Introduction
High mortality and morbidity rates are associated with bloodstream infections (BSIs) [
1,
2]. In polymicrobial BSIs, which are not uncommon (ranging from 10 to 20% of all diagnosed BSIs), the mortality rates are even higher than with monomicrobial BSIs [
3‐
6]. Therefore, rapid administration of effective antimicrobial treatment is crucial for patient survival [
7,
8]. To support this, timely and accurate identification of microorganisms are central to the optimal management of BSIs.
The “gold standard” for screening of BSIs is still blood culture followed by Gram-staining and conventional culture-based techniques. The conventional techniques have the advantage of providing accurate species identification and antimicrobial susceptibility profile but the disadvantage of requiring 1–2 additional days after signal-positive blood cultures. During the last few years, various methods have been developed to optimize the detection of the etiological agent of BSI such as fluorescence hybridization probes [
9‐
11], pathogenic-specific PCR [
12,
13], and MALDI-TOF MS [
14‐
16]. Despite the advantages of these approaches, many of them, however, comprise multiple assay steps for the purification and extraction of the microbes and their target protein or DNA. Moreover, most of them provide limited information from polymicrobial blood culture samples. Modern automatization and advances in molecular technology, however, have recently led to the development of molecular-based high multiplex platforms, such as the FilmArray (bioMérieux, France), the ePlex (GenMark Diagnostics, USA), and the Verigene (Luminex, USA), for direct and rapid multimicrobial detection from signal-positive blood culture bottles [
17‐
19]. These assays allow significantly shorter time to detection than conventional methods and a high performance even with polymicrobial samples, although some limitations in the first-generation assay panel content were evident (e.g., absence of the most common resistance genes from the panel) [
17‐
19].
Here, the performance and utility of the recently launched updated version of the FilmArray Blood Culture Identification 2 panel, the BCID2 assay (bioMérieux, France), were evaluated. According to the manufacturer, the BCID2 panel identifies simultaneously 33 species/genus targets and 10 antimicrobial resistance genes. This is nine microbial targets and seven genes coding for antimicrobial resistance more than with the first-generation BCID [
20] assay.
Results
From the simulated sample mixes, the BCID2 panel provided a correct identification in 99.9% (95% CI, 99.7–100%) of all target microbes and in 100% (95% CI, 98.7–100%) of all resistance marker genes (Table
1). The false detections consisted of one
Klebsiella aerogenes isolate, which was repeatedly undetected from different sample mixes and
Shigella flexneri, which was misidentified as
Escherichia coli. In addition,
Bacteroides ovatus, a member of
B. fragilis group, was not detected as
B. fragilis, whereas
B. thetaiotaomicron, also a member of
B. fragilis group, was detected as
B. fragilis. This was considered as a false-negative detection for the
B. ovatus isolate. In general, the BCID2 panel showed excellent performance from the multipositive samples providing correct identification even from samples containing 11 different targets.
Of the 103 clinical blood culture samples, 80 were signal-positive containing either Gram-negative bacteria, Gram-positive rods, diplococci, cocci in chains, yeasts, or polymicrobial growth by Gram-staining. The proportion of polymicrobial samples was 15.0% (
n = 12). The remaining 23 samples, included in the study, were signal- and growth-negative. In total, 36 Gram-positive bacteria, consisting of 12 different species, 36 Gram-negative bacteria, consisting of nine different species, and five yeasts, consisting of two different species, were correctly identified by the BCID2 panel (Table
3). The most common microbes recovered were
E. coli (
n = 15),
Streptococcus spp. (
n = 11),
Klebsiella spp. (
n = 10), and
L. monocytogenes (
n = 6). The BCID2 panel covered 81.9% of the microbes found from the blood culture samples. The missing (off-panel) species were mainly anaerobic Gram-negative rods and aerobic Gram-positive rods other than
Listeria spp. One unresolved discordant result was observed from a sample containing
Enterococcus avium,
Pseudomonas aeruginosa, and
E. coli from where the BCID2 panel yielded
Streptococcus spp.,
P. aeruginosa, and
E. coli positive result. Growth appropriate for streptococci was not observed on any agar plates nor was the isolate confirmed by any other method. Along with the Gram-negative and Gram-positive bacteria, 10 resistance genes (six
blaCTX-M and four
mecA/C genes) were detected and correctly identified by the BCID2 panel (Table
3). No additional resistance markers were found by the routine reference methods. Thus, the overall sensitivity and specificity of the BCID2 panel from the clinical sample set were 100% (95% CI, 95.9–100%) and 99.9% (95% CI, 99.8–100%), respectively.
