Rickettsia felis in Fleas, Southern Ethiopia, 2010

To the Editor: Fleas (order Siphonaptera) are obligate hematophagous insects. They are laterally flattened, holometabolous, and wingless ectoparasites. More than 2,500 species of flea, belonging to 16 families and 238 genera, have been described. A minority of these genera live in close association with humans (synanthropic), including fleas of these species: Pulex irritans, Ctenocephalides felis, Ctenocephalides canis, Xenopsylla cheopis, Nosopsyllus fasciatus, Echidnophaga gallinacea, and Tunga penetrans (1). Many fleas are capable of transmitting the following pathogens to their hosts: bacteria (e.g., Rickettsia typhi, R. felis, Yersinia pestis, and many Bartonella spp.); viruses (e.g., myxoma virus); protozoa (e.g., Trypanosoma spp.); or helminths (e.g., Hymenolepis spp.) (2). Ctenocephalides spp. fleas are of special interest as main reservoirs and vectors of R. felis, because this agent causes an emerging disease, fleaborne rickettsiosis. The distribution and prevalence of this disease have not been well studied. Symptoms of this disease range from mild to moderate and include fever, cutaneous rash, and sometimes an inoculation eschar (3,4). R. felis can also infect at least 10 other species of arthropods, including P. irritans fleas, trombiculid and mesostygmata mites, hard and soft ticks, and booklice (5,6). 
 
In Africa, the presence of R. felis in fleas has been documented in Algeria, Tunisia, Egypt, Ethiopia, Gabon, Cote d’Ivoire, and the Democratic Republic of Congo (5). Recent studies conducted in Senegal (3) and Kenya (4) have shown that as much as 4.4% and 3.7%, respectively, of acute febrile diseases in these regions may be caused by R. felis infections. We conducted a study to determine the distribution and prevalence of R. felis in fleas in Ethiopia. 
 
In our study, 55 fleas were collected in 2010 in 2 villages in Ethiopia; 25 fleas were collected from Tikemit Eshet (6°51′837″N and 35°51′348″E; altitude2,121 m), and 30 fleas were collected from Mizan Teferi (6°59′640″N and 35°35′507″E; altitude 1,700 m). The specimens were collected by using a plate filled with soapy water with a candle in the middle of the plate. Because fleas are thermotropic, they jumped toward the candle and fell onto the plate, where they rapidly drowned in the soapy water. The fleas were identified by morphologic features and stored in 90% ethanol until DNA extraction. 
 
To confirm the phenotypic identification, we designed primers and probes for quantitative real-time PCR (qPCR) that were specific for 2 species of flea (P. irritans and C. felis) based on the sequences of mitochondrial cytochrome oxidase gene published in GenBank (Table). All of the identifications made by morphologic appearance were confirmed by qPCR because some specimens were damaged and difficult to identify. We found that most (52/55) of the fleas collected in human dwellings were P. irritans, and 3 specimens were C. felis. A screening by amplification using primers and probes specific for the 16S–23S internal transcribed spacer of Bartonella spp (7). produced no positive results. 
 
 
 
Table 
 
Sequences of primers and probes used to identify fleas by quantitative real-time PCR, southern Ethiopia, 2010 
 
 
 
We screened rickettsial DNA by using qPCR with a Rickettsia-specific, gltA gene-based RKND03 system (8) and a bioB-based qPCR system specific for R. felis. We found that the 3 specimens of C. felis fleas contained the DNA of R. felis; however, 23 (43%) of 53 P. irritans specimens also contained DNA of R. felis. We amplified and sequenced nearly the entire rickettsial gltA gene from 3 C. felis and 10 P. irritans specimens and found that the sequence was identical to that of R. felis (GenBank accession no. {"type":"entrez-nucleotide","attrs":{"text":"NC_007111","term_id":"67459862","term_text":"NC_007111"}}NC_007111). 
 
During the field collection of the fleas, the conservation of specimens may be difficult. Degradation of specimens may pose a problem for the ensuing morphologic identification. For fleas, a specific preparation is required that destroys internal organs and produces a chitin complex of the insect. This type of preparation makes it difficult, and sometimes impossible, to use the insect later for molecular studies. The development of qPCR specific for P. irritans and C. felis fleas facilitated the identification of damaged samples and also precluded the laborious and time-consuming procedure of identification by morphologic features. 
 
We conclude that the reservoirs of R. felis in Ethiopia include both C. felis and P. irritans fleas. In Ethiopia, P. irritans fleas have been reported to be prevalent (9). P. irritans fleas have been shown to be infected by R. felis in several locations, notably in the Democratic Republic of the Congo and in the United States, and another rickettsia phylogenetically similar to R. felis has been detected in P. irritans fleas in Hungary (10). Reports attributing substantial numbers of acute febrile illnesses to fleaborne rickettsiosis caused by R. felis in Senegal and Kenya (3,4) place fleaborne rickettsiosis among emerging diseases with the potential for adverse public health effects. Furthermore, the identification of the vectors of R. felis in Ethiopia reveals the epidemiologic background for the fleaborne spotted fever in this region. We speculate that the elucidation of the full range of possible vectors of R. felis may facilitate the development of prevention measures that will help control this disease.

