Comparing BioFire FilmArray BCID2 and BCID Panels for Direct Detection of Bacterial Pathogens and Antimicrobial Resistance Genes from Positive Blood Cultures

Among molecular assays currently developed for detection and identification of pathogens (and their antimicrobial resistance genes) in positive blood cultures (BCs) (1), the BioFire FilmArray blood culture identification (BCID) panel (bioMérieux, Marcy l’Etoile, France), a multiplex PCR assay with <2 min of hands-on time and an ∼1-h turnaround time, allows for syndromic diagnosis of bloodstream infection (BSI) (2, 3). Previously, the panel could identify 24 etiological agents of BSI (11 Gram-negative bacteria, 8 Gram-positive bacteria, and 5 yeast species), as well as three antimicrobial resistance genes (mecA, vanA/B, and bla_KPC, which encodes Klebsiella pneumoniae carbapenemase). Now, the BioFire FilmArray BCID2 panel encompasses 43 molecular targets associated with BSI, including 15 Gram-negative bacteria, 11 Gram-positive bacteria, 7 yeast species, and 10 antimicrobial resistance genes (https://www.biomerieux-diagnostics.com/biofire-bcid-panel). The last targets include genes encoding carbapenemases (IMP, KPC, OXA-48-like, NDM, and VIM), colistin resistance (mcr-1), extended-spectrum β-lactamase (ESBL) (CTX-M), methicillin resistance (mecA/C and, specifically for methicillin-resistant Staphylococcus aureus [MRSA], mecA/C and MREJ [mec right-extremity junction]), or vancomycin resistance (vanA/B). Unlike BCID, no published studies to date reported on BCID2 performance. This study evaluated and compared the accuracy of BCID2 with that of BCID to identify bacterial species and relative antimicrobial resistance genes directly from positive BCs.

We used archived samples from positive BCs (1.5 ml thereof mixed with 100 μl of dimethyl sulfoxide and stored at −80°C), which had prospectively been processed with the BD Bactec 9240 (BD Diagnostic Systems, Sparks, MD), BacTAlert 3D (bioMérieux), or BacTAlert Virtuo (bioMérieux) system at two hospital microbiology laboratories from January 2018 to August 2020. For BCID2 testing, samples were thawed and allowed to reach room temperature, and a 200-μl aliquot was lysed and processed following the manufacturer’s instructions. We compared BCID2 results with those of conventional identification or detection (i.e., MALDI BioTyper analysis and, relative to antimicrobial resistance genes, PCR sequencing, both performed on microbial isolates subcultured from positive BC samples), which was considered the reference method in this evaluation, or with those of the BCID. Both the reference method and the BCID had been part of a previously implemented laboratory workflow for BSI diagnosis (4). Frozen aliquots of the original BC samples were submitted to subculturing and testing of the organism(s) grown from the aliquots to confirm samples’ previous state.

We studied 90 BCs that grew clinically relevant bacterial species, of which 55 monomicrobial (55 species in total) and 35 polymicrobial (77 species in total, including 1 Candida albicans) with antimicrobial resistance determinants (53 genes in total) were identified (see Table S1 in the supplemental material). We selected BCs to ensure testing of organisms either belonging to bacterial species (e.g., Bacteroides fragilis) or carrying antimicrobial resistance genes (e.g., bla_CTX-M) that were not included in the BCID. Both BCID2 and BCID assays were compared with the reference method specified above. Regarding results with on-panel organisms (Table 1), the percent agreement was 100.0% (95% confidence interval [CI], 97.2% to 100.0%) for the BCID2 and 100.0% (95% CI, 96.8% to 100.0%) for the BCID. All off-panel organisms yielded a negative result with the BCID2 (1 Ochrobactrum anthropi and 1 Brevibacterium casei) or the BCID (8 B. fragilis, 1 O. anthropi, 8 Stenotrophomonas maltophilia, and 1 B. casei). Regarding results with on-panel genes (Table 1), the percent agreement was 100.0% (95% CI, 92.3% to 100.0%) for the BCID2 and 100.0% (95% CI, 85.8% to 100.0%) for the BCID. No genes that encode ESBL- or carbapenemase-mediated β-lactam resistance (i.e., 16 CTX-M, 3 NDM, 4 OXA-48-like, and 4 VIM) were detected by the BCID. With respect to organisms targeted by both BCID2 and BCID assays, no discrepancies between the assays’ results were observed with the polymicrobial BC samples analyzed as a whole (Table S1).

Regarding apparently discrepant results (all for methicillin resistance genes), three mecA genes were not detected by the BCID2, one in monomicrobial and two in polymicrobial BC samples (see IDs 54, 83, and 87 in Table S1). These samples grew Staphylococcus haemolyticus (identified as Staphylococcus) alone in one sample or together with S. aureus (1 mecA negative and 1 mecA positive) in two remaining samples. This was because the BCID2, contrary to the BCID, does not perform mecA detection for an organism not identified at the species level. Furthermore, the BCID2 detected mecA genes from two other polymicrobial BC samples that grew mecA-negative Staphylococcus epidermidis together with MRSA, for which a positive mecA/C and MREJ detection signal was provided (IDs 84 and 85 in Table S1). This was because the BCID2 also signals the mecA/C gene alone in samples that grow MRSA together with S. epidermidis or Staphylococcus lugdunensis, the two staphylococci other than S. aureus (SOSA) identified by the panel.

Here, we report that the BCID2 panel identified more bacterial species and relative antimicrobial resistance genes in positive BCs than the BCID panel while retaining similar performance for targets already included in the BCID. This was the case for BSI pathogens, such as Escherichia coli, K. pneumoniae (with the KPC gene), S. aureus (with the mecA gene), or Enterococcus faecalis (with the vanA gene), which are the most common bacterial pathogens (5). With the BCID2, the opportunity to identify MRSA (5) or S. epidermidis and S. lugdunensis (6) among SOSA may be clinically useful (7), whereas the incapacity to identify SOSA such as S. haemolyticus or other coagulase-negative Staphylococcus species may be detrimental, particularly in the presence of mecA-positive isolates (8, 9). Because the reference method helped us to interpret BCID2 (or BCID) results, we believe that the FilmArray BCID2 panel may be a valuable adjunct to culture-based diagnostics.

We thank Franziska Lohmeyer for English revision of the manuscript.

bioMérieux (Marcy l’Etoile, France) provided FilmArray BCID2 panels for this evaluation, but the company had no role in study design, data collection or interpretation, or manuscript preparation.

We have no other conflicts of interest to report.