Evaluation of a Commercial Multiplex Molecular Panel for Diagnosis of Infectious Meningitis and Encephalitis

ABSTRACT

Rapid and accurate laboratory tests are important for the timely diagnosis and treatment of central nervous system infections. The FilmArray meningitis/encephalitis (ME) panel (BioFire Diagnostics, Salt Lake City, UT) is an FDA-cleared, multiplex molecular panel that allows the detection of 14 pathogens (bacterial [n = 6], viral [n = 7], and fungal [n = 1] pathogens) from cerebrospinal fluid (CSF). In this study, we evaluated the performance characteristics of the FilmArray ME panel using clinical, residual CSF samples (n = 291) that tested positive by a routine method(s) (e.g., bacterial culture, individual real-time PCR assay) for a pathogen represented on the ME panel. Of note, a subset (n = 76) of the CSF specimens was collected during the prevaccine era and had been characterized as positive for a bacterial pathogen. The FilmArray ME panel demonstrated an overall percent positive agreement (PPA) of 97.5% (78/80) for bacterial pathogens, 90.1% (145/161) for viruses, and 52% (26/50) for Cryptococcus neoformans/C. gattii. Despite the low overall agreement (52%) between the ME panel and antigen testing for detection of C. neoformans/C. gattii, the percent positive agreement of the FilmArray assay for C. neoformans/C. gattii was 92.3% (12/13) when the results were compared directly to the results of routine fungal smear or culture. The FilmArray ME panel offers a rapid (∼60-min), syndrome-based approach for the detection of select meningitis and encephalitis pathogens.

INTRODUCTION

Infectious meningitis and encephalitis are serious, potentially life-threatening clinical conditions involving inflammation of the central nervous system (CNS). These infections are associated with significant morbidity and mortality, and therefore, rapid diagnosis is important for the most effective and appropriate management of patients (1, 2). However, it is estimated that in up to 50% of encephalitis cases and up to 60% of meningitis cases, the etiology is unidentified (1, 2). Enteroviruses (EVs) and herpes simplex virus (HSV) are the most common causes of infectious meningitis and encephalitis, respectively, with children and the elderly representing the populations at the highest risk (2, 3). The rates of bacterial meningitis, most commonly caused by Streptococcus pneumoniae, Neisseria meningitidis, and Haemophilus influenzae type B, have declined significantly since the widespread implementation of childhood immunization programs, with the majority of cases now occurring in older adults (4, 5).

Delays in the diagnosis and treatment of bacterial meningitis and HSV meningoencephalitis are associated with morbidity and mortality (6, 7). Conversely, unnecessary antimicrobial treatment and hospitalization can be associated with the selection of antimicrobial resistance, dysbiosis, and increased health care costs. The clinical diagnosis of meningitis or encephalitis in patients can be challenging, due in part to nonspecific signs and symptoms. Routine chemistry and cellular analyses of cerebrospinal fluid (CSF) may assist in rapidly differentiating bacterial versus viral infections; however, these methods are not able to identify a specific etiology. To establish a definitive diagnosis, microbiology tests are required. Despite the high specificity of culture-based tests, recovery of bacterial, fungal, and viral pathogens from CSF may take several days, and the results may be negative if antimicrobials are administered empirically or if the infection is caused by fastidious or nonculturable microorganisms. In recent years, nucleic acid amplification tests (NAATs) (e.g., real-time PCR) that allow the rapid detection and identification of infectious agents in CSF have been developed; however, to date, these tests typically have targeted a single (or sometimes two) pathogen(s), and therefore, providers must order a number of assays together on the basis of a patient's clinical presentation and other laboratory and radiology findings.

