Do we need serodiagnosis?
Chlamydial serology is undeniably useful for comparing sample populations for prior or current exposure to chlamydial infection, perhaps as a prelude for a prospective study. There are a large number of studies which show that e.g. patients with tubal factor infertility have a higher prevalence of elevated chlamydial antibody than comparable fertile women. Many gynaecologists wonder whether chlamydial serology could be useful for screening women for so-called ‘silent’ pelvic chlamydial infection. Some even suggest chlamydial serology could be used instead of invasive and costly laparoscopy in the investigation of infertile women. Akande, 2002 states “—- an appropriate Chlamydia antibody titre that would distinguish women at risk of tubal pelvic damage should be determined using diagnostic test analysis and clinical judgement. Identification by serology of women who are likely to have damage would enable these women to undergo a diagnostic test such as laparoscopy sooner, allowing treatment to be provided earlier”. This does not seem a practicable approach because of the following unresolved issues:
1. Which serological test? We have seen that the “gold standard” Micro IF test for chlamydial antibody is not a good test for diagnostic purposes because of its lack of specificity, objectivity, reproducibility and agreed protocol. While expert chlamydiologists might be happy to use the test for comparative epidemiology, most would agree it is quite a different matter to be confident about the meaning of a result in an individual patient. A priori one might expect some of the newer tests with defined antigens to be more specific, but there is little evidence for this at the moment and it is a difficult thing to prove. In studies of the association of antibody to C. pneumoniae with coronary heart disease, the choice of test appears to be very important [Halvorsen et al, 2002; Schumacher et al., 2001]. The newer serological tests based on synthetic or recombinant antigens have largely been evaluated against the classic MIF as gold standard. It is important to remember that no test can perform better than the selected gold standard and, since MIF is a far from perfect test, it may be that the apparent performance of some of the newer tests is constrained by this choice of gold standard. However further evaluation of these new tests and their application is required before they enter routine diagnostic practice.
2. What endpoint? We have already seen from consideration of test endpoints, that sensitivity and specificity are crucially related to the endpoint chosen. Positive predictive value will depend on the specificity of the test plus the actual prevalence of infection in the test population. For a number of reasons, the positive and negative predictive values of even a ‘good’ serological test are likely to be much lower for serology than for good detection assays. There is a lot of disagreement on the serological criteria to be applied to enzyme immunoassay techniques and their association with actual chlamydial infection [Tuuminen et al., 2000].
3. What is the effect of differing antibody and infection kinetics? When individuals acquire a chlamydial infection, there is a lag period before they make an antibody response. Similarly antibody persists for long after an infection is resolved. High or persisting antibody in an individual does not necessarily mean current or persistent infection. [Nor do persisting symptoms]. Persisting or lagging antibody will further confound the positive and negative predictive value of serology. The evidence, such as it is, is that serology and direct evidence of infection are not well correlated [Rabenau et al., 2000]. In female sex workers serologic testing had quite good ability to exclude active infection of the genital tract with negative predictive value of the order of 95% [Comment: it would probably be less reliable in those making a primary antibody response or those less professionally exposed to infection]. However the authors considered that detection of C. trachomatis could only be reliably achieved by nucleic acid amplification assays [Rabenau et al., 2000].
