Bartonella Endocarditis in Spain: Case Reports of 21 Cases

Blood culture negative endocarditis (BCNE) is frequent in infective endocarditis (IE). One of the causes of BCNE is fastidious microorganisms, such as Bartonella spp. The aim of this study was to describe the epidemiologic, clinical characteristics, management and outcomes of patients with Bartonella IE from the “Spanish Collaboration on Endocarditis-Grupo de Apoyo al Manejo de la Endocarditis infecciosa en España (GAMES)”cohort. Here we presented 21 cases of Bartonella IE. This represents 0.3% of a total of 5590 cases and 2% of the BCNE from the GAMES cohort. 62% were due to Bartonella henselae and 38% to Bartonella quintana. Cardiac failure was the main presenting form (61.5% in B. hensalae, 87.5% in B. quintana IE) and the aortic valve was affected in 85% of the cases (76% in B. henselae, 100% in B. quintana IE). Typical signs such as fever were recorded in less than 40% of patients. Echocardiography showed vegetations in 92% and 100% of the patients with B. henselae and B. quintana, respectively. Culture was positive only in one patient and the remaining were diagnosed by serology and PCR. PCR was the most useful tool allowing for diagnosis in 16 patients (100% of the studied valves). Serology, at titers recommended by guidelines, only coincided with PCR in 52.4%. Antimicrobial therapy, in different combinations, was used in all cases. Surgery was performed in 76% of the patients. No in-hospital mortality was observed. One-year mortality was 9.4%. This article remarks the importance for investigating the presence of Bartonella infection as causative agent in all BCNE since the diagnosis needs specific microbiological tools and patients could benefit of a specific treatment.


Introduction
Infective endocarditis (IE) is a fatal condition without treatment with an incidence oscillating worldwide between 1.5 and 9.6 cases/100,000 inhabitants [1] although in developed countries it is estimated between 3.1 and 3.7 episodes per 100,000 inhabitants per year [2,3]. The highest incidence occurs in elderly patients with degenerative heart disease and comorbidities such as chronic liver disease, kidney failure and neoplasms [4,5]. In our environment and according to a prospective multicenter registry of 25 Spanish hospitals that analyzed 1804 patients, it has been observed that it mainly affects men (68%) of 55-77 years age, associated, in a considerable percentage, with health procedures or nosocomial environment [6]. Regarding the etiological agents involved in its etiology, 80% of the cases are due to Gram-positive bacteria, mainly due to Staphylococcus spp., followed by Streptococcus spp. and a long list of other bacterial species [6,7].
Blood culture remains the most frequently used technique for etiological diagnosis, although other techniques such as molecular biology have been generalized for microbiological diagnosis. In the Spanish series, 14.7% were blood culture negative endocarditis (BCNE) and in 9.1%, despite using other techniques such as molecular and serological, the etiology was unknown [6]. In the large series of the International Collaboration on Endocarditis in which 2781 patients were collected, BCNE accounted for 11.1% [7]. However, depending on the place and the available techniques where the studies were carried out, BCNE can increase up to 68 and 70% [8,9]. This fact is important since the absence of a microbiological diagnosis, at least in our environment, could be an independent predictor of mortality [10]. There are many causes of BCNE, including difficult-to-culture microorganisms among which are Bartonella spp.
Bartonella spp. are fastidious Gram-negative bacilli, facultative intracellular bacteria, belonging to the Alpha-2 subgroup of the phylum Proteobacteria and includes more than 35 validated species and three subspecies [11,12]. These bacteria are usually transmitted to humans through animal bites or scratches (cats, dogs and other animals) or by scratch inoculation of infected flea and body louse feces [13]. In addition, sand-fly vectorcompetence was proven for transmission of Bartonella bacilliformis in Peru during the last century [14]. Other arthropods such as ticks, head lice, bedbugs, bat flies and mites have been associated with human Bartonella spp. infections, but their role as vectors remains to be verified [13,[15][16][17].
Bartonella spp. have been associated with an expanding clinical spectrum in humans and animals. Some species, such as Bartonella bacilliformis (Oroya fever, Verruga Peruana or Carrión's disease) cause potentially life-threatening illness, but are limited geographically by the transmitting vector (confined to the highlands of Peru, Colombia and Ecuador) whereas other species such as Bartonella quintana are transmitted by lice under poor hygienic conditions throughout the world. B. quintana the etiological agent of trench fever, also is cause of bacillary angiomatosis and peliosis hepatis in HIV patients, chronic bacteremia, chronic lymphadenopathy, as well as BCNE [13,18]. Bartonella henselae is the most frequent etiological agent of a sub-acute and chronic lymphadenopathy named cat-scratch disease (CSD) in children and teenagers and occurs throughout the world. A subset of CSD patients develop severe or systemic disease manifestations, including endocarditis, osteomyelitis, granulomatous hepatitis and hepatosplenic abscess [19][20][21][22][23][24][25][26]. Besides these, at least 15 other Bartonella spp. have been associated with human diseases or have been detected in humans.
Regarding endocarditis due to Bartonella spp., at least eight species have been involved. B. henselae and B. quintana are present in more than 99% of the cases, although other Bartonella spp. such as Bartonella elizabethae, Bartonella vinsonii subsp. berkhoffii, B. vinsonii subsp. arupensis, Bartonella koehlerae, Bartonella alsatica and "Candidatus Bartonella mayotimonensis", have been occasionally reported [15,[27][28][29][30][31][32][33][34]. In any case, there are few published series of Bartonella IE and when searching about this condition, there are some differences in the epidemiology, such as the distribution of the involved Bartonella spp., and the clinical course.
The aim of this study was to describe the epidemiologic, clinical characteristics, management and outcomes in a cohort of patients with IE due to Bartonella spp. from Spain.

