Persons included in our study had clinical signs and symptoms of meningitis or encephalitis, or both, and were selected from the 11 patients with cases of B. crocidurae TBRF that were reported to and confirmed by the National Reference Center for Borrelia (NRCB) in France during 2009–2011. The NRCB is the reference laboratory responsible for the epidemiologic surveillance of TBRF in France. Clinical meningitis or encephalitis was defined as previously reported (2). Borrelia species were detected in Giemsa-stained thin blood smears by microscopy and quantitative buffy coat analysis (Becton Dickinson, Le Pont de Claix, France) when available (4). Borrelia spp. were detected and identified by using 16SrRNA PCR and subsequent sequencing as described (8). We tested serum and CSF samples with standardized antibody assays for detection of Borrelia spp. that cause Lyme disease (Table).

Among the 11 TBRF cases reported to NRCB during the 3-year study, we identified 4 (36%) cases of clinical meningitis or encephalitis, or both. The epidemiologic, clinical, and laboratory findings and the treatment of the 4 patients are documented in the Table. Three of the 4 patients were adult men, 26–57 years of age, and 1 was a 7-year-old girl. None of the patients were immunocompromised. They were all given appropriate antimalarial chemoprophylaxis. Patients 1 and 4 experienced their first febrile episode in Africa and were empirically treated with antimalarial drugs without biological confirmation of Plasmodium infection.

At the time of admission to health care facilities, all patients had fever and headache. Patients 2, 3, and 4 had signs of meningitis, including neck stiffness; patients 2 and 4 also had phonophobia and photophobia. Patients 1 and 4 had encephalitis with drowsiness which for patient 4 was accompanied by cerebellar syndrome (dysarthria and dysmetria). All patients except patient 3 underwent computed tomography scanning of the brain; no abnormality was detected. The 2 patients with encephalitis were examined by magnetic resonance imaging; in patient 4, a predominant positive contrast of the cerebellum leptomeninges on the right side was observed.

For all patients, 16S rRNA PCR and sequencing identified B. crocidurae (8) in blood samples. Laboratory analysis of the 4 CSF samples showed a lymphocytic pleocytosis, high protein concentrations, and a glucose value within reference range (Table). The molecular methods applied to CSF samples confirmed neurologic B. crocidurae infection in patients 1, 2, and 3. Serum samples collected from patients 1 and 4 at the time of diagnosis were tested by using Lyme disease serodiagnostic assays. ELISA detected substantial levels of IgM and IgG in samples from both patients; 1 was confirmed by Western blot analysis. The CSF sample from patient 4 showed a low level of IgG (Table).

All cases were treated with doxycycline or ceftriaxone, or both (Table). In all patients, fever resolved within 3 days of the beginning of the appropriate treatment, and the outcomes were favorable. No Jarisch-Herxheimer reaction was observed.

B. crocidurae–associated TBRF is an emerging disease that is considered to be benign (1,9). However, the series of infections reported here suggest that severe neurologic complications, notably, meningitis and encephalitis, occur more frequently than previously thought and could be particularly common in travelers who acquired this infection in West Africa. For the patients we studied, the earliest neurologic signs occurred during the second febrile episode, confirming previous studies reporting the onset of neurologic complications after the first episode (2). However, facial palsy, often considered to be among the main clinical signs and symptoms of neuroborreliosis caused by TBRF-associated Borrelia species, was not observed in these patients (2). A similar clinical manifestation described in a recent case report of B. crocidurae encephalitis is entirely consistent with our observations (3).

Functional and experimental studies have focused on the capacity of TBRF-associated Borrelia species to cross the blood–brain barrier and to persist in the brain (2–7,10,11). These studies have established B. crocidurae as the most neurotropic species, an observation consistent with this and other case series and case reports. In animal models, this feature has been associated with the presence of vascular microemboli in the brain of infected animals and the particular ability of B. crocidurae to form and bind to erythrocyte rosettes, a phenomenon also involved in cerebral malaria pathogenesis. Erythrocyte aggregation might prevent host–pathogen interactions and thereby protect the spirochetes from the specific immune response (10,12,13).

The rather high frequency and severity of neurologic complications associated with B. crocidurae infection raise the problem of distinguishing it from cerebral malaria, because the areas of endemicity of these 2 diseases largely coincide (1,9). Indeed, relapsing fever is frequently misdiagnosed as malaria, as it was for 2 of the patients we studied, who were initially treated with antimalarial drugs (14). In this context, quantitative buffy coat analysis that can effectively detect each pathogen in blood might be of particular interest (4). In addition, our study confirms the usefulness of molecular methods applied to blood and CSF samples to confirm Borrelia infection (3,8). The negative result obtained by PCR of CSF from patient 4 could have been the consequence of inappropriate storage of the sample at high room temperature for 72 hours before analysis.

Lyme disease serodiagnostic testing of serum and CSF samples might be helpful. Indeed, cross-reacting IgG and IgM were detected by ELISAs and in Western blot assays. Because Lyme disease is endemic to France, our results could have been caused by the actual detection of B. burgdorferi sensu lato antibodies, although none of the patients had a known history of Lyme disease.

No specific recommendations have been proposed for the treatment of patients with TBRF neuroborreliosis. Erythromycin and penicillin have been reported to be ineffective (5,6). In our series, all patients were prescribed either ceftriaxone or doxycycline, or both (Table), resulting in successful treatment of the disease. Thus, from the literature and our own experience, we suggest that TBRF with neurologic involvement should be treated with ceftriaxone or doxycycline for at least 21 days.

Our study highlights the frequent occurrence of meningitis or encephalitis in patients with B. crocidurae TBRF acquired in West Africa. The clinical and radiologic manifestations suggest that this infection could be more severe than previously thought. Consequently, travelers returning from West Africa with febrile neurologic disorders should be tested immediately for biological confirmation of Borrelia infection through blood and CSF analyses, including molecular methods.