Treponemal Tests for Neurosyphilis—Less Accurate Than What We Thought?

Abstract

Treponema pallidum subsp. pallidum (hereafter termed T. pallidum) is the bacterium that causes syphilis. Neuroinvasion by T. pallidum occurs early in the course of disease (1-3) and can lead to asymptomatic meningitis that, untreated, can progress to symptomatic neurosyphilis. Early forms of symptomatic neurosyphilis include meningitis, hearing loss, vision loss, and stroke. Late forms include dementia (general paresis) and sensory ataxia (tabes dorsalis). Cerebrospinal fluid (CSF) analysis is crucial for diagnosis of asymptomatic and symptomatic neurosyphilis, but there is no perfect “gold standard” diagnostic test. Typical CSF abnormalities include mild mononuclear pleocytosis and elevated protein concentration. The CSF-Venereal Disease Research Laboratory (VDRL) test is specific but not sensitive, and particularly in early neurosyphilis, may be nonreactive. The “party line” has been that CSF treponemal antibody tests, while lacking the specificity of the CSF-VDRL, are diagnostically sensitive and can be used to rule out the diagnosis of neurosyphilis in equivocal cases. A systematic review by Harding and Ghanem (4) published in this issue of Sexually Transmitted Diseases challenges this tenet. In their review, the authors assessed studies that described the performance of CSF treponemal tests. They focused on the negative predictive value of these tests and restricted the analysis to studies in which controls were individuals who had syphilis but not neurosyphilis, an approach that is most relevant to clinical practice. Perhaps not surprisingly for a disease that lacks a gold standard diagnostic test, the authors found that the sensitivity of the CSF treponemal tests varied with how neurosyphilis was defined. If CSF-VDRL reactivity was used to establish the diagnosis, the sensitivity of the CSF treponemal antibody tests consistently approached 100%, but if broader criteria were used, particularly including clinical manifestations, the sensitivity was lower and more variable (22.2%-100%). Treponemal antibodies are directed against T. pallidum proteins, as detected by the fluorescent treponemal antibody absorbed (FTA-ABS) test, the T. pallidum particle agglutination (TPPA) test, the T. pallidum hemagglutination (TPHA) test or various enzyme immunassays (EIAs). Another treponemal test included in the analysis, the microhemagglutination-T. pallidum (MHA-TP), is no longer commercially available. The FTA-ABS, TPPA and TPHA tests measure IgG and IgM antibodies to native T. pallidum antigens, while the EIA assays detect IgG and IgM antibodies to recombinant treponemal proteins, usually TpN15, TpN17, and TpN47. For the FTA-ABS, TPPA and TPHA tests, the patient's serum or CSF is mixed with sorbent to remove cross-reactive antibodies to non-pathogenic treponemes. CSF samples that are nonreactive by FTA-ABS test are often reactive when the absorption step is omitted (5). Because EIA does not require the pre-absorption step, it could potentially be a more accurate test. However, there are no currently published studies using EIA on CSF to test this hypothesis. In the studies analyzed by Harding and Ghanem, it is difficult to determine whether the treponemal antibodies measured in the CSF were made intrathecally or in the periphery. Some investigators have argued that CSF:serum treponemal antibody ratios or indicies corrected for CSF and serum albumin or IgG concentrations can differentiate between intrathecal and peripheral antibody production and thus distinguish true CNS infection from passive diffusion of antibody from the peripheral blood (6-8). Sera from patients with reactive CSF FTA-ABS and CSF-TPPA tend to have higher treponemal antibody titers than sera from those with non-reactive tests, suggesting that a reactive CSF FTA-ABS or CSF-TPPA may simply be a surrogate for higher antibody concentration in sera (6, 9, 10). A study of the diagnostic utility of the CSF-TPHA suggested that a CSF titer >1:320 was as good as an elevated antibody index for neurosyphilis diagnosis (8), suggesting that there may be a CSF treponemal antibody concentration above which intrathecal antibody production is more likely. Distinguishing the source of CSF treponemal antibodies is important in assessing the accuracy of CSF treponemal tests. However, Harding and Ghanem did not include studies using such indices in their review. Syphilis and HIV co-infection occurs commonly worldwide (11, 12). Diagnosing neurosyphilis in HIV-infected patients continues to be a challenge because HIV infection itself can cause CSF pleocytosis and elevated protein concentration. Only four of the studies (1, 3, 7, 13) assessed by the authors included HIV-infected patients, and one of the four studies had very small patient numbers (7), making it difficult for the authors to assess the accuracy of CSF treponemal tests in this patient population. This article highlights the challenges clinicians caring for patients with syphilis continue to face – imperfect tests for neurosyphilis that are neither very sensitive nor specific. As the authors point out, while a positive CSF-VDRL “rules in” neurosyphilis, a negative CSF FTA-ABS does not necessarily “rule out” the diagnosis, especially when patients have neurological symptoms or signs. As the incidence of syphilis continues to increase, especially among HIV-infected patients, further work to develop better neurosyphilis diagnostic tests is sorely needed.