Clinical Features
According to the medical records, 14 of the 21 patients had previous heart disease (9/13 (69.2%) of the patients with IE due to B. henselae and 5/8 (62.5%) of the patients with IE due to B. quintana). Ten of these patients (41.6%) had previous valve disease (5/13 (38.4%) of the patients with B. henselae IE and 5/8 (62.5%) of the patients with B. quintana IE). Prosthetic valve was present only in one of the B. henselae patients (1/13; 7.7%) and two of this group also had a coronary disease (2/13; 15.4%), three had congenital heart disease (3/13; 23.1%). Two of the patients with B. quintana IE (2/8; 25%) and one of the B. henselae group (1/13; 7.7%) presented previous heart failure. Other medical conditions are shown in Table 2.
Physical examination focusing on cardiac auscultation showed a new regurgitant murmur in 4/13 (30.8%) and 5/8 (62.5%) of patients with B. henselae and B. quintana IE, respectively, while classical clinical signs of endocarditis, such as petechiae, were only observed in one patient of the B. henselae group. Embolisms were observed in 4/13 (30.8%) and 1/8 (12.5%) of patients with B. henselae and B. quintana, respectively, (all of them affecting CNS in the B. quintana group while only one affected this structure in the B. henselae patients). New renal function impairment was observed in four of the 13 patients infected by B. henselae (30.8%) and in 5/8 (62.5%) by B. quintana. Two patients had splenomegaly, one of each group (1/13 (7.7%) of B. henselae and 1/8 (12.5%) of B. quintana). Only one patient with B. henselae IE presented glomerulonephritis (1/13; 7.7%). One patient of the B. quintana group presented a new conduction abnormality (1/8; 12.5%) and one of the B. henselae group had septic shock (1/13; 7.7%).   Cardiac failure    3%) and 100% of the patients with B. henselae and B. quintana, respectively. The size of the vegetations was between 5 and 25 mm in patients infected by B. henselae (mean ± SD: 13.8 ± 6.5 mm) and between 7 and 13 mm in those infected by B. quintana (mean ± SD: 10.6 ± 2.2 mm).
The involved valve was the aortic in 18/21 cases (85.7%), although seven of them (7/18; 38.9%) also presented mitral valve involvement. In three patients the unique valve involved was the mitral (3/21; 14.3%). According to the Bartonella spp. implicated, in patients with B. henselae IE, the aortic valve was affected in ten (10/13; 76.9%), three of them also presented mitral valve involvement. In three patients (3/13; 23.7%) the only valve affected was the mitral. Nevertheless, in patients infected by B. quintana the aortic valve was affected in all patients with half of them also presenting mitral valve involvement.
Native Laboratory records showed alterations in C-reactive protein and erythrocyte sedimentation rate values. These data are shown in Table 2. Anemia (hemoglobin (Hb) levels < 12.0 g/dL in women and < 13.0 g/dL in men) was observed in ten of the thirteen patients infected with B. henselae (76.9%) and in six of the eight patients infected with B. quintana (75%).
Surgery was performed in 16/21 patients (76.2%). Urgent surgery was performed in three cases (3/16; 18.8%) and emergency surgery in one (1/16; 6.3%). Table 3 shows the techniques used for the microbiological diagnosis, the results obtained depending of the applied technique and the coincidence among them. It should be noted that serology was carried out in 17/21 patients (80.9%) and, initially, 6/17 (35.3%) had titers < 800 and 9/17 (53%) had titers ≥ 512. In the five cases in which the diagnosis was only made by serology, either the serologies were positive only for B. henselae while negative for B. quintana, or there was a greater rise in titer for B. henselae (more than three titers).

