During the study period, 153 positive blood cultures gave a positive signal and were assessed for inclusion in our analysis. Overall, eight positive blood cultures were excluded from the study, due to a negative initial Gram-stain result. In total, 137 positive blood cultures were included in our study (Fig.
1). Among those 98 episodes were monomicrobial BSI with Gram-positive organisms, 33 were monomicrobial with Gram-negative organisms and 6 episodes were caused by yeast strains (Tab.
1). In addition, in eight samples, more than one organism was detected by culture. One example was not included in the study as the presence of multiple organisms was already suspected by the Gram stain, the results of the remaining seven samples are summarized in Table
S2.
Table 1
Identification results for bacteria and yeast (monomicrobial BSI)
ID of Gram-positive organisms |
S. aureus (n = 18) | 18/0 | 18/0 | 119/0 | 119/0 | |
S. epidermidis (n = 35) | 35/0a | 35/2 | 102/0 | 100/0 | FilmArray® ID only on Staphylococcus spp. level, false ePlex® ID of S. hominis as S. epidermidis, false ePlex® ID of S. warneri as S. epidermidis |
S. haemolyticus (n = 8) | 8/0a | 7/0 | 129/0 | 129/1 | FilmArray® ID only on Staphylococcus species level, no ePlex® ID for one S. haemolyticus isolate |
S. hominis (n = 6) | 6/0a | 5/0 | 131/0 | 131/1 | FilmArray® and ePlex® ID only on Staphylococcus spp. level, false ePlex® ID of S. hominis as S. epidermidis |
S. warneri (n = 2) | 1/0a | 1/0 | 135/1 | 135/1 | FilmArray® and ePlex® ID only on Staphylococcus spp. level, no FilmArray® ID for 1/2 isolates, false ePlex® ID of one S. warneri isolate as S. epidermidis |
S. capitis (n = 2) | 2/0a | 2/0 | 135/0 | 135/0 | FilmArray® and ePlex® ID only on Staphylococcus spp. level |
E. faecium (n = 4) | 4/0a | 4/2 | 133/0 | 131/0 | FilmArray® ID only on Enterococcus spp. level, false ePlex® ID for two E. faecalis isolates as E. faecium |
E. faecalis (n = 5) | 5/0a | 3/0 | 132/0 | 132/2 | FilmArray® ID only on Enterococcus spp. level, false ePlex® ID for two E. faecalis isolates as E. faecium |
S. pneumoniae (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | |
S. gallolyticus (n = 2) | 2/0 | 2/0 | 135/0 | 135/0 | FilmArray® and ePlex® ID only on Streptococcus spp. level |
S. mitis group (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | FilmArray® and ePlex® ID only on Streptococcus spp. level |
S. oralis (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | FilmArray® and ePlex® ID only on Streptococcus spp. level |
S. anginosus (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | FilmArray® ID only on Streptococcus spp. level |
Streptococcus sp. (group C) (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | FilmArray® and ePlex® ID only on Streptococcus spp. level |
C. acnes (n = 5) | n/a | 5/0 | n/a | 132/0 | Not included in the FilmArray® panel |
A. meyeri (n = 1) | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex® panel |
Actinomyces sp. (n = 1) | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex® panel |
P. faecalis (n = 1) | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex® panel |
Lactobacillus sp. (n = 1) | n/a | 1/0 | n/a | 136/0 | Not included in the FilmArray® panel, ePlex® ID only on Lactobacillus spp. level |
Corynebacterium sp. (n = 1) | n/a | 1/0 | n/a | 136/0 | Not included in the FilmArray® panel |
A. parvulum (n = 1) | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex® panel |
Total number of organisms (n = 98) | 86/0 | 89/4 | 50/1 | 39/5 | |
ID of Gram-negative organisms |
E. coli (n = 15) | 15/0 | 15/0 | 122/0 | 122/0 | |
K. pneumoniae/variicola (n = 5) | 5/0 | 5/0 | 132/0 | 132/0 | Two K. variicola identified as K. pneumoniae by FilmArray® and ePlex® |
K. oxytoca (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | |
S. marcescens (n = 2) | 2/0 | 2/0 | 135/0 | 135/0 | |
E. cloacae Komplex (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | |
E. ludwigii (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | FilmArray® and ePlex® ID as E. cloacae complex |
P. aeruginosa (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | |
H. influenzae (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | |
M. liquefaciens (n = 1) | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex®panel |
