INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Serological screening with either cardiolipin or treponemal antigen tests is crucial for the detection and control of syphilis. In Europe screening is based mainly on treponemal antigen tests such as the Treponema pallidum hemagglutination assay (TPHA) or more recently by enzyme immunoassays (EIAs) (20) which are based on extracts of the Nichols strain of T. pallidum. In contrast, in the United States the rapid plasma reagin cardiolipin antigen test is the recommended screening test in hospital laboratories, while screening by treponemal antigen tests is restricted to blood banks (13). Cardiolipin tests such as the rapid plasmin reagin test, although technically simple, are labor-intensive, may occasionally give false-negative reactions due to the prozone phenomenon (2, 8, 12, 19), and are insensitive with samples from patients with late-stage infection (20) and the results can be difficult to interpret.

In contrast, EIA is ideally suited for the screening of large numbers of specimens because it can be readily automated, the results are read objectively, and reports may be generated electronically, removing any risk of transcriptional errors. Treponemal antigen EIAs are also able to detect syphilis in, and hence assess, patients with all stages of untreated and treated syphilis (21), which is particularly important in countries where the incidence of infectious syphilis is low because the majority of new diagnoses will be in those with late-stage infection (4). Although the United States experienced a high incidence of syphilis during the late 1980s and early 1990s, the level reported in 1996 (4.4 cases/100,000) (3) was close to the lowest rate recorded (3.9 cases/100,000 in 1956 and 1957). In the United States treponemal antigen tests have already been advocated for the screening of certain groups such as psychiatric patients (18) and are recognized as important in determining whether human immunodeficiency virus (HIV)-infected patients have ever been exposed to syphilis (7, 11, 22).

When large numbers of specimens are screened, a high degree of specificity is essential in order to minimize the number of nonproductive tests, i.e., initially reactive sera that are found not to contain antitreponemal antibodies on subsequent confirmatory testing, because confirmatory tests are expensive in terms of both labor and reagents. The use of recombinant T. pallidum antigens in place of a poorly defined mixture of antigens from wild-type T. pallidum which may be contaminated with rabbit testicular components has the potential for improving the specificities of serological tests (10). This report assesses the performance characteristics (specificity and sensitivity) of a new recombinant antigen-based EIA (Syphilis ICE; Murex, Dartford, United Kingdom) when performed in parallel with our existing routine screening test, the Captia SelectSyph-G EIA (Centocor, Malvern, Pa.), which uses wild-type T. pallidum antigen (21).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Sera. A total of 1,184 unselected blood specimens submitted to the STD Diagnostic Laboratory for routine screening for syphilis and a panel of 101 serum specimens that were known to contain antitreponemal antibodies (stored at 70°C) and that were from patients with various stages of syphilis were included in the study. After the specimens had clotted, a sample of serum was removed and was heat inactivated (56°C for 30 min) before testing. Aliquots of sera were stored at 70°C for further analysis in the case of positive screening results.

Serologic methods. Inactivated sera were tested in parallel by the Captia SelectSyph-G EIA (Centocor) and the ICE Syphilis EIA (Murex). Each test was performed according to the manufacturer's specifications and included the appropriate kit controls. Results of the SelectSyph-G EIA were scored as described previously (21): an antibody index (absorbance of test serum/mean absorbance of low-titer positive control) of <0.9 was considered negative, an antibody index of 0.9 to 1.1 was considered equivocal, and an antibody index of >1.1 was considered positive.

Confirmatory testing and classification of sera. Sera reactive on screening by one or both EIAs were retested as appropriate and were subjected to confirmatory testing by a quantitative Venereal Disease Research Laboratory (VDRL) test (Murex), a quantitative TPHA (Fujirebio, Tokyo, Japan), a fluorescent treponemal antibody-absorbed (FTA-abs) test (Mast Laboratories, Merseyside, United Kingdom), and the Captia Syphilis M EIA (Centocor). Sera were classified as negative for antitreponemal antibodies if they were negative by both screening tests. Sera positive on screening were classified as positive for antitreponemal antibodies if the screening test was repeat reactive and confirmatory tests (TPHA and FTA-abs) were positive. The panel of 101 serum specimens containing antitreponemal antibodies was tested by the ICE Syphilis EIA and the SelectSyph-G EIA. These sera had a range of antibody levels and were from patients with different stages of untreated and treated syphilis who had attended the Genitourinary Medicine Clinic in Edinburgh, Scotland. The sera had previously been characterized by a quantitative VDRL test, a quantitative TPHA, the FTA-abs test, and the Syphilis M EIA, and stage of disease and treatment status were determined from a review of the case records. Clinical stage was assessed by standard criteria (note, however, that early latent syphilis is where infection occurred <2 years previously, while late latent is where infection occurred >2 years previously). The majority of patients had been treated within the Genitourinary Medicine Clinic in Edinburgh, and the treatment was well documented. Sera from this panel that were negative by the SelectSyph-G EIA or the ICE Syphilis EIA, or both, were retested as appropriate.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