Treatment and Outcome
According to European guideliness, the proposed therapy is doxycycline 100 mg/12 h orally for four weeks plus gentamicin (3 mg/24 h) i.v. for two weeks. However, several therapeutic regimens have been reported, including aminopenicillins (ampicillin or amoxicillin, 12 g/24 h i.v.) or cephalosporins (ceftriaxone, 2 g/24 h i.v.) combined with aminoglycosides (gentamicin or netilmicin) [35]. Antimicrobial therapy for each patient is shown in Table 4.  . Two patients were readmitted (2/21; 9.5%). Neurological or cardiac sequelae (e.g. dysarthria, hemiparesis, hemiplegia, heart failure) were present in three patients (3/21; 14.3%). None of the patients had a relapse. None of the patients died during the hospital stay. Two patients died two and three months after the process in relationship with the IE.

Discussion
To date, IE remains a serious disease all over the world, including developed countries. In most cases, Gram-positive cocci are the causative agents but fastidious organisms such as Bartonella spp., Coxiella burnetii, Tropheryma whipplei and others must be considered. In the GAMES cohort, Bartonella spp. is the third most common cause of BCNE behind C. burnetii and T. whipplei [36]. There a few Bartonella series and here we present a new one that is of the most numerous series from a developed country.
The prevalence of Bartonella endocarditis varies among countries. A majority of cases of Bartonella IE have been reported from Europe and the Americas. However, cases from Asia, Africa, and Australia have been also documented, suggesting a worldwide distribution [37][38][39][40][41][42][43][44][45][46][47]. In Europe, Bartonella spp. account for 0-4.5% of the infective endocarditis and 20-30% of BCNE endocarditis [48][49][50][51][52]. In UK Bartonella IE represent 1.1% of IE [53]. However, Bartonella spp. as a causative agent of infective endocarditis reaches 9.8% in Tunisia or even 15.6% in Algeria [39,54]. In Spain, last published data of the GAMES cohort, which included patients recruited beteween 2008 and 2012, estimated that Bartonella spp. were the cause of endocarditis in 0.2% of the cases [6,36]. Here we report the main characteristics of 21 patients affected by Bartonella spp. Taking into account that just 17 of them were from hospitals belonging to the GAMES Cohort, the rate of Bartonella spp. as a causative agent of endocardits recorded in the GAMES Cohort between 2008 and 2020, is 0.3% of the total of IE, and 2% of the BCNE registered. These data are far below those reported in the previously cited series. Nevertheless, the prevalence of Bartonella endocarditis could be underestimated due to the difficulties involved in its identification. Probably, data here do not represent the true incidence of Bartonella IE in our country but in our opinion is the best approximation.
To diagnose endocarditis due to Bartonella spp. is not easy since the clinical presentation is not typical. Fever is not presesent in more than half of the cases and patients usually come to the hospital due to complications such as cardiac failure. This fact is very important since it reinforces the need to request echocardiography in these patients, that is the gold standard for diagnosing endocarditis. Other typical signs such as embolic events are also rare in these patients. However, as in other IE, the existence of previous valve disease is a risk factor for endocarditis due to Bartonella spp. [19]. Preexisting valvular abnormalities are a common condition in patients with Bartonella endocarditis, in more than 40% of the cases, reaching 62.5% of the patients with B. quintana IE. As previously described, results presented here show that the aortic valve is the most affected one, affecting all cases of B. quintana here described and almost 80% of the endocarditis due to B. henselae [19,24,39,43,47,48,[52][53][54][55][56][57].
The epidemiology in the cases presented here differs from most series. According to the literature, B. quintana is the causative agent in approximately 75% of all cases, while B. henselae occurs in around 25% of all infective endocarditis cases [47]. However, it is important to note that in the data presented here, B. henselae accounts for almost 62% of the cases while the remaining are due to B. quintana. B. henselae as the main Bartonella spp. causing IE has been only reported in just two other series [43,55].
