Case series of 12 Bartonella quintana endocarditis from the Southwest Indian Ocean

Background Bartonella spp. are fastidious bacteria frequently identified as the cause of blood culture-negative (BCN) endocarditis. However, Bartonella infections are difficult to diagnose in routine laboratory testing and their incidence is probably underestimated. We investigated the epidemiological and clinical features of Bartonella endocarditis cases diagnosed between 2009 and 2021 on Reunion Island (Southwest Indian Ocean). Method We retrospectively included all patients diagnosed with Bartonella endocarditis at Reunion Island University Hospital during this period. Endocarditis was diagnosed on the basis of microbiological findings, including serological tests (IFA) and PCR on cardiac valves, and the modified Duke criteria. We used then the multispacer typing (MST) method to genotype the available Bartonella strains. Findings We report 12 cases of B. quintana endocarditis on Reunion Island (83.3% in men, median patient age: 32 years). All the patients originated from the Comoros archipelago. The traditional risk factors for B. quintana infection (homelessness, alcoholism, exposure to body lice) were absent in all but two of the patients, who reported head louse infestations in childhood. Previous heart disease leading to valve dysfunction was recorded in 50% of patients. All patients underwent cardiac valve surgery and antimicrobial therapy with a regimen including doxycycline. All patients presented high C-reactive protein concentrations, anemia and negative blood cultures. The titer of IgG antibodies against Bartonella sp. exceeded 1:800 in 42% of patients. Specific PCR on cardiac valves confirmed the diagnosis of B. quintana endocarditis in all patients. Genotyping by the MST method was performed on four strains detected in preserved excised valves and was contributive for three, which displayed the MST6 genotype. Conclusions Bartonella quintana is an important cause of infective endocarditis in the Comoros archipelago and should be suspected in patients with mitral valve dysfunction and BCN from this area.


Findings
We report 12 cases of B. quintana endocarditis on Reunion Island (83.3% in men, median patient age: 32 years).All the patients originated from the Comoros archipelago.The traditional risk factors for B. quintana infection (homelessness, alcoholism, exposure to body lice) were absent in all but two of the patients, who reported head louse infestations in childhood.Previous heart disease leading to valve dysfunction was recorded in 50% of patients.

Introduction
Bartonella quintana is a small Gram-negative bacillus from class Alphaproteobacteria.These fastidious bacteria are generally transmitted by arthropod vectors, predominantly in the feces of body lice, and can cause various human infections.B. quintana was initially identified as the cause of trench fever in soldiers during World War I, but now mostly affects people living in poor sanitary conditions worldwide [1].This re-emerging bacterium causes bacteremia and infective endocarditis (IE) in homeless and alcohol-dependent individuals in Europe and America.It is now often referred to as "urban trench fever" [2].B. quintana can also cause bacillary angiomatosis and peliosis hepatitis in immunocompromised patients [3].
Very few data are available concerning B. quintana endocarditis in South-East Africa and the Indian subcontinent, and no case has ever before been reported in the Southwest Indian Ocean area.Moreover, very few studies on B. quintana endocarditis have reported data for strain genotyping.Such analyses are useful for epidemiological investigations aiming to determine the source of the infection [16,17].
The aim of this study was to review the epidemiological and clinical data, and outcomes of the documented cases of B. quintana endocarditis diagnosed on Reunion Island from 2009 to 2021, and to discuss possible links between cases or a common source of infection.

Ethics statement
Since this was a retrospective study based on archived patient records, it falls into the category of research not involving the human person (article L.1121-1 of the French Public Health Code, deliberation no.2018-155 of the French Data Protection Authority (CNIL)).In addition, this research was performed in accordance with the Declaration of Helsinki.Consequently, committee approval was not required.

Patient recruitment
All definite cases of Bartonella endocarditis diagnosed at Reunion Island University Hospital from 2009 to 2021 were retrospectively included in this study.All patients diagnosed with infective endocarditis (IE) on Reunion Island are treated at this hospital.For each case, medical records were analyzed and various data were extracted, including age, sex, year of diagnosis, country of residence, medical history, clinical presentation, presence of fever, possible exposure, suspected infection pathway, affected valve, surgical intervention, determination of C-reactive protein and hemoglobin levels, antibacterial treatment and outcome.

Case definition
The diagnosis of definite Bartonella endocarditis is based on at least two major Duke score criteria: (i) echocardiogram positive for endocarditis, as defined in the modified Duke criteria [18,19] and (ii) Bartonella-positive PCR on infected cardiac tissue, as defined in the recent update of the modified Duke criteria [18,20,21].According to the guidelines followed in our medical laboratory [19], blood is collected for culture in cases of suspected endocarditis and incubated for 15 days.In cases of negative blood culture, serological testing is performed for Bartonella spp.and Coxiella burnetii.PCR analysis is also performed in patients undergoing valve surgery.

