INTRODUCTION
Members of the genus
Bartonella are fastidious Gram-negative facultative intracellular bacteria that are typically transmitted by arthropod vectors (
1).
Bartonella spp. are distributed worldwide and have been found to cause clinically overt disease, mainly in humans and dogs (
2). Further, they have been detected in a large array of potential reservoir hosts, in which they establish a persistent intraerythrocytic bacteremia (
3). The most relevant species for human health are
Bartonella bacilliformis,
B. quintana, and
B. henselae. However, in recent years other
Bartonella spp. were recognized as human pathogenic as well, although they have been reported to affect humans only infrequently. The most prominent disease that they cause is culture-negative endocarditis (
1).
Only two human infections with
B. washoensis have been reported so far.
B. washoensis was isolated in North America from the blood of two patients with myocarditis (
4) and meningitis (
5), respectively. Further, it was cultured from the blood of a Californian dog with endocarditis (
6). Squirrels were suggested to be the reservoir hosts for human infection, because the 16S rRNA,
groEL, and
gltA sequences of
B. washoensis strain Sb944nv cultured from a Californian ground squirrel (
Spermophilus beecheyi) were 100% identical to those from the patient with myocarditis mentioned above (
4). In North America, one of the arthropod vectors that might transmit
B. washoensis to humans is the
S. beecheyi-parasitizing flea
Oropsylla montana, because
gltA sequences from fleas of this species were 100% identical to those from human infection (
5).
Here, we report the case of a 75-year-old German woman with prosthetic valve endocarditis due to B. washoensis, the molecular characterization of the infectious agent using a multilocus sequence typing (MLST) approach, and its detection in European red squirrels (Sciurus vulgaris).
DISCUSSION
We report here what was, to our knowledge, the first human infection with
B. washoensis in Europe, as the two other known cases occurred in North America (
4,
5). Further, this report represents the first recognition of
B. washoensis as a causative agent of human endocarditis, although it had been isolated earlier from a Californian dog with endocarditis (
6). Interestingly, humans and dogs have been described to suffer from similar disease manifestations due to
Bartonella species infections (
2), including endocarditis (
11). Apart from the common species
B. quintana and
B. henselae, several other
Bartonella spp. have been shown to cause culture-negative endocarditis (
1). Thus, this seems to be a general feature of the genus
Bartonella.
In a small case series of 48 patients, preexisting valvular heart disease was associated with
B. henselae but not with
B. quintana endocarditis (
12). However, as valvular heart disease is a major risk factor for endocarditis in general, the mitral valve stenosis of her prosthetic valve probably predisposed our patient to such disease. Typically, patients with
Bartonella endocarditis present with nonspecific symptoms such as fever, fatigue, and weight loss (
1), but such was not the case in our patient. Echocardiographic evidence of vegetation has been described previously to be not as easily detectable in
Bartonella endocarditis as in other forms of endocarditis (
1). This was also true for the patient presented here, as no echocardiographic signs of endocarditis were evident preoperatively. However, intraoperatively, cauliflower-like plaques were observed on the mitral valve leaflets and the histopathological analysis was consistent with active endocarditis.
A
Bartonella IgG titer of ≥1:800 is regarded as a cutoff value associated with a high positive predictive value for endocarditis (
1). Broad serological cross-reactivity between members of the genus
Bartonella has been observed previously (
1). Thus, in our patient,
B. henselae and
B. quintana were used as surrogate antigens, yielding an IgG antibody titer of 1:2,048 for
B. henselae in the immunofluorescence test, which was substantially above the cutoff level of 1:800. The immunofluorescence test is considered to be not species specific and shows considerable cross-reactivity between
Bartonella spp. (
1), as further supported by our findings. Although there is no clinical experience in the management of
B. washoensis endocarditis, treatment for
Bartonella species endocarditis according to current guidelines was recommended (
13).
