Bartonella and Coxiella infections presenting as systemic vasculitis: case series and review of literature Free

Abstract

• Bartonella infections in the context of endocarditis can mimic ANCA-associated vasculitis.

• Coxiella infections can mimic vasculitis of all vessel sizes, through cryoglobulinaemia or direct vascular infection.

• Knowing these features can avoid wrongfully treating patients exclusively with immunosuppressors instead of antibiotics.

Introduction

According to the 2012 International Chapel Hill Consensus Conference, vasculitides are dichotomized into primary vasculitides and secondary vasculitides with probable aetiology. In the latter, aetiologies mostly include drugs and infections such as hepatitis B virus-associated PAN, hepatitis C virus-associated cryoglobulinaemia vasculitis (CryoVas) and syphilis-associated aortitis [1]. Distinguishing primary from secondary vasculitides is of the utmost importance as timely diagnosis is necessary for appropriate care. Coxiella burnetti, the causative agent of Q fever, is known for its ability to infect endothelial cells especially endocardium, native vascular tissue or even prosthetic grafts. In addition to its infectious potency, observations suggest a unique ability to induce autoimmunity [2]. Bartonella henselae and Bartonella quintana also display a particular tropism for endocardial tissue and are responsible for 1–4% of endocarditis, depending on geographical area [3]. Recently, it has been reported that ANCA were frequently positive in Bartonella endocarditis [4].

Given the vascular tropism of these bacteria and their ability to induce autoimmunity, rare cases of secondary vasculitides caused by Coxiella and Bartonella infections that may sometimes present as primary vasculitides have previously been reported. There is to date no comprehensive review describing vasculitides associated with these infectious agents. In the present study, we combine a significant case series and a comprehensive literature review of Coxiella and Bartonella infections to outline their clinical, biological and histological presentation, as well as their outcome.

Methods

Patients

We performed a retrospective study based on a national survey to identify patients with vasculitis manifestations revealing Coxiella or Bartonella infections, supported by the French Vasculitis Study Group (FVSG), a nationwide research collaboration in the field of systemic vasculitis. Patients were included if they had clinical manifestations of vasculitis (purpura, arthralgia, myalgia, kidney involvement, peripheral neuropathy), in association with suggestive immunological (ANCA and cryoglobulinaemia positivity), radiological (CT, PET-CT, MRI and/or angiography) and/or histological features (leukocytoclastic, granulomatous or necrotizing vasculitis, crescentic GN), together with a proven infection by Coxiella burnetti or Bartonella spp. A proven infection was defined as a positive bacteriological test (serology, PCR and/or culture) with organ specific damage and/or ineffectiveness of immunosuppressors alone.

Next-generation sequencing

To determine whether in situ tissue infection was responsible for organ damage, available tissue was tested with a shotgun metagenomic procedure, an unbiased and highly sensitive accredited method (ISO15189) to detect any microbial DNA [5]. Briefly, tissues were extracted using a pre-treatment combining bead beating, chemical and enzymatic lysis with extraction by DNA Blood kit on QiaSymphony instrument (Qiagen, Hilden, Germany). DNA and RNA were prepared for sequencing by means of Nextera XT and TruSeq Stranded Total RNA library kits (Illumina, San Diego, CA, USA), respectively, and sequenced with NextSeq 500/550 High Output Kit v2.5 (300 cycles) (Illumina) on NextSeq500 (Illumina). Analysis was performed with metaMIC software (https://gitlab.com/mndebi/metamic.git) in order to document and quantify microorganisms. All runs included environmental (sterile water) and positive control Microbial community Standard (ZymoBiomics, Irvine, California, USA).

Literature review

We performed a comprehensive literature search of cases with Coxiella or Bartonella infection and clinical manifestations of vasculitis. We searched MEDLINE via PubMed for all articles in English, French or Spanish using the key words or mesh terms ‘coxiella’, ‘bartonella’, ‘q fever’, ‘vasculitis’, ‘purpura’, ‘ANCA’, ‘cryoglobulinemia’, ‘GCA’, ‘Takayasu’, ‘polyangiitis’, ‘polyarteritis nodosa’ and ‘leukocytoclastic vasculitis’. Only reports with clinical, biological, bacterial and histological data when performed were included.

Statistics

Descriptive statistics included mean (s.d.) or median [interquartile range (IQR)] for continuous variables and frequency (percentage) for categorical variables.

Ethics

The collection of data was retrospective and observational. Data collection and analyses were approved by the Ethic Review Committee of Cochin University Hospital (decision number AAA-2021-08020).