Genotyping
We used the multispacer typing (MST) genotyping method to identify and compare B. quintana strains.The culture of Bartonella spp. is challenging, and the MST approach makes it possible to circumvent the difficulties of culture, by making it possible to use directly extracted DNA, with sufficient resolution to discriminate between Bartonella clones.The MST method for Bartonella was set up by Foucault et al. through the amplification and sequencing of 34 spacers from a collection of 71 isolates, and can be used to characterize several genotypes based on the combined single-nucleotide polymorphism (SNP) profiles of two spacers (336 and 894) [23].This method is suitable for use in epidemiological and evolutionary studies.In this study, DNA was extracted from preserved B. quintana PCR-positive cardiac samples, when available.The analysis was performed by the French National Reference Center (IHU Me ´diterrane ´e Infection, Marseille, France).

Characteristics of the patients and epidemiological data (Table 1)
In total, 12 cases were included in the study.Most of the infected patients were male (83.3%) and the median age was 32 years (IQR: 23.75-40.5 years).The recorded sites of residence of the patients were: Mayotte (n = 6, 50%), Comoros (n = 5, 41.7%) and Reunion Island (n = 1, 8.3%).Five of the six patients from Mayotte had recently traveled to or originated from Comoros.The only patient living on Reunion Island had also recently moved there from Mayotte (Table 1).None of the patients was immunocompromised or homeless, although information about living conditions was not available for all patients.However, three patients (25%) reported living in poor socioeconomic conditions: sheet metal house, clay ground and absence of running water.The population of Comoros is mostly Muslim, so alcohol consumption is uncommon.None of the patients had an alcohol use disorder.A history of contact with head lice was reported for only two patients (16.7%), and two patients, both farmers, reported contact with animals.

Clinical features (Table 2)
Eight patients (66.7%) had fever at initial presentation.Heart failure was the main presenting form of endocarditis caused by B. quintana in our cohort, reported in 11 patients (91.7%).
Heart failure was mostly due to valvular insufficiency (10/11 patients).Acute renal failure complicating severe chronic renal failure was observed in one of the 12 patients (8.3%).Embolic phenomena were recorded in seven patients (58.3%, Table 2), most frequently taking the form of splenic (3 patients, 25%) or intracranial mycotic aneurysms (2 patients, 16.6%).All patients had native valves during the first episode of IE.The mitral valve was affected in 10 patients (83.3%), four of whom also had aortic valve involvement, with one patient also having tricuspid valve involvement.In two patients (16.7%)only the aortic valve was involved.

Treatment and outcome (Table 2)
In this study, all patients received doxycycline for at least four weeks.Seven patients (58.3%) received concomitant gentamicin for at least two weeks.Two patients received rifampicin, and one received amoxicillin in place of gentamicin, in combination with doxycycline.Cardiac valve surgery was performed in all patients.Significant postoperative complications occurred in eight patients: six suffered from cardiac rhythm disorders, and two had mediastinitis.One of the two cases of postoperative mediastinitis involved Staphylococcus epidermidis, whereas the causal organism was not documented in the other case.None of the patients died from IE, and none suffered a relapse of B. quintana IE.However, one patient experienced three episodes of other types of bacterial endocarditis and sepsis in the following months and years.One patient had to undergo a second operation to replace the mitral valve seven years after his initial IE.Two patients developed heart failure sequelae: follow-up heart scintigraphy on patient 6 showed an altered post-stress left ventricular ejection fraction one month after medical care, whereas patient 8 presented grade 1 aortic regurgitation and grade 2 mitral regurgitation three years after surgery.

Diagnosis
All patients had B. quintana infection confirmed by molecular testing on a cardiac specimen.Blood cultures were performed and were negative for all patients.Serological testing was performed for 11 patients, and nine patients (75%) had antibody titers >1:100.The French National Reference Center considers IFA results to be positive if the IgG antibody titer is �1:100, and uses a titer of 1:800 as the cutoff for IE diagnosis [5].Six patients (54.5%) had antibody titers >1:800.Serological results were negative for two patients (16.7%).Laboratory tests revealed high C-reactive protein levels (median: 61 mg/L, range: 14.3-189 mg/L) and anemia (median Hb: 9.35 g/dL, range: 7.6-11.1 g/dL) in all patients.
Echocardiography records were obtained for all patients.In Table 3, echocardiograms are described as TTE if transthoracic, TEE if transesophageal and ECHO if the type of echocardiogram was not specified.Vegetations were observed in 10 patients (83.3%).
All patients required valve replacement, due to cardiac valve destruction or insufficiency.The results of PCR for B. quintana on the DNA extracted from cardiac valve tissue or vegetations were positive for all 12 patients.