In Africa, the presence of R. felis in fl eas has been documented in Algeria, Tunisia, Egypt, Ethiopia, Gabon, Côte d'Ivoire, and the Democratic Republic of Congo (5). Recent studies conducted in Senegal (3) and Kenya (4) have shown that as much as 4.4% and 3.7%, respectively, of acute febrile diseases in these regions may be caused by R. felis infections. We conducted a study to determine the distribution and prevalence of R. felis in fl eas in Ethiopia.
In our study, 55 fl eas were collected in 2010 in 2 villages in Ethiopia; 25 fl eas were collected from Tikemit Eshet (6°51′837″N and 35°51′348″E; altitude2,121 m), and 30 fl eas were collected from Mizan Teferi (6°59′640″N and 35°35′507″E; altitude 1,700 m). The specimens were collected by using a plate fi lled with soapy water with a candle in the middle of the plate. Because fl eas are thermotropic, they jumped toward the candle and fell onto the plate, where they rapidly drowned in the soapy water. The fl eas were identifi ed by morphologic features and stored in 90% ethanol until DNA extraction.
To confi rm the phenotypic identifi cation, we designed primers and probes for quantitative real-time PCR (qPCR) that were specifi c for 2 species of fl ea (P. irritans and C. felis) based on the sequences of mitochondrial cytochrome oxidase gene published in GenBank (Table). All of the identifi cations made by morphologic appearance were confi rmed by qPCR because some specimens were damaged and diffi cult to identify. We found that most (52/55) of the fl eas collected in human dwellings were P. irritans, and 3 specimens were C. felis. A screening by amplifi cation using primers and probes specifi c for the 16S-23S internal transcribed spacer of Bartonella spp. (7) produced no positive results.
We screened rickettsial DNA by using qPCR with a Rickettsia-specifi c, gltA gene-based RKND03 system (8) and a bioB-based qPCR system specifi c for R. felis. We found that the 3 specimens of C. felis fl eas contained the DNA of R. felis; however, 23 (43%) of 53 P. irritans specimens also contained DNA of R. felis. We amplifi ed and sequenced nearly the entire rickettsial gltA gene from 3 C. felis and 10 P. irritans specimens and found that the sequence was identical to that of R. felis (GenBank accession no. NC_007111).
During the fi eld collection of the fl eas, the conservation of specimens may be diffi cult. Degradation of specimens may pose a problem for the ensuing morphologic identifi cation. For fl eas, a specifi c preparation is required that destroys internal organs and produces a chitin complex of the insect. This type of preparation makes it diffi cult, and sometimes impossible, to use the insect later for molecular studies. The development of qPCR specifi c for P. irritans and C. felis fl eas facilitated the identifi cation of damaged samples and also precluded the laborious and time-consuming procedure of identifi cation by morphologic features.
We conclude that the reservoirs of R. felis in Ethiopia include both C. felis and P. irritans fl eas. In Ethiopia, P. irritans fl eas have been reported to be prevalent (9). P. irritans fl eas have been shown to be infected by R. felis in several locations, notably in the Democratic Republic of the Congo and in the United States, and another rickettsia phylogenetically similar to R. felis has been detected in P. irritans fl eas in Hungary (10). Reports attributing substantial numbers of acute febrile illnesses to fl eaborne rickettsiosis caused by R. felis in Senegal and Kenya (3,4) place fl eaborne rickettsiosis among emerging diseases with the potential for adverse public health effects. Furthermore, the identifi cation of the vectors of R. felis in Ethiopia reveals the epidemiologic background for the fl eaborne spotted fever in this region. We speculate that the elucidation of the full range of possible vectors of R. felis may facilitate the development of prevention measures that will help control this disease. Five months after treatment initiation, the patient experienced severe abdominal pain, diarrhea, and continued weight loss. Lymph node biopsy was repeated; results showed intramacrophagic coccobacilli tinted with Ziehl-Neelsen, Gram, and periodic acid-Schiff (PAS) stains. Two diagnoses were considered: malakoplakia and Whipple disease (WD). Screening results from quantitative real-time PCR (qPCR) for Tropheryma whipplei were negative for blood, saliva, stools, urine, and lymph nodes.

Oleg Mediannikov, Alemseged Abdissa, Georges Diatta, Jean-François Trape, and Didier Raoult
Although no characteristic Michaelis-Gutmann bodies were seen, the staining characteristics of the intracellular coccobacilli were compatible with Rhodococcus equi, a pathogen associated with malakoplakia. Combined treatment with ertapenem, teicoplanin, and amikacin was implemented but failed to induce clinical improvement. Culture of the biopsy specimen failed to grow R. equi or mycobacteria, and