Detection of invasive Bartonella infections with next-generation sequencing of microbial cell-free DNA

Abstract

We report 9 patients with invasive Bartonella infections, including 5 with endocarditis, who were diagnosed with microbial cell-free DNA next-generation sequencing and Bartonella serology studies. Diagnosis with plasma mcfDNA NGS enabled a faster clinical and laboratory diagnosis in 8 patients. Prompt diagnosis impacted antibiotic management in all 9 patients.

Introduction

Disseminated Bartonella infections have a broad spectrum of clinical presentations, and the organism represents a particular diagnostic challenge due to its fastidious nature, long incubation times, and the limited sensitivity of Bartonella cultures. ^1,2 While more sensitive than cultures, Bartonella serology is not specific as cross-reactivity with Coxiella species has been reported and may adversely impact therapy and outcomes. ¹ Additionally, due to differing assay sensitivities, negative serology does not exclude active infection in the context of high clinical suspicion. ¹ Polymerase chain reaction (PCR) for Bartonella is sensitive and specific, ¹ but some commercial tests may only detect one or a limited number of species.

Particularly concerning with disseminated bartonellosis is pathogen seeding heart valves and other end organs. In the past 2 decades since its initial recognition as a causative organism for infective endocarditis, it has been increasingly demonstrated as a common pathogen in culture-negative endocarditis, representing 28% of cases in a French reference center. ³ Management commonly includes the combination of doxycycline or a beta-lactam and gentamycin or rifampin, ⁴ but more data is needed on these regimens.

Over the last decade, next-generation sequencing (NGS) of plasma microbial cell-free DNA (mcfDNA) has become an open-ended, noninvasive testing tool for diagnosing various pathogens. The mcfDNA NGS assay used in this study was the Karius Test™ by Karius (Redwood City, CA), a College of American Pathologists-accredited, Clinical Laboratory Improvement Amendments-certified mcfDNA NGS laboratory. This case series aimed to characterize the presentations of patients with disseminated bartonellosis and describe the role of mcfDNA NGS in diagnosis and tailoring adequate antimicrobial management.

Method

This retrospective case series was performed at Baylor St. Luke’s Medical Center (BSLMC), a quaternary academic medical center, between 2017 and 2022. Patients with Bartonella serologies and Bartonella species concentrations at or above the mcfDNA NGS commercial threshold were included. These thresholds were established in a previously described cohort of 684 healthy adult patient sera used as healthy controls. ⁵ This study was approved by the Baylor College of Medicine Institutional Review Board.

The electronic medical record for each patient was assessed for presenting symptoms and laboratory studies. The clinical turnaround time (TAT) and laboratory TAT were calculated for mcfDNA NGS, Bartonella serologies, and PCR. Clinical TAT was defined as the time from the physician’s order to reported results. Laboratory TAT was defined as the time from laboratory receipt of the patient sample by the send out laboratory to reported results. Modified Duke criteria were used to diagnose infective endocarditis. ⁶

Results

Thirty-eight patients with both mcfDNA NGS and Bartonella serology were reviewed.

Both tests were positive in 9 patients (8 with B henselae and 1 with B vinsonii speciated by mcfDNA NGS). There were no discrepancies found between them yielding a 100% accuracy for the mcfDNA NGS test compared with serology.

The median age for the 9 included patients was 45, and 7 of 9 patients were male. Their clinical characteristics and initial labs are shown in Table 1. Four patients had prosthetic valves. 2 had immunocompromising conditions, including 1 with HIV and 1 with a kidney transplant on immunosuppression. Two patients reported exposure to cats.

Table 1. Clinical features of patients infected with Bartonella

Note. RV-PA: right ventricle-pulmonary artery; LVOT: left-ventricular outflow tract; PWID: person who injects drugs; ESRD: end-stage renal disease; TOF: tetralogy of Fallot; WBC: white blood cells; BUN: blood urea nitrogen; PR3-ANCA: proteinase 3-antineutrophil cytoplasmic antibodies; RBC: red blood cells; PCR: polymerase chain reaction; mcfDNA: microbial cell-free DNA; NGS: next-generation sequencing; MPM: molecules per microliter; TAT: turnaround time; ICU: intensive care unit; CMV: cytomegalovirus; EBV: Epstein-Barr virus; HHV: human herpesvirus; AFB: acid-fast bacillus; ANA: antinuclear antibody; ASO: antistreptolysin O; RF: rheumatoid factor; TPO: thyroid peroxidase; ANCA: antinuclear cytoplasmic antibody; TSH: thyroid-stimulating hormone.