N. polysaccharea (n = 1) | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex® panel |
C. canimorsus (n = 1)a | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex®panel; aonly identified by 16S-rRNA-PCR |
B. fragilis (n = 2) | n/a | 2/0 | n/a | 135/0 | Not included in the FilmArray® panel |
B. vulgatus/dorei (n = 1) | n/a | n/a | n/a | n/a | Not included in the FilmArray® and ePlex® panel |
Total number of organisms (n = 33) | 27/0 | 29/0 | 110/0 | 108/0 | |
ID of yeasts |
Candida albicans (n = 2) | 2/0 | 2/0 | 135/0 | 135/0 | |
Candida glabrata (n = 2) | 2/0 | 2/0 | 135/0 | 135/0 | |
Candida parapsilosis (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | |
Candida tropicalis (n = 1) | 1/0 | 1/0 | 136/0 | 136/0 | |
Total number of organisms (n = 6) | 6 | 6 | 131/0 | 131/0 | |
Identification of Gram-positive BSI pathogens
In summary 21 different Gram-positive organisms (98 isolates in total) were determined by routine-diagnostic methods. Identification by MALDI-TOF was compared to identification results created by the BioFire FilmArray® and the GenMark ePlex® system.
Correct identification by the FilmArray® was achieved for 86/98 (87.8%) isolates on genus and species level, as defined in the system panel for the respective target organism. The system scored 12 samples negative (12.2%). These 11 samples contained organisms that are not included in the system panel, namely
C. acnes (
n = 5
), Actinomyces meyeri (
n = 1),
Actinomyces species (n = 1),
P. faecalis (n = 1),
Lactobacillus species (n = 1),
Corynebacterium species (n = 1) and
Atopobium parvulum (n = 1). One
Staphylococcus warneri isolate was not detected as
Staphyloccus species by the system. ID failure of
S. warneri has been reported before and was explained by a reduced sensitivity of the FilmArray® system for some CoNS including
S. warneri [
19], which is also mentioned in the package insert of the panel. The FilmArray® therefore identified 86/87 (98.9%) correctly taking only the organisms included in the panel into account. The results are shown in detail in Table
1. For the Gram-positives, the FilmArray® showed a sensitivity of 98.9% and a 100% specificity (Table
4) for organisms covered by the panel.
In contrast to the FilmArray® system, the ePlex® system aims at identifying enterococci and coagulase negative staphylococci to the species level. In total identification to the species level was achieved in 89/98 (90.8%) of the Gram-positive bacterial isolates. Two
Staphyloccus species were misidentified by the system:
S. hominis as well as
S. warneri were identified as
S. epidermidis. A larger sample size is required to assess if this assay is more prone to misidentification of CoNS. Furthermore, it is difficult to rule-out that one or more of these specimens contained multiple
Staphylococus species. In addition, as all these three CoNS species belong to the
S. epidermidis group as part of the normal skin flora with similar pathogenicity [
20] it is not mandatory to identify these CoNS on the species level. Worth mentioning is the inclusion of
S. lugdunensis in the Genmark ePlex® panel as it is a common pathogen with species specific virulence factors [
21] associated with aggressive causes of infective endocarditis [
22‐
24].
In addition, two
E. faecalis isolates were misidentified as
E. faecium. The correct identity of these two
E. faecalis isolates was verified by MALDI-TOF and confirmed by next generation sequencing (NGS) of the whole genome on an Illumina Nextseq platform followed by calculating the average nucleotide identity (ANI) [
25]. The ePlex® cartridges used in this study were for Research Use Only and according to the package insert of the test assay in a subsequent clinical study evaluating the performance of the CE-IVD and U.S. FDA cleared assays, the ePlex® demonstrated high sensitivity and specificity for identification of
Enterococcus faecalis.
Enterococcus species cause approximately 10% of nosocomial blood stream infections [
26] caused by a variety of different entries of the pathogen [
27]. Treatment of
Enterococcus related BSI is strongly dependent on correct species identification. Whereas
E. faecalis is mostly susceptible to the small-spectrum antibiotic ampicillin,
E. faecium is resistant [
28].
