|Home | About | Journals | Submit | Contact Us | Français|
Respiratory tract infections (RTIs) are caused by a plethora of viral and bacterial pathogens. In particular, lower RTIs are a leading cause of hospitalization and mortality. Timely detection of the infecting respiratory pathogens is crucial to optimize treatment and care. In this study, three U.S. Food and Drug Administration-approved molecular multiplex platforms (Prodesse ProFLU+/FAST+, FilmArray RP, and Verigene RV+) were evaluated for influenza virus detection in 171 clinical samples collected during the Belgian 2011-2012 influenza season. Sampling was done using mid-turbinate flocked swabs, and the collected samples were stored in universal transport medium. The amount of viral RNA present in the swab samples ranged between 3.07 and 8.82 log10 copies/ml. Sixty samples were concordant influenza A virus positive, and 8 samples were found to be concordant influenza B virus positive. Other respiratory viruses that were detected included human rhinovirus/enterovirus, respiratory syncytial virus, parainfluenza virus type 1, human metapneumovirus, and coronavirus NL63. Twenty-five samples yielded discordant results across the various assays which required further characterization by sequencing. The FilmArray RP and Prodesse ProFLU+/FAST+ assays were convenient to perform with regard to sensitivity, ease of use, and low percentages of invalid results. Although the limit of sensitivity is of utmost importance, many other factors should be taken into account in selecting the most convenient molecular diagnostic assay for the detection of respiratory pathogens in clinical samples.
Respiratory tract infections (RTIs) have an enormous social economic impact, with a high incidence of hospitalization and high direct and indirect costs. Because of similar clinical symptoms (1) and simultaneous circulation of several different viruses (2, 3), the etiology is often unknown. Timely detection and discrimination of the infecting pathogens is crucial in order to optimize treatment and care, to prevent unnecessary antibiotic use, and to prevent the secondary spread of infection (2, 4–6).
Conventional diagnostic methods have several limitations, such as the time and labor-intensive viral culture method (7), for which even more specific cell lines for different respiratory viruses are needed. Although there are already several rapid immunoassays on the market for respiratory viruses, these assays lack sensitivity, making the predictive value of these methods rather poor when used outside of the epidemic period (8). Even during the epidemic period, because of the lack of sensitivity, the negative predictive value of these tests is low, which can result in the withholding of treatment inappropriately.
Molecular assays are rapid and more sensitive than conventional methods (7, 9). PCR methods (10) that detect all common respiratory viruses in one assay are widely available, but although highly sensitive, they are complex and generally require several hours to perform nucleic acid extraction, PCR amplification, and detection of multiple viral targets (11).
During the last 3 years, the U.S. Food and Drug Administration (FDA) has granted approval for several simplified molecular tests and systems (Prodesse ProFLU+/FAST+, FilmArray RP, and Verigene RV+) that were designed to detect several different types of respiratory viruses. These assays are easy to perform (sample-to-result automation) (12) and can detect coinfections, but their disadvantages include high costs and the need for specialized equipment (13).
The Prodesse ProFLU+ (GenProbe, San Diego, CA) assay enables the detection and differentiation of influenza A virus, influenza B virus, and respiratory syncytial virus (RSV; types A and B). The Prodesse ProFAST+ assay (GenProbe, San Diego, CA) is a multiplex real-time reverse transcription-PCR (RT-PCR) in vitro diagnostic test for the rapid and qualitative detection and discrimination of influenza A virus subtypes: seasonal A/H1, seasonal A/H3, and 2009 H1N1. The Prodesse ProFLU+ and ProFAST+ assays were cleared for in vitro diagnostic use by the FDA in 2008 and 2010, respectively (2).
The FilmArray respiratory panel (RP; BioFire Diagnostics, Salt Lake City, UT) is a multiplexed nucleic acid test using DNA melting analysis intended for use with the FilmArray instrument for the simultaneous qualitative detection and identification of multiple respiratory viral nucleic acids in nasopharyngeal swabs obtained from individuals suspected of RTIs. The following pathogen types and subtypes are identified using the FilmArray RP: adenovirus, bocavirus, coronavirus 229E, coronavirus HKU1, coronavirus NL63, coronavirus OC43, human metapneumovirus, influenza A virus subtype H1, influenza A virus subtype H3, influenza A virus subtype 2009 H1, influenza B virus, parainfluenza virus types 1 to 4, rhinovirus/enterovirus, RSV, Bordetella pertussis, Chlamydophila pneumoniae, and Mycoplasma pneumoniae. In 2011, the FilmArray RP received FDA clearance for the detection of the pathogens mentioned above (1, 7, 11, 12, 14, 15).