Multiplex molecular tests represent an advancement in the ability of clinical laboratories to rapidly and reliably detect and identify causes of infectious diseases. In 2015, the FilmArray meningitis/encephalitis (ME) panel (BioFire Diagnostics, Salt Lake City, UT) received FDA clearance as an aid in the diagnosis of CNS infections. The FilmArray ME panel includes 14 targets (Escherichia coli K1, H. influenzae, Listeria monocytogenes, N. meningitidis, Streptococcus agalactiae, S. pneumoniae, cytomegalovirus [CMV], enterovirus [EV], herpes simplex virus 1 [HSV-1], herpes simplex virus 2 [HSV-2], human herpesvirus 6 [HHV-6], human parechovirus [HPeV], varicella-zoster virus [VZV], and Cryptococcus neoformans/C. gattii) and generates results in ∼1 h. A multicenter, clinical trial performed to support the FDA submission of the multiplex assay reported 84.4% positive agreement and >99.9% negative agreement between the ME panel and routine methods (8). However, this previously published study included prospectively collected CSF samples, and therefore, the number of cases due to bacterial pathogens (n = 22) and C. neoformans/C. gattii (n = 5) was too low to provide a robust assessment of the clinical performance of the ME panel for these targets. In this study, we sought to further evaluate the ME panel using convenience CSF specimens known to be positive for a bacterial, viral, or fungal target(s) represented on the FilmArray ME multiplex assay. We included a number of CSF specimens (n = 76) collected from patients with bacterial meningitis in the prevaccine era.

MATERIALS AND METHODS

Clinical samples.

Cerebrospinal fluid specimens (n = 291) originally submitted for routine microbiology testing at the Mayo Clinic that were positive for at least one target represented on the FilmArray ME panel were selected for this study. Samples met the inclusion criteria if they had not been centrifuged and had a residual volume of ≥200 μl following completion of routine testing. Repeat specimens from the same patient were not excluded, although this occurred in only seven patients who were being monitored for their response to antifungal therapy following an initial diagnosis of cryptococcal CNS disease (see Table S1 in the supplemental material). Following routine testing, samples were stored either refrigerated (4°C) or frozen (≤−70°C) until testing by the FilmArray ME panel could be completed. For the majority of CSF specimens that were positive for viral or C. neoformans/C. gattii targets, the storage period was ≤7 days. However, the majority (76/80) of CSF specimens positive for a bacterial target had been stored frozen for >3 decades (with the collection dates ranging from March 1975 to June 1997, with 58/76 [76.3%] having been collected between 1980 and 1986). This study was reviewed and approved by the Institutional Review Board at Mayo Clinic.

Routine testing.

CSF specimens that were known to be positive for bacterial pathogens (H. influenzae [n = 40], S. pneumoniae [n = 25], N. meningitidis [n = 10], or E. coli [n = 1]) had been tested by counterimmunoelectrophoresis (CIE), as previously described (9, 10). In addition, the results of routine Gram stain, bacterial culture, and bacterial serotyping were documented, if available (Table S2). Some CSF samples positive for S. pneumoniae had been tested by a bacterial antigen assay (BinaxNOW S. pneumoniae antigen card; Alere, Waltham, MA). Four CSF samples were positive for a bacterium (S. agalactiae [n = 2], S. pneumoniae [n = 2]) during the study period and were also included. Routine viral testing for EV, CMV, HHV-6, and VZV was performed using a combination of targeted, laboratory-developed tests (LDTs) that were validated according to Clinical Laboratory Improvement Amendments (CLIA) requirements (11–15). Routine testing of CSF for HSV-1/2 was performed using an LDT on samples received up until September 2016 and using an FDA-cleared, commercial method (Simplexa HSV 1&2 Direct; DiaSorin, Cypress, CA) thereafter (11). The LDT was verified to have a limit of detection (LoD) of 10⁴ copies/ml for both HSV-1 and HSV-2. The Simplexa HSV 1&2 Direct assay has a reported LoD of 5 50% tissue culture infective doses (TCID₅₀)/ml for HSV-1 (McIntyre strain) and 1.25 TCID₅₀/ml for HSV-2 (G strain). Cerebrospinal fluid samples submitted for cryptococcal antigen (CrAg) testing were analyzed by a lateral flow assay (LFA; IMMY, Norman, OK) according to the manufacturer's instructions and subsequently stored at either 4°C or ≤−70°C prior to FilmArray ME panel testing. In a subset of CSF samples (n = 20), routine fungal culture (brain heart infusion agar; BD BBL Fisher Scientific, Waltham, MA) and smear (KOH-calcofluor white preparation) results were available and compared to those of the FilmArray ME panel.