Classifying CoNS and enterococci only to the genus level, correct identification could be achieved for 93/98 (94.9%) of the Gram-positive bacterial isolates. Based on the broader panel
C. acnes (
n = 5),
Lactobacillus species (
n = 1) and
Corynebacterium species (
n = 1) were detected in contrast to the FilmArray® system. Four bacterial isolates were not identified as they were not part of the panel:
Actinomyces meyeri (
n = 1),
Actinomyces species (
n = 1),
P. faecalis (
n = 1) and
Atopobium parvulum (
n = 1). Moreover one
S. haemolyticus strain was not detected by the system. Regarding only organisms covered by the panel, the ePlex® correctly identified 93/94 (98.9%) strains to genus level and 89/94 (94.7%) to the species level. For two blood cultures, the ePlex® detected a second Gram-positive organism (
Micrococcus in addition to
S. aureus and accordingly
Staphylococcus spp. in addition to
E. faecium). None of these additional organisms could be confirmed by culture (Table
1). The ePlex® panel allows in contrast to the FilmArray® identification of potential contaminants namely
Corynebacterium spp.,
Micrococcus spp., and
C. acnes usually resulting in no need for treatment unless there is direct evidence for infection by the organism. Fast identification of such contaminants can help in quick de-escalation of the ongoing therapy and prevent redundant administration of anti-infectious therapies [
29‐
31].
The ePlex® system showed a sensitivity of 94.7% and a specificity of 90.7% for Gram-positive pathogens included in the panel (Table
4).
Percentages of detected organisms covered by the respective blood culture identification panel as well as for all microorganism detected by the reference method are shown in Table
2.
Table 2
Resistance genes identified for Gram-positive bacteria by FilmArray® and ePlex®
Staphylococcus species (total n=66) | mecA in house PCR | mecC in house PCR | mecA FilmArray® | mecA ePlex® | mecC ePlex® | comment |
S. aureus (n=18) | negative (n=18)a | n/a | negative (n=18) | negative (n=18) | negative (n=18) | |
Coagulase-negative Staphylococcus spp. (n=48) | negative (n=15)b | n/a | negative (n=16) | negative (n=15) | negative (n=48) | False-negative mecA result by FilmArray® (n=1) |
positive (n=33)c | | positive (n=31) | positive (n=33) | | No target detected by FilmArray® (n=1) |
Genotypic detection of vancomycin resistance (vanA/vanB) |
Enterococcus species (total n=9) | vanA/vanB in-house PCR | vanA/vanB FilmArray® | vanA/vanB ePlex® | comment |
Enterococcus vancomycin sensitive (n= 8) | negative (n=8) | negative (n=8) | negative (n=8) | E. faecalis (n=5), E. faecium (n=3) |
Enterococcus vancomycin resistant (n= 1) | positive van B (n=1) | positive vanA /vanB (n=1) | positive vanB (n=1) | vanB positive E. faecium isolate |
Identification of Gram-negative BSI pathogens
We further assessed identification of Gram-negative organisms for the BioFire FilmArray® and the Genmark ePlex® system. Routine-diagnostic analysis provided evidence for 14 different species within 33 samples in total. Due to the lack of bacterial growth, one sample with
Capnocytophaga canimorsus was not verified by MALDI-TOF, but by 16S rRNA PCR out of the positive blood culture bottle. Two
Klebsiella variicola isolates were reported as
Klebsiella pneumoniae by the FilmArray® and GenMark ePlex®, but were not assessed as misidentification, as MALDI-TOF based differentiation of these two species has only recently become available [
32].
Klebsiella variicola was described as a new species genetically isolated from
K. pneumoniae and can be phenotypically distinguished by the inability of adonitol fermentation. The pathogenicity potential and analysis of virulence factors of
K. variicola are still under investigation and high-risk antibiotic resistance genes are already described for
K. variicola [
33‐
35]. Thus, identification of
K. variicola on species level and dissociation from
K. pneumoniae could be of relevance for a future panel update.
In contrast to the FilmArray® panel, the ePlex® system can identify the
Bacteroides fragilis, which is of clinical importance since the
B. fragilis group belongs to the most prevalent anaerobic pathogens causing BSI [
36].
In total, the FilmArray® identified 27/33 isolates (81.8%). Six samples included organisms outside the panel (Moraxella liquefaciens (n = 1), Neisseria polysaccharea (n = 1), Capnocytophaga canimorsus (n = 1), B. fragilis (n = 2) and B. vulgatus/dorei (n = 1)). This resulted in a correct identification by the FilmArray® instrument for all 27/27 (100%) organisms included in the panel.
