Possible Role of Rickettsia felis in Acute Febrile Illness among Children in Gabon

Infection is widespread but most prevalent among young, rural residents with fever.

O ver the past decade, reported cases of malaria and associated deaths have declined in Africa (1). This decrease has led to a search for other causes of fever in Africa, where unexplained febrile illnesses are one of the major health problems. In some sub-Saharan Africa countries, malaria treatments are still administered without a biologic diagnosis. For example, an assessment of complicated malaria and other severe febrile illness cases in a pediatric ward in Libreville, Gabon, showed that 43.5% of the children who received an antimalarial treatment had microscopy test results negative for malaria (2).
Other studies have shown that, in addition to malaria, other bacterial infections are a major cause of fever in Africa (3)(4)(5)(6). Staphylococcus aureus, Streptococcus pneumoniae, nontyphoidal Salmonella spp., Klebsiella pneumoniae, and Escherichia coli are the bacteria most often detected in sub-Saharan Africa by the culture method (7,8). The use of molecular tools has enabled the identification of the following fastidious bacteria as a cause of unexplained fevers in Africa: Rickettsia spp., including R. felis (3)(4)(5)9); Coxiella burnetii (10); Tropheryma whipplei (11); and Borrelia spp. (12,13). However, the epidemiology of many fastidious bacteria, such as R. felis, remains poorly understood. In rural areas of Senegal, the prevalence of R. felis was generally higher (7%-24%) than that in urban areas of sub-Saharan African, such as Franceville, Gabon (10%) (14).
R. felis is a gram-negative bacterium belonging to the spotted fever group of Rickettsia spp. In Gabon, the bacterium has been reported in arthropods, including Ctenocephalides felis cat fleas (15) and Aedes albopictus mosquitoes (16), and in humans (14). Similar to many African countries, Gabon has a strong disparity between health care in urban and rural areas; in rural areas, little is known about the epidemiology of infectious diseases. The aim of our study was to evaluate the prevalence of R. felis infection among febrile and afebrile children in rural and urban areas of Gabon and the possible role of R. felis in acute febrile illness.

Study Area
Gabon is a central African country located on the equator along the Atlantic Coast ( Figure 1). The country has a low coastal plain and hilly inland areas and savannas to the east and south; 80% of Gabon is covered by forest. The tropical climate is hot and humid, and the seasons alternate in precipitation and length: short dry season, long rainy season, long dry season, short rainy season.
Center of Lastourville, are in semiurban and rural areas, respectively, of Ogooué Lolo Province. The fourth center, the Medical Research Unit of Ngounie in Fougamou, is in a rural area of Ngounié Province.
The National Ethics Committee of Gabon approved this prospective study (no. 0023/2013/SG/CNE). Written informed consent forms and questionnaires were completed by parents or legal guardians upon a child's enrollment in the study.
During April 2013-January 2014, a total of 525 children <15 years of age were recruited for the study; 465 of the children were febrile (axillary temperature >37.5°C), and 60 were afebrile (controls). Febrile children were recruited from the pediatric outpatient clinics at the 4 health care centers. The control group was recruited from children who had accompanied their sick parents to the health care centers. Children in the control group had to be free of fever for at least 1 week before study inclusion.

Sample Collection and Molecular Analysis
Molecular analyses were performed on DNA extracts from blood samples from each child; blood smears, serologic testing, and culture were not done. After a child's parent or legal guardian was interviewed, a blood sample was collected into an EDTA tube. World Health Organization guidelines for blood collection were followed, including guidelines for hand hygiene, use of sterile tubes, and skin disinfection with 70% alcohol. The International Center of Medical Research of Franceville, which has a well-trained staff with expertise in infectious diseases, performed DNA extraction by using the E.Z.N.A. Blood DNA Maxi Kit (Omega Bio-tek, Norcross, GA, USA) according to the manufacturer's protocol (17). A total of 150 μL of DNA extract was obtained from each sample. The extracts were stored at -20°C before being sent on ice to URMITE (Unité de Recherche sur les Maladies Infectieuses et Tropicales, Marseille, France) for molecular analyses.
Specific quantitative PCR (qPCR) was performed by using a CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories, Marnes-la-Coquette, France). qPCR Master Mix (Eurogentec, Liege, Belgium) was prepared according to the manufacturer's instructions; for each reaction, 15 μL of Master Mix was added to 5 μL of DNA. The quality of extracted DNA and the lack of PCR inhibitors were systematically checked by targeting a housekeeping gene, human β-actin (14). Positive (R. felis DNA) and negative (mix alone) controls were also systematically used for each PCR assay. All samples were screened by Rickettsia spp.-specific qPCR targeting the gltA gene and by R. felis-specific qPCR targeting the bioB, orfB, and vapB1 genes (14) ( Table 1). Samples positive for at least 2 genes were considered positive.

Statistical Analysis
Data were analyzed by using Epi Info software version 7.0.8.0 (Centers for Disease Control and Prevention, Atlanta, GA, USA). Mantel-Haenszel χ 2 and Fisher exact tests were used to compare the prevalence between groups (i.e., febrile and afebrile children), seasons, and geographic areas and by the participants' sex and age. A 2-tailed p value <0.05 was considered statistically significant.

