BioFire® FilmArray® Pneumonia Panel versus bacterial culture in adult community acquired pneumonia: treatment implications

A prospective study was conducted at Port Said University, Ain Shams University Hospitals and Ain Shams University Specialized Hospital from February 2024 to November 2024. Participants were older than 18 years and were randomly selected from patients hospitalised with sCAP in the ICU, diagnosed on clinical and radiological grounds according to the ATS/IDSA 2019 guideline criteria (10). Patients who presented with radiological features inconsistent with pneumonia and immunosuppressed individuals due to any cause, including malignancies, were excluded from the study.

After providing written informed consent, all patients underwent a full medical history assessment and clinical examination upon admission. The study was approved by the Ethical Committee of the Faculty of Medicine at Port Said University (MED [4/2/2024], S.No. 141, CHS_002).

In all participants, mini-bronchoalveolar lavage (mini-BAL) specimens were collected and submitted to the microbiology laboratory for both BFPP and SOC testing.

Antibiotic selection was guided by the results of the BFPP. For patients with negative BFPP results, empirical antimicrobial therapy was initiated based on the ATS/IDSA 2019 guideline recommendations (10), and modifications were made following the standard sensitivity testing results. Hospital length of stay (LOS), ICU LOS, duration of mechanical ventilation and patient outcomes (discharge or mortality) were recorded.

LRT specimens were evaluated using the BioFire® FilmArray® 2.0 system BioFire Diagnostics, LLC (Biomérieux) system with the BFPP, a multiplex PCR test with an approximate turnaround time of 60–75 min. Qualified laboratory technicians conducted the testing in accordance with the manufacturer’s instructions and institution-specific laboratory protocols to ensure proper handling of respiratory samples and preservation of quality. All initial specimens were processed in a biological safety cabinet (BSC) and then transferred into a FilmArray injection vial (FAIV) containing sample buffer. Subsequently, the mixture was injected into the BFPP test pouch. Technicians then inserted the inoculated pouches into the FilmArray instrument for analysis. Each specimen was processed individually, and the BSC was surface disinfected prior to handling subsequent samples (11, 12).

The BFPP test pouch includes all necessary reagents for specimen lysis, nucleic acid extraction, reverse transcription, amplification and detection of genomic sequences specific to each of the 33 panel targets (Table 1). Additionally, the test pouch contains two internal controls that assess proper function of the pouch and enables calculation of the semi-quantitative results. The user hydrated the BioFire pouch using the manufacturer-supplied hydration solution, followed by loading the sample mixture (respiratory specimen and buffer) and inserting the pouch into the BioFire instrument. The user then scanned the pouch, inserted it into the BioFire FilmArray® 2.0 and initiated the run. Each BFPP pouch includes two process controls (an Ribonucleic acid (RNA) process control and a quantified standard material control), both of which must yield positive results for the run to be considered valid. Runs that failed the internal control criteria were repeated using a new test pouch (13).

Specimens were Gram-stained and cultured according to established clinical laboratory protocols. Acceptable samples were concentrated using Cytospin and assessed via conventional Gram staining before inoculation. Bacterial cultures were prepared by inoculating portions of the specimen onto various selective and differential media including blood agar, chocolate agar and MacConkey agar using the streak plate method with a 0.001-mL calibrated loop. Blood agar and chocolate agar plates were incubated at 35°C, while MacConkey agar was incubated at 35°C in a 5% CO₂ atmosphere. Bacterial growth on each plate was evaluated daily. The VITEK® 2 system (bioMérieux, Marcy l’Étoile, Lyon, France) was used per the manufacturer’s instructions to identify bacterial isolates and determine antibiotic susceptibility breakpoints (4, 12, 14). There were made no specific tests for identifying viruses due to unavailability.

Laboratory personnel performing BFPP and SOC culture were blinded to clinical data to avoid bias.

When BFPP detected organisms were not identified by SOC, clinical judgement relied on clinical data, radiological evidence and laboratory biomarkers.

When BFPP discovered organisms not recognised by SOC culture, the results were evaluated based on clinical correlation. This included a multidisciplinary team (pulmonologists and microbiologists) review of the patient’s clinical presentation (symptoms, signs), radiological findings (new infiltrates on Chest X-ray (CXR) or Computed tomography (CT) and laboratory biomarkers (e.g. procalcitonin >0.5 ng/mL or C-reactive protein [CRP] >50 mg/L). A result was considered a true infection if the clinical context was strong to support it.

Statistical analysis was performed using SPSS software (version 25; SPSS Inc., Chicago, IL, USA). Quantitative data were reported as mean ± standard deviation (SD), while categorical data were expressed as frequency and percentage. The correlation between BFPP and SOC testing was evaluated using positive percent agreement (PPA), negative percent agreement (NPA) and overall percent agreement (OPA), with 95% confidence intervals (CIs) calculated by the modified Wald method in GraphPad Prism® (version 10.2.0; GraphPad, San Diego, CA, USA).