The ePlex® system identified 29/33 isolates correctly (87.9%), whereas M. liquefaciens (n = 1), N. polysaccharea (n = 1), C. canimorsus (n = 1) and B. vulgatus/dorei (n = 1) were not included in the panel. Thus, the remaining 29/29 specimens (100%) were correctly identified.
Identification of yeast strains as BSI pathogens
During the study period, six BSI episodes with different yeast strains occurred. Both Biofire FilmArray® and GenMark ePlex® identified all 6/6 strains (100%) correctly, namely C. albicans (n = 2), C. glabrata (n = 2), C. parapsilosis (n = 1) and C. tropicalis (n = 1), which were all confirmed by MALDI-TOF analysis.
Detection of polymicrobial BSI pathogens
In total, for seven polymicrobial BSI episode samples, both test systems were performed (summarized in Table
S2). The FilmArray® and ePlex® detected all pathogens covered by the panels at least on genus level with only one exception: the FilmArray® missed detecting a
Streptococcus constellatus isolate although covered by the panel, which was found in addition to
Proteus mirabilis in one blood culture bottle. In the initial Gram-stain, only Gram-negative rods were observed. Furthermore, culture-based diagnostic showed only a few colonies of
Streptococcus on CNA agar plates, indicating a low burden of Gram-positives in this blood culture.
For the ePlex® the cartridge for Gram-negatives was used for the analysis of that blood culture resulting in the detection of the Gram-positive PAN target indicating the precence of the Streptococcus constellatus isolate. The Gram-positive PAN target was also detected in a polymicrobial BSI episode containing S. epidermidis in addition to K. pneumoniae.
In another polymicrobial blood culture, Fusobacterium nucleatum and Dialister pneumosintes were identified by culture. In the initial Gram-stain, the Gram-negative rods appeared Gram-positive, whereupon the ePlex® cartridge for Gram-positives was used. Both species are not covered by the FilmArray® panel and were not detected accordingly. However, no PAN target for Gram-negatives was detected by the ePlex®.
Detection of genotypic resistance markers
In total, 71
Staphylococcus species were identified in the study. Five
Staphylococcus isolates could not be verified for the presence of
mecA by in-house PCR as the isolates were not stored by the reason of routine laboratory workflow. Nevertheless, the susceptibility results of these five
Staphylococcus isolates correlate with the detection results for
mecA generated by the FilmArray® and the ePlex®. For the residual 66
Staphylococcus isolates in-house
mecA-PCR results were available. All
S. aureus isolates (
n = 18) were methicillin-susceptible and negative for
mecA as reported by the FilmArray®, the ePlex® and the in-house PCR (Table
3). Moreover, no
mecC was detected by the ePlex® system. In the 48 BSI episodes with CoNS, 15 isolates were methicillin-susceptible and 33 methicillin-resistant according to VITEK 2 results and
mecA in-house PCR. This is in accordance with the
mecA-PCR results of the ePlex® system. The FilmArray® produced one false-negative
mecA result for an oxacillin-resistant
S. epidermidis isolate positive for
mecA in the in-house PCR. For the
S. warneri isolate not identified by the system no PCR result was produced. Other, recently evaluated option for accelerated AST results include the culture-based EUCAST rapid antimicrobial susceptibility testing (RAST) allowing interpreting inhibition zones after 4, 6, or 8 h of growth (The European Committee on Antimicrobial Susceptibility Testing. Zone diameter breakpoints for rapid antimicrobial susceptibility testing (RAST) directly from blood culture bottles. Version 1.0, 2018.
http://www.eucast.org.).