Sequence typing
Residual biological material (cardiac specimens or serum samples) was available for sequence typing for only four of the 12 cases.MST analysis was informative in three of these cases.It was performed on three cardiac samples, from patients 8, 10 and 12. Spacers 336 and 894 were amplified by PCR and their nucleotide sequence was determined.All three strains were found to have the same MST profile, with type 1 as the sequence profile for spacer 336, and type 5 for spacer 894.Based on this combination of sequence types, we classified the three isolates as genotype MST 6 (S1 Table ).

Discussion
We report here a first series of 12 cases of B. quintana infective endocarditis occurring in the Southwest Indian Ocean area from 2009 to 2021.Cases of Bartonella infective endocarditis have been reported worldwide, mostly in the Americas and Europe with bacteria of this genus accounting for 0 to 4.5% of infective endocarditis cases.The most frequently reported species in such cases is B. quintana [5,15,.
The Cardiothoracic Surgery Department of Reunion Island University Hospital is the reference center for the diagnosis and treatment of endocarditis in the Southwest Indian Ocean area, providing care for all patients from this region.Based on the data collected at Reunion Island University Hospital for 2020, the prevalence of Bartonella endocarditis was 3.1% among the documented cases of IE.However, the prevalence of Bartonella IE in this region is almost certainly underestimated, as only the most serious clinical cases benefit from medical evacuation from Mayotte to Reunion Island, as demonstrated by the high rates of heart failure and valve damage in our series of patients (all of whom required surgery).
Our findings confirm published data indicating a male preponderance and low median age [41,[55][56][57].Clinical presentation was similar to that for other types of subacute bacterial endocarditis.However, severe acute symptoms (acute pulmonary edema, acute coronary syndrome, subarachnoid hemorrhage) were often present at initial presentation, as previously reported [15,45].This finding may be explained by the subacute pathophysiology of Bartonella IE, and by late diagnosis and care, as access to healthcare is often very difficult in poor areas, such as Mayotte and Comoros.It is important to remind that patients in this cohort are transferred from Mayotte/Comoros to Reunion Island at an advanced stage of the disease, thus presenting symptoms reported might be those of late evolution of endocarditis.
Half the patients in our study already had damaged valves at presentation.Interestingly, no prior valvular disease was documented for most of the published patients with B. quintana endocarditis [4,42,46].However, some studies reported rheumatic heart disease as the risk factor most frequently associated with B. quintana IE [15,50,53].This may reflect differences in the clinical course of the disease between the Southwest Indian Ocean region and Europe, where studies have suggested that B. quintana IE followed chronic bacteremia in patients without prior valve defects [42].Moreover, the mitral valve was the valve most frequently affected (75%) in our patients, whereas the aortic valve was the valve most frequently affected in published cases [57].One hypothesis could be an association between previous presence of rheumatic heart disease (41.7%) in the cohort, and mitral valve affect.A predominant mitral valve involvement has already been observed in cases reporting rheumatic heart disease in medical history [15,50,53].
The recommended treatment for B. quintana endocarditis according to European guidelines is oral doxycycline (100 mg/12 h) for four weeks plus gentamicin (3 mg/24 h) for two weeks [19].All the patients in our cohort received at least one of these two antibiotics.However, three patients were not treated with the recommended double-therapy regimen: patients 6, 9 and 12 (patients 5 and 10 received gentamicin for 2 weeks before Bartonella infection diagnosis, as initial antibacterial treatment).They received doxycycline alone, or associated to amoxicillin or rifampicin.Although not in accordance with the guidelines, theses regimens showed clinical success.Gentamicin is known to cause renal toxicity, nevertheless, this complication has not been observed in the cohort following the use of this antibiotic.Bartonella IE is known to cause significant destruction of valve tissues, necessitating valve surgery more frequently than IE caused by other pathogens [58].Our study confirmed these findings, with all 12 cases requiring surgical treatment.This greater need for surgical treatment may be due to delayed management or a pre-existing altered state potentially aggravated by undernutrition and poor socioeconomic conditions and underlying the higher rates of postoperative complications (66.7%) than for other types of bacterial IE [59].