For pelvic inflammatory disease the key question is what are the predictive values for serology in the diagnosis of pelvic inflammatory disease? An interesting study by Cherneskey et al., 1998 compared five different anti chlamydial antibody assays on sera from 14 women with pelvic pain whose de-paraffinised endometrial biopsy tissue was PCR-positive for C. trachomatis IgM MIF 78.6% and 93.6%, and for heat shock protein-60 enzyme immunoassay 42.9% and 100% respectively. The specificity of a commercial recombinant anti-lipopolysaccharide enzyme immunoassay was low. [Given the design of this study it is not possible to conclude what the real false positive and false negative results of decisions based on serology would have been]. DNA, versus 31 such women whose tissue was PCR-negative. Accepting the presence of plasmid DNA as the gold standard, no single test had total diagnostic accuracy. The sensitivity and specificity for antibody measured by whole inclusion fluorescence was 100% and 80.6%; for
Akande et al., 2003 in Bristol, UK compared serum chlamydial antibody titre with tubal status and pelvic findings in 1006 women undergoing laparoscopy for infertility. Not surprisingly they observed a highly significant association between chlamydial antibody status and the likelihood of tubal damage. Women with positive titres were more likely to have pelvic adhesions than tubal occlusion unless titres were very high, when tubal damage was likely to be more severe. It was concluded that chlamydial antibody tests might be useful as a screening test for the likelihood of tubal damage in infertile women and might facilitate decisions on which women should proceed with further investigations without delay. [This was an uncontrolled study which looked at a selected subgroup of infertile women (those undergoing laparoscopy and therefore thought likely to have tubal damage) and which did not examine chlamydial antibody titres in fertile women. Other screening methods for damage were not evaluated]. In contrast, Guerra-Infante et al., 2003 in Mexico in a similar but smaller study of infertile women undergoing laparoscopy concluded that circulating IgG antibody to C. trachomatis was not useful for the identification of patients with periadnexal adhesions as a cause of infertility.
A considerable number of commercial assays based on allegedly more specific antigens are now available. It is important that these should be rigorously evaluated. Mouton et al., 2002 compared the C. trachomatis enzyme immunoassay (EIA), Labsystems (CtL); SeroCT, Savyon (CtS); pELISA, Medac (CtMp); and a reference assay rELISA, Medac (CtMr)) in two study populations. The first consisted of 134 female patients of whom 63% had associated tubal factor infertility. Percent positivity rates for C. trachomatis IgG in tubal obstruction versus non tubal obstruction infertility patients was 41% vs 10%, 57% vs 18% and 55% vs 25% respectively for the CtL, CtS and CtMp tests. The specificity of Ct-specific IgA and IgG in this patient group varied between 92 to 98% and 76 to 90%, respectively. The second group consisted of 107 consecutive gynaecology patients with fertility problems or suspected pelvic inflammatory disease among whom the specificity of the peptide based tests were around 80% – 90% for specific IgA antibody and 75 – 85% for IgG. The negative predictive values exceeded 90%, while the positive predictive values varied from 30% to 47% for IgA antibody and around 30% for IgG. IgG tests added no useful information to that obtained by IgA tests. It was concluded that the synthetic peptide-based EIA tests are able to detect species-specific antibodies and that specific IgA antibodies may be useful in the serodiagnosis of tubal factor infertility caused by C. trachomatis [This paper involves a number of assumptions about the reference sera tested. The high negative predictive values and low positive predictive values suggest that such tests are most useful for identifying infertility patients in whom it is unlikely that infection is a significant factor; the reverse of what is normally attempted]. Prevalence of C. trachomatis antibody was negatively correlated with age, concording with C. trachomatis specificity. The tests were non-reactive with C. pneumoniae antibody. In patients with a polymerase chain reaction positive (n = 324) C. trachomatis infection, the CtMp, CtL and CtS!IgG!IgA [Verkooyen et al., 2002]. tests respectively were positive for specific antibody in 75%, 70% and 68% of cases versus 45%, 38% and 47%, for
In a more extensive study, Jones et al., 2003 compared the performance of a number of commercial tests against their own preferred “gold standard”, whole inclusion immunofluorescence (WIF) [essentially this is a microscope-based indirect immunfluorescence assay using infected chlamydial cells as antigen]. Assays compared with WIF were: SeroCT, C tracho(pep); Medac p-EIA; Vircell and Labsystems C trachomatis IgG EIAs which use a major outer membrane protein-derived antigen; an inactivated organism EIA (Genzyme Virotech EIA); and a genus specific EIA (Platelia Chlamydia IgG). They examined 90 sera from patients presenting with ectopic pregnancies; 187 sera from patients with a variety of types of infertility; 33 sera from cases of PID where a fourfold rise in WIF titre occurred; 90 sera from antenatal clinic subjects and a panel of 36 sera from laboratory-diagnosed cases of C. psittaci or C. pneumoniae infection. The Genzyme Virotech EIA showed the highest rank correlation coefficient (0.82) with WIF, particularly at high WIF titres but showed poor specificity (5.6%; 95% confidence interval (CI), 0.68% to 18.7%), being reactive with 34 of the panel of 36 sera from suspected C. psittaci or C. pneumoniae infections. The MOMP-specific assays had rank correlation coefficients against WIF ranging from 0.70 (Medac p-EIA) to 0.80 (Vircell EIA) and showed high specificity, particularly the Medac p-ELISA (97.2%; 95% CI, 85.5% to 99.9%) where only 1 serum from the C. psittaci / C. pneumoniae panel was reactive [Jones et al., 2003].
