Pathogen Detection in Infective Endocarditis Using Targeted Metagenomics on Whole Blood and Plasma: a Prospective Pilot Study

ABSTRACT

Initial microbiologic diagnosis of infective endocarditis (IE) relies on blood cultures and Bartonella and Coxiella burnetii serology. Small case series and one prospective study have preliminarily reported application of metagenomic sequencing on blood or plasma for IE diagnosis. Here, results of a prospective pilot study evaluating targeted metagenomic sequencing (tMGS) for blood-based early pathogen detection and identification in IE are reported. Subjects diagnosed with possible or definite IE at a single institution were prospectively enrolled with informed consent from October 2020 to July 2021. Blood was drawn and separated into whole blood and plasma. Both specimen types were subjected to nucleic acid extraction and PCR targeting the V1-V3 region of the 16S ribosomal RNA gene, followed by next-generation sequencing on an Illumina MiSeqTM platform. 35 subjects, 28 (80%) with definite and 7 (20%) with possible IE were enrolled, including 6 (17%) with blood culture-negative endocarditis (BCNE). Overall, 20 whole blood (59%) and 16 plasma (47%) samples tested positive (P = 0.47). When results of whole blood and plasma testing were combined, a positive tMGS result was found in 23 subjects (66%). tMGS identified a potential pathogen in 5 of 6 culture-negative IE cases. Although further study is needed, the results of this pilot study suggest that blood-based tMGS may provide pathogen identification in subjects with IE, including in culture-negative cases.

INTRODUCTION

Infective endocarditis (IE) has an incidence of 7.9 cases per 100,000 persons-year (1) and a fatality rate of 25% (2, 3). Causes vary by region, with most being bacterial; Staphylococcus aureus and viridans group streptococci are implicated in ~80% of cases (2). Diagnosis relies on modified Duke criteria, which include a combination of clinical, echocardiographic, histologic, and microbiologic criteria (4, 5). Using these criteria, preoperative identification of causative pathogen(s) is limited to information provided by blood cultures and serologic tests for Coxiella burnetii and Bartonella species (although the latter is not part of the modified Duke criteria) (6). Despite recommendations to obtain blood cultures prior to antibiotic therapy, collect adequate blood volumes and use extended incubation for recovery of Cutibacterium acnes, blood cultures are negative in 2– 40% of IE cases (3, 7–9).

The most frequent cause of blood culture-negative endocarditis (BCNE), responsible for an estimated 30 to 40% of culture-negative cases, is administration of antibiotics prior to blood culture collection (9, 10). In one study, antibiotics were administered prior to blood culture collection in more than 63% of cases (11). Difficult-to-grow pathogens such as C. burnetii, Bartonella species, Tropheryma whipplei and C. acnes, also contribute to BCNE. For C. burnetii and Bartonella species, serology may provide a diagnosis, although cross-reactivity and seroreactivity from prior infection may obfuscate results interpretation (6). Histopathologic and molecular evaluation of cardiac valves can be helpful but require cardiac surgery which is delayed or not performed in many cases (9).

Molecular methods increase the yield of bacterial detection and identification in IE, but at the current time, these assays, such as 16S ribosomal RNA (rRNA) gene PCR/Sanger sequencing, are only recommended on cardiac valves (6, 12–14). The narrow spectrum of specific PCR assays limits usage in clinical practice. In addition, even for known causes of IE for which specific PCR assays are available, sensitivity may be limited when performed on blood-based specimens. For example, Bartonella PCR performed on valve tissue was found to have a 92% sensitivity compared to 30% when tested on blood or serum of subjects with Bartonella endocarditis (15).

Recently, shotgun metagenomics sequencing (sMGS) has been proposed as a potential diagnostic method for infectious diseases (16–18). In IE, identification of pathogens in culture-negative valve tissues using sMGS of cardiac valves has been reported (19–23), although background human DNA has been a limitation to optimizing sensitivity of this approach (23). Alternatively, 16S rRNA gene-based targeted metagenomic sequencing (tMGS) of cardiac valves can be performed (24).