Results

Bartonella infections

We included six original cases of Bartonella infection and 18 patients from the literature review [6–23]. All cases are detailed in Table 1. Most patients were men (16/24, 67%), median age was 62.5 (IQR 46.5–69) years and 13 (54%) had a pre-existing valvulopathy or prosthetic valve. Median delay between symptoms onset and diagnosis was 2.8 months (1.6–5). Twenty-one (88%) patients had negative blood culture for common infectious endocarditis, with pathogen documented by serology in most cases (79%) with IgG positive in 92% and IgM in 80%, or PCR, performed on excised tissue (38%) or blood samples (17%). Causative pathogens were B. henselae in 16 (67%) patients, B. quintana in seven (29%) and B. bacilliformis in one case.

Bartonella infection mimicked ANCA-associated vasculitis (AAV) in 20 (83%) cases and presented as cryoglobulinaemic vasculitis in two (8%) and leukocytoclastic vasculitis in one (4%). The last patient, with an underlying IgA vasculitis, presented a new vasculitis flare concomitant with Bartonella infection. Main symptoms were fever (59%), purpura (41%) and arthralgia (23%). Kidney involvement was present in 22 (92%) cases, proteinuria and haematuria respectively in 20 (83%). Median serum creatinine at presentation was 206 (IQR 186–308) µmol/l. Among ANCA-positive patients, proteinase 3 (PR3) was the most frequent specificity (80%). Unusual for AAV, complement fraction levels were low in 14/17 (82%) of cases and rheumatoid factor was positive in 11/13 (85%). The most common finding in kidney biopsies was cellular proliferation in 18/20 (90%) cases, with crescents in 14/18 (78%) but also endocapillary proliferation in 4/18 (22%), membranous proliferation and mesangial proliferation in 2/18 (11%) patients. Immunofluorescence analysis on renal lesions showed glomerular immune deposits in 12/17 (71%). Only five (29%) patients displayed typical pauci-immune crescentic GN, all having positive ANCA and low complement levels.

All patients received antibiotics, in association with glucocorticoids in 13/24 (54%), and cyclophosphamide in 7/24 (29%). Therapeutic plasma exchanges were performed in 3/24 (13%) cases, and 10 (42%) patients required valvular replacement. Strikingly, all cases treated with immunosuppressors alone without proper antibiotic regimen deteriorated or relapsed.

No death from infectious cause or vasculitis was reported. Renal outcome was good with a median serum creatinine of 108 (IQR 96.5–122.5) µmol/l at last follow-up.

Coxiella infections

We included five new cases of Coxiella infection and 16 patients from the literature review. Detailed case descriptions are reported in Table 2. Most patients were male (14/21, 67%), and median age was 61 (IQR 54.5–65.5) years. Animal exposure or rural habitat was documented in 8/21 (38%) cases. All patients had positive Coxiella serology, two had positive PCR in the blood and six positive within tissues. Eleven (52%) patients had definite infectious endocarditis. Median delay between symptoms onset and diagnosis was 5.5 (2.5–6.5) months.

Sixteen (71%) patients had small-vessel vasculitis presenting as purpura (n = 10), GN (n = 7), livedo (n = 1), arthralgia (n = 1) and/or peripheral neuropathy (n = 1). Mixed cryoglobulinaemia was detected in 12/16 (75%) patients, and PR3-ANCA was positive in 2/16 (13%) cases. Two patients with negative immunological work-up had perivascular neutrophilic infiltrate with fibrinoid changes on skin biopsy, consistent with leukocytoclastic vasculitis without any aspect of thrombotic vasculopathy.

One patient had medium-vessel vasculitis presenting as PAN with fever, myalgia and hepatic artery aneurysm. Finally, four patients had large-vessel vasculitis mimicking either GCA (n = 3) or Takayasu arteritis (n = 1). One patient had temporal artery biopsy showing features suggestive of GCA, and one patient had giant cells on histological analysis of the aorta with a positive Coxiella PCR on tissue. None of these five patients had endocarditis.

All patients received antibiotics, including tetracyclines in 100% associated with hydroxychloroquine in 86% of cases. Eight patients also received immunosuppressive agents, four with cryoglobulinaemic vasculitis, three with large-vessel vasculitis and one with PR3-ANCA vasculitis. Only one patient with cryoglobulinaemic vasculitis and positive blood Coxiella PCR was successfully treated with rituximab after inefficiency of antibiotics and glucocorticoids. Outcome was favourable in all but one patient.

Tissue next-generation sequencing results

Four skin biopsies of purpuric lesions and one kidney biopsy were tested using next-generation sequencing, including three from Coxiella-infected patients and two from Bartonella infections. No microbial DNA was detected in any of these five samples despite a good depth of sequencing above recommendations for this test (>20 million sequences per samples) [37].