The Verigene respiratory virus plus nucleic acid test (Verigene RV+; Nanosphere, Northbrook, IL) is a nucleic acid multiplex test intended to simultaneously detect and identify multiple respiratory pathogens, namely, influenza A virus, influenza A virus subtypes H1, H3, and 2009 H1N1, influenza B virus, and RSV subtypes A and B. For specimens positive for influenza A virus subtype H1 (also 2009 H1N1), the Verigene RV+ determines the presence or absence of an H275Y mutation in the neuraminidase (NA) gene, which confers resistance to oseltamivir. The Verigene RV+ is a multiplex PCR with gold nanoparticle probe-based hybridization of virus-specific amplicons on a microarray, which achieved FDA clearance in January 2011 (13).
Subtyping of influenza viruses is based on the hemagglutinin (HA) and NA sequences. The degree of resistance to oseltamivir is specific to the subtype, i.e., most seasonal H1N1 viruses are oseltamivir-resistant, whereas most of the pandemic H1 2009 and seasonal H3N2 influenza viruses are sensitive to oseltamivir (16). Subtyping could therefore be useful for making treatment decisions. With the discovery of new broadly neutralizing antibodies, such as CR6261 and CR8020, which have activity against, respectively, cluster I (e.g., H1, H2, H5, and H9) and cluster II (e.g., H3, H7) influenza viruses (17–19), subtyping prior to treatment might become more important. The Prodesse ProFAST+, FilmArray RP, and Verigene RV+ platforms are capable of HA subtyping influenza A virus (influenza A virus subtypes H1, H3, and 2009 H1).
In the present study, three FDA-approved molecular assays (Prodesse ProFLU+/FAST+, FilmArray RP, and Verigene RV+) were evaluated for influenza virus detection and discrimination between influenza virus clusters I and II in clinical samples collected during the Belgian 2011-2012 influenza season, and the performance characteristics were compared.
The clinical study was a prospective, biological specimen collection study from patients presenting with influenza-like illness (ILI) and/or acute RTI in an outpatient setting for a network of 15 general practitioners. The probability of enriching the sample collection with influenza virus-containing samples was augmented by implementing an influenza epidemiological surveillance guided sampling campaign, assuring that the sampling time window corresponded to the period closest to the peak of the influenza epidemics. The Belgian Flu National Surveillance System was used as the main source of information for the identification of the optimum start and end timing for this study. Patients suffering from RTI and/or ILI, presenting for medical support at a primary care physicians' office, were enrolled between February and March 2012. The diagnosis of ILI and/or acute respiratory infection was based on European Center for Disease Control criteria. The patient inclusion criteria allowed the inclusion of both sexes and all age groups in the study. To enhance influenza virus detection, only patients symptomatic for fewer than 3 days were included in the study. Patients who received treatment for influenza virus infection (zanamivir, oseltamivir, amantadine, and rimantadine) during the last 7 days and/or were vaccinated against flu within the prior 3 weeks were excluded. Any clinical symptoms and underlying respiratory condition(s) were registered. Sampling was performed using mid-turbinate flocked swabs (i.e., a mid-turbinate adult or pediatric flocked swab with a nylon tip, adult [80 mm] or pediatric [50 or 80 mm]; Copan, Italy), and the collected samples were stored in 3 ml of universal transport medium (UTM; Copan) at −80°C. A total of 171 swab samples (1 per patient) were available. This sampling method is noninvasive and user-friendly and allows sampling from patients of different ages, without causing major discomfort. To keep the freeze-thaw cycle for all samples for the different assays identical, samples were stored at −80°C in aliquots until specific testing was performed. Because of the limited stock volume per swab sample, a 1:1.5 dilution was created in UTM and used for all assays except for the viral load quantification. The study was approved by the Committee on Medical Ethics of the University Hospital (Leuven, Belgium), and written consent from all participants was obtained.