FilmArray ME panel testing.

Testing by the FilmArray ME panel was performed according to the manufacturer's instructions. In brief, 200 μl of uncentrifuged CSF and hydration solution was drawn into the FilmArray ME reagent pouch by vacuum. The reagent pouch was then placed in a FilmArray (version 2.0) instrument (BioFire) and tested. According to the manufacturer, the FilmArray ME panel has been cleared for CSF specimens stored under refrigeration (4°C) for up to 7 days.

Data analysis.

The results of the FilmArray ME panel were considered true positive (TP) or true negative (TN) if they agreed qualitatively (positive or negative) with the result(s) of routine testing. Percent positive agreement (PPA) and percent sensitivity were calculated using standard methods. The 95% confidence interval (CI) was calculated using the modified Wald method. Because specimens were not tested for all targets represented on the FilmArray ME panel during routine testing, percent negative agreement and specificity were not calculated.

Discordant analysis.

Results were considered discordant when the result of the FilmArray ME assay did not agree with that of routine testing for the specific target(s) represented on the multiplex panel. For example, the result for a sample with a positive result for HSV-1 by the FilmArray ME panel and a negative result for HSV-1 by routine testing was considered discordant. However, the result for a sample with a positive result for HSV-1 by the FilmArray ME panel and a positive result for HSV-1 and Epstein-Barr virus (EBV) by a routine assay would not have been considered discordant, as EBV is not represented on the multiplex panel. In situations where discordant results were identified or when the FilmArray ME panel was positive for a target not initially tested for by routine methods, discordant analysis was performed using an alternative method, when available. In such cases, molecular testing for S. pneumoniae, N. meningitidis, and H. influenzae was performed using a real-time PCR LDT at the Minnesota Department of Health (St. Paul, MN). Molecular testing for C. neoformans/C. gattii was performed by NAAT at the University of Washington, Department of Laboratory Medicine (Seattle, WA). In some cases, discordant testing could not be performed due to an inadequate volume of sample remaining or the lack of a confirmatory method. In these situations, the results were considered discordant and the lack of discrepant analysis was noted. Whenever possible, medical record review was performed when the results of the FilmArray ME panel and routine testing were discordant. Information collected during the medical record review included CSF chemistries, protein and glucose concentrations, white blood cell count and differential, as well as CSF testing and the results thereof, antimicrobial therapy, and the final clinical diagnosis.

RESULTS

A total of 293 CSF specimens were tested by the FilmArray ME panel, with 5 (1.7%) yielding an invalid result following initial testing. Three of these 5 had sufficient volume for repeat testing, which generated a valid result, yielding 291 samples in the final data set. Overall, the FilmArray ME panel demonstrated a PPA with routine testing of 85.6% (249/291) (Table 1). When the results for C. neoformans/C. gattii were excluded, the adjusted PPA was 92.5% (223/241). Although the majority of samples were positive for only one target by the FilmArray ME panel, coinfections (2 pathogens in the same sample) were identified in 8 (2.7%) samples (Table 2). No CSF specimens were positive for >2 analytes by the FilmArray ME panel.

TABLE 1.

Comparison of FilmArray ME panel results to results of routine testing using a convenience set of CSF samples^j