Table 3
Results of the FilmArray BCID2 assay from clinical blood culture samples (n = 103, one per patient)
Gram-positive bacteria |
Enterococcus faecalis | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Enterococcus faecium | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Listeria monocytogenes | 6 | 97 | 0 | 0 | 100% (54.1–100%) | 100% (96.3–100%) |
Staphylococcus spp. | 1a | 102 | 0 | 0 | 100% (2.5–100%) | 100% (96.5–100%) |
Staphylococcus aureus | 5 | 98 | 0 | 0 | 100% (47.8–100%) | 100% (96.3–100%) |
Staphylococcus epidermidis | 4 | 99 | 0 | 0 | 100% (39.8–100%) | 100% (96.3–100%) |
Staphylococcus lugdunensis | 1 | 102 | 0 | 0 | 100% (2.5–100%) | 100% (96.5–100%) |
Streptococcus spp. | 11b | 91 | 0 | 1c | 100% (71.5–100%) | 98.9% (94.1–100%) |
Streptococcus agalactiae (group B) | 4 | 99 | 0 | 0 | 100% (39.8–100%) | 100% (96.3–100%) |
Streptococcus pneumoniae | 2 | 101 | 0 | 0 | 100% (15.8–100%) | 100% (96.4–100%) |
Streptococcus pyogenes (group A) | 2 | 101 | 0 | 0 | 100% (15.8–100%) | 100% (96.4–100%) |
Gram-negative bacteria |
Acinetobacter calcoaceticus-baumannii complex | 1 | 102 | 0 | 0 | 100% (2.5–100%) | 100% (96.5–100%) |
Bacteroides fragilis | 1 | 102 | 0 | 0 | 100% (2.5–100%) | 100% (96.5–100%) |
Enterobacterales | 2d | 101 | 0 | 0 | 100% (15.8–100%) | 100% (96.4–100%) |
Enterobacter cloacae complex | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Escherichia coli | 15 | 88 | 0 | 0 | 100% (78.2–100%) | 100% (95.9–100%) |
Klebsiella aerogenes | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Klebsiella oxytoca | 5 | 98 | 0 | 0 | 100% (47.8–100%) | 100% (96.3–100%) |
Klebsiella pneumoniae group | 5 | 98 | 0 | 0 | 100% (47.8–100%) | 100% (96.3–100%) |
Proteus spp. | 1 | 102 | 0 | 0 | 100% (2.5–100%) | 100% (96.5–100%) |
Salmonella spp. | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Serratia marcescens | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Haemophilus influenzae | 1 | 102 | 0 | 0 | 100% (2.5–100%) | 100% (96.5–100%) |
Neisseria meningitidis | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Pseudomonas aeruginosa | 5 | 98 | 0 | 0 | 100% (47.8–100%) | 100% (96.3–100%) |
Stenotrophomonas maltophilia | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Yeasts |
Candida albicans | 3 | 100 | 0 | 0 | 100% (29.2–100%) | 100% (96.4–100%) |
Candida auris | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Candida glabrata | 2 | 101 | 0 | 0 | 100% (15.8–100%) | 100% (96.4–100%) |
Candida krusei | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Candida parapsilosis | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Candida tropical | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Cryptococcus neoformans/gattii | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Resistance markers |
CTX-M | 6 | 97 | 0 | 0 | 100% (54.1–100%) | 100% (96.3–100%) |
IMP | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
KPC | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
NDM | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
OXA-48-like | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
VIM | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
mecA/C | 1 | 102 | 0 | 0 | 100% (2.5–100%) | 100% (96.5–100%) |
mecA/C and MREJ (MRSA) | 3 | 100 | 0 | 0 | 100% (29.2–100%) | 100% (96.4–100%) |
mcr-1 | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
vanA/B | 0 | 103 | 0 | 0 | N/A | 100% (96.5–100%) |
Species identification with the routine culture-based methods was available for 68.0% of the samples within 24–48 h, when short-incubation (si-)MALDI-TOF could not be utilized (e.g., with polymicrobial growth and slowly growing organism), and for 32.0% of the samples within 6–8 h, when si-MALDI-TOF could be utilized. For antimicrobial susceptibilities, the total turnaround time was 24–48 h with the routine methods. The direct analysis of the clinical blood culture samples with the automated BCID2 panel enabled species identification and detection of the most common antimicrobial resistance genes within 70 min. This provided significant (P value < 0.0001) time reduction as compared with the routine methods.
The BCID2 panel invalidity rate from the clinical blood culture bottle samples was 1.0% (1/103).
Discussion
In the current study, the performance and usability of the FilmArray BCID2 panel for rapid detection of organism from blood culture samples were evaluated. The new BCID2 panel has been improved greatly as it detects nine microbial targets and seven resistance genes more than the first-generation BCID panel [
20,
23]. In addition, the assay run time has been reduced from 90 to 70 min. Hence, the panel provides a detection of 33 most common microbial species/genus causing BSI and 10 antimicrobial resistance marker genes reducing the time to species identification and detection of major resistance genes by at least a full day as compared with conventional culture-based methods.