The internal extraction control B (IEC-B) was produced as described earlier (20). For the external quantification control (EQC), a standard RNA dilution series was tested in duplicate on each real-time PCR plate with the same reaction mix (influenza virus A) used to process the samples in order to determine the dynamic range, the performance of the real-time PCR reagents during each run, and absolute quantification. The viral loads of the processed samples were determined as the log10 copies/ml through extrapolation on the EQC standard curve. A consensus sequence of the matrix gene derived from 113 matrix sequences of different subtypes (H1, H2, H3, H5, H7, and H9) was assembled and prepared by using synthetic biology. This 142-bp fragment was cloned into a standard pMA-T vector (Life Technologies, USA). Plasmid DNA was transformed in Top10 competent Escherichia coli cells according to the manufacturer's instructions (Life Technologies). Plasmid isolation was achieved with a PureLink HiPure plasmid filter purification kit (Invitrogen), and the DNA pellet was resuspended in 200 μl of Tris-EDTA buffer. Plasmid DNA was linearized with the restriction enzyme PvuII (Fermentas, Belgium) using 1 μl of DNA, 2 μl of buffer G, 15 μl of H2O, and 2 μl of PvuII (10 U/μl) for 1 h at 37°C. Purification was performed with a QIAquick PCR quantification kit (Qiagen) and eluted in 30 μl of prewarmed elution buffer. In vitro RNA transcription was performed with a MEGAscript kit according to the manufacturer's instructions (Ambion), generating a RNA fragment of 269 bases. DNase treatment was performed with 1 μl of Turbo DNase (Ambion) for 15 min at 37°C, and RNA was recovered and eluted in 30 μl of RNase-free water with an RNeasy kit (Qiagen). From these EQC RNA pools, a 1:10 dilution series was made in extraction buffer 3 (bioMérieux, Netherlands), and the RNA concentration of the 1:20 dilution of the EQC RNA pool was measured using a Quant-iT RNA assay kit (Invitrogen) (average of 5 measurements). An extrapolation was made for the eight EQC dilutions to define the RNA concentration. A standard curve was constructed by plotting each cycle threshold (CT) value against the log10 quantity of the standard RNA copy numbers. To obtain amount of viral load present in each clinical sample, the test CT value was used to determine the viral load using a linear regression equation for the standard curve. The amplification primers for the EQC (i.e., the FluA forward and reverse primers and probe) are specified in Table 1.
Viral RNA was isolated starting from 100-μl sample. The IEC-B was added to all samples prior to the RNA extraction to monitor the RNA extraction efficiency, as described previously (20). IEC-B is a nonspecific artificial sequence that can be detected with the IEC-B amplification primer sequences (Table 1). A total of 2 ml of NucliSENS extraction lysis buffer (bioMérieux) was added, and the solution was incubated for 10 min prior to extraction on an automated easyMAG extraction platform (bioMérieux). A total of 100 μl of extraction control mix (for 8 samples: 550 μl of NucliSENS magnetic beads, 110 μl of IEC-B, and 440 μl of NucliSENS extraction buffer 3) was added to each well prior to extraction. Elution of the RNA was done in 100 μl, and samples were stored at −80°C until further use (i.e., viral load quantification and HA and NA sequencing).
Real-time PCR was performed according to the CDC protocol for influenza A virus (targeting the matrix gene) with a panel of oligonucleotide primers and dually labeled hydrolysis (TaqMan) probes (20a). Each sample was tested in duplicate. The sequences of the primer and probe sets used here are shown in Table 1. Amplification and detection were performed on a LightCycler 480 instrument (Roche Applied Science). The 25–μl reaction mixture of real–time RT–PCR was comprised of 5 μl of total RNA, 12.5 μl of 2× PCR master mix, 0.5 μl of SuperScript III RT–Platinum Taq mix (Life Technologies), 0.5 μl of forward and reverse primers (40 μM), 0.5 μl of labeled probe (10 μM), and 5.5 μl of RNase/DNase–free water. The IEC–B reaction mixture was composed of 5 μl of total RNA, 12.5 μl of 2× PCR master mix, 0.5 μl of SuperScript III RT–Platinum Taq mix, 1.0 μl of forward and reverse primers (20 μM), 0.2 μl of labeled probe (25 μM), and 4.8 μl of RNase/DNase–free water. The thermal cycling conditions for both mixtures were as follows: reverse transcription at 50°C for 30 min, 95°C for 2 min for Taq polymerase activation, followed by 45 cycles of PCR amplification (95°C for 15 s and 55°C for 30 s). The following formula was used for normalization of the data: CT sample − (IEC–B CT sample − IEC–B CT average of plate).
