From June 2010 through October 2011, fingerstick blood samples (200 μL) were collected from 1,549 febrile patients (axillary temperature >37.5°C) at 14 dispensaries and 91 randomly selected healthy villagers in Senegal. Samples were subjected to DNA extraction. Digesting, binding, and washing were performed directly in the village dispensaries by use of the QIAamp kit (QIAGEN, Hilden, Germany) as previously reported (6). DNA elution was performed at Aix-Marseille Université in Marseille, France. Quantitative PCR (qPCR) was performed by using primers and probes specific for the genus Borrelia. All samples positive for Borrelia spp. were subjected to standard PCR (flaB gene) (4). To determine the quality of the extracted DNA, we also measured the human actin gene (5). The samples were considered positive only if qPCR and flaB-based PCR results were positive; the sequencing of all flaB amplicons demonstrated that they belonged to B. crocidurae.

Isolation of borreliae involved intraperitoneal inoculation of laboratory BALB/c mice with 100 μL of patient capillary blood. Borreliae in mice were detected by microscopic examination of Giemsa-stained peripheral blood smears followed by qPCR of blood samples.

The sex ratio for the 1,549 patients did not differ significantly among sites (772 male and 777 female patients). An analysis of 6 age groups (<12 months, 1–3 years, 4–6 years, 7–15 years, 16–29 years, and >30 years) showed no differences among sites. All tested samples from clinically healthy persons had negative qPCR results for borreliae.

The incidence rate was calculated as the number of febrile episodes divided by the person-time multiplied by 1,000 (data available only for the Sine-Saloum site). The incidence rate of febrile episodes was 0.80 in Dielmo and 0.36 in Ndiop (p<0.05). Among the 1,566 samples tested, 115 (7.3%) were positive for B. crocidurae. The incidence rate for TBRF was 9.7 cases/100 persons in Dielmo and 2.4 cases/100 persons in Ndiop. The first autochthonous cases in Ndiop, which was previously considered borreliosis free, were observed in October 2010; incidence was significantly lower in Ndiop than in Dielmo (p<0.05). All cases registered in Ndiop before October 2010 were included in the epidemiologic investigation and considered to be imported. The proportion of the Borrelia-positive samples was significantly higher for northern sites with a drier Sudanian climate; positivity reached 19.1% (33/173) in Niakhar (Table). By analyzing the epidemiologic questionnaires completed by families of ill persons, we determined that the 2 TBRF cases in Casamance were imported from the northern regions of Senegal by seasonal workers.

Patients most frequently infected were 7–15 years of age (13.5%, 43/318), unlike in eastern Africa where younger persons are more frequently infected (9). No positive results were found among the 155 children <12 months of age, but positive results were found for 16 (4.8%) of the 352 children 1–3 years of age (p<0.05). Unlike in other northern regions, in Sine-Saloum, the proportion of Borrelia-positive samples was significantly higher for samples collected during the dry (16.9%, 40/237) than the rainy season (6.9%, 30/432); p<0.0001 (Figure 1).

Figure 2

Figure 2. Thick smear of mouse blood showing Borrelia crocidurae isolate 02–03Giemsa stain; original magnification x900.

We identified 20 patients (49 samples) for whom 2–4 samples were positive for B. crocidurae. The interval between the sample collections was short (5–30 days) for 13 persons, average (30–66 days) for 4 persons, and long (102–381 days) for 3 persons. Two Borrelia isolates (no. 03–02 from Ndiop and no. 19/31 from Dielmo) were recovered from the peripheral blood of 2 febrile patients. The bacteria had a morphologic appearance that was typical for borreliae (Figure 2). A BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) search for the sequenced flaB gene (JX119098) demonstrated that the isolates were nearly identical with the type Achema strain of B. crocidurae (CP003426).

We detected an alarmingly high proportion of Borrelia DNA in the blood of febrile patients in Senegal. The presence of this DNA is strongly and specifically linked to the fever because no Borrelia DNA was identified among the 90 control participants. In Tanzania, however, borreliae have been identified in up to 33% of blood samples obtained from asymptomatic blood donors who lived in similar conditions as ill persons (13).

The geographic repartition of TBRF is linked to drier climates (9). We observed autochthonous cases only in northern Senegal, roughly north of the 13°30′ parallel. We noted the recent extension of B. crocidurae into the village of Ndiop, which had been free of B. crocidurae. This extension might be linked to recent climate changes (14). The person-year incidence of borreliosis in our study (6.1 cases/100 population) is similar to that reported by Vial et al. (4 cases/100 population) for the interepidemic period (8).

We report a unique series of cases in which Borrelia DNA was identified several times consecutively in the blood of the same patient. For 17 patients for whom the time between positive samples was short or average (up to 66 days), repeated detection of Borrelia DNA during repeated episodes of fever could be explained by relapses. However, reinfection is strongly suspected in 3 patients because the interval between 2 positive samples was >100 days. To the best of our knowledge, reinfection with relapsing fever borreliae has not been previously reported in Africa. The phenomenon of easy reinfection after treatment with tetracycline has been reported for the relapsing-fever group B. hermsii in vervet monkeys, which could be reinfected 12–36 weeks after primary infection (15).

In conclusion, the incidence of TBRF and the proportion of borreliosis cases among febrile patients in Senegal is very high and, in at least 1 region (Niakhar), exceeds that of malaria. This considerably high incidence rate should lead to the development of new therapeutic strategies that could be based on treating febrile patients in Senegal with doxycycline.