Squirrels were suggested to be the reservoir for human infection, because the 16S rRNA,
groEL, and
gltA sequences of
B. washoensis strain Sb944nv cultured from a Californian ground squirrel (
S. beecheyi) were 100% identical to those from a human patient (
4). The
B. washoensis human_1487_18 strain described here showed the highest nucleotide identities with red squirrel strain ER14-3 from
S. vulgaris orientis from China and with red squirrel strain SC12uk from
S. vulgaris from the United Kingdom. The red squirrel strain from the United Kingdom was described as
Bartonella sp. because it did not group with known
Bartonella spp. in the phylogenetic analysis at the time of publication (
14). However, we compared it to GenBank strains and found that its
gltA sequence was 99.4% (329/331 bp) identical to that of red squirrel strain ER14-3 (GenBank accession number
AB444974) from China and 94.0% (311/331 bp) identical to a
B. washoensis strain from
S. beecheyi from North America (GenBank accession number
AF470616). A similar observation was made for its
groEL sequence, which showed identity of 98.4% (1,166/1,185 bp) to red squirrel strain ER14-3 (GenBank accession number
AB519097) and of 96.1% (1,139/1,185 bp) to a
B. washoensis strain from a Siberian chipmunk (
Tamias sibiricus) (GenBank accession number
AB519090). Therefore, red squirrel strain SC12uk might represent
B. washoensis as well.
On the other hand, red squirrel strain ER14-3 and our human 1487_18 strain formed a separated cluster in the phylogenetic analysis (
Fig. 2). Thus, these strains might belong to a subspecies that has yet to be defined. A similar observation was made for
B. washoensis isolates from black-tailed prairie dogs (
Cynomys ludovicianus) that were previously proposed to form the subspecies “
Candidatus B. washoensis subsp. cynomysii” (
15). The name of this subspecies was later corrected to “
Candidatus B. washoeensis subsp. cynomyis” (
16).
To prove that the
B. washoensis strain described here was present in the European red squirrel (
S. vulgaris) population, we investigated 62 animals from the Netherlands. A relevant proportion of 39% (24/62) was found to be infected, and their
gltA sequences were 100% identical to our patient strain. Thus, red squirrels might serve as reservoir hosts for human infection. Note that the widely used primers for screening of
Bartonella infection described by Norman et al. (
9) did not amplify the
B. washoensis gltA variant we described here. An infection rate of 69% (97/140) with
B. washoensis was reported for the Californian ground squirrel (
S. beecheyi) (
17), which is somewhat higher than the infection rate of 39% that we detected in red squirrels. Carnivores such as the Japanese marten (
Martes melampus) might not be the reservoir for human infection with the
B. washoensis variant that we describe here, because its concatenated housekeeping gene sequence (16S rRNA,
ftsZ,
gltA,
groEL,
ribC, and
rpoB) did not cluster with red squirrel strain ER14-3 (
18).
The
gltA sequences from squirrels and humans available at GenBank were compared to the unique
gltA sequence found here, and the results showed that the strains from humans and squirrels from North America as well as from Europe and Asia clustered together (
Fig. 2). Thus, instead of a separate subspecies, strain human_1487_18 could represent a Eurasian variant of
B. washoensis pathogenic for humans.
In North America, one of the arthropod vectors that might transmit
B. washoensis to humans is the
S. beecheyi-parasitizing flea
O. montana because
gltA sequences from fleas of this species were 100% identical to those from human infection (
5). However, flea transmission of
B. washoensis has not been proven experimentally (
3). In Europe, red squirrels (
S. vulgaris) are mainly parasitized by the flea
Ceratophyllus sciurorum. It has to be shown in the future which
B. washoensis variants might be present in
C. sciurorum to gain evidence about its vector competence for
B. washoensis.
At the least, patients with
B. henselae endocarditis reported significantly more frequent exposure to cats or cat fleas than the controls (
12). Our patient denied contact with animals and their ectoparasites. This, however, does not exclude the possibility that she was infected via transmission by fleas, because it is a common finding in vector-borne diseases in general that the exposure to the vector is not recognized. For example, only 60% of patients with erythema migrans due to
Borrelia burgdorferi infection recalled the tick bite (
19).
In conclusion, we present the first human case of B. washoensis endocarditis. Further studies are needed to show whether other reservoir hosts apart from red squirrels (S. vulgaris) might harbor human-pathogenic B. washoensis variants and whether fleas parasitizing red squirrels might be involved in their transmission.