Next-generation sequencing (NGS) methods applied to blood-based specimens offer the potential for culture-independent, noninvasive diagnosis of IE, but have been incompletely evaluated. To our knowledge, two studies have evaluated the use of metagenomics for diagnosis of IE in plasma. Both used the Karius test, a shotgun metagenomic sequencing-based test performed on plasma, including a retrospective study of 12 pediatric IE cases (25), and a prospective study of 24 definite IE cases published in abstract form (26), making it hard to define performance and clinical utility. In this pilot prospective study, the yield of a tMGS assay performed on both plasma and whole blood for IE diagnosis and factors influencing test performance were preliminarily evaluated.

MATERIALS AND METHODS

Subject recruitment and inclusion.

Subjects were prospectively enrolled at an early phase of IE diagnosis. Subjects were included if they had been admitted to Mayo Clinic with a possible or definite diagnosis of IE based on the modified Duke criteria (5). Minors (<18 years old), inmates, and pregnant women were excluded. Subjects were also excluded if they refused consent or were discharged or died before blood collection (Fig. 1). Once written consent was collected, eligible subjects had one single draw 20 mL blood specimen collected, divided into two 10 mL EDTA tubes (Vacutainer, BD). The first 10 mL of whole blood was aliquoted into 10 cryovials of 1 mL each and frozen at −80°C. The second 10 mL of whole blood was centrifuged and plasma harvested and aliquoted into 5 cryovials of 1 mL each which were frozen at −80°C. All samples were analyzed after being frozen at –80°C. Subject characteristics and laboratory results were collected through review of the electronic medical record, including clinical notes, and laboratory and imaging findings.

FIG 1.

Study subject characteristics. IE, infective endocarditis.

Nucleic acid extraction.

2 mL of prewarmed (40°C) NUCLISENS easyMAG lysis buffer (bioMérieux, Marcy-l’Étoile, France) was mixed with whole blood or plasma. Then, it was left at room temperature for 10 min and loaded onto an EMAG nucleic acid extraction system (bioMérieux, Marcy-l’Étoile, France) using Specific B program. For whole blood, 200 μL input sample and 140 μL magnetic silica were used. For plasma, 1 mL input sample and 50 μL magnetic silica were used. With 1 mL of whole blood input, 50% of whole blood samples clotted in the extraction instrument, hence 200 μL was used. Eluate volume was 50 μL for both plasma and whole blood. Primers targeting the V1-V3 region of the bacterial 16S rRNA gene, coupled to dual priming oligonucleotides (DPOs) were used to reduce human DNA contamination (27).

Amplification.

Amplification was performed on a LightCycler 480II instrument (Roche Diagnostics, Basel, Switzerland), resulting in amplification of ~530 bp of the bacterial 16S rRNA gene. A positive control (Corynebacterium glutamicum DNA ATCC 13032) and a negative control (double autoclaved Qiagen Buffer EB, Hilden, Germany) were included in each run, from extraction to sequencing. To assess for inhibition, each sample was tested in parallel with C. glutamicum ATCC 13032 spiked-in.

Sequencing.

For NGS of amplified DNA, the 16S Metagenomic Sequencing Library Preparation protocol from Illumina was used to obtain an 8 pM DNA library (28). Paired-end sequencing was done on an Illumina MiSeq using a 2 × 300 cycle V3 Nano kit (Illumina), followed by onboard Illumina processing (adapter trimming, index demultiplexing, and fastq generation). For bioinformatic analysis, files were processed using Pathogenomix RipSeq NGS software that uses the “Pathogenomix Prime 16S database” for de novo clustering into Operational Taxonomic Units (OTUs) using a 99% homology threshold, as described in the Supplemental Methods. Sequences with <100 reads were rejected. For species-level identification, a cutoff of ≥99.2% homology was applied, except for Mycobacterium species and aerobic actinomycetes (e.g., Nocardia species), which required 100% homology. Between 98.0 and 99.1% homology, genus-level identification was made.

Repeat testing.

tMGS-negative samples were extracted a second time using a MagNA Pure 96 (Roche, Basel, Switzerland), using 200 μL input volume for whole blood and plasma, and 100 μL elution volumes. Amplification was performed as described above. Paired-end NGS was performed on an Illumina MiSeq^TM using a 2 × 250 V2 nano kit as described above.

Interpretation of results.