^a Some patients seen in 2017 received an earlier version of mcfDNA NGS studies that did not quantify the MPM of mcfDNA. Instead, the samples were tested with a negative buffer control.

^b Epstein-Barr virus and torque teno virus were found in small concentrations in 2 patients, but were not considered clinically significant by treatment teams and may have been the result of mild viremia in the setting of a secondary source of critical illness.

^c The patient was discharged after 10 days and was later readmitted for 16 days resulting in death.

^d Tests for an organism not detectable by mcfDNA NGS assay.

Seven patients presented with fever, and four with septic shock. Eight had available transthoracic echocardiograms, two had valve thickening, and three had valvular masses or densities. Two patients had transesophageal echocardiograms, one of which was abnormal with a valvular density. Five patients met the modified Duke criteria for definite infective endocarditis, while one met the criteria for possible infective endocarditis, and three had neither.

Eight of nine patients were hospitalized for an average of 18.5 days (range 6–46 days). All patients had both positive mcfDNA NGS and Bartonella serology—for four, Bartonella was mentioned as part of the differential, and serologies were sent before mcfDNA NGS results, while for five, serologies were not sent until after mcfDNA NGS results. One had a valve tissue PCR, and another had a whole blood PCR that was negative for Bartonella. mcfDNA NGS was the first positive test for Bartonella in 8 patients, while serology was the first positive result in 1.

Serology had a median clinical TAT of 4 days (range 3–8) and a median laboratory TAT of 2 days (range 1–8). mcfDNA NGS had a median clinical TAT of 2 days (range 2–7) and a median laboratory TAT of 1 day (range 1–3). In the 8 cases where mcfDNA was the first to result, it impacted antimicrobial management. After diagnosis with Bartonella infection, 6 patients started doxycycline and rifampin, and 2 started doxycycline alone. One was already taking doxycycline for culture-negative endocarditis and had an adjustment in the gentamycin dose. Five discontinued unnecessary antibiotics.

Discussion

Few studies exist on the incidence of disseminated bartonellosis or Bartonella endocarditis, potentially owing to the difficulty of diagnosing both with conventional laboratory equipment and their relatively recent description in the literature. Here, we present a case series of 9 patients with invasive bartonellosis, including 5 with endocarditis, diagnosed with mcfDNA NGS and Bartonella serologies. Similar to a study of French and Canadian patients with Bartonella endocarditis, we found a primarily male patient population with high rates of preexisting valvular disease, and we seldom identified zoonotic exposure. ⁷

One of France’s largest reports of culture-negative endocarditis described 28% of cases with Bartonella as the etiology. However, 78% of these were associated with Bartonella quintana, which was not seen in our case series. ³ By comparison, Bartonella henselae cases were more commonly associated with preexisting valvular disease and contact with cats ³ —both seen in this study. There is scarce literature on the proportion of invasive Bartonella infections associated with endocarditis, but a recent pediatric study showed 10 of 23 patients diagnosed with bartonellosis having endocarditis, consistent with the proportion seen in this series. ⁸

The patients in this series presented with nonspecific symptoms and a broad differential. Culture-negative endocarditis was only suspected in three cases (Table 1). As a result, they received a battery of laboratory studies and were hospitalized for up to eight days before Bartonella was identified, and antimicrobial therapy was adjusted accordingly. These adjustments included starting targeted antibiotic coverage for Bartonella and discontinuing unnecessary antibiotics, including nephrotoxic treatment, in patients with compromised renal function. mcfDNA NGS could offer a significant advantage in diagnosing Bartonella infections, especially in those with critical presentations requiring prompt, directed antimicrobial therapy.

Echocardiogram findings in Bartonella-associated endocarditis can be variable and nonspecific. It is believed to cause larger vegetations and significant valvular damage, although this was not always observed in our limited case series. ⁹

This study is limited by the small number of patients, the lack of a control group to compare clinical outcomes in patients diagnosed by conventional methods and the retrospective nature of the analysis. Additionally, clinical TATs may vary among different institutions, and laboratory TATs depend on the procedures of outside laboratories. Moreover, we did not perform a financial evaluation of this technique, which made us unable to assess its cost-effectiveness.

In conclusion, plasma mcfDNA NGS is an additional tool for diagnosing Bartonella species infections and may lead to improved patient outcomes. Still, larger studies with control groups are needed to evaluate the impact of these rapid diagnostics on clinical care, diagnostic stewardship, and overall cost.