Several molecular assays were evaluated for respiratory pathogen detection. All reagents and assays were purchased directly from the manufacturers. The FilmArray and Verigene equipments were leased for the time of this study.
The Prodesse ProFLU+ and ProFAST+ real–time PCR assays were performed according to the manufacturer's instructions. Briefly, 20 μl of internal control was added to a 180–μl sample (1/1.5 dilution) or to a 180–μl UTM (negative control) to monitor inhibition. Isolation and purification of nucleic acids were performed using a NucliSENS easyMAG System (bioMérieux), and the samples were eluted in 55 μl. Then, a 5–μl portion of the purified nucleic acids from the samples was added to 20 μl of ProFAST+/ProFLU+ Supermix (19.45 μl of ProFAST+ or ProFLU+ mix, 0.30 μl of Moloney murine leukemia virus reverse transcriptase, and 0.25 μl of RNase inhibitor). A negative control and a positive control were also included in each run. Reverse transcription of the RNA into cDNA and subsequent amplification of the DNA was performed in a Cepheid SmartCycler II instrument. Data analysis was performed using the SmartCycler software (version 3.0b).
The FilmArray RP assay was performed based on the manufacturer's protocol. The FilmArray pouch stores all of the necessary reagents for sample preparation, RT–PCR, and detection in a freeze–dried format. Approximately 1 ml of hydration solution was injected into the cartridge, together with an unprocessed sample (300 μl of the 1:1.5 diluted sample). In the present study, we used FilmArray RVP IVD v1.6 pouches (which detect, in addition to the pathogens mentioned above, bocavirus, coronavirus HKU1 and OC43, Bordetella pertussis, Chlamydophila pneumoniae, and Mycoplasma pneumoniae). However, no information on these pathogens was obtained due to limitations of the version of the software used.
The Verigene RV+ assay was performed as described in the user manual. In brief, single–use disposables, including an extraction tray, amplification tray, Verigene test cartridge, and a separate tip holder assembly were used. After the consumables were loaded, the sample (200 μl of the 1:1.5 diluted sample) was pipetted into the extraction tray, and the test protocol was initiated, results were collected within 2.5 h.
To confirm the presence or absence of influenza virus, Sanger sequencing of the influenza A virus HA and NA genes was performed on the 23 discordant samples (see Table 3). RNA was extracted as described above. Full–length HA was amplified, whereas for NA just a short fragment (from amino acids I65 to D304 according to the reference sequences described below), covering most NA inhibitor resistance mutations including E119V and H275Y, was amplified (for a table showing the resistance mutations, see http://www.isirv.org/site/images/stories/avg_documents/Resistance/mutations_18.04.12.pdf). cDNA was generated using ThermoScript reverse transcriptase (Life Technologies) and the primer Uni12(M) at 10 μM (Table 1), which is complementary to the conserved 3′ ends of all influenza A viral RNA segments. The HA/NA gene was amplified by nested PCR (see the primers and probes in Table 1). Both outer and inner PCRs were performed using Platinum Taq DNA polymerase (Life Technologies) with 10% of the cDNA synthesis reaction volume (2 μl). Thermal cycling for HA consisted of a denaturation step of 94°C for 4 min, followed by 35 cycles of 30 s at 94°C, 30 s at 50°C, and 2 min at 72°C and then by 1 cycle of 7 min at 72°C. The inner PCR was conducted with 1 μl of outer PCR product. Thermal cycling for HA (inner PCR) consisted of 30 cycles, and annealing occurred at a temperature of 60°C. Thermal cycling for part of the NA (I65 till D304) gene consisted of a denaturation step at 94°C for 4 min, followed by 40 cycles of 30 s at 94°C, 30 s at 55°C, and 2 min at 72°C and then by 1 cycle of 7 min at 72°C. The inner PCR was conducted with 2 μl of outer PCR product, and the same PCR conditions were used. Amplicons were purified using a QIAquick PCR purification kit. Sequencing reactions were performed using a BigDye Terminator cycle sequencing kit and run on an ABI3730×/automated sequencer (Applied Biosystems, Foster City, CA). The thermal cycling conditions were as follows: 25 or 30 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C, ending at 12°C. Purification of the sequence products was performed with a DyeEx sequence purification kit (Qiagen). Sequence analysis was performed with the SeqScape v2.5 software (Applied Biosystems). The obtained HA proteins were aligned with the HA proteins, including the following reference strains: A/California/07/2009 (H1N1; ACP41953.1), A/Puerto Rico/8/34 (H1N1; NP_040980.1), A/South Carolina/1/1918 (H1N1; AAD17229.1), A/New York/392/2004 (H3N2; YP_308839.1), and A/Hong Kong/1/1968 (H3N2; ACU79871.1). The obtained NA proteins were aligned with the NA proteins from the reference strain A/New York/392/2004 (H3N2; NC007368).