All of the test runs met the appropriate validation criteria for each kit. The mean cutoff for the ICE Syphilis EIA was 0.32, with a coefficient of variation of 3.7%: the SelectSyph-G EIA cutoff was 0.72, with a coefficient of variation of 17.4%. On routine screening of the 1,184 specimens, 13 (1.1%) were positive or equivocal (antibody index, 0.9) by the SelectSyph-G EIA and 7 (0.6%) were positive or equivocal by the ICE Syphilis EIA. After repeat and confirmatory testing, 4 serum samples were classified as containing antitreponemal antibodies (3 were initially reactive by both the SelectSyph-G and the ICE Syphilis EIAs and 1 was reactive only by the ICE Syphilis EIA; on repeat testing, all 4 serum samples were reactive by both tests and by the TPHA and FTA-abs test); 10 were false positive by the SelectSyph-G EIA (all were reactive by the SelectSyph-G EIA and negative by the TPHA and the FTA-abs test on repeat testing); 1 was false positive by the ICE Syphilis EIA (it was reactive by the ICE Syphilis EIA and negative by the TPHA and the FTA-abs test on repeat testing); two of the serum samples initially reactive by the ICE Syphilis EIA were negative on repeat and confirmatory testing). The distribution of antibody indices for the initial testing of the 1,184 routine serum specimens and the 101 serum specimens known to contain antitreponemal antibodies is presented in Table 1.

After screening by the SelectSyph-G EIA, 0.85% (10 of 1,180) of the serum samples not containing antitreponemal antibodies required further investigation, whereas after screening by the ICE Syphilis EIA, 0.25% (3 of 1,180) required further testing; this is a statistically significant difference (P < 0.05). After repeat testing the specificity was 99.2% (1,170 of 1,180) for the SelectSyph-G EIA and 99.9% (1,179 of 1,180) for the ICE Syphilis EIA (P < 0.02).

The sensitivities of the ICE Syphilis EIA, the SelectSyph-G EIA, and the other tests applied to the 101 serum specimens containing antitreponemal antibodies from our serum bank and the 4 serum specimens containing antitreponemal antibodies detected on screening are given in Table 2.

Equivocal reactions were considered positive reactions in calculating the sensitivity data presented in Table 2. Nine serum specimens containing antitreponemal antibodies were negative and nine were in the equivocal zone on initial testing by the SelectSyph-G EIA, whereas only one was negative and none were in the equivocal zone by the ICE Syphilis EIA (Table 1). The specimen initially nonreactive by the ICE Syphilis EIA was from a patient infected with HIV who had been treated for primary syphilis more than 8 years previously; all other treponemal tests were negative. The difference in sensitivity between the ICE Syphilis EIA, the SelectSyph-G EIA, and the FTA-abs test is statistically significant (P < 0.01), but the difference between the sensitivity of the ICE Syphilis EIA and TPHA is not statistically significant (0.5 > P > 0.3). The poor sensitivity of the VDRL test is in keeping with the large proportion of patients in the study population treated for syphilis. The sensitivities of the ICE Syphilis EIA and the SelectSyph-G EIA (after repeat testing) and other serological tests are given in Table 3 in relation to the stage of disease and treatment status.

The VDRL test had a sensitivity of 67.9% for the detection of untreated infection, although this increased to 90.9% (10 of 11) for the detection of untreated early infection (primary, secondary, and early latent). The main difference between the ICE Syphilis EIA and the SelectSyph-G EIA was in the detection of treated infection; the SelectSyph-G EIA was negative only for one patient with untreated syphilis. The study population imposed an extremely stringent test of sensitivity: 74 (71.2%) of the serum samples were from treated patients, and many of them had low TPHA titers because they had been treated many years previously. The mean period of time since treatment was 107.4 months (range, 1 to 529 months) for 68 of the 74 treated patients for whom the date of treatment was known. Negative reactions by the SelectSyph-G EIA were significantly associated with negative or low TPHA titers and equivocal or negative FTA-abs test scores. The incidence of false-negative SelectSyph-G EIA results was 28% (7 of 25) for specimens with a TPHA titer of 80 compared with 1.3% (1 of 80) for specimens with a TPHA titer of 160 (P < 0.001): the corresponding values for the FTA-abs test were 38.9% (7 of 18) for specimens with an FTA-abs test result of equivocal or negative versus 1.2% (1 of 87) for specimens with an FTA-abs test result of weakly positive or better (P < 0.001).

