Secondary Logo

Journal Logo


HIV and Syphilis: When to Perform a Lumbar Puncture

Libois, AgnÈs MD*; De Wit, StÉphane MD, PhD*; Poll, BÉnÉdicte*; Garcia, Felipe MD, PhD; Florence, Eric MD, PhD; Del Rio, Ana MD; Sanchez, Paquita MD§; Negredo, Eugenia MD; Vandenbruaene, Marc MD; Gatell, JosÉ M. MD, PhD; Clumeck, Nathan MD, PhD*

Author Information
Sexually Transmitted Diseases: March 2007 - Volume 34 - Issue 3 - p 141-144
doi: 10.1097/01.olq.0000230481.28936.e5
  • Free

THE CENTRAL NERVOUS SYSTEM IS frequently and rapidly affected (at least 40% of the cases) during infection with Treponema pallidum. This invasion is usually moderate and asymptomatic and seldom progresses to neurosyphilis.1 HIV-negative patients treated for an early syphilis rarely progress to neurosyphilis, although intramuscular benzathine penicillin does not achieve treponemicidal levels in the cerebrospinal fluid (CSF).2 Alteration of the immunity during HIV infection appears to promote the persistence of T. pallidum in the central nervous system and increases the risk for neurosyphilis. Case reports have suggested that progression to neurosyphilis could be more frequent and faster among HIV-infected patients. More treatment failures were also described.2–4 Prevalence of neurosyphilis in HIV-infected patients with late-latent syphilis is estimated to be between 9.1% and 23.5%.5,6

Neurosyphilis warrants a specific therapeutic approach. Standard treatment with intramuscular penicillin is suboptimal for neurosyphilis. The recommended treatment is intravenous penicillin, but treating systematically all coinfected patients with this treatment would lead to overtreatment of a significant number of patients. Currently, the Center for Disease Control and Prevention and the European guidelines recommend a lumbar puncture (LP) among all HIV-infected patients with a late-latent syphilis or a syphilis of unknown duration.7,8 Some other experts advise a LP for all patients coinfected by HIV and syphilis.5,9 Because of the lack of clear indication of LP in HIV-infected patients with syphilis, in practice, LPs are performed at the discretion of the treating physician in many centers.

This study aims to determine predictive factors for neurosyphilis among patients coinfected with HIV and syphilis, the objective being to identify and treat patients with higher probability for neurosyphilis and to decrease unnecessary LP in this population.

Such a selective approach would be relevant at the present times in both Western countries where a recent increase of syphilis is observed in men who have sex with men, particularly those infected with HIV, and in resource-limited settings where the use of the resources could be optimized.



HIV-infected patients with a reactive serum treponemal test who underwent a LP to evaluate syphilis were retrospectively evaluated. Patients who had an LP for another indication were excluded from the study. The study was carried out in 5 HIV reference centers. A total of 112 patients were included in the study: 44 patients were retrieved from Saint-Pierre Hospital in Brussels (1988–2004), 18 at the University Hospital of Antwerp (1998–2003), and 50 in Barcelona (Spain) (1999–2003) (10 patients were found in “Hospital Clinic,” 10 in “Hospital Germans Trias I Pujol,” and 30 in “Hospital del Mar”). Time periods for patients’ inclusion were different in each hospital according to the laboratory database available at each site. Data on demographic characteristics (age, sex, ethnic group), neurologic manifestations (headache, stiff neck, photophobia, cranial nerve abnormalities, lightning pains, neuropsychiatric features, focal neurologic deficits, or pupillary change), cutaneous manifestations (generalized macular, maculopapular, papular or pustular rash with or without palmar–plantar involvement), stage of syphilis, HIV status (CD4, viral load), sexual behavior and HIV acquisition factor, serum RPR (or Veneral Research Disease Laboratory [VDRL]) and T. pallidum hemagglutination (TPHA), results of the LP (cells, proteins, RPR [or VDRL], TPHA, and index ITPA) were derived from patients’ files.

The criteria for diagnosis of neurosyphilis were a CSF–white blood cell (WBC) count ≥20/μL and/or a reactive CSF–VDRL and/or a positive ITPA index, being defined by TPHA-CSF IgG titer/total CSF IgG >/=3 with an undamaged blood–brain barrier9 TPHA-serum IgG titer/total serum IgG.

Statistical Analysis

Univariate analysis (Fisher exact test and Kruskal-Wallis) and multivariate analysis (logistic regression) were performed to identify predictive factors for neurologic involvement. The multivariate analysis included variables with significant level of P <0.10 on univariate analysis. Serum RPR being a quantitative measure with serial 2-fold dilution, RPR was log2 transformed for analysis.