None of our patients referred body lice contact or an undomiciled/homeless situation and alcohol abuse was observed only in very few cases. These factors, which have been classically associated with B. quintana infection [57], could explain the low prevalence of B. quintana endocarditis in our environment campared with other series such as the French, Tunisian or Algerian ones where the prevalence of B. quintana IE is high [19,39,54]. It should also be noted that, in our series, a not insignificant percentage of patients with B. quintana IE were immigrants. These patients were from Moldavia, Pakistan and Morocco but although they were diagnosed in Barcelona, Spain, the infection most likely happened in their country of origin since two of them had just arrived in Barcelona and, since in the case of the Moroccans, they spent long periods in Morocco. Other classical epidemiological factors such as cat contact, which has been associated with an increased risk of B. henselae infection, is not a prominent factor in our series. Moreover, both species of Bartonella described here affect men more frequently than women and, the age of presentation (49 years) is lower than in the endocarditis caused by the "usual agents" [6,7] and even lower than in patients with Bartonella IE recorded in the French series [19].
The systematic use of serological and molecular assays in the study of BCNE has improved the diagnosis of this entitity. When Bartonella IE is suspected, it can be diagnosed by blood or valve culture (not very sensitive with the usual techniques), molecular assays such as PCR on blood or excised tissue, that in our series is the most usuful technique, and by serological procedures. Other techniques such as silver nitrate-based staining (Warthin-Starry stain) in affected tissues are also very suggestive [58]. Before the incorparation of molecular techniques, most of the cases of Bartonella endocarditis were diagnosed using serological tests, especially immunofluorescence assay (IFA) that to date are recommended in the guideliness and by experts [35,59]. In the series presented here, it is remarkable that serology has a poor correlation for the diagnosis if the cut-off point recommended in the guideliness is used [35,59]. As previously described [19], we have also observed that 35% of the patients with definitive endocarditis produced by Bartonella spp. confirmed by PCR do not have the recommend cut-off point of IgG ≥ 800 by IFA. In this sense, molecular procedures are likely to become a key tool to improve BCNE diagnostics and contribute to a better understanding of the aetiology [60]. Moreover, the limitations of serological tests should be recognized due to the high level of serological cross-reactions between Bartonella and Chlamydia/Chlamydophila [61] and the difficulty to distinguish between species of Bartonella which further complicates the diagnosis of this type of endocarditis. In the last few years, new methods improving the sensitivity to diagnose Bartonella infections have been described that may be validated for the use in clinical practice, e.g., a method based on a liquid enrichment Bartonella alphaproteobacteria growth medium (BAPGM) followed by PCRs for the amplification of Bartonella spp. [62].
Most of the patients in this series, although in different combinations, had antimicrobial therapy according to guideliness. It is important to emphasize that most of the patients presented here required surgical treatment, perhaps since the diagnosis of Bartonella endocarditis, and in general of BCNE, was later and optimal treatment was not always included early. Mortality due to Bartonella IE in the first descriptions was significant with rates ranging 7 to 30%. However, in recent times mortality rates closer to 7% have been shown, probably due to the improved diagnostic and therapeutic measures [9,47,51,[63][64][65][66][67][68]. These data are in agreement with those reported here (mortality rate observed here was 9.4%).