In cases of LRT infections, conventional culture methods alone are inadequate; additional molecular techniques are required to detect viral pathogens and unculturable bacteria. Therefore, the terms PPA, NPA and OPA are preferred over sensitivity, specificity and accuracy, respectively (4, 15). The formulas used were as follows (15):

• PPA = (true positives/[true positives + false negatives]) × 100%

• NPA = (true negatives/[true negatives + false positives]) × 100%

• OPA = ([true positives + true negatives]/[true positives + true negatives + false positives + false negatives]) × 100%

The study included 236 mini-BAL samples collected from 236 subjects who were admitted to the ICUs of Ain Shams University and Ain Shams University Specialized Hospitals with a diagnosis of CAP. The majority of participants were male (63.6%), with a mean age of 62.89 ± 18.61 years. Most were current smokers (59.3%) and had various comorbidities. Table 2 presents demographic characteristics, comorbidity distribution, hospital and ICU LOS, duration of mechanical ventilation and clinical outcomes.

Mini-BAL samples were collected from 236 subjects for microbiological evaluation using both the BFPP and SOC culture methods. BFPP detected bacteria exclusively in 60.59% (143/236) of samples, identifying multiple bacterial species in 30.51% (72/236) and a single bacterial species in 30.08% (71/236). Viruses were exclusively detected in 25% (59/236) of samples, while both bacteria and viruses were co-detected in 72.03% (170/236).

By contrast, culture identified a single bacterium in 30.08% (71/236) and multiple bacteria in 10.17% (24/236) of samples. BFPP detected atypical bacteria in 2.12% (5/236), whereas SOC culture identified fungal species alone in 35.59% (84/236) and co-occurring bacteria and fungi in 75.85% (179/236). Normal flora was reported in 20.76% (49/236) of SOC samples. Negative results were observed in 39.41% (93/236) of BFPP tests and 59.75% (141/236) of SOC cultures. The distribution of single and co-detections of respiratory pathogens identified by both methods is illustrated in Figure 1.

Figure 1.

Overview of BFPP results in comparison to the standard culture method. BFPP, BioFire® FilmArray® pneumonia panel; SOC, standard of care.

Notably, several bacterial pathogens were detected exclusively by BFPP, that is, not identified by culture such as Streptococcus agalactiae, Streptococcus pyogenes, Moraxella catarrhalis, Serratia marcescens and Klebsiella aerogenes. In addition, certain clinically significant pathogens were detected only once by SOC supreme detection culture (true positives), but were detected more frequently by BFPP (suggesting potential false positives), including Streptococcus pneumoniae, Proteus spp., Haemophilus influenzae, Klebsiella oxytoca and Enterobacter cloacae complex (Table 3, Figure 2).

Figure 2.

Percentages of detected bacteria by BFPP and SOC. BFPP, BioFire® FilmArray® pneumonia panel; SOC, standard of care.

The most frequently detected bacteria (organism/236 samples) by BFPP and SOC methods, respectively, were Klebsiella pneumoniae group (61 vs 49), Pseudomonas aeruginosa (49 vs 25), Escherichia coli (43 vs 25), Staphylococcus aureus (32 vs 10) and Acinetobacter baumannii complex (27 vs 9) (Figure 2). The most frequently identified viruses by BFPP included human rhinovirus/enterovirus (9.32%), followed by Influenza A virus (5.93%), human metapneumovirus (2.97%) and Influenza B virus (2.54%) (Figure 3).

Figure 3.

Percentages of viruses detected by BFPP. BFPP, BioFire® FilmArray® pneumonia panel.

Compared with traditional culture methods, the BFPP demonstrated an overall sensitivity of 81.97% (95% CI: 73.98–88.34). The overall specificity was 94.44% (95% CI: 93.59–95.21), and the OPA was 93.98% (95% CI: 93.11–94.77). The (PPA, equivalent to sensitivity) for individual bacterial targets ranged from 73.5% to 100%, while the (NPA, equivalent to specificity) ranged from 86.6% to 99.6%. The OPA for these targets ranged from 83.9% to 99.6%. The performance characteristics of BFPP and standard culture methods in detecting respiratory pathogens are summarised in Table 3.

A total of 214 antimicrobial resistance (AMR) genes were detected using the BFPP. The most frequently identified were genes associated with carbapenemase-producing bacilli, including 60 NDM, 42 OXA-48, 9 VIM, 2 KPC and 2 IMP genes (n = 115). These were followed by extended-spectrum β-lactamase (ESBL) genes, predominantly Cefotaximase-M-type β-lactamase (CTX-M) (n = 75), and methicillin-resistant Staphylococcus aureus (MRSA)-associated genes, specifically mecA/Cassette (or right-extremity) (C-MREJ) (n = 24) (Table 4).