Recruitment
A total of 465 febrile children were recruited from Franceville (n = 80), Koulamoutou (n = 167), Lastourville (n = 155), and Fougamou (n = 63) ( Table 2). However, 55 of these children were excluded from the statistical analysis because their sex and age data were unavailable; 3 of the excluded children were from Franceville, 26 from Koulamoutou, 21 from Lastourville, and 2 from Fougamou. Of the 410 children included in the statistical analysis, 52.7% (216/410) were recruited during a rainy season, and 47.3% (194/410) were recruited during a dry season. A total of 60 afebrile children were recruited and enrolled in parallel from Franceville (n = 24), Fougamou (n = 20), and Lastourville (n = 16); no afebrile children were enrolled from Koulamoutou. The 410 febrile patients included in the statistical analysis consisted of 212 boys and 198 girls (sex ratio 1.07).

R. felis in Febrile and Afebrile Children in Gabon
R. felis DNA was detected in 42 (10.2%) of 410 analyzed samples from febrile children ( Table 3). The bacterium was detected significantly more frequently during the rainy season (  R. felis prevalence also did not differ significantly by age, even though prevalence was higher among 1-to 3-year-old children (   (Table 3; Figure 2). The infected febrile child was a 2-year-old boy, and the infected afebrile child was a 3-year-old girl; both children became infected during the dry season.