Target	No. of CSF specimens with:					PPA (95% CI)	Adjusted PPA (95% CI)
	Expected positive result	Positive result by FilmArray ME panel	TP result	FP result	FN result
EV	43	41	41	0	2a	95.4 (83.7, 99.6)	95.4 (83.7, 99.6)
HSV-1	26	20	19	1b	7c	73.1 (53.7, 86.5)	82.6 (62.2, 93.6)
HSV-2	55	49	48	1b	7d	87.3 (75.7, 94.0)	96.0 (85.8, 99.7)
CMV	3	3	3	0	0	100 (38.3, 100)	100 (38.3, 100)
VZV	29	30	29	1e	0	100 (86.1, 100)	100 (86.5, 100)
HHV-6	5	7	5	2f	0	100 (51.1, 100)	100 (55.7, 100)
HPeV	0	0	0	0	0	UTC	UTC
Streptococcus agalactiae	2	3	2	1g	0	100 (29.0, 100)	100 (29.0, 100)
Streptococcus pneumoniae	27	30	27	3h	0	100 (85.2, 100)	100 (85.7, 100)
Neisseria meningitidis	10	10	9	1e	1b	90.0 (57.4, 99.9)	100 (67.9, 100)
Haemophilus influenzae	40	39	39	0	1b	97.5 (86.0, 99.9)	100 (89.3, 100)
Listeria monocytogenes	0	1	0	1g	0	UTC	UTC
Escherichia coli K1	1	1	1	0	0	100 (16.8, 100)	100 (16.8, 100)
Cryptococcus neoformans/C. gattii	50	26	26	0	24i	52.0 (38.5, 65.2)	57.8 (43.3, 71.0)
Total	291	260	249	11	42	85.6 (81.0, 89.2)	90.4 (86.3, 93.3)

^aBoth samples were retested and confirmed to be positive for enterovirus by molecular testing.

^bThis sample tested negative by an alternate molecular method.

^cAmong these 7 samples, 3 tested negative and 1 tested positive for HSV-1 by an alternate molecular method; for 3 of these 7 samples, discordant analysis was unable to be performed.

^dAmong these 7 samples, 5 tested negative and 1 tested positive for HSV-2 by an alternate molecular method; discordant analysis was unable to be performed on 1 sample.

^eThis sample was confirmed to be positive by an alternate method.

^fAmong these 2 samples, 1 tested positive and 1 tested negative for HHV-6 by an alternate molecular method.

^gDiscordant analysis was unable to be performed for this sample.

^hAmong these 3 samples, 1 tested positive and 2 tested negative for S. pneumoniae by an alternate method.

ⁱFive of these samples were tested by an alternate molecular method, and all 5 were negative.

^jData are for 291 samples. CSF, cerebrospinal fluid; ME, meningitis/encephalitis; TP, true positive; FP, false positive; FN, false negative; PPA, percent positive agreement; EV, enterovirus; HSV, herpes simplex virus; CMV, cytomegalovirus; VZV, varicella-zoster virus; HHV-6, human herpesvirus type 6; HPeV, human parechovirus; UTC, unable to calculate.

TABLE 2.

Apparent coinfections detected by FilmArray ME panel and results following confirmatory testing^a

Case	Expected result from routine testing	Result(s) of FilmArray ME panel	Result(s) following confirmatory testing
1	Haemophilus influenzae	H. influenzae, Listeria monocytogenes	H. influenzae
2	Cryptococcus neoformans	C. neoformans/C. gattii, S. pneumoniae	C. neoformans/C. gattii
3	Neisseria meningitidis	N. meningitidis, S. pneumoniae	N. meningitidis
4	H. influenzae	H. influenzae, HHV-6	H. influenzae
5	C. neoformans	C. neoformans/C. gattii, HSV-1	C. neoformans/C. gattii
6	HSV-1	HSV-1, HHV-6	HSV-1, HHV-6
7	CMV	CMV, VZV	CMV, VZV
8	H. influenzae	S. pneumoniae, N. meningitidis	S. pneumoniae, N. meningitidis

^aHHV-6, human herpesvirus type 6; HSV-1, herpes simplex virus 1; CMV, cytomegalovirus; VZV, varicella-zoster virus.

Viral targets.

Among the 291 CSF samples included in the final sample set, 161 (55.3%) were selected due to a virus being identified during routine testing. Of these 161 specimens, the FilmArray ME panel correctly identified the virus in 145 (90.1%). The FilmArray assay was uniquely positive (i.e., routine testing was negative or not initially performed) for a viral target in 5 samples (HSV-1 [n = 1], HSV-2 [n = 1], VZV [n = 1], and HHV-6 [n = 2]) and was negative for a viral target in 16 CSF specimens that were positive for a virus (EV [n = 2], HSV-1 [n = 7], and HSV-2 [n = 7]) by routine testing. After discordant analysis, the adjusted, overall PPA for viral targets was 94.8% (147/155) (Table 1). Coinfections involving a virus(es) were detected in 4 CSF samples by the FilmArray ME panel (Table 2). One CSF sample was positive for both CMV and VZV, a second one was positive for HSV-1 and HHV-6, a third one was positive for HHV-6 and H. influenzae, and the fourth one was positive for HSV-1 and C. neoformans/C. gattii.