Vainas et al., 2003 reported good agreement for C. pneumoniae antibody in cardiovascular disease between a commercial enzyme immunoassay and micro IF, but this is a small study which offers no insights as to whether serology is useful in this context.
Bas et al., 2002 tried to determine the most sensitive and specific method of measuring antibody to C. trachomatis in reactive arthritis. A panel of serum samples were chosen from 17 patients with C. trachomatis reactive arthritis and twenty patients with other inflammatory arthritic conditions not involving chlamydiae. Chlamydial IgG, IgM and IgA antibody was measured by immunoblotting, by various enzyme-linked immunosorbent assays using six synthetic peptides or recombinant antigens and microimmunofluorescence. The best association of sensitivity (76%) and specificity (85%) was obtained when IgG and/or IgA reactivity to two species-specific antigens was determined. These antigens were synthetic peptides, derived from species-specific epitopes in the variable domain IV of the major outer membrane protein (Labsystems, Finland) and recombinant polypeptide encoded by open reading frame 3 of the plasmid (pgp3). For the impact of sensitivity and specificity on predictive values see: worked examples. [ Comment: This was a good study based on a chronic condition where it might reasonably be expected that specific antibody should be associated with infection. Given that the best that could be achieved using multiple tests was a sensitivity of 76%, (which would not give an impressive negative predictive value) and a specificity of 85% (which would give poor positive predictive value) this reviewer concludes, unlike the authors, that serology has little role to play in the laboratory diagnosis of reactive arthritis].
Conclusion: There is inadequate scientific justification for making serious clinical decisions about patient management on the basis of chlamydial serology. New generation recombinant or peptide-based tests for antibody to C. trachomatis are likely to be more specific than tests based on whole chlamydial antigens. Where other diagnostic methods are lacking or it is not possible to obtain relevant samples, the new serology tests with their high negative predictive value may be of use for identifying patients in whom it is unlikely that C. trachomatis infection is playing a role. Given that much better methods are available for detecting current chlamydial infection, clinicians are strongly recommended to focus their efforts on seeking direct evidence of chlamydial infection wherever possible, using the best locally available test, preferably one using nucleic acid amplification. Chlamydial serology continues to have a role in research [Moss & Darougar, 2001].
MeWard September 2003
NEXT: Molecular diagnostics: Nucleic acid hybridization
Akande, V. (2002). Tubal pelvic damage: prediction and prognosis. Human Fertility (Cambridge) 5, (1 Suppl):, S15 – S20.
Akande, V. A., Hunt, L. P., Cahill, D. J., Caul, E. O., Ford, W. C. & Jenkins, J. M. (2003). Tubal damage in infertile women: prediction using chlamydia serology. Human Reproduction 18, 1841 – 1847.
Bas, S., Genevay, S., Schenkel M. C. & Vischer, T. L. (2002). Importance of species-specific antigens in the serodiagnosis of Chlamydia trachomatis reactive arthritis. Rheumatology (Oxford) 41, 1017 – 1020. [ A thorough study by Sylvette Bas embodying newer synthetic or recombinant antigens].
Chernesky, M., Luinstra, K., Sellors, J., Schachter, J., Moncada, J., Caul, O., Paul, I., Mikaelian, L., Toye, B., Paavonen, J. & Mahony, J. (1998). Can serology diagnose upper genital tract Chlamydia trachomatis infections? Studies on women with pelvic pain, with or without chlamydial plasmid DNA in endometrial biopsy tissue. Sexually Transmitted Disease 25, 14 – 19.