Sequencing results were interpreted by a clinical microbiology laboratory director with expertise in clinical metagenomic testing who was blinded to results of other testing, as described in Fig. 2. Then, clinical significance of the sequence-based findings was evaluated by at least 2 unblinded metagenomics experts using the algorithm outlined in Fig. 2. The threshold to call C. acnes positivity was >90% relative abundance. tMGS results were classified based on clinical significance. A result was considered positive and significant if the identification provided by tMGS matched that of blood culture, and/or any other routine testing, and/or yielded a potential new pathogen consistent with IE. Only clinically significant tMGS results were considered positive results. Clinical providers were not informed of results of the tMGS assay.

FIG 2.

Decision tree for sequencing results interpretation. If an organism was found in the negative control, it was not reported as positive. Percentage (%) of abundance refers to the number of reads mapping to the same sequence divided by the number total reads.

Statistical analysis.

Fisher’s exact test was used to compare qualitative values. For quantitative values, normality was assessed using the Kolmogorov-Smirnov test and results compared using the Mann-Whitney U test or t test. GraphPad Prism 9.0 (San Diego, CA) was used for statistical analysis.

Ethics.

This study was reviewed and approved by the Mayo Clinic Institutional Review Board (number 18-006497).

Data availability.

Detailed sequencing data is available upon request to the corresponding author.

RESULTS

Between October 2020 and July 2021, 40 subjects met inclusion criteria, 3 declined participation and 2 were discharged before blood draw. Of the 35 remaining subjects, 28 (80%) had definite IE and 7 (20%) had possible IE, according to modified Duke criteria (Fig. 1). For one subject, plasma was not prepared and for another, whole blood was not saved. The median subject age was 69 years old, 8 (23%) were female, and 22 (63%) had a prosthetic valve (Table 1). Only 15 (43%) underwent valve replacement surgery; valve culture was positive in 6 of 12 subjects (50%), and 3 of 5 (60%) had a positive 16S rRNA gene PCR/Sanger sequencing result on valve tissue, as performed routinely in our institution (24). Overall, the median time between the first antibiotic received and the tMGS blood draw was 6 days (interquartile range [IQR] 4 to 15 days). A total of 29 (83%) subjects had positive blood cultures; the median time between the last positive blood culture and the tMGS blood draw in this group was 4 days [IQR 0.75 to 7 days]. Six subjects (17%) had BCNE.

TABLE 1.

Subjects and sample characteristics^a

Subjects characteristics	Values
Demographics, n = 35
Age, yrs, median (IQR)	69 [55–82]
Sex, female, (%)	8 (23)
BMI, kg/m2, median (IQR)	29 [23–34]
Prosthetic valve, (%)	22 (63)
Previous infective endocarditis, (%)	9 (26)
Injection drug use, (%)	2 (6)
Duke criteria for endocarditis, n = 35
Definite, (%)	28 (80)
Possible, (%)	7 (20)
Laboratory findings
Hemoglobin, mg/dL, median (IQR), n = 35	9.5 [8.4−12]
Leukocyte count, 109/L, median (IQR), n = 35	9.4 [7.4−13]
Neutrophil count, 109/L, median (IQR), n = 34	7.2 [5.7−11]
Platelets, 109/L, median (IQR), n = 35	210 [98-277]
C-reactive protein (CRP), mg/L (reference range, 0–8.0), median (IQR), n = 24/35	105 [51-204]
Blood culture negative endocarditis, (%)	6 (17)
Echocardiography, n = 35
Transthoracic echocardiogram (TTE) positive for IE/total of TTE, (%)	12/30 (40)
Transesophageal echocardiogram (TEE) positive for IE/total of TEE, (%)	32/35 (91)
Complications, n = 35
Heart failure, (%)	17 (49)
Perivalvular or septal/myocardial abscess, (%)	11 (31)
Peripheral septic emboli, (%)	17 (49)
Death, (%)	6 (17)
Treatment, n = 35
Valve replacement surgery, (%)	15 (43)
Laboratory testing
Positive valve culture/valve sent for culture, (%)	6/12 (50)
Positive 16S ribosomal RNA (rRNA) gene PCR with sequencing on valve/valve sent for 16S rRNA gene PCR with sequencing, (%)	3/5 (60)
Inflammation present in valve tissue/valve sent for histopathologic analysis, (%)	12/13 (92)
Timing of blood draw for tMGS test
Number of days between tMGS and last positive culture, median [IQR], n = 29	4 [0.75–7]
Number of days between tMGS and first antimicrobial treatment, median [IQR], n = 35	6 [4−15]

^aIE, infective endocarditis; BMI, body mass index; tMGS, targeted metagenomic sequencing; IQR, interquartile range.