Bacterial targets.

A total of 80 (27.5%) CSF specimens in the final sample set were positive for a bacterial pathogen by routine testing, and these pathogens were accurately detected by the FilmArray ME panel in 78 (97.5%) (Table 1). Target-specific PPAs ranged from 90% (9/10) for N. meningitidis to 100% for S. agalactiae (2/2) and E. coli K1 (1/1). S. pneumoniae was uniquely detected by the FilmArray ME panel in 3 CSF specimens, while S. agalactiae, N. meningitidis, and L. monocytogenes were each detected in one specimen each that was negative for these targets by routine methods (Table 1). Interestingly, the unique detections of each of S. pneumoniae, N. meningitidis, and L. monocytogenes were part of coinfections, as determined by the FilmArray ME panel (Table 2). After discordant analysis, the adjusted, overall PPA for bacterial targets was 100% (80/80), and 4 targets were uniquely detected by the FilmArray ME panel (Table 1).

Cryptococcus neoformans/C. gattii.

The FilmArray ME panel was positive for C. neoformans/C. gattii in 26 (52%) of 50 CSF specimens that were positive by the CrAg LFA (Table 1). Five (20.8%) of the 24 discordant specimens were tested for C. neoformans/C. gattii nucleic acid by an alternate molecular method (University of Washington), and all 5 (100%) were negative. Of the 50 total CrAg LFA-positive CSF specimens selected for inclusion in this study, the results of routine fungal smear and fungal culture and the LFA titers were available for 20 samples collected from 15 unique patients (Table 3). Among these 20 CSF samples, 9 were collected prior to initiation of antifungal therapy, while 11 were collected from patients who were on antifungal therapy for a prior diagnosis of cryptococcal meningitis. Six of these 20 CSF specimens were negative by both the FilmArray ME assay and fungal smear/culture. Among the 14 specimens positive for Cryptococcus species by fungal smear or culture, the FilmArray ME cryptococcal assay was positive for 13 (92.9%) (Table 3). The multiplex panel was negative for 1 CSF sample that was negative for C. neoformans by smear but positive for C. neoformans by culture.

TABLE 3.

Comparison of results of FilmArray ME assay for Cryptococcus neoformans/C. gattii with results of routine fungal culture and smear for 20 CSF specimens

Result by FilmArray ME assay for C. neoformans/C. gattii	No. of CSF specimens with indicated result by routine fungal culture and/or smear		% sensitivity (95% CI)
	Positive	Negative
Positive	13	0	92.8 (66.5, 99.9)
Negative	1a	6b

^aThis sample was negative by routine fungal smear, but C. neoformans was identified from fungal culture.

^bFive of these 6 CSF specimens were collected from patients who had a prior diagnosis of cryptococcal infection and were on antifungal therapy.

DISCUSSION

In this study, we investigated the performance of the FilmArray ME panel for the detection of bacterial, viral, or fungal pathogens in CSF. Overall, the multiplex panel demonstrated a good correlation with routine testing, detecting 97.5% (78/80) of the bacterial pathogens and 90.1% (145/161) of the viruses identified by conventional methods. These results are similar to those reported in previous publications on the FilmArray ME panel (8, 16–18). However, our study included the largest number of total positive samples (n = 291) tested by the multiplex assay, especially for bacterial targets (n = 80), and provides a novel and unique assessment of the FilmArray ME panel that would not likely be feasible using a prospective study design.