The performance of the BCID2 panel was excellent providing overall 100% sensitivity and 100% specificity for on-panel antimicrobial resistance markers and 98.8% sensitivity and 99.9% specificity for on-panel microbial targets. Of the four cases from which the BCID2 panel failed to correctly identify the isolates, two contained off-panel organisms (
E. avium and
S. flexneri) according to reference methods. Of these,
S. flexneri in monomicrobial growth was falsely identified as
E. coli which is a severe misidentification, due to the low incidence but potentially life-threatening outcomes of
S. flexneri bacteremia. The misidentification may occur if, e.g., β-glucuronidase gene (uidA), which is present in both species, is used for
E. coli detection [
24,
25]. Although these two organisms might be difficult to distinguish, there are genes that could be used for accurate differentiation and species detection [
24]. Here, the
S. flexneri isolate was used in simulated sample and from pure culture to ensure the result reliability.
The sample containing E. avium, however, remained unresolved as it contained polymicrobial growth of E. avium, P. aeruginosa, and E. coli of which P. aeruginosa and E. coli were correctly identified. With the reference methods, the reported Streptococcus spp. result could not be confirmed and was thus considered as a false detection from the E. avium isolate. However, since the sample contained polymicrobial growth, it is possible that streptococci, which could have been overgrown by the other species, was present in the sample but not detected by the culture-based reference methods. Sadly, the Enterococcus genus level detection, which was included in the first-generation BCID panel, has been removed from the BCID2 panel. It could have been used for the differentiation of Enterococcus spp. and Streptococcus spp. and used here to conclude whether or not the sample contained all four species (P. aeruginosa, E. coli, Enterococcus spp., and Streptococcus spp.). The differentiation of enterococci and streptococci genus from BSI samples is important, in general, due to their different natural resistance traits and thus different treatment options.
Of the other two samples yielding false-negative results with the BCID2 panel, one contained
K. aerogenes (on-panel target) isolate which was confirmed as
K. aerogenes by MALDI-TOF and 16S rDNA sequencing and the other one
B. ovatus which is a member of the
B. fragilis group. The BCID2 panel yielded no result from the
B. ovatus while providing
B. fragilis positive results from a sample containing
B. thetaiotaomicron, another member of the
B. fragilis group. All of these species are abundant in colon and important pathogens in polymicrobial infections [
26]. Moreover, according to Brook,
B. thetaiotaomicron and
B. ovatus covers approximately 21% of all infections caused by the
B. fragilis group and are associated with high mortality rate (20–30%) when inducing bacteremia [
26].
Regarding the above-mentioned discrepancies it is interesting to note that three of the four misidentifications observed in this study are related to targets introduced in the panel of the new version of the kit. Consequently, further evaluation with more isolates within this group of microbes would be highly beneficial.
In addition to the FilmArray BCID2 panel, there are only few other fully automated assays that provides a broad spectrum of species identification and detection of antimicrobial resistance genes directly from signal-positive blood culture bottles. Of these, the ePlex® Blood Culture Identification (BCID) panel (Genmark Diagnostics, USA) can detect multiple microbes and resistance genes (i.e.,
blaKPC,
blaNDM,
blaVIM,
blaIMP, and
blaOXA-48-like genes in BCID-GN panel) within 90 min but is limited to detect only one type of organisms with a one type of test panel (e.g., BCID-GN for Gram-negative bacteria and BCID-GP for Gram-positive bacteria), and none of the panels are yet CE-IVD marked [
27]. The Verigene blood culture assay (Luminex, USA) has the same limitation with test panels than the ePlex and requires approximately 2.5 h of run time per sample [
28].
The limitation of this study was the low number of positive genes coding for carbapenem and vancomycin resistance in clinical samples, due to the low prevalence of carbapenem-producing organisms and vancomycin-resistant enterococci in BSI in Finland [
29‐
31].
In conclusion, our results show that the FilmArray BCID2 panel proved to be a well-performing and invaluable tool for the early detection of common clinical isolates causing BSI and their antibiotic resistance genes while providing rapid aid for clinicians and helping them to administer effective antimicrobial therapy more hastily and therefore reducing patient mortality [
32]. Whether to use FilmArray BCID2 panel to test all positive blood culture bottles regardless of the gram stain result or not depends on multiple factors. In an ideal situation where the most accurate and timely microbiological diagnosis is paramount FilmArray BCID2 panel could be used to test all positive blood culture bottles. However, this kind of approach can easily lead to excessive overall costs and increase the laboratory workload unnecessarily. A more frugal manner of testing only one blood culture bottle per patient might be a recommendable approach for routine diagnostic practices in most laboratories. In surroundings where resources are scarce, or the capacity of the test system is limited a more controlled approach for FilmArray BCID2 panel use could be implemented focusing on the cases for which the test provides maximal clinical impact. Before implementing this kind of diagnostic test to the routine practices, we encourage to assess the need for the test and the proposed use for the test locally with infectious disease specialists to integrate the new diagnostic means to local antimicrobial stewardship guides and to maximize the benefits of the test without increasing the overall costs of diagnostics unnecessarily. However, the overall costs of diagnostics would increase with the FilmArray BCID2 panel as compared with the conventional methods.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.