Guerra-Infante, F. M., Carballo-Perea, R., Zamora-Ruiz, A., Lopez-Hurtado, M., Flores-Medina, S., Contreras, G. M. (2003). Evaluation of an indirect immunofluorescence assay for detecting Chlamydia trachomatis as a method for diagnosing tubal factor infertility in Mexican women. International Journal of Fertility and Womens Medicine 48, 74 – 82.
Halvorsen, D. S., Borvik, T., Njolstad, I., Gutteberg, T. J., Vorland, L. H. & Hansen, J. B. (2002). Chlamydia pneumoniae_ IgA- and IgG antibodies in young survivors of myocardial infarction. A comparison of antibody detection by a microimmunofluorescence test and an enzyme immunoassay. Journal of Internal Medicine 251, 142 – 147.
Jones, C. S., Maple, P. A., Andrews, N. J., Paul, I. D. & Caul, E. O. (2003). Measurement of IgG antibodies to Chlamydia trachomatis by commercial enzyme immunoassays and immunofluorescence in sera from pregnant women and patients with infertility, pelvic inflammatory disease, ectopic pregnancy, and laboratory diagnosed Chlamydia psittaci / Chlamydia pneumoniae infection. Journal of Clinical Pathology 56, 225 – 229. [ An important study comparing newer commercial tests but marred by a somewhat eccentric choice of WIF as gold standard. WIF was a particular favourite of this group, little used by others, but reportedly correlates well with MIcro IF].
Moss, T. R. & Darougar, S. (2001). Human genital infections with Chlamydia trachomatis – is there a role for serology? In: International handbook of Chlamydia [ed. Moss, T. R.] pp 33 – 47. Euromed Publications Ltd., Haslemere UK, ISBN 1 899015 43 4
Mouton, J. W., Peeters, M. F., van Rijssort-Vos, J. H. & Verkooyen, R. P. (2002). Tubal factor pathology caused by Chlamydia trachomatis: the role of serology. International Journal of STD and AIDS. 13 Suppl 2, 26 – 29.
Nikkari, S., Puolakkainen, M., Narvanen, A., Aakre, O., Toivanen, P. & Leirisalo-Repo, M. (2001). Use of a peptide based enzyme immunoassay in diagnosis of Chlamydia trachomatis triggered reactive arthritis. Journal of Rheumatology 28, 2487 – 2493.
Rabenau, H. F., Kohler, E., Peters, M., Doerr, H. W. & Weber, B. (2000). Low correlation of serology with detection of Chlamydia trachomatis by ligase chain reaction and antigen EIA. Infection 28, 97 – 102.
Schumacher, A., Lerkerod, A. B., Seljeflot, I., Sommervoll, L., Holme, I., Otterstad, J. E. & Arnesen, H. (2001) Chlamydia pneumoniae_ serology: importance of methodology in patients with coronary heart disease and healthy individuals. Journal of Clinical Microbiology 39, 1859 – 1864. Full article
Tuuminen, T., Palomaki, P. & Paavonen, J. (2000). The use of serologic tests for the diagnosis of chlamydial infections. Journal of Microbiological Methods 42, 265 – 279.
Vainas, T., De Graaf, R., Stassen, F. R., Kurvers, H. A., Grauls, G. E., Kitslaar, P. J. & Bruggeman CA. (2003). Chlamydia pneumoniae serology: comparing a commercial enzyme immunoassay and microimmunofluorescence test in patients with cardiovascular disease. APMIS. 111, 363 – 369.
Verkooyen, R. P., Peeters, M. F., van Rijsoort-Vos, J. H., van der Meijden, W. I. & Mouton, J. W. (2002). Sensitivity and specificity of three new commercially available Chlamydia trachomatis tests. International Journal of STD and AIDS. 13 Suppl 2, 23 – 25.
NEXT: Molecular diagnostics: Nucleic acid hybridization