BioFire FilmArray Pneumonia Panel enhances detection of pathogens and antimicrobial resistance in lower respiratory tract specimens

This study investigated the diagnostic utility of the BioFire FilmArray Pneumonia Panel (PN panel), an automated and multiplexed nucleic acid detection system that rapidly detects 26 pathogens (18 bacteria and eight viruses) and seven antimicrobial resistance markers in a single assay. We analyzed the targets in lower respiratory tract specimens using the PN panel and compared the detection results with those of bacterial culture methods and antimicrobial susceptibility testing. Of the 57 samples analyzed, the PN panel detected 97 targets (84 bacteria, four viruses, and nine antimicrobial resistance markers). Detection of bacteria and antimicrobial resistance was three times greater than that of the bacterial culture (25 bacteria and two resistant isolates) against the targets available in the panel. The overall positive and negative percent agreements between the PN panel and culture methods for bacterial detection were 100.0% and 92.9%, respectively. Multiple pathogens were detected by the PN panel in 24 samples (42.1%), ranging from two pathogens in 11 samples (19.3%) to six pathogens in one sample (1.8%). The PN panel semiquantitatively detected higher copies (≥ 106 copies/mL) of bacterial targets if the bacteria were positive by the culture method. In contrast, the semiquantitative values obtained by the panel varied (104 to 107 ≤ copies/mL) among bacteria that were negative by the culture method. The PN panel enhanced the detection of pathogens and antimicrobial resistance markers in lower respiratory tract specimens.


Background
Lower respiratory tract infections (LRIs) are a significant public health concern and a leading cause of death from infection worldwide [1]. Therapeutic strategies for LRIs, mainly antimicrobial selection, consist of empirical and definitive therapies [2][3][4]. Initial empirical therapy covers the major causative pathogens that are predicted, while antimicrobials for definitive therapy are selected against identified pathogens in consideration of antimicrobial susceptibility. Several guidelines and studies have reported the isolation frequency of pathogens and the risk factors for the acquisition of antibiotic-resistant bacteria [3][4][5]. However, it may be difficult to accurately predict etiological agents at the time of empiric therapy because of differences in regional epidemiology and medical settings. Additionally, culture-based bacterial

Annals of Clinical Microbiology and Antimicrobials
identification and antimicrobial susceptibility testing (AST) are useful for definitive therapy but are time intensive.
The FilmArray system is an automated and multiplexed nucleic acid detection system, and the BioFire FilmArray Pneumonia Panel (PN panel) simultaneously detects 33 targets (26 pathogens and seven antimicrobial resistance markers) in approximately 75 min in a simple assay procedure.
In this study, we analyzed these targets in lower respiratory tract specimens using the PN panel and compared the results with those of bacterial culture and AST.

Study design and samples
Sputum, tracheal aspirates, bronchoalveolar lavage fluid (BAL), and mini-BAL specimens, which were submitted to the clinical laboratory for microbiological evaluation, were collected at the Nagasaki University Hospital between October 2020 and January 2021. The hospital had 874 beds, and the average number of inpatients and outpatients were 669 and 1653 patients per day, respectively, during this period. The quality of sputum and tracheal aspirates was assessed according to the Miller and Jones classification [6]. Samples were randomly selected from specimens P1 (purulent, grade 1-pus amounting to less than one-third of the specimen), P2 (purulent, grade 2-pus amounting to one-third to two-thirds of the specimen), and P3 (purulent, grade 3-pus amounting to more than two-thirds of the specimen). De-duplication was performed if samples were repeatedly collected from a single patient, and samples from unique patients were included in this study. The specimens were stored at − 80 °C until assay using the PN panel.