A total of 171 mid–turbinate flocked samples were collected from patients presenting with acute RTI and/or ILI. Each sample was analyzed using four different FDA–approved assays: the Prodesse ProFLU+ and ProFAST+ assays, the FilmArray RP assay, and the Verigene RV+ assay. In addition, the viral load was determined with a qRT–PCR assay.
Using the Prodesse ProFLU+ assay, we found that 84 samples were positive for influenza A virus, 10 samples were positive for influenza B virus, 2 samples were positive for RSV, and coinfection of influenza A and B viruses was found in 1 sample. Two samples had invalid results due to failure of the internal control (Table 2).
Using the Prodesse ProFAST+ assay, we found that 78 samples tested positive for seasonal influenza virus A/H3, 3 samples contained 2009 H1N1 Influenza virus, and no samples with seasonal influenza virus A/H1 were detected. Two samples had invalid results due to failure of the internal control (Table 2).
When we tested the FilmArray RP assay, two samples had invalid results due to one software and one control error. Two samples were equivocal for influenza A virus, since only 1 pan–influenza A virus assay (either FluA–pan1 or FluA–pan2) was positive for these samples, and all of the subtyping assays were negative. Two other samples were equivocal for influenza virus A/H3, with only a positive result for FluA–H3, but no positive result for both FluA–pan assays. Among the remaining 165 samples, 70 were identified as influenza A virus H3 and 3 samples contained influenza A virus H1 2009. Eleven samples were identified as influenza B virus. Other detected respiratory viruses included human rhinovirus/enterovirus (n = 8), RSV (n = 1), parainfluenza virus type 1 (n = 1), human metapneumovirus (n = 7), and coronavirus NL63 (n = 5). In one sample, human rhinovirus/enterovirus was simultaneously detected with influenza A virus H3, and in another sample coinfection of human rhinovirus/enterovirus with RSV was detected. A total of 57 samples were negative for the respiratory viruses that were included in the FilmArray respiratory panel (Table 2).
A high percentage of invalid results was found when we tested the Verigene RV+ assay (n = 27; 15.8%), mainly due to the failure of an internal control. Of the remaining 144 samples, 62 samples were identified as influenza A virus, 10 samples were identified as influenza B virus, 2 samples were identified as RSV (subtype B), and the remaining 70 samples were negative for both influenza virus and RSV. Considering the 62 influenza A virus samples, 55 were subtyped as influenza A virus H3, 3 contained influenza A virus, subtype H1 2009, and for 4 no subtype could be determined. No oseltamivir (H275Y) resistance was found in the 3 influenza A virus H1 2009 samples (Table 2).
The viral load in 171 samples was determined with qRT–PCR assay using universal influenza A virus primers and probe, which are located in the matrix gene. All CT values were corrected for the loss of RNA during extraction by use of the IEC–B. The EQC standard curve has the following characteristics: y = −3.0295x + 40.168 with R2 = 0.9876, and a linear range of 6 log10 input copies (from 3 to 9 log10 copies/ml). A total of 92 (53.8%) of 171 samples were positive for influenza A virus. The amount of viral RNA present in the swabs (after correction for the 1:1.5 dilution) ranged between 3.07 and 8.82 log10 copies/ml, with an average of 5.93 log10 copies/ml (Fig. 1).