Although the overall agreement between the multiplex panel and routine tests was high, there were discrepancies. Among 161 CSF samples that were positive for a viral target by conventional methods, 21 (13.0%) showed a discordant result when analyzed by the FilmArray ME panel (Table 1). The majority (16/21; 76.2%) of these were for HSV-1 (n = 8) and HSV-2 (n = 8). Real-time PCR for HSV-1/2 that had been ordered for routine clinical purposes identified 14 CSF samples that were positive for either HSV-1 (n = 7) or HSV-2 (n = 7), and each of these was negative for HSV-1/2 by the FilmArray ME panel. The majority of these samples were stored for ≤7 days at 4°C prior to testing by the FilmArray ME panel. Discrepant analysis was performed on 10 of the 14 samples, with 8 of the 10 testing negative for HSV-1/2 by an alternate molecular test. One of the two CSF samples that were positive by the alternate molecular test had a cycle threshold (C_T) value of 33, while the second one was positive only by melting curve analysis (i.e., the reaction did not generate a C_T value). Furthermore, the 8 specimens that were negative by the alternative method had an initial C_T value of >37.5 on routine testing, indicating the presence of a very small amount of target nucleic acid. The reported LoD for the FilmArray ME panel HSV-1 and HSV-2 targets (250 TCID₅₀/ml and 50 TCID₅₀/ml, respectively) is ∼10-fold higher (less sensitive) than the reported LoD for our laboratory's routine HSV-1/2 assay (Simplexa HSV 1&2 Direct), and therefore, this may account for these discordant results. In addition, there were 2 CSF samples that were uniquely positive for either HSV-1 (n = 1) or HSV-2 (n = 1) by the FilmArray ME panel, and both of these tested negative for HSV-1/2 by an arbiter assay (Table 1).

Of the remaining 5 CSF specimens showing a discordant result for a viral target, 2 CSF specimens were positive (as determined by melting curve analysis) for EV by routine real-time PCR but negative by the FilmArray ME panel. The presence of enterovirus nucleic acid was confirmed upon retesting. It should be noted that both the FilmArray ME panel and the routine EV real-time PCR may cross-react with rhinoviruses. Unfortunately, sufficient sample was unavailable for further testing/typing, and therefore, we cannot rule out the possibility that these specimens were positive for a rhinovirus serotype that may not be detected by the FilmArray ME panel.

Finally, three CSF specimens were observed to be positive by the FilmArray ME panel for either VZV (n = 1) or HHV-6 (n = 2), but testing for these viruses had not been originally ordered by the health care provider. After the results of the FilmArray ME panel were obtained, these CSF specimens were tested by the laboratory's routine real-time PCR for VZV or HHV-6. Varicella-zoster virus nucleic acid was detected in the sample that was positive for VZV by the FilmArray, while 1 of 2 HHV-6-positive results was confirmed by additional testing. Following discordant analysis, the overall adjusted PPA among viral targets was 94.8% (147/155) (Table 1).

To address the limited data on the ability of the FilmArray ME panel to detect bacterial causes of meningitis, we accessed a convenience panel of clinical CSF samples (n = 76) that had been stored at the Mayo Clinic at −70°C for >3 decades. Unique positive results by the FilmArray panel were observed for S. agalactiae (n = 1), L. monocytogenes (n = 1), N. meningitidis (n = 1), and S. pneumoniae (n = 3) (Table 1). Unfortunately, there was an insufficient remaining volume of CSF to perform discordant analysis for the samples positive for S. agalactiae and L. monocytogenes. The single sample that was uniquely positive for N. meningitidis by the FilmArray ME panel was also positive by an alternate molecular method. Among the three samples testing uniquely positive for S. pneumoniae by the FilmArray panel, one was confirmed to be positive by an alternate molecular method. The remaining 2 CSF samples that were uniquely positive for S. pneumoniae were also positive for either N. meningitidis (sample A) or C. neoformans/C. gattii (sample B) by the FilmArray ME panel (Table 2). Clinical information was unavailable for these specimens; however, only N. meningitidis was identified on Gram stain and culture from sample A, while the CrAg titer was ≥1:2,560 in sample B.

Previous studies have demonstrated the potential for positive results for certain bacterial targets on the FilmArray ME panel, especially S. pneumoniae, which are not able to be confirmed by alternate methods. Given that the incidence of bacterial meningitis has declined significantly, positive results need to be interpreted with caution. It will be essential for laboratory staff and health care providers to adhere to strict procedures for the collection, transport, processing, and testing of CSF specimens so that the potential for contamination may be reduced (19). It is also important to emphasize that the continued use of routine Gram stain and bacterial cultures will be important to help confirm the results of molecular tests.