Comparing between the resistance gene profiles detected by BFPP and the phenotypic resistance profiles determined by VITEK® 2 for common pathogens revealed: in 87.75% of P. aeruginosa isolates, the carbapenemase genes detected by BFPP (NDM, OXA-48, etc.) were concordant with phenotypic carbapenem resistance. Similarly, all mecA/MREJ-positive S. aureus isolates were confirmed as MRSA by culture-based methods.

Regarding the timing outcome of antibiotic changes, patients with BFPP-guided therapy had modifications (escalation or de-escalation) on average 45.2 hr earlier than those whose antibiotics were adjusted based only on standard culture and susceptibility testing.

A total of 77.96% (184/236) of participants used antibiotics before symptom deterioration and hospital presentation; 48.37% of them used azithromycin (89/184) and 32.61% used amoxicillin/clavulanic acid (60/184), while 19.02% (35/184) were using levofloxacin. Duration of use ranged from 3 days to 10 days, at an average of 6.5 days.

Comorbid diseases associated with participating pneumonia subjects as diabetes mellitus (DM) (53.81%), chronic respiratory diseases (15.68%) and chronic kidney disease (CKD) (15.25%) are known to be risk factors for pneumonia with Gram-negative bacteria, especially P. aeruginosa, K. pneumoniae, E. coli, A. baumannii and Gram-positive cocci, especially S. aureus.

DM represents 53.81% of the total study subjects, while patients with DM constitute 22% of patients having K. pneumoniae in their BFPP results, 18.9% of Pseudomonas aeronginosa, 11.8% of E. coli, 13.4% of A. baumannii and 11% of S. aureus. Patients with DM also constitute 2.4% of VIM, 1.6% of IMP, 7.1% of mecA_C_and_MREJ, 22.8% of NDM, 14.2% of OXA_48_like and 27.6% of CTX_M resistance genes detected in BFPP test results.

Patients with chronic respiratory disease (15.68% of all participants) form 21.6% of patients having K. pneumoniae in their BFPP results, 35.1% of P. aeronginosa, 10.8% of E. coli, 5.4% of A. baumannii and 21.6% of S. aureus. Chronic respiratory disease constitutes also 8.1% of VIM, 2.7% of KPC, 10.8% of mecA_C_and_MREJ, 24.3% of NDM, 13.5% of OXA_48_like and 27.0% of CTX_M resistance genes.

Patients with CKD (15.25% of all pneumonia subjects) comprise 30.6% of patients having K. pneumoniae in their BFPP results, 13.9% of P. aeronginosa, 30.6% of E. coli, 16.7% of A. baumannii and 5.6% of S. aureus. CKD constitutes also 2.8% of VIM, 2.8% of mecA_C_and_MREJ, 33.3% of NDM, 22.2% of OXA_48_like and 27.8% of CTX_M resistance genes.

The association of different comorbid diseases with the BFPP-resulting bacteria and resistant genes is presented in Tables 5 and 6.

The average time from ICU admission to Bronchoalveolar lavage (BAL) collection, which was 6.2 ± 3.4 hr, and the average time to availability of BFPP results was 2.1 ± 0.5 hr after BAL collection. These timelines allowed early identification of pathogens and resistance genes, contributing to antibiotic modification decisions.

S. aureus was detected in 32 specimens by BFPP, of which 10 (31.25%) were positive for S. aureus in the corresponding SOC cultures. The mecA_C_and_MREJ resistance gene was detected in 24 BFPP samples with S. aureus results indicating MRSA. These samples proved MRSA in SOC cultures, which agrees with the BFPP results.

P. aeruginosa was detected in 49 (20.76%) specimens by BFPP, of which 25 (10.59%) were positive for P. aeruginosa in the corresponding SOC cultures. The carbapenemase resistance gene was detected in 43 (87.75%) BFPP samples with P. aeruginosa results. These samples proved to be resistant to carbapenems in SOC cultures, which agrees with the BFPP results.

Finally, viruses were detected exclusively in 59/236 patients (25%) and in combination with bacteria in 170/236 (72.03%) of BFPP results, among which BFPP detected influenza A virus in 14 samples (5.93%) and influenza B virus in 6 samples (2.54%), while no specific physician order for an influenza virus test had been made. This allowed rapid initiation of antiviral therapy such as oseltamivir, as well as implementing specific infection control measures such as droplet isolation.

In the patients whose antibiotics were guided by BFPP results, the mean time to escalation or de-escalation was 9.4 ± 2.7 hr from ICU admission, compared with 54.6 ± 12.1 hr in the group that depended only on culture.

Normal flora (e.g., Candida spp., Coagulase-negative Staphylococci, Viridans group Streptococci), which are not targeted by the BFPP panel, was reported in 20.76% (49/236) of SOC samples.