There was no significant difference in tMGS bacterial detection based on sample type; 20 (59%) whole blood and 16 (47%) plasma samples were positive (P = 0.47). When results of testing of both sample-types were combined, 23 subjects (66%) had at least one positive tMGS result (Table 2).

TABLE 2.

Results of targeted metagenomics sequencing (tMGS) of whole blood, plasma, and both combined^a

Subjects	Positive tMGS test
	Whole blood n = 34	Plasma n = 34	Combined n = 35
All infective endocarditis, n positive/n total (%)	20/34 (59%)	16/34 (47%)	23/35 (66%)
Blood culture positive infective endocarditis, n positive/n total BCPE (%)	17/28 (61%)	13/29 (45%)	18/29 (62%)
Blood culture positive on day tMGS collected, n positive/subgroup total (%)	7/7 (100%)	6/7 (86%)	7/7 (100%)
Blood culture negative on day tMGS collected, n positive/subgroup total (%)	10/21 (48%)	7/22 (32%)	11/22 (50%)
Blood culture negative infective endocarditis, n positive/n total BCNE (%)	3/6 (50%)	3/5 (60%)	5/6 (83%)

^aPositivity was determined according to the decision tree shown in Fig. 2. For one subject, there was no plasma and for another no whole blood available for testing. BCPE, Blood culture-positive endocarditis. BCNE, Blood culture-negative endocarditis.

In the subgroup with blood culture positive endocarditis (BCPE), tMGS was positive in 17 of 28 whole blood (61%) and 13 of 29 plasma (45%) samples. In the subgroup of 7 subjects in whom blood cultures (which were positive) and blood for tMGS were collected on the same day, there was 100% tMGS positivity for whole blood (7/7), and 86% tMGS positivity for plasma (6/7) samples. In contrast, when blood cultures collected the same day as blood for tMGS were negative in the context of having prior positive blood cultures, only 10 of 21 whole blood (48%) were tMGS positive, and 7 of 22 plasma (32%) samples were tMGS positive (Table 2).

In the 29 subjects with BCPE, 11 had negative tMGS results (including a single definite IE case with blood cultures positive for Candida albicans). Of the 18 positive tMGS cases, 10 yielded concordant results from plasma and whole blood, and blood culture, including 3 Staphylococcus aureus complex, 2 Enterococcus faecalis, and 1 each Streptococcus bovis group, Streptococcus mitis, Staphylococcus epidermidis, Staphylococcus hominis, and Staphylococcus lugdunensis cases. For the 8 remaining cases, tMGS results differed between whole blood and plasma, but in all cases, at least one detected organism was found in blood cultures. For one subject, tMGS detected S. aureus complex (as did blood cultures) from plasma but not from whole blood. In one sample, plasma tMGS and blood cultures detected S. mitis group and S. aureus complex, whereas whole blood tMGS detected only S. aureus complex. Finally, one sample was positive for Streptococcus salivarius group from blood cultures with the same detected by tMGS of whole blood, while tMGS of plasma additionally identified Rothia mucilaginosa, a finding of unclear clinical significance. There were four subjects with positive blood cultures and positive tMGS results from whole blood but negative tMGS findings from plasma, including 3 S. aureus complex (including one subject who was 31 days following a positive S. aureus blood culture) and 1 Corynebacterium amycolatum (after 44 days of adequate antibiotic therapy and 7 days from the last positive culture). Details are found in the Supplemental Results where reported results are indicated in bold and results from second sequencing are highlighted in beige.