Discussion

We describe here characteristics of Bartonella and Coxiella infections presenting as systemic vasculitis through an original case series and a comprehensive literature review. This study provides original insights into the presentation, treatment and prognosis of infection-associated vasculitis.

Bartonella infections most commonly induced small-vessel vasculitis with clinical features mimicking renal-limited PR3-ANCA vasculitis in the setting of endocarditis. Endocarditis from any origin has been shown to induce ANCA positivity in 19–24% of cases [38], whereas this prevalence can be as high as 60% in Bartonella endocarditis [4]. In our study, ANCA positivity was more frequently associated with Bartonella than Coxiella infections, as was GN. Expression of complementary epitopes at the surface of Bartonella spp. could induce a molecular mimicry phenomenon leading to the production of pathogenic autoantibodies, as was discovered for anti-LAMP-2 antibodies induced by gram-negative bacteria [39].

Whether ANCA induce vascular damage in endocarditis is up for debate. Although MPO-ANCA can induce vasculitis in murine models [40], a direct pathogenic role of PR3-ANCA has not been clearly demonstrated so far [41]. Although a retrospective study has suggested that kidney involvement tends to be more frequent in ANCA-positive endocarditis [38], our study does not support a pathogenic role of ANCA. Immunofluorescence analysis in kidney biopsies showed immune deposits in the vast majority of cases in association with hypocomplementaemia, suggesting the role of immune complexes in organ damage. Moreover, renal outcome was usually favourable in our study despite limited use of immunosuppressors, supporting a non-pathogenic role for ANCA.

Three major red flags should alert physicians towards an infectious origin when considering GN with ANCA positivity. First, was the absence of ENT involvement and a single occurrence of pulmonary nodules in a context of tricuspid endocarditis in our series, whereas it is prevalent in up to 73% and 45%, respectively, of patients with granulomatosis with polyangiitis [42]. Second, low complement fraction levels were observed in the majority of cases, even in pauci-immune crescentic GN, which is highly unusual in granulomatosis with polyangiitis. Deshayes et al. reported hypocomplementaemia in 5% of AAV [43]. This finding is similar to data from the FVSG database, showing low C3 and C4 levels in only 1.3% and 3.9%, respectively (unpublished data). Third, in cases of GN, renal biopsy seems essential in reconsidering AAV diagnosis in cases of complement and immunoglobulin deposition. Physicians should therefore suspect an underlying Bartonella infection in case of PR3-ANCA GN with low complement fraction levels, lack of ENT involvement and positive immunofluorescence on kidney biopsy.

Mixed cryoglobulinaemia was another frequent cause of small-vessel vasculitis in our study, especially during Coxiella endocarditis. Coxiella is able to induce multiple forms of autoimmunity through autoantibody formation [44]. It seems therefore plausible that, in predisposed subjects, the immune response to Coxiella infection could transiently trigger formation of autoantibodies. To support this hypothesis, our study did not find any trace of in situ tissue infection on three small vessel vasculitis biopsies, suggesting immune-mediated organ damage. Only one case from the literature showed an unfavourable outcome with antibiotics alone and required rituximab infusions. The remaining cases had a favourable outcome with antibiotic treatment alone, as expected in infectious cryoglobulinaemia [45]. These cases show the need for a thorough work-up to eliminate an infectious origin in cases of cryoglobulinaemic vasculitis.

Medium and large vessel vasculitides were observed only in the setting of Coxiella infections and without associated endocarditis. Specific PCR was positive on aortic tissue in one patient. Hagenaars et al. described the presence of necrotizing granulomas in more than half of patients with Q fever aortic aneurysms, with the presence of giant cells in two specimens [46]. We therefore suggest that in contrast with small-vessel vasculitis, it rather is in situ microbial infection of medium or large vessels that results in a granulomatous immune response mimicking granulomatous vasculitis such as GCA or Takayasu arteritis.

Overall, our study shows distinct vasculitis patterns between Bartonella and Coxiella infections. Bartonella infections mainly mimic primary small-vessel PR3-ANCA vasculitis with predominant or exclusive kidney involvement, whereas Coxiella infections can induce small vessel mixed cryoglobulinaemia vasculitis, medium vessel vasculitis and granulomatous large vessel vasculitis.

Funding: No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: B.T. received some consulting fees and/or grants from Roche/Chugaï, AstraZeneca, GlaxoSmithKline, Bristol-Myers Squibb, Lilly, Vifor Pharma, LFB, Grifols and Terumo BCT. L.G. received consulting fees and/or grants from Roche, Novartis, GSK, Lilly, Sanofi, Novo Nordisk, Biogen, Boehringer Ingelheim and UCB. X.P. received consulting fees and/or grants from Roche, GSK, LFB and Pfizer. C.R. has served as an advisor and/or speaker for Illumina and Vela Diagnostics.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References