Culture methods
Sputum and tracheal aspirate samples, which were prepared using Sputazyme (Kyokuto Pharmaceutical Industrial Co., Ltd.), and BAL and mini-BAL samples were streaked using 10 µL loop and cultured on Nissui Separated Plate Sheep Blood Agar/Chocolate Agar EXII (NIS-SUI PHARMACEUTICAL CO., LTD.), CHROMagar Candida/BTB agar (Prepared media) (KANTO CHEMI-CAL CO., INC.). Additionally, if the physician specifically requested testing for methicillin-resistant Staphylococcus aureus (MRSA), the samples were cultured on Pourmedia MRSA SELECTIVE AGAR II (Eiken Chemical Co., Ltd.). All bacteria that were considered pathogens, excluding normal flora, were picked up and identified even if they were cultured in rare or few quantities. After pure culture, isolates were identified using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS), MALDI Biotyper system (Bruker Daltonik, GmbH). Streptococcus pneumoniae was identified using the optochin disc diffusion test.
Antimicrobial susceptibility was determined using a BD Phoenix M50 (Becton Dickinson). MRSA and extendedspectrum β-lactamase (ESBL) producers were automatically determined using BD Phoenix M50. Antimicrobial susceptibility of Haemophilus influenzae was determined using minimal inhibitory concentration (MIC) plates customized by Eiken Chemical Co., Ltd.

Analysis using the BioFire FilmArray Pneumonia Panel
The BioFire FilmArray Pneumonia Panel (bioMérieux Japan Ltd.), which has been approved by Food  The BioFire FilmArray Pneumonia Panel pouch, which is a closed and disposable system, contains all the reagents required for nucleic acid extraction and purification, reverse transcription, and PCR. The stored specimens were analyzed using the PN panel. Briefly, the pouch was hydrated by the hydration injection vial via the pouch hydration port. Approximately 200, 50, or 10 µL of the specimen was mixed with the sample buffer using a vortex mixer. After flashing in a centrifuge to remove foam, the mixture was transferred to the sample injection vial and injected into the pouch via the pouch sample port. The sample preparation process, including the sample volume used for the assay, was off-label. The prepared pouch was inserted into the instrument, and subsequent assay steps (nucleic acid extraction and purification, nested multiplex PCR, and detection of each target in the array) were automatically performed on the system. The system reports the detection results of pathogen and antimicrobial resistance markers and the semiquantification results of the detected bacterial targets excluding three atypical bacteria.

Comparison between the PN panel and bacterial culture-based methods
The detection results of the PN panel were compared with those of bacterial culture methods and AST, and positive and negative percent agreements (PPA and NPA) were calculated as follows: PPA, the number of concordant positive results divided by the number of all positive results by culture-based methods; NPA, the number of concordant negative results divided by the number of all negative results by culture-based methods. The 95% confidence intervals were calculated using R (version 3.5.2) [7].

Results
A total of 57 specimens (35 sputum, nine tracheal aspirate, and 13 BAL or mini-BAL specimens) were included in this study. Of the 44 sputum and tracheal aspirate specimens, one, 13, and 30 were identified as P1, P2, and P3, respectively, according to the Miller and Jones classification. The average patient age was 66.7 years, and 36 patients (63.2%) were male.
The frequencies of the detected bacteria and antimicrobial resistance are presented in Table 1. The PN panel detected 84 bacteria and nine antimicrobial resistance markers. The panel detected P. aeruginosa and S. aureus (17 targets each) most frequently, followed by the 10 K. pneumoniae group, six A. calcoaceticus-baumannii complex, six E. cloacae complex, and six H. influenzae. With respect to antimicrobial resistance markers, the PN panel detected five mecA/mecC and MREJ genes, as well as four CTX-M genes.
Meanwhile, 25 bacteria were isolated using culture methods, with P. aeruginosa (eight isolates) isolated most frequently, followed by five H. influenzae, three strains in K. pneumoniae group, and three S. aureus. Two ESBL producers were isolated based on their AST results.
The PN panel detected four rhinovirus/enterovirus targets, that were not tested using the reference method in this study. Table 2 shows the concordance of the detection results for bacteria between the PN panel and culture methods. The PN panel correctly detected all bacterial targets  Table 3 presents the number of samples in which multiple pathogens were detected by the PN panel and culture methods. Of the 57 samples included in the study, multiple pathogens were detected in 24 samples (42.1%) using the PN panel, ranging from two pathogens in 11 samples (19.3%) to six pathogens in one sample (1.8%). Conversely, the culture method detected two pathogens in four samples (7.0%). Table 4 summarizes the semiquantitative values of the bacteria measured using the PN panel. The PN panel detected bacterial targets of 10 6 copies/mL or greater if the bacteria were positive by the culture method.