Among the 112 patients included in the study, there were 102 men and 10 women. One hundred seven were white. Mean age was 37 years (range, 19–69 years). Median CD4 cells count at the time of the diagnosis of syphilis was 499/μL (range, 4–1,292/μL). Eighty-four patients were men who have sex with men and the 28 others were heterosexual. Sixty-six patients had late-latent syphilis or a syphilis of unknown duration, 39 secondary syphilis, 4 an early-latent syphilis, and 3 primary syphilis. Median RPR for the 112 patients was 1/32 (range, 0–1/1,024).

Twenty-six of 112 (23.2%) patients had a diagnosis of neurosyphilis. The diagnosis of neurosyphilis was based on a CSF–WBC count ≥20/μL for 24 patients, a reactive CSF–VDRL test result for 8 patients (7 had both CSF–WBC count ≥20/μL and a reactive CSF–VDRL, and one had a reactive CSF–VDRL test result only) and a positive ITPA index for one patient. For 17 patients, the diagnosis of neurosyphilis was based only a CSF–WBC count ≥20/μL. Among those, 6 had neurologic manifestations. The median CSF–WBC count was 30 cells/μL in patients with neurosyphilis and 4 cells/μL in patients without neurosyphilis (P <0.0001). The median CSF protein was 0.45 mg/dL in patients without neurosyphilis and 0.82 mg/dL in patients with neurosyphilis (P <0.0001), but 59 patients had a CSF protein level above the upper limit of normality, which constitutes the only anomaly of the CSF for 37 of them (median protein rate, 0.75 mg/dL; range, 0.51–1 mg/dL). Eleven of 26 patients with neurosyphilis had a negative CSF–TPHA.

Among the 26 patients with neurosyphilis, there were 21 men and 5 women. Eleven patients had neurologic manifestations (6 with concomitant cutaneous manifestations) and 15 had no neurologic manifestations (4 with cutaneous manifestations).

In the univariate analysis (see Table 1), neurologic manifestations and serum RPR were associated with neurosyphilis (P = 0.036 and P = 0.018, respectively) as well as female sex (P = 0.05). There was no significant association between neurosyphilis and age, cutaneous manifestations, CD4 cells count, or stage of syphilis. After controlling for sex and neurologic manifestations (see Table 2), log2 RPR serum was still significantly associated with neurosyphilis (P = 0.005) with an adjusted odds ratio of 1.4 (95% confidence interval, 1.1–1.7) for each RPR dilution. In the multivariate analysis, neurologic manifestations (P = 0.0007) and female sex (P = 0.007) were significantly associated with neurosyphilis.

Risk Factors for Neurosyphilis (univariate analysis)
Risk Factors for Neurosyphilis (multivariate analysis)

To evaluate the relationship between serum RPR and risk for neurosyphilis for patients without neurologic manifestations and to establish whether a cutoff point could be determined for the performance of an LP, a receiver operating characteristic (ROC) curve was built (Fig. 1). The ROC curve shows the sensibility and the specificity as the position of the RPR value varies; 1/32 combines the best sensitivity (100%) with an acceptable specificity (40%). The negative and positive predictive values were, respectively, 100% and 26%. A logistic regression model linking log2 RPR and risk of neurosyphilis was built for these patients (Fig. 2). This model indicates the probability of neurosyphilis according to serum RPR. The probability of neurosyphilis was calculated based on coefficients of logistic regression.

Figure 1
Figure 1:
Receiver operating characteristic curve for rapid plasma reagin serum value in patients without neurologic manifestations.
Figure 2
Figure 2:
Probability and proportion of neurosyphilis according to serum rapid plasma reagin in neurologic asymptomatic patients. Numbers on the graph correspond to the numbers of patients with syphilis or neurosyphilis. NS = neurosyphilis.

The risk increases slowly until it reaches a RPR of 1/32 and then increases sharply. Statistical analysis confirmed that patients with a RPR ≥1:32 are significantly more likely to have neurosyphilis (P = 0.002). None of the patients without neurologic manifestations and a RPR <1/32 had neurosyphilis.


The diagnosis of neurosyphilis in HIV-infected patients is important because of the risk of persistence and reactivation linked to immunodeficiency and because standard treatment with intramuscular penicillin does not achieve treponemicidal levels in the CSF. Systematic treatment with intravenous penicillin or the performance of an LP in all HIV-infected patients with syphilis is impractical and unnecessary. A controversy exists on whether to perform an LP in HIV-infected patients.5,7–9

The determination of predictive factors for neurosyphilis in these patients would allow identifying patients with high probability for neurosyphilis for whom an LP should be performed.