BCNE is associated with a higher rate of in-hospital adverse events than blood culture-positive endocarditis, although recent publications have suggested similar long-term outcomes [60].Two patients developed postoperative mediastinitis.Only one other case has ever been reported to develop post-sternotomy mediastinitis following B. quintana IE [44].Despite the short-term complications observed, none of the patients died, contrasting with the previous estimate of 7-30% for Bartonella IE-related mortality [26].Patient 2 developed recurrent IE caused by Streptococcus spp.months and years after initial B. quintana IE diagnosis.This case highlights the risk factor that chronic B. quintana IE represents for additional endocarditis, and the possibility of underestimating B. quintana if the infection was concomitant.This has already been described by Boodman et al. who mentioned "subacute B. quintana infection creates valvular damage and large vegetations, which provide a substrate for seeding when acute bacteremia occurs with a different pathogen" [30,61].Serological results were positive (IgG �1:100) in 81.8% of patients, slightly below the sensitivity of 91% estimated for IFA by Edouard et al. [5].Nevertheless, serology guided diagnosis in 6 of 11 cases (54.5%), with IgG titers �1:800, the recommended cutoff for Bartonella IE diagnosis [5].Serological assay showed low performance in this study, compared to other studies that show a sensitivity of 63-94% [5,47].Maurin et al. has already showed the low sensitivity of the IFA-Focus Diagnostics kit in detecting B. quintana IgG on B. quintana endocarditis patients [22].In addition, we noted that patient 11 presented a negative serology, possibly due to his severe undernutrition that could cause relative immunosuppression.Moreover, it is noteworthy the serological titers measured against both B. quintana and B. henselae cross-react between these two species, as already published [22,27,29,33,37,47].Also, recent studies have demonstrated cross-seroreactivity at low level between Bartonella spp.and other species in cases of blood culture positive endocarditis.Consequently, anti-Bartonella serology may be insufficient without confirmation by another technique, such as PCR in this study [47,62].PCR therefore remains the best diagnostic tool, provided that sufficient tissue is available.However, RT-PCR has a low sensitivity when applied to blood and serum samples (33% and 36%, respectively) [5].Overall, molecular assays on tissues and serological testing perform well for the diagnosis of B. quintana IE.An increase in C-reactive protein concentration and anemia were observed in all patients.The anemia could be attributed to endocarditis-related inflammation.Thus, given that confirmation analyses can take a long time, we believe that any patient from Mayotte or Comoros presenting with symptoms of endocarditis, a biological inflammatory syndrome, and vegetations detectable on an echocardiogram but with negative blood cultures should be considered a possible case of B. quintana (or Coxiella burnetti) IE, until proven otherwise.In such cases, the early administration of doxycycline, before confirmation of the diagnosis, may be beneficial.
The B. quintana genotype circulating in the Southwest Indian Ocean region appears to be different from those circulating in Europe and Africa.Interestingly, the MST 6 genotype identified in our cohort was previously identified only once, in aortic valve tissues from a patient with B. quintana endocarditis in Australia [51].All three strains successfully analyzed by MST had identical sequence profiles, and two of these strains were isolated two years apart.Despite the very small population analyzed, these findings suggest that there may be only one B. quintana genotype (MST 6) circulating in the Comoros archipelago, or that there is a potential common source of contamination on Mayotte.This exposure factor has yet to be identified.One of these patients was a farmer who may have come into contact with animals and lived in a rural community.The known risk factors for B. quintana endocarditis are homelessness, alcohol dependence and exposure to body lice [1].It is reasonable to assume that these markers are linked to poor socioeconomic status [4].Exposure to head lice was identified only in two cases, but we believe that information about head louse infestation history were missing, probably because patients had been infected months or years before their admission to hospital for IE.Moreover, Sangare ´et al. demonstrated the effective presence of B. quintana vectors in the Southwest Indian Ocean area (in 4.5% of body and 2.6% of head lice in Madagascar, with even higher rates in other African countries, including East African countries) [63].Furthermore, all of the patients studied here originated from Mayotte or Comoros, where a high proportion of the population has an income below the threshold defining the national poverty line and there are many exchanges of population with Southern Africa [64].However, one of the limitations of this retrospective study is that the bacterial diagnosis was established a long time after admission.It was therefore not possible to identify and record possible sources of infection for most of the cases.

Conclusion
We report here an original series of 12 cases of B. quintana IE from the Southwest Indian Ocean, where this disease has never before been described.Our data suggest that B. quintana is not an uncommon cause of native valve endocarditis in Comoros archipelago, particularly on Mayotte.Moreover, immigration from the Comoros archipelago to the French islands in this region is continuing, rendering this case series of particular interest.Local epidemiological and seroprevalence studies are required to determine the frequency of B. quintana infection in the Southwest Indian Ocean area.This study should raise local awareness of this neglected disease.B. quintana IE should be suspected in patients from Mayotte or Comoros with mitral valve dysfunction and blood culture-negative endocarditis, even in the absence of the traditional risk factors for Bartonella infection.

Table 1 . Demographic characteristics of the 12 patients with documented B. quintana endocarditis. Pat. No. Sex Origin Year of initial presentation Age at initial presentation (years) Medical history Possible exposure, links to rural community
*:No relevant medical history connected with cardiac or infectious disease; F: female; M: male; U: unknown.https://doi.org/10.1371/journal.pntd.0011606.t001