Discussion
Our study demonstrated that the PN panel rapidly and effectively detected a variety of pathogens in the lower respiratory tract specimens. The PN panel detected 97 targets (84 bacteria, four viruses, and nine antimicrobial resistance markers), and the detection of bacteria and antimicrobial resistance was three times greater than that of the culture-based methods (25 bacteria and two resistant isolates) ( Table 1). Several previous studies have shown similar results in that the panel detected more bacterial targets than the culture method [8,9]. A recent study reported that the PN panel identified nearly twice as many total bacterial targets as standard-of-care culture in BAL specimens [8]. Another study reported that the PN panel detected one or more bacterial targets in an additional 20% of patients compared to culture methods [10]. In addition, our results also showed that the PN panel correctly detected all bacteria that were detected by the culture methods (PPA, 100%) if they were included in the targets of the panel (  [8,9,11]. In this study, only four bacterial species (E. faecalis, E. faecium, S. maltophilia, C. propinquum) and one C. albicans, which were unavailable in the PN panel, were isolated using the culture method. A recent study from Taiwan reported that the PN panel may cover approximately 70-90% of the most prevalent pathogens causing moderate to severe community-acquired pneumonia in adults, as well as 70-80% of those causing healthcare-associated pneumonia [9]. The current diagnostic strategies require multiple tests to detect possible pathogens. Bacterial culture is the gold standard for bacterial detection, but it is time consuming. The PN panel, which can simultaneously detect major pathogens in a short time, can be useful for the diagnosis of LRIs.
Although atypical bacteria were not detected by the PN panel in this study, recent studies have reported that the PN panel detected bacteria such as L. pneumophila and M. pneumoniae in a few samples [9,12].
We found four samples in which rhinovirus/enterovirus was simultaneously detected with the bacteria. Previous studies have reported that co-infection with viruses and bacteria contributes to increased disease severity and mortality [13,14]. Although we could not assess the relationship between co-infection and disease severity, comprehensive pathogen detection by the PN panel might contribute to elucidating the epidemiology and pathophysiology of co-infection.
The PN panel can semiquantify the bacterial targets. In this study, of the 59 analytes with negative results by the culture methods, bacterial targets of 10 4-5 copies/mL were detected in 36 analytes (61.0%) ( Table 4). Quantitative values below the defined thresholds, such as 10 3 CFU/mL for the protected specimen brush and 10 4 CFU/mL for BAL, might be used as indicators for discontinuing antibiotics against ventilator-associated pneumonia (VAP) [3]. Therefore, the semiquantitative values determined by the PN panel may be useful for the management of VAP.
Meanwhile, we should consider the possibility of bacterial colonization, regardless of the detection method used. However, because P. aeruginosa and MRSA colonization are reported as the risk factors for subsequent infection [15,16], detecting colonized bacteria would be helpful for prompt initiation of treatment when infection occurs.
Detection using culture methods depends on pathogen viability. Therefore, it is possible that the PN panel detected nonviable bacteria in this study, as one of the causes of discrepant results in bacterial detection between the PN panel and culture methods. There are other possibilities that the PN panel detected bacteria that were low-abundance or not cultured because of fastidious growth characteristics [8]. Therefore, the results of molecular assays should be carefully interpreted when used for patient management, in consideration of the differences in characteristics between molecular assays and culture methods.
Additionally, even if antimicrobial resistance markers are not detected, antimicrobial susceptibility can be decreased through other resistance mechanisms. Therefore, culture methods are necessary to test antimicrobial susceptibility. Further studies are required to establish practical strategies for complementary use of PN panel and traditional culture-based methods.
This study had some limitations. First, we did not collect all the samples submitted to our laboratory during the study period. Therefore, the data did not represent