The diagnosis of neurosyphilis is difficult in patients coinfected with syphilis and HIV; whereas an HIV-negative patient with a reactive serum treponemal test and an increase of CSF–WBC count or CSF proteins is considered as having neurosyphilis, this rule is not valid when the patient is infected with HIV. Studies showed an increase of cells and/or proteins in the CSF resulting from the HIV infection itself in 38% to 69% of cases.10,11 The differential diagnosis between neurosyphilis and abnormalities resulting from HIV infection is sometimes difficult, and criteria of neurosyphilis in HIV-infected patients vary from study to study. Some authors define neurosyphilis by strict criteria: a reactive serum syphilis serology and a reactive CSF–VDRL.5,12 CSF–VDRL is highly specific but the sensitivity ranges from 30% to 70%,13 indicating that a negative test cannot rule out neurosyphilis. The ITPA index could be used if the CSF–VDRL is negative.9,14,15 Another study used broader criteria: a reactive CSF–VDRL and/or a CSF–WBC count ≥20/μL and/or CSF proteins >50 mg/dL.18 In accordance with the most recent studies,16 criteria used in our study were CSF–WBC count ≥20/μL and/or a reactive CSF–VDRL and/or a positive ITPA index.

Although patients with neurosyphilis had a significantly more important increase of CSF proteins, 37 patients without neurosyphilis had a moderate increase of CSF proteins. In those patients, this increase of the protein was probably the result of HIV infection. The increase of CSF proteins was not used as criteria for neurosyphilis in our study. This criterion could be probably used with a more elevated cutoff that must be determined. The value of a negative CSF–TPHA remains controversial.13,17 In this study, 11 of 26 patients with neurosyphilis had a negative CSF–TPHA, and among these 11 patients, 5 had neurologic manifestations. We think that a negative CSF–TPHA cannot rule out neurosyphilis.

In our study, the prevalence of neurosyphilis was 23.2%. A significant association was found between neurosyphilis and neurologic manifestations and between neurosyphilis and serum RPR. Female sex was also a risk factor for neurosyphilis. Among the 10 women of this series, 5 had neurosyphilis and all 5 came from the same center. This unexpected observation, not described elsewhere, could be the result of a selection bias or the small number of women included in the study. This small number of women is not unexpected; indeed, mainly HIV-infected patients with syphilis are men who have sex with men. There was no association between neurosyphilis and age, cutaneous manifestations, CD4+ T cell count, history, or stage of syphilis. It is difficult to compare these results with other studies because the criteria of neurosyphilis are not consistent. An association between neurosyphilis and CD4 cell count has been described in 2 recent studies, but not in another one.6,16,18 In the study by Marra et al, neurosyphilis was significantly more common in patients with CD4+ T cell count ≤350/μL.16

Our results indicate that there is a link between neurosyphilis and serum RPR; this association was confirmed after controlling for sex and neurologic manifestations. How can these results help in clinical practice to establish which HIV patients coinfected with syphilis and who have no neurologic manifestations could have a high probability of neurosyphilis?

A serum RPR of 1/32 seems to be the best cutoff point to decide whether to perform an LP. The risk for neurosyphilis is low when the serum RPR is <1/32 and increases sharply and significantly after this point. The use of this criterion would avoid unnecessary LB while avoiding as much as possible the underdiagnosis of this severe complication. If we applied this sample, the recommendations of performing an LP in case of neurologic manifestations and/or a serum RPR ≥1/32, 27 of 112 patients should have avoided an LP without missing a diagnosis of neurosyphilis. In practice, for the few patients with a RPR <1/32 who fail with standard therapy, an LP should be performed later when treatment failure is noted. These results confirm a recent study in which the risk of neurosyphilis increased 5.98-fold if the serum RPR is ≥1/32.16

Our study has some limitations. It is a retrospective study and results came from different laboratories, which can influence the measurements. LP was performed at the discretion of the treating physician, without predetermined indications, and the criteria for diagnosis of neurosyphilis were not prospectively established. One could say that individuals with reactive serum treponemal tests but nonreactive serum RPR do not have active syphilis and should not be included in the study. Nine patients had a negative serum RPR; all had late-latent syphilis or syphilis of unknown duration. Among those patients, 7 had neurologic manifestations and 2 had symptomatic neurosyphilis. In untreated patients with late-stage syphilis, serum treponemal tests can eventually become nonreactive.13 It could be argued that the CSF pleocytosis was caused by HIV rather than by syphilis, but 7 of 24 patients with CSF–WBC >20 also had a reactive CSF–VDRL test; and our cutoff for defining abnormal CSF–WBC counts is above the mild pleocytosis resulting from HIV alone.11,19