Two potentially false-negative results by the FilmArray ME panel, one each for N. meningitidis and H. influenzae, were investigated by performing a review of the patients' medical records. Both specimens had been stored at −70°C for >3 decades. A CSF sample that was characterized as positive for N. meningitidis by routine testing had been collected from a pediatric patient admitted for fever and decreased feeding. A total of 3 CSF specimens were collected from the patient, and all were negative by Gram stain and routine bacterial culture. Only 1 of the 3 CSF specimens, which was the sample included in this study, was positive for N. meningitidis (serogroup Y) by CIE. Therefore, it is possible that the CIE result was falsely positive. Additionally, one CSF specimen tested positive for H. influenzae by CIE but was negative by the FilmArray ME panel. Upon chart review, no organisms were observed on Gram stain. Following the negative result by the FilmArray ME panel, this specimen was tested by an alternate molecular assay and was negative for H. influenzae. These results raise the possibility that for these 2 CSF samples, the original CIE result may have been misinterpreted or the target nucleic acid may have degraded during storage.

The highest rate of discordant results in our study was observed when comparing the FilmArray assay for C. neoformans/C. gattii to routine CrAg testing by LFA. Previous studies have demonstrated that CrAg may be detectable for extended periods of time following initiation of therapy, even after routine fungal culture and smear results become negative (20). Additionally, studies have compared the FilmArray ME panel to other molecular assays and antigen testing and have shown that CrAg remains detectable after the FilmArray and comparator molecular assays are negative (8, 17). Therefore, it may be more clinically relevant to compare the results of the FilmArray ME panel to those of routine fungal culture/smear or only to those of CrAg testing performed for initial diagnostic purposes prior to initiation of antifungal therapy. To this end, the FilmArray assay for C. neoformans/C. gattii was positive for 13 (92.9%) of 14 samples that were also positive by routine fungal culture and/or smear (Table 3).

Despite the higher correlation of the FilmArray C. neoformans/C. gattii target results with those of fungal culture/smear, the role of the FilmArray ME panel for the first-time diagnosis of cryptococcal meningitis remains undefined. To date, studies evaluating the performance of the FilmArray ME panel have included a relatively small number of CSF samples collected from patients with suspected (i.e., not previously diagnosed) cryptococcal meningitis (8, 17, 21). A recent study from Uganda demonstrated the 100% sensitivity of the FilmArray cryptococcal assay using CSF (n = 18) collected to establish an initial diagnosis of cryptococcal meningitis (21). However, false-negative FilmArray ME panel results for cryptococci have been reported in both the FilmArray ME panel package insert and a case report (22), especially in patients with a low burden of disease. Further studies are needed to better define the performance of the FilmArray ME panel for establishing an initial diagnosis of cryptococcal meningitis.

An advantage of multiplex molecular tests is the ability to detect coinfections that might go undetected with individual, targeted assays. In our study, coinfections were detected in 8 (2.7%) of 291 samples tested (Table 2). In 7 of the 8 cases, the expected, previously characterized target was detected by the FilmArray ME panel. For the additional targets detected, discordant analysis and chart review were performed, where possible. In one CSF sample (case 1) that was positive for H. influenzae by routine testing (CIE), L. monocytogenes was also detected by the FilmArray ME panel (Table 2). However, the additional detection of L. monocytogenes was unable to be confirmed due to the lack of an available confirmatory test and no access to the patient's medical record. Among the remaining six samples, four of the additional detections (S. pneumoniae [n = 2; cases 2 and 3], HHV-6 [n = 1; case 4], and HSV-1 [n = 1; case 5]) were not confirmed by alternate molecular testing. In two cases, an additional target (HHV-6 [n = 1; case 6] and VZV [n = 1; case 7]) for which routine testing was not initially ordered was detected. The presence of these viral targets was confirmed by an alternate molecular test; however, clinical information was unavailable in these cases, and therefore, we were unable to determine if these coinfections were of clinical significance. Finally, in case 8, the CSF was noted to be positive for H. influenzae by CIE, but the FilmArray ME panel detected N. meningitidis and S. pneumoniae (Table 2). Interestingly, the laboratory record indicated that no organisms had been seen on CSF Gram stain, and no culture had been performed. Additionally, discordant analysis using an alternate molecular test confirmed the positive results for N. meningitidis and S. pneumoniae but was negative for H. influenzae. In total, 3 of 8 coinfections detected by the FilmArray ME panel were able to be confirmed. Detection of multiple organisms in CSF is likely a rare event, and coinfections reported from this source by multiplex molecular tests should be interpreted with caution and potentially confirmed using a second assay.

