Skip to main content

Table 1 Main Results. Diagnostic methods for neurosyphilis and their performance

From: Diagnostic tools for neurosyphilis: a systematic review

Type of diagnostic method Test Performance (%) Gold standard used to define neurosyphilis Sample size (n) HIV (+)
Commentaries Ref.
Sensitivity Specificity NS (+) NS (−) Control
Clinical Neurological symptoms 46% 33% CSF-VDRL (+) OR
a. CSF-WBC > 5 cells/μL OR
b. CSF-protein > 450 mg/L OR
c. Neurological symptoms without other known cause
50 50 0 0% Case-control study that evaluates the usefulness of TPPA as a diagnostic tool, either alone or associated with other criteria. Consider this test when there is clinical suspicion and negative CSF-VDRL. Possible selection bias due to retrospective inclusion of patients who underwent lumbar puncture. Another limitation is that some of the diagnostic tests evaluated are also included in the gold standard employed by the study, thus generating performance analysis confusion. 15
Neurological symptoms AND
CSF-protein > 497 mg/L AND CSF-WBC > 3 cells/μL
89% 98%
Neurological symptoms AND
CSF-protein > 497 mg/L AND
CSF-WBC > 3 cells/μL
CSF-TPPA > 1:160
92% 40%
Laboratory (CSF) CSF-protein > 497 mg/L 54% 85%
CSF-WBC > 3 cells/μL 48% 82%
CSF-TPPA > 1:160 90% 84%
CSF-TPPA 1:80 95% (I)
76% (II)
68% (III)
Three gold standards used:
II. T. pallidum RT-PCR (+)
III. New loss of vision or hearing
105 86 0 78% Evaluated the accuracy of CSF-TPPA, by comparing it with the sensitivity of CSF-FTA-ABS in a first group (n = 191), and with the specificity of VDRL in a second group (n = 380). A reported limitation is that the sample does not represent the population of patients with syphilis. Most patients were HIV (+). We considered that the clinical definition (III) of NS lacked the precision to evaluate any diagnostic test. 16
CSF-TPPA 1:640 97% (I)
94% (II)
93% (III)
120 260 0
Reactive CSF-FTA-ABS 89% 22% ITPA index (TPPA CSF/serum ratio) > 2 AND:
b. CSF-RPR > 1:1 OR
c. CSF pleocytosis > 4 cells/μL OR
d. CSF-protein > 500 mg/L OR
e. Albumin quotient (CSF/serum) > 7.8
38 29 0 52% The study correlates pleocytosis and albumin quotient with NS (+) (independent of HIV co-infection). Highlights the importance of lumbar puncture in diagnosing asymptomatic patients, especially in the HIV (+) population. There is a possible selection bias due to the high clinical suspicion in the patient’s inclusion in the study. 17
Reactive CSF-RPR 21% 97%
Reactive CSF-RPR 100% 100% Clinical suspicion, serological treponemal reactive test AND
a. CSF-VDRL (+) OR
b. CSF-WBC > 5 cells/μL OR
c. CSF-protein > 450 mg/L
NS confirmed if the CSF-VDRL is reactive.
NS suspected if it is nonreactive.
(confir-med NS)
(sus-pected NS)
50 Not reported Compared the performance of different treponemal tests (RPR and USR) with each other using CSF-VDRL as a standard. There was perfect qualitative agreement (kappa value = 1) between evaluated tests and VDRL; sensitivity and specificity were both 100%. These values should be understood as evidence of diagnostic equivalence between these tests and the standard (VDRL), which has its own limitations. Considering this NS definition, USR and RPR are as good as VDRL to differentiate between confirmed and suspected NS. The study did not report HIV status and only included patients with neurological symptoms. 18
Reactive CSF-USR 100% 100%
Reactive CSF-VDRL 54% (I)69% (II) 75% (I)73% (II) The authors mentioned
CSF-VDRL as gold standard (a), but used the following definitions for the performance analysis:
I. CSF-WBC > 20 cells/μL (regardless of other variables)
II. Vision or hearing loss
(regardless of other variables)
54 (a)
152 (I)
145 (II)
163 (a)
65 (I)
72 (II)
0 70% Concludes that the specificity of CSF-SYPHICHECK with cutoff, and sensitivity without cutoff perform similar to CSF-VDRL and remarks that titers rapidly normalize after treatment. Reports impaired patient humoral response due to high prevalence of HIV coinfection. There was no comparison with healthy or control patients. Definitions used for evaluation were not justified with references and we considered them to be imprecise for test performance evaluation. 19
Reactive CSF-FTA-ABS 70% (I)
81% (II)
54% (I)51% (II)
CSF-SYPHICHECK 62% (I)64% (II) 57% (I)53% (II)
CSF-SYPHICHECK 1:4 37% (I)44% (II) 81% (I)79% (II)
CSF-VDRL 85% 100% a. Two reactive/positive tests (regardless if treponemal or nontreponemal) OR
b. Reactive PCR.
18 0 14 38% Among study limitations were the small sample size and the fact that the tests being evaluated were used as diagnostic criteria for NS (+), which increased its accuracy. Not all cases were tested with all methods due to the small volume of some specimens. 20
CSF-MAXISYPH 100% 100%
CSF-INNO-LIA 92% 100%
CSF-TPPA 83% 100%
Laboratory (blood) RPR   1:4 77% 80% I. Confirmed NS:
II. Probable NS:
Syphilis of any stage with:
a. CSF-protein > 500 mg/L OR CSF-WBC > 10 cells/μL (without another cause) AND
b. Signs/symptoms consistent with NS (without another cause).