Patients' Recruitment
Patients diagnosed with Bartonella IE were recruited from a registry of the "Spanish Collaboration on Endocarditis -Grupo de Apoyo al Manejo de la Endocarditis infecciosa en España (GAMES)". GAMES is formed by 25 Spanish hospitals that makes a prospective registry including aspects related to IE. Furthermore, four cases from the Vall d Hebron Hospital (Barcelona), which is not included in the Games Cohort, were also recruited. The necessary data for this study of patients from Vall d Hebron Hospital were obtain retrospectively from their medical records by their physician. Between 1 January 2008 and 31 December 2020, multidisciplinary teams prospectively completed a standardized case report form containing epidemiological, clinical, biological (including main hematological and biochemical values) and therapeutic data for each patient. This case report form was designed to record any type of endocarditis and does not include specific variables for Bartonella endocarditis, therefore, some data of interest for this type of endocarditis, such as contact with cats, fleas, body lice, undomiciled/ homeless situation or history of alcohol abuse (considering alcohol abuse, according to the WHO, the consumption of more than 40 g for men and more than 25 g for women), were retrospectively reviewed from the clinical record of each patient. Due to this information was restricted to what the clinician has entered into the medical records some data could be underestimated. Regional and local ethics committees approved the study (Comité Ético para la Investigación Clínica-Regional de la Consejería de Sanidad de la Comunidad de Madrid, code: 18/07; Date: 11 January 2008), and patients gave their informed consent for entering the cohort [6].

Definitions
IE and Bartonella infection as causative agent was defined according to the modified Duke criteria [69]. In our clinical practice a routine protocol that includes serology to Coxiella burnetii, Bartonella spp., Legionella spp. and Brucella spp. was completed in all patients with BCNE. When valve surgery was carried out molecular assays (PCR) or culture were performed. Molecular tests were performed using Bartonella spp. rpoB gene and broadrange 16S rRNA gene and subsequent sequencing [70,71]. Serological studies were carried out in each clinical laboratory included in the study using commercial immunofluorescence assays. Taking into account the possibility of cross-reactions between Bartonella spp., the species was determined based on serological assays if results were positive only for one species or if there was at least a three-fold difference in titer against one species compared to the other one.

Statistics
Statistical analysis was performed using SPSS, version 26.0 (IBM Corporation, Armonk, NY, USA).

Conclusions
Bartonella IE is not a frequent cause of IE that should be suspected in all patients with BCNE. In these cases, the index of suspicion should be high even if contact with animals has not been reported or there are no risk factors such as body lice parasitization. A serology for Bartonella spp. negative or lower than the recommended cut-off does not exclude the possibility of a Bartonella IE. The prognosis of Bartonella IE was better in this study than in IE due to other causes observed in the Games Cohort [6,10], and involves valve replacement in a high percentage of the cases.  Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki, and approved by the Comité Ético para la Investigación Clínica-Regional de la Consejería de Sanidad de la Comunidad de Madrid (code: 18/07; Date: 11 January 2008).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Not applicable.