In the BCNE cases, tMGS found 3 potential pathogens in the 6 whole blood samples (50%), 3 in the 5 plasma samples (60%), and 5 of 6 when both sample types were combined (83%). In one subject, Kingella species most closely related to Kingella denitrificans was detected by whole blood tMGS only in a subject with definite aortic prosthetic mechanical valve IE returning to the hospital 2 months after previously treated Cardiobacterium hominis IE, with fever, renal insufficiency, splenic emboli, and new aortic valve vegetations (Table 3.). Bartonella species was found by both whole blood and plasma tMGS in a subject with a mechanical aortic valve and three cats presenting with a 2 months history of fever and weight loss, anemia, cryoglobulinemia and heart failure. Blood cultures were negative; Bartonella henselae serology was IgG positive (titer 1:32768). The diagnosis was later confirmed by clinically performed 16S rRNA PCR (Ct value of 25) followed by Sanger sequencing on cardiac valve tissue (24). C. acnes was found by plasma tMGS in a person who injected drugs, with possible native valve IE; C. acnes was also found in whole blood but at an abundance not meeting the 90% threshold. The subject died shortly following diagnosis. Corynebacterium species was found by plasma tMGS in a subject with a second episode of IE involving an aortic bioprosthetic valve presenting with heart failure and a 6 × 5 mm vegetation; the first IE episode involved the same valve and Staphylococcus aureus. tMGS was repeated from extraction and the result confirmed. Streptococcus salivarius/thermophilus was found by tMGS in whole blood of a subject with definite native tricuspid IE. The subject had a history of Streptococcus dysgalactiae IE involving the same valve 2 months prior, with the last dose of antibiotic treatment 15 days before the relapse. Unfortunately, no plasma was available from this subject; a clinically performed commercially available shotgun metagenomics test performed on plasma (Karius test [16]), detected S. dysgalactiae.

TABLE 3.

Description of cases and targeted metagenomic sequencing (tMGS) results in blood culture negative infective endocarditis (IE) cases

Subject	Case description	tMGS-whole blood	tMGS-plasma	Confirmation by duplicate tMGS testing
2	Definite mechanical aortic valve IE with new vegetations two months after treatment for Cardiobacterium hominis IE	Kingella sp. (most closely related to Kingella denitrificans)	Negative	No
12	Definite mechanical aortic valve IE	Bartonella sp.	Bartonella sp.	Yes
13	Possible native tricuspid valve IE in person who injected drugs	Negative	Cutibacterium acnes	Yes
17	Possible bioprosthetic aortic valve IE three months after previously treated Staphylococcus aureus IE	Negative	Corynebacterium sp.	Yes
26	Definite mechanical aortic and mitral valve IE plus native tricuspid valve IE after previously treated S. aureus aortic IE	Negative	Negative	Not applicable
35	Definite native tricuspid valve IE two months after treated Streptococcus dysgalactiae IE in the presence of a pacemaker	Streptococcus salivarius /thermophilus	Not tested	Yes

As shown in Fig. 3A to C, using whole blood, plasma and both combined, tMGS positivity was not significantly different between definite and possible IE, between BCPE and BCNE, or between subjects with and without vegetations (P > 0.05). Positivity of blood culture at time of tMGS specimen collection significantly increased the likelihood of a positive tMGS result from whole blood, plasma or both combined (P < 0.05). Samples with Ct values <37 had higher likelihoods of tMGS positivity from plasma (P = 0.04), but not whole blood or both combined (P > 0.05). For culture positive cases, the median time between the last positive culture and the tMGS blood draw was 2 days (IQR 1 to 6) for positive and 7 days (IQR 5 to 16) for negative tMGS whole blood samples (P = 0.017), 5 days (IQR 1 to 7) and 6.5 days (IQR 3.250 to 15) for positive and negative tMGS plasma samples, respectively (P = 0.003), and 2 days (IQR −1 to 6) and 7 days (IQR 5 to 16) for positive and negative tMGS whole blood and plasma samples combined, respectively (P = 0.002, Fig. 3D).

FIG 3.

Factors influencing targeted metagenomic sequencing (tMGS) positivity. (A) Whole blood; (B) Plasma; and (C) Whole blood and plasma combined (i.e., positive result from either). (D) Time between tMGS blood draw and most recent antecedent positive culture in the 28 culture-positive samples. (E) Time between the tMGS blood draw and the first antibiotic dose in the entire cohort. IE: infective endocarditis, Ct: cycle threshold, BCPE: blood culture positive endocarditis, BCNE: blood culture negative endocarditis.

The time from first antibiotic dose to tMGS specimen collection did not impact the probability of having positive tMGS results from whole blood (P = 0.309), plasma (P = 0.065) or both combined (P = 0.338), although the median duration was numerically longer for negative than positive samples across all three groups (Fig. 3E).

Repeat sequencing of negative tMGS samples provided 5 additional positive whole blood results (of 19 repeated) and 1 additional positive plasma result (of 19 repeated).