191 179 0 0% Test performances were evaluated for NS (+) general detection (I OR II being the exposed values) and discriminating between confirmed (I) and probable (II), with a better accuracy being described for (I). A multivariate analysis found another biomarker, plasmatic CK-MB. The study included only HIV (−) patients with neurological symptoms, without control groups NS (−) or asymptomatic patients. RPR was used as the gold standard, which differs from most studies analyzed in this review, which used CSF-VDRL. 21
TPPA 1:2560 83% 83%
RPR   1:2 OR TPPA 1:1280 96% 46%
RPR ≥ 1:16 32% 88% Asymptomatic NS:
a. No neurologic symptoms/signs AND
b. CSF-RPR (+) OR
c. WBC > 5 cells/μL OR
d. CSF-protein > 450 mg/dL
139 263 0 0% The sample included syphilis patients with persistent RPR titles after treatment. ANS was most frequent between ages 51–60 years, and the best cutoff value was 1:16. This study recommended lumbar puncture in patients with persistent RPR titles. Study limitations: the absence of HIV (+) population; patient’s outcome was not reported. 22
RPR   1:32 67% 59% Asymptomatic NS:
a. CSF-RPR (+) OR
b. CSF-WBC > 20 cells/μL AND CSF-TPPA > 1:640
12 19 0 100% This study has a small sample size and restricted population characteristics (only latent syphilis, HIV (+), and asymptomatic patients). Uses RPR as diagnostic criteria, possibly interfering with the reported specificity/sensibility values. 23
CD4   350 75% 82%
RPR   1/32 AND CD4   350 50% 67%
Molecular TP 47 PCR 76% 87% CSF-TPHA/FTA-ABS (+) AND
a. CSF-WBC > 5–10 cells/μL OR
33 91 0 Mostly positive Addresses PCR as a promising technique for NS diagnosis. The majority of the patients presented with latent syphilis. Study limitations: small sample size; no differentiation between latent syphilis stages (which interferes in the differentiation between late and early NS/meningitis); patient outcome not reported. 24
POL A PCR 70% 92%
TPP 47 Nested PCR 42% 97% Mentioned CSF-VDRL as gold standard, but used the following definitions for the analysis:
a. Serological reactive non-treponemal and treponemal tests AND;
b. Signs/symptoms AND;
c. CSF abnormalities such as VDRL (+), FTA-ABS (+), elevated WBC, elevated proteins)
40 0 0 45% Study considerations valid only for symptomatic patients (exclusion of patients without ophthalmic and neurologic symptoms). The study tested Nested PCR in samples of patients with confirmed NS according to the gold standard used. The study describes problems with sample preservation that could affect sensitivity. CMV coinfection was a confusion factor present. 25
Immunological biomarkers CSF-CXCL13 > 256.4 pg/mL 85%
82% (ANS)
88% (ANS)
Confirmed NS: CSF-VDRL (+) AND CSF-TPPA (+)
Presumed NS: CSF-VDRL
(−), CSF-TPPA (+), AND:
a. CSF-WBC > 8 cells/μL or CSF-protein > 450 mg/L without another cause OR
b. Signs/symptoms consistent with NS without another cause
191 123 92 0% Chemokine levels were useful for patient follow-up (decreased after treatment). They may change due to other inflammatory conditions and previous treatments/medications. Not useful for HIV co-infection. Control serum and CSF samples were from different individuals. 26
CSF-CXCL8 > 48.1 pg/mL 79%
71% (ANS)
89% (ANS)
CSF-CXCL10 > 163.1 pg/mL 80%
69% (ANS)
90% (ANS)
CXCL13 (CSF/serum) > 4.36 87%
83% (ANS)
99% (ANS)
CXCL8 (CSF/serum) > 10.3 79%
68% (ANS)
72% (ANS)
CXCL10 (CSF/serum) > 1.02 86%
77% (ANS)
CSF-CXCL13 > 76.3 pg/mL 50% 90% CSF-RPR (+) 16 87 0 54% The study is limited by the lack of clinical data about previous patient’s treatment, and by the sole inclusion of patients that underwent lumbar puncture. CXCL13 added more diagnostic value to RPR when evaluating patients with HIV co-infection. There is a possible classification bias, as the gold standard disregards CSF abnormalities such as protein and WBC count. 27
CSF-CXCL13 > 4.87 pg/mL 80% 81% Not described in the article, just referenced [15]:
Syphilis positive serologies AND
b. CSF-TPPA (+); an otherwise unexplained neurologic manifestation consistent with NS; CSF-proteins > 50 mg/dL or CSF-WBCs > 10 cells/μL
40 57 49 0% No difference was reported for different clinical manifestations. There was evidence of intrathecal CXCL13 production. Controls did not undergo lumbar puncture, limiting comparisons. 28
Quotient a > 2.408 88% 69%
  1. Main results of the 14 studies that evaluated diagnostic tests and criteria for neurosyphilis and their performance. Abbreviations: NS Neurosyphilis, ANS Asymptomatic neurosyphilis, NS (+) Positive neurosyphilis diagnosis, NS (−) Negative neurosyphilis diagnosis, HIV Human immunodeficiency virus, CSF Cerebrospinal fluid, WBC White blood cells, VDRL Venereal disease research laboratory, RPR Rapid plasma reagin, USR Unheated serum reagin, TPPA T. pallidum particle agglutination, TPHA T. pallidum hemagglutination, FTA-ABS Fluorescent treponemal antibody absorption, PCR Polymerase chain reaction, RT-PCR Reverse transcriptase polymerase chain reaction, CMV Cytomegalovirus, CXCL Chemokine CXC ligand
  2. a Quotient = (CSF-CXCL13 / CSF-albumin) / (Serum-CXCL13 / Serum-albumin)