The details for all sera negative by the SelectSyph-G EIA after repeat testing are given in Table 4. Of the negative specimens for which data are presented in Table 4, three were known to be from HIV-positive patients. The SelectSyph-G EIA was positive for 12 other serum specimens from patients known to be HIV positive and included in the evaluation. For each specimen, however, the ICE Syphilis EIA antibody index was much higher than that given by the SelectSyph-G EIA (Table 5).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The performance of the ICE Syphilis recombinant-antigen test is superior to that of the SelectSyph-G EIA, which uses the Nichols strain of T. pallidum as the source of antigen. The specificity (99.8%) which was determined in an "in-use" evaluation with a wide spectrum of specimens from various hospital wards and clinics including genitourinary-medicine wards and antenatal clinics was comparable to that found by other recombinant-antigen EIAs evaluated with specimens from more restricted populations. An EIA with recombinant antigen TpN44.5 (TmpA) had a specificity of 99.6% when blood from 938 donors was tested (9), while the ELISyph-rG EIA, based on the TpN15 and TpN47 antigens, gave 99.8% specificity when blood from 1,822 blood donors was evaluated (23). The 100% sensitivity of the ICE Syphilis EIA is all the more remarkable given that the population of treponeme-infected patients tested included 70% treated patients, with the mean period of time since treatment being 107.4 months (range, 1 to 529 months), and 30% patients with late latent infection. The TmpA EIA gave a sensitivity of detection of untreated syphilis of 76% for 42 patients with primary infection, 100% for 39 patients with secondary infection, and 98% for 54 patients with early latent infection (9). The results of this test tended to become negative following treatment, which makes the test unsuitable as a marker for past syphilis. Although sera from patients with late latent infection were not included in the evaluation, it is known that antibodies to TpN44.5 are not among the major antibodies detected in patients with late infection (1, 10). The sensitivity of the ELISyph-rG EIA was stated as virtually 100% with a reference panel of 50 "syphilitic samples," although no details as to stage of infection or treatment status were given (23).

The use of the three recombinant antigens (TpN15, TpN17, and TpN47), which are detectable throughout the course of syphilis and which are known to predominate in patients with late latent infection (1, 10, 16), contributes to the ability of the ICE Syphilis EIA to detect all stages of infection. Antibodies to TpN47 are particularly important and were predominant on Western blot analysis of serum from a patient with asymptomatic HIV infection who acquired syphilis: the serological response was abnormal, with antibodies to far fewer treponemal antigens than normally present (17). Gerber and colleagues (5) noted that assays with three antigens combined resulted in improved diagnostic sensitivity: an EIA based on TpN17, TpN44.5, and TpN47 was reactive for 17 of 18 serum specimens from patients with all stages of syphilis, whereas 42 normal human serum specimens were nonreactive.

While the ability of a new test to detect low levels of antibody in patients with adequately treated syphilis may not be significant for the majority of patients, it is useful in providing a complete clinical picture for HIV-infected patients. Particularly with respect to neurologic disease, the clinical manifestations of syphilis and HIV infection are similar; acute meningitis, cranial neuropathies, encephalopathy, myelopathy, and cerebrovascular disease may be features of both infections (15). In some countries, most notably, the United States, a single intramuscular dose of benzathine penicillin has been the preferred treatment for early syphilis. Treatment failures, however, have been reported in HIV-infected individuals (6), and treponemes have been detected in the cerebrospinal fluid of patients treated with this penicillin regimen (14). Because specific treatments for neurosyphilis and HIV encephalopathy exist, it is important for the physician to differentiate between these infections as the cause of neurologic disease in an HIV-infected individual. The baseline screening of individuals with newly diagnosed HIV infection should include serological tests for syphilis to detect not only untreated syphilis but also syphilis that had apparently been treated. The ICE Syphilis EIA is ideally suited to excluding or detecting treponemal infection in HIV-infected patients because it gives a much higher antibody index (absorbance of test serum/kit cutoff) with treponeme-containing sera than the SelectSyph-G EIA. A total of 101 samples containing antitreponemal antibodies gave a high antibody index (>2.5) by the ICE Syphilis EIA, whereas only 13 serum samples tested by the SelectSyph-G EIA gave a high antibody index (P < 0.001). The ICE Syphilis EIA was positive for sera from all 15 HIV-infected patients tested in the study (13 with an index of >2.5), whereas the SelectSyph-G EIA was positive for sera from 12 HIV-infected patients (2 with an index >2.5). We conclude that the high sensitivity and high specificity of the ICE Syphilis EIA combined with its suitability for automation make it an ideal screening test.