In conclusion, in HIV-infected patients with early or late syphilis, there is a significant association between the likelihood of neurosyphilis and the serum RPR. Among patients without neurologic manifestations, the risk of neurosyphilis becomes clinically relevant if serum RPR is ≥1/32. In patients coinfected with HIV and syphilis, it thus appears reasonable to propose to perform an LP when serum RPR is ≥1/32 and to avoid it when serum RPR is <1/32 if the clinical context does not suggest neurologic involvement.


1. Lukehart SA, Hook EW III, Baker-Zander SA, et al. Invasion of the central nervous system by Treponema pallidum: Implications for diagnosis and treatment Ann Intern Med 1988; 109:855–862.
2. Musher DA, Hamill RJ, Baughn RE. Effect of human immunodeficiency virus (HIV) infection on the course of syphilis and on the response to treatment. Ann Intern Med 1990; 113:872–881.
3. Johns DR, Tierney M, Felsenstein D. Alteration in the natural history of neurosyphilis by concurrent infection with the human immunodeficiency virus. N Engl J Med 1987; 316:1569–1572.
4. Berry CD, Hooton TM, Collier AC, et al. Neurologic relapse after benzathine penicillin therapy for secondary syphilis in a patient with HIV infection. N Engl J Med 1987; 316:1587–1589.
5. Holtom PD, Robert AL, Leal ME, et al. Prevalence of neurosyphilis in human immunodeficiency virus-infected patients with latent syphilis. Am J Med 1992; 93:9–12.
6. Bordon J, Martinez-Vazquez C, Alvarez M, et al. Neurosyphilis in HIV-infected patients. Eur J Clin Microbiol Infect Dis 1995; 14:864–869.
7. Centers for Disease Control and Prevention. Sexually transmitted diseases treatment guidelines 2002. MMWR Recomm Rep 2002; 51:1–78.
8. Goh BT, van Voorst Vader PC. European guidelines for the management of syphilis. Int J STD AIDS 2001; 12(suppl 3):14–26.
9. Tramont EC. Treponema pallidum (syphilis). In: Mandell, Douglas, Benett. Principles and Practice of Infectious Diseases, 5th ed. 2000:2474–2490.
10. Appleman ME, Marshall DW, Brey RL, et al. Cerebrospinal fluid abnormalities in patients without AIDS who are seropositive for the human immunodeficiency virus. J Infect Dis 1988; 158:193–199.
11. García F, Niebla G, Romeu J, et al. Cerebrospinal fluid HIV-1 RNA levels in asymptomatic patients with early stage chronic HIV-1 infection: Support for the hypothesis of local virus replication. AIDS 1999; 13:1491–1496.
12. Flood JM, Weinstock HS, Guroy ME, et al. Neurosyphilis during the AIDS epidemic, San Francisco, 1985–1992. J Infect Dis 1998; 177:931–940.
13. Hook EW III, Marra CM. Acquired syphilis in adults. N Engl J Med 1992; 326:1060–1069.
14. Tomberlin MG, Holtom PD, Owens JL, et al. Evaluation of neurosyphilis in human immunodeficiency virus-infected individuals. Clin Infect Dis 1994; 18:288–294.
15. Luger AF, Schmindt BL, Kaulich M. Significance of laboratory findings for the diagnosis of neurosyphilis. Int J STD AIDS 11:224–234.
16. Marra CM, Maxwell CL, Smith SL, et al. Cerebrospinal fluid abnormalities in patients with syphilis: Association with clinical and laboratory features. J Infect Dis 2004; 189:369–376.
17. Jaffe HW, Larsen SA, Peters M, et al. Tests for treponemal antibody in CSF. Arch Intern Med 1978; 138:252–255.
18. Hopkins S, McDermott H, Lyons F, et al. Syphilis and the central nervous system in HIV infected patients [A79]. Presented at IAS Paris, 2003.
19. Collier AC, Marra C, Coombs RW, et al. Central nervous system manifestations in human immunodeficiency virus infection without AIDS. J Acquir Immun Defic Syndr 1992; 5:229–241.
© Copyright 2007 American Sexually Transmitted Diseases Association