DISCUSSION

This is one of the first prospective cohort studies to test tMGS of whole blood and plasma in subjects with IE. This approach may provide a strategy to detect bacteria in patients with IE, when blood cultures have returned negative. The assay can be completed in ~60 h, or ~120 h if repeat sequencing is necessary. Using the protocol established here, tMGS detected and identified potential pathogens in both culture positive and negative IE without significant differences between the two. The positivity rate was influenced by the time since the last positive culture; all tMGS specimens collected within 24 h of a positive blood culture were positive. tMGS trended toward negativity with more days after first antibiotic usage, although not being significantly influenced by time since first antibiotic dose; notably, this study was underpowered to detect such differences. Although tMGS of whole blood was positive in 59% of cases compared to 47% of cases when plasma was tested, the difference was not statistically significant, and results were complementary, increasing positivity to 66% when results of testing of both sample types were combined. Of 6 BCNE cases, potential pathogens were found in five.

In a retrospective study of 10 selected pediatric IE cases, To et al. reported 8 positive results from a commercial plasma shotgun metagenomic sequencing test (Karius Test), including 5 of 7 BCNE cases (25). However, in a larger retrospective study on the use of the same test in a pediatric population, plasma shotgun metagenomics did not identify a pathogen in the 4 BCNE cases tested (29). Shotgun metagenomic sequencing (Karius Test) was assessed in a prospective cohort of 24 adults presenting with definite IE; of four BCNE cases, two were negative and one each was positive for Enterococcus faecalis and Escherichia coli (1) (26).

In the current study, sequencing negative results a second time increased assay sensitivity, as did sequencing both plasma and whole blood. The addition of tMGS of whole blood and not just plasma might increase yield since it can detect intracellular DNA and not just cell free DNA. An advantage of using tMGS compared to shotgun metagenomic sequencing is that human DNA does not contribute to the sequences generated, simplifying bioinformatic analysis and data interpretation. The downside is that this approach cannot be used to predict antimicrobial resistance or to detect nonbacterial organisms, such as Candida albicans, as found in one case in this study.

Limitations of the current study include the small number of subjects studied (although to the best of our knowledge, this is the largest prospective adult cohort published to date). Because Mayo Clinic is a tertiary reference center, 14 subjects were referred from local hospitals, preventing collection of blood for tMGS at initial presentation. Subjects were included regardless of the date of their last positive culture or first antibiotic dose, to reflect real clinical practice; doing so may have decreased the positivity rate. Indeed, samples for tMGS were often obtained several days after the last positive blood culture and after antibiotic treatment had commenced. In subjects with BCPE for whom the blood cultures were negative at time of specimen collection for tMGS testing, there were 10/21 positive tMGS results from whole blood and 7/22 from plasma, suggesting potential applicability in BCNE or to follow up treatment efficacy as suggested by Eichenberger et al. in bloodstream infection (17). Analytical sensitivity and specificity of the assay should be analyzed in a larger study. Additionally, testing included a second round of sequencing for 19 samples of plasma and 17 samples of whole blood that initially tested negative. Due to unavailability of the same reagents at the time of secondary sequencing, a different extraction method was used, which prevented assessment of the ideal extraction approach. The second round of sequencing was valuable for 4 whole blood and 1 plasma samples that were negative in the first round. Further development of the assay is needed to address which specimen type and testing algorithm (including volume of input sample and extraction method) will achieve the best performance.

For samples with low DNA concentrations, it can be challenging to discriminate between background and pathogens. For this reason, a decision tree was used for data interpretation of raw sequencing results. Organisms such as C. acnes cannot be differentiated from background unless present in high abundance. Polymicrobial sequence results resembling skin or oral microbiota must be interpreted carefully, as they do not necessarily represent pathogens, but contaminants introduced during sampling or sample processing or constituting the normal blood microbiome (e.g., for oral microorganisms) (30). Finally, results must be correlated with the clinical context. The role of tMGS outside possible and definite IE is not addressed by the current study.

Finally, tMGS could possibly be used to understand, follow, and guide treatment of IE, as it identified a pathogen in a subject (subject number 4) under antibiotic therapy 31 days after the last positive blood culture. However, this potential application requires further study.

In conclusion, tMGS might improve IE diagnosis, identifying pathogens in BCNE. Additionally, this approach might provide a faster diagnosis than blood culture for slow growing pathogens, and/or reduce the number of tests required for diagnosis (e.g., serology, valve culture).