This study has several important limitations. First, the majority (76/80; 95%) of CSF samples that were positive for a bacterial target were originally characterized by CIE and not by current molecular methods (e.g., real-time PCR or sequencing). CIE has been noted to lack specificity in certain cases (23). Second, the number of CSF specimens positive for parechovirus (n = 0), CMV (n = 3), L. monocytogenes (n = 1), and S. agalactiae (n = 2) was low, limiting the conclusions that can be made regarding the performance of the FilmArray ME panel for these targets. Third, this was a retrospective study performed using a convenience panel of fairly well-characterized, positive samples. Therefore, we were unable to assess the specificity of the multiplex panel or calculate percent negative agreement. However, a prior multicenter evaluation of the FilmArray ME panel included over 1,500 prospective CSF specimens and demonstrated the high overall specificity of the panel (8). Finally, although we attempted to further analyze all discordant results, we were unable to resolve some discrepancies due to an inadequate volume of remaining CSF, the lack of an available confirmatory test, or limited access to patient or laboratory records.

Despite these limitations, our study includes the largest number of positive CSF samples (n = 291) used to date to assess the performance of the FilmArray ME panel. Syndrome-based testing represents a paradigm shift in the diagnosis of infectious diseases, and our results, along with those of previously published studies, suggest that the use of multiplex molecular panels may assist in the diagnosis of CNS infections. Specifically, multiplex panels may play an important role when antimicrobials are administered empirically prior to collection of a CSF sample, thereby potentially impacting the sensitivity of culture-based methods (24, 25). However, routine implementation of multiplex CNS panels, such as the FilmArray ME panel, will require careful consideration and thoughtful utilization management strategies. Consideration of patient risk factors, including age, exposure history, seasonality, and immune status, will be important in determining the appropriate diagnostic tests to perform (26). For example, HHV-6 and CMV do not typically cause CNS infections in immunocompetent individuals. Therefore, detection of these viruses in CSF may be difficult to interpret, even in immunocompromised hosts, as a positive result may be attributable to latently infected leukocytes in the CSF (27). In cases of suspected bacterial meningitis, routine CSF chemistries, cellular analysis, and Gram stain remain important initial tests. In addition, routine bacterial culture will continue to be necessary for susceptibility testing and serotyping. Further, important causes of CNS infection, including but not limited to West Nile virus, Mycobacterium tuberculosis, Borrelia burgdorferi, and Histoplasma capsulatum, are not included on the ME panel, and therefore, health care providers will need to carefully consider factors such as seasonality, geographic location, prevalence, and patient risk factors when selecting tests for CSF and, if using the FilmArray ME panel, may need to collect sufficient CSF to enable this testing if the panel results are noninformative (27).

In summary, the FilmArray ME panel tests for 14 of the most common causes of infectious meningitis/encephalitis and may provide a rapid, front-line diagnostic option for laboratories that do not have the capacity to perform routine real-time PCR and currently rely on send-out testing. This study demonstrated that the results of the assay show a high overall correlation (90.4% after discordant resolution; 96.6% when cryptococcal results are excluded) with the results of routine tests. Furthermore, the assay provides rapid results (∼1 h). Although the assay is promising, future studies are needed to define the impact of the FilmArray meningitis/encephalitis panel on treatment decisions, cost-effectiveness, and, ultimately, patient outcomes.