Table 3
Resistance genes identified for Gram-positive bacteria by FilmArray® and ePlex®
Genotypic detection of methicillin resistance (mecA/mecC) |
Staphylococcus species (total n = 66) | mecA in house PCR | mecC in house PCR | mecA FilmArray® | mecA ePlex® | mecC ePlex® | Comment |
S. aureus (n = 18) | Negative (n = 18)a | n/a | Negative (n = 18) | Negative (n = 18) | Negative (n = 18) | |
Coagulase-negative Staphylococcus spp. (n = 48) | Negative (n = 15)b | n/a | Negative (n = 16) | Negative (n = 15) | Negative (n = 48) | False-negative mecA result by FilmArray® (n = 1) |
Positive (n = 33)c | | Positive (n = 31) | Positive (n = 33) | | No target detected by FilmArray® (n = 1) |
Genotypic detection of vancomycin resistance (vanA/vanB) |
Enterococcus species (total n = 9) | vanA/vanB in-house PCR | vanA/vanB FilmArray® | vanA/vanB ePlex® | Comment | | |
Enterococcus vancomycin sensitive (n = 8) | negative (n = 8) | negative (n = 8) | negative (n = 8) | E. faecalis (n = 5), E. faecium (n = 3) |
Enterococcus vancomycin resistant (n = 1) | positive van B (n = 1) | positive vanA /vanB (n = 1) | positive vanB (n = 1) | vanB positive E. faecium isolate |
During the study period, nine
Enterococcus species (
E. faecalis [
n = 5] and
E. faecium [
n = 4]) were isolated. Eight isolates were negatively tested for the presence of
vanA and
vanB by routine-diagnostic in-house PCR as well as by the FilmArray® and ePlex® system. One vancomycin resistant isolate (positive for
vanB) was identified by FilmArray® and ePlex® (Tab.
3). None of the 33 isolated Gram-negative microorganisms were carbapenem-resistant, being in line with negative PCR results from the evaluated systems and reflecting the low prevalence of BSI with carbapenem-resistant bacteria in our hospital [
37]. However, due to increasing bacterial resistance, rapid detection of genes encoding carbapenemases is a desirable characteristic of molecular-based blood culture identification panels. In this regard the carbapenemase resistance genes covered by the ePlex® panel for Gram-negatives can be highlighted as the most common genes [
38]. The ePlex® detected the presence of
blaCTX-M for three study isolates (
E. coli (
n = 2) and
K. pneumoniae (
n = 1)), phenotypically confirmed by the presence of an ESBL phenotype in the cultured strains.
A summary of the overall performance of the two test systems is illustrated in Table
4. Concerning the detection of Gram-positive bacteria included in the panels of the FilmArray® and the ePlex®, the best performance with slightly higher sensitivity, specificity, PPV and NPV values was shown by the FilmArray® (Table
4). However, due to the broader identification panel of the ePlex®, putative contaminants or frequent anaerobes can be identified, resulting in on overall higher number of identified Gram-positive organisms (93/98) as compared to the FilmArray® with identification of 86 of 93 specimens. Regarding the identification of Gram-negative organisms and yeast strains, no major differences were found between the two systems.
Table 4
Overview of sensitivity, specificity, positive predictive values (PPV), and negative predictive values (NPV) for in-panel organisms only
Sensitivity (%) |
Gram-positives | 98.9 | 94.7 |
Gram-negatives | 100 | 100 |
Specificity (%) |
Gram-positives | 100 | 90.7 |
Gram-negatives | 100 | 100 |
Positive predictive value (PPV) (%) |
Gram-positives | 100 | 95.7 |
Gram-negatives | 100 | 100 |
Negative predictive value (NPV) (%) |
Gram-positives | 98 | 88.6 |
Gram-negatives | 100 | 100 |
In conclusion and in view of our study aims, both blood culture identification systems showed good results for fast pathogen recognition directly from positive blood cultures as well as for resistance gene detection. The broad coverage three panel approach of the ePlex® system implicate a drawback as it necessitates a Gram-staining prior to decide which cartridge to use for identification. However, the intended use of the other system also requires a Gram-stain as part of their approved instructions for use. Both systems require only short hands-on time, can easily be implemented in a routine microbiological diagnostic workflow and cause similar costs.
The Accelerate Pheno® system, which is already implemented for high risk patients in our blood culture workflow [
12], allows accelerating reports for ID and especially full AST for hard to predict resistance profiles for Gram-negatives. The panels of the FilmArray® and ePlex® systems with their ability for detecting common antimicrobial resistance genes are suitable for fast identification and rough genotypic resistance characterization of Gram-positive organisms, specifically for
Staphylococcus species demarcating
S. aureus and MRSAs and
Enterococcus species defining VREs. Specific performance characteristics of the assays, cost-benefit ratio, laboratory operating hours, manning and state of knowledge of the laboratory personnel, patient population as well as effect on patient care from individual identification panels affect the advantages of rapid PCR-based blood culture diagnostics implicating the need of a considered selection and implementation of a rapid molecular ID system [
39].