missed, or high prevalence groups, where almost 50% of negative results will be false. Thus, the test fails to detect the majority of cases of true infection, and wrongly identifies many individuals to be infected, who are actually not. It is interesting to know that others in Russia (St. Petersburg) have been disappointed with the sensitivity of Russian DFA tests when compared with an in-house PCR.¹⁴

We used the LCR assay because it was in routine use in London at the time and was a test with which we were familiar. The fact that the LCR assay is no longer commercially available does not diminish the significance of the results. A similar outcome would have been seen if another NAAT, currently available, of similar high sensitivity had been used. As far as the DFA test is concerned, it is not the one to be used for routine testing on a large scale, although it may still be valuable in situations where small numbers of specimens are to be tested. This being so and given the major epidemics of STDs and HIV that are currently occurring in the former Soviet Union, the importance of chlamydial infection in the pathogen-esis of female infertility, and the concerns within Russia about the decline in fertility, there is a need for the Russian Ministry of Health to require registration of DFA tests based on systematic evaluation of performance, restrict use of unregistered tests, and for Russian manufacturers to seek ways of improving test performance. This is a plea endorsed by other Russian investigators¹⁴ and not to be seen as encouragement for widespread use of DFA testing. Indeed, the results overwhelmingly indicate that the focus of attention should be on the use of NAATs.

Footnotes

Acknowledgements

The authors thank the Wellcome Trust for the funding kindly provided to support this study and Natasha Anwar for technical assistance.

References

Tichonova

, Borisenko

, Ward

, Meheus

, Gromyko

, Renton

. Epidemics of syphilis in the Russian Federation: trends, origins, and priorities for control. Lancet 1997; 350: 210–3.

Rhodes

, Ball

, Stimson

. HIV infection associated with drug injecting in the Newly Independent States, eastern Europe: the social and economic context of epidemics. Addiction 1999; 94: 1323–36.

Rhodes

, Lowndes

, Judd

. Explosive spread and high prevalence of HIV infection among injecting drug users in Togliatti City, Russia. AIDS 2002; 16: F25–31.

UNAIDS. 2004 Global Report on the Global HIV/AIDS Epidemic: 4th Global Report. Geneva: UNAIDS; 2004.

Benotsch

, Somlai

, Pinkerton

. Drug use and sexual risk behaviours among female Russian IDUs who exchange sex for money or drugs. Int J STD AIDS 2004; 15: 343–7.

World Health Organisation. Core Health Indicators - Russia; 2004. See http://www.who.int/whosis/en (last accessed October 2008).

Fenton

, Lowndes

. Recent trends in the epidemiology of sexually transmitted infections in the European Union. Sex Transm Infect 2004; 80: 255–63.

Westrom

. Incidence, prevalence and trends of acute pelvic inflammatory disease and its consequences in industrialized countries. Am J Obstet Gynecol 1980; 138: 880–92.

Horner

, Boag

. 2006 UK National Guideline for the Management of Genital Tract Infection with Chlamydia trachomatis. BASHH 2006. See http://www.bashh.org/guidelines/2006/chlamydia_0706.pdf.

10.

Thomas

, Evans

, Hawkins

, Taylor-Robinson

. Sensitivity of detecting Chlamydia trachomatis elementary bodies in smears by use of a fluorescein labelled monoclonal antibody: comparison with conventional chlamydial isolation. J Clin Pathol 1984; 37: 812–6.

11.

Hay

, Thomas

, Horner

. Chlamydia trachomatis in women: the more you look, the more you find. Genitourin Med 1994; 70: 97–100.

12.

Michel

, Sonnex

, Carne

. Chlamydia trachomatis load at matched anatomic sites: implications for screening strategies. J Clin Microbiol 2007; 45: 1395–402.

13.

Mallinson

, Hopwood

, Mutton

. Resolution of the recent performance problem of Abbott LCx Chlamydia trachomatis assay. Issues of repeat testing for confirmation of chlamydial infection. Sex Transm Infect 2002; 78: 225–6.

14.

Shalepo

, Savicheva

, Shipitsyna

, Unemo

, Domeika

. Diagnosis of Chlamydia trachomatis in Russia-in-house PCR assays may be effective but overall optimization and quality assurance are urgently needed. APMIS 2006; 114: 500–7.