Discussion
The lack of molecular tools in many health centers in countries in sub-Saharan Africa limits the management of all febrile illnesses in these areas. In this study, we used molecular tools (PCR assays) to assess the prevalence of R. felis in blood specimens from febrile and afebrile children from rural, semiurban, and urban areas of Gabon. One of the most frequent pitfalls of PCR assays is the contamination of samples, which can occur any time during or after collection of the samples, including during their use in the laboratory. Over the years, the URMITE laboratory has developed strategies and applied rigorous procedures to prevent and detect contamination (18)(19)(20)(21). For example, we require that 2 different PCR assays show positive results before we conclude that a sample is positive (21). In this study, we used a Rickettsia spp.-specific qPCR targeting the highly conservative gltA gene and an R. felis-specific qPCR targeting the bioB, orfB, and vapB1 genes. The Rickettsia spp.-specific assay can amplify almost all Rickettsia spp., including R. conorii and R. rickettsii, but it is less sensitive than the R. felis-specific assay. When a sample in our study was positive by the Rickettsia spp.-specific qPCR, it was also positive by the R. felis qPCR. Several samples had positive results for only 2 of the 3 genes targeted by the R. felis-specific qPCR. This discrepancy might be explained by possible genetic diversity of the R. felis strains. The 3 genes used for the R. felis-specific assay are not all conservative, so it is possible that some strains have a certain degree of genetic diversity that may occasionally cause false-negative PCR results. Our results were also systematically validated by using rigorous criteria. To check the quality of each PCR run, we used negative controls (i.e., PCR mix without template) with every tenth sample, and we used 2 positive controls (R. felis DNA) per run. In addition, we required that all negative and positive controls be systematically correct (i.e., negative and positive, respectively) before validating each PCR run, and a sample was not considered R. felis-positive unless confirmed by at least 2 of the 4 sequences of targeted DNA. Thus, we consider our results to be valid. The fact that R. felis DNA has not been detected in samples (from febrile patients in France and Tunisia) previously analyzed in our laboratory (14) provides further support of our team's ability to prevent and detect contamination during PCR runs. Furthermore, it has been reported that the prevalence of R. felis is low in northern Africa countries (France, Algeria, Morocco, Tunisia) and increases in southern Africa countries (Mali, Senegal, Gabon) (14).
Consistent with our previous findings from Franceville in Haut-Ogooué Province (14), our findings from this study confirmed the presence of R. felis bacteremia in febrile children in Gabon and showed that the prevalence of infection was higher in rural than urban and semiurban areas. This fastidious bacterium was previously found in arthropods in Franceville (15), including the cat flea, C. felis. The bacterium was also detected in A. albopictus mosquitoes in Libreville in Estuaire Province (16). Data from our study also confirm the presence of R. felis in children in Ogooué lolo and Ngounié Provinces. Therefore, R. felis is widespread in Gabon, and its prevalence should be assessed in other areas of the country.
Common microorganisms involved in bacteremia (S. aureus, Streptococcus pyogenes, E. coli, K. pneumoniae, Salmonella spp., and S. pneumoniae) were previously assessed in Gabon by using standard culture methods (8), but the prevalence of fastidious bacteria, which are mainly detected by using molecular techniques, was not studied. There is a need to include these sensitive methods in diagnostic determinations. Our findings, plus those from studies in Senegal (14,22), Mali (14), Kenya (5), Ethiopia (23,24), Cameroon (25), Democratic Republic of the Congo (26,27), Ivory Coast (Côte d'Ivoire) (6), and Zimbabwe (28), show that R. felis is widespread in sub-Saharan Africa countries (29). However, its prevalence changes according to the season, year, area (rural, urban, and semiurban), country, and age of those infected.
A comparison of our findings with previously reported data showed that R. felis prevalence among febrile children in Franceville decreased from 10% in 2012 (14) to 1.3% in 2014. In addition, the variation in the R. felis prevalence between urban (1.3%) and rural (39.7%) areas of Gabon showed that R. felis is unequally distributed in the country. Differences in the prevalence of a possible vector or reservoir, or both, and disparate health care-associated conditions, including environmental conditions, poverty, and the availability and quality of health care facilities, between rural and urban areas may influence the distribution of R. felis in Gabon; however, these factors have not been determined. An increased prevalence of R. felis was observed during the rainy season. This same finding was described in Senegal, where the prevalence of infection in the rural areas was 24 times higher than that in urban areas of Algeria (14). Together, the finding suggests that R. felis is more prevalent in rural areas and during the rainy season in sub-Saharan Africa. R. felis was previously found in C. felis fleas from a pet monkey in Gabon, but the data concerned only 1 region, Franceville (15). R. felis has been reported to be absent from C. felis fleas in rural areas of Senegal, where R. felis is common (30). The factors explaining the spread of R. felis in Gabon should be evaluated in further studies. Although R. felis is widespread in Africa and unequally distributed in Gabon and Senegal (14), its prevalence varies by country: 15.0% in Senegal (14), 3.0% in Mali (14), and 7.2% in Kenya (5). In our study, R. felis was mainly detected in young febrile children <5 years of age (primarily in those 1-3 years of age). In a rural area of Senegal (Dielmo and Ndiop), the incidence of R. felis infection has also been reported to be higher (reaching 36%) among febrile children 1-3 years of age, but the incidence was lower (0.1%) in persons >15 years of age (14). It has been reported that the seroprevalence of R. felis increases with age in areas where the bacterium is endemic (5). This increase is probably due to exposure to the bacterium during the course of a lifetime, leading to protection by a progressive development of immunity against this bacterium in adults.
Most of the fever-associated studies conducted in Africa failed to use a control group of afebrile persons. Consequently, when a pathogen was detected in febrile patients, it was systematically and automatically considered as the cause of fever (7). In some cases, we have also observed a mistake in methodology: data comparisons were performed between samples from afebrile persons in an occidental area and from febrile persons in Africa (31). The epidemiology of microorganisms depends on the studied areas, and the positive predictive value of a disease depends on its symptoms and epidemiology. For example, Plasmodium falciparum, the primary agent of malaria, is commonly detected in blood specimens from apparently healthy, afebrile persons in Sub-Saharan Africa; prevalence can reach 20% in Ethiopia and 32% in Senegal (32,33). Respiratory viruses, including influenza virus, have also been found in 12% of nasopharyngeal samples of asymptomatic Hajj pilgrims (34). More recently in Tanzania, the prevalence of S. pneumoniae DNA was less frequently detected in febrile (5.1%) than afebrile (6.3%) persons (35). Thus, these examples show that even wellknown pathogens may be detected in blood or respiratory secretions of afebrile persons. The inclusion of control groups of participants in studies is indispensable to a better understanding of infectious diseases; the use of controls has shown the existence of carriers of well-known pathogens, emphasizing that it is not easy to interpret data about the potential pathogenic role of a microorganism. The finding of R. felis in febrile versus afebrile persons is not well characterized. The overall prevalence of R. felis in Gabon was higher in febrile (10.2%) than afebrile (3.3%) children, but the difference was not statistically significant (p = 0.09). Of more interest, in rural Fougamou, the difference in prevalence was significantly higher in similarly aged children with and without fever (39.7% vs. 5.0%, respectively; p = 0.004). Therefore, the presence of R. felis in febrile and afebrile persons should not exclude that this bacterium is a cause of fever in sub-Saharan Africa.
In 2010, independent research teams detected R. felis in blood specimens from febrile patients in 2 different areas of Africa (eastern and western) (3,4). In other studies, these teams confirmed and extended the preliminary data: 1 team showed that the presence of R. felis was 2.2 times higher in blood specimens from febrile persons compared with afebrile persons in Kenya (5), and the other team showed that the prevalence of R. felis was significantly higher in febrile (15.0%) than afebrile (4.0%) persons in Senegal (14). In Senegal, an 8-month-old febrile girl was cured of R. felis infection after treatment with doxycycline (36). The presence of R. felis has also been observed in blood specimens from febrile patients in Asia (37). The higher prevalence of R. felis among febrile persons compared with healthy persons in our study led us to suspect that this microorganism plays the role of pathogen. However, the presence of the microorganism may be a cofactor or the cause of a previous event not yet determined. Another hypothesis would be that blood specimens may be contaminated by surface bacteria, including R. felis, which has been detected on the skin of healthy persons in Senegal (38).
In summary, the R. felis bacterium is widespread in Gabon, but it primarily occurs in rural areas and is most prominent during the rainy season. R. felis is also more prevalent among febrile than afebrile children in rural areas of Gabon. More studies will help to better understand the pathogenic role of R. felis in this part of the world.
Search past issues of EID at wwwnc.cdc.gov/eid