The main laboratory characteristics of these assays are summarized in Table 3. The combination of the Prodesse ProFLU+ and ProFAST+ assays allows for the detection of influenza A virus, influenza B virus, and RSV and the subtyping of influenza A virus. Since the internal control of both assays is identical, they can be combined by a single nucleic acid extraction. The FilmArray RP panel allows the simultaneous detection of 21 respiratory pathogens, including bacteria, in a single assay. One strength of the Verigene RV+ test lies in its ability to detect oseltamivir resistance for H1 and H1 2009 influenza A viruses, an advantage that is not afforded by any of the other evaluated assays. One main advantage of the FilmArray and Verigene systems is that the sample preparation is included in the assays. This is not the case for the Prodesse ProFLU+ and ProFAST+ assays, which means that a specified lab environment is necessary to perform RNA extraction. The turnaround time varies between 1 and 3 h, which is comparable for all of the assays. However, major differences in hands–on time were observed. Whereas the Prodesse ProFLU+ and ProFAST+ assays require ~1.5 h (which is mainly due to the RNA extraction), only 2 to 5 min of hands–on time are necessary for the FilmArray and Verigene systems. A total of 14 samples (plus a positive and a negative control) can be performed in one run for the Prodesse ProFLU+ and ProFAST+ assays compared to only one sample per instrument for the FilmArray and Verigene assays. The throughput of the Verigene system can be enhanced by adding multiple processors to the same reader, and the throughput of the Prodesse ProFLU+ and ProFAST+ assays depends on the number of SmartCyclers available. The percentages of invalid results were small for the Prodesse ProFLU+, Prodesse ProFAST+, and FilmArray assays, i.e., 1.2%. However, for the Verigene system, as many as 15.8% of the results were invalid, mainly due to failure of the internal controls.
Since 31 samples had an invalid result for one or more assays, they were excluded from further analysis. For the remaining 140 samples, the detection rate of influenza virus was evaluated. Sixty samples (42.86%) were identified as influenza A virus in the Prodesse ProFLU+, the FilmArray, and Verigene assays. The viral RNA loads of these 60 influenza A virus–positive samples ranged between 3.60 and 8.60 log10 copies/ml. Eight samples (5.71%) were identified as concordant influenza B virus. Fifty samples (35.71%) were negative for both influenza A and B viruses in all assays.
Influenza A virus subtyping was also evaluated using the ProFAST+, FilmArray RP, and Verigene RV+ assays. Of the 60 influenza A virus–positive samples, 3 were determined to be influenza A virus, H1 2009 subtype, and 54 were determined to be influenza A virus, H3 subtype, across all three influenza A virus subtyping assays. The other 3 samples were influenza A virus positive across all of the assays, but no subtyping could be obtained using the Verigene RV+ assay (see below).
In all, 25 samples (17.86%) had discordant results across the various assays either for influenza virus detection or influenza A virus subtyping. Subsequently, the presence or absence of influenza virus was confirmed by amplification and sequencing of the influenza A virus HA and NA genes (Table 4). Three samples (samples 1 to 3, either with viral loads of 3.5 log10 copies/ml or negative) were positive for influenza A virus (Prodesse ProFLU+) but no influenza A virus subtype could be detected with the Prodesse ProFAST+, FilmArray RP, or Verigene RV+ assay. Eight samples (samples 4 to 11) were correctly identified as influenza virus A/H3 by the ProFLU+/FAST+ and FilmArray RP assays but were either determined to be negative by the Verigene RV+ assay or no subtyping could be obtained. Two samples (samples 12 and 13) were determined to be positive for influenza virus A/H3 by the Prodesse ProFLU+/FAST+ assays, but this could not be confirmed by the FilmArray RP or Verigene RV+ assays. However, influenza virus A/H3 could be sequenced in one sample. Five other samples (samples 14 to 18; viral load < 4 log copies/ml), which were determined to be negative by the Prodesse ProFLU+ assay, were positive for one or more of the other assays. Coinfection of influenza viruses A and B in sample 19 (as determined by the Prodesse ProFLU+ assay) could not be confirmed by the FilmArray RP and Verigene RV+ assays. The last six samples were positive for influenza A virus in the qRT–PCR assay but were negative for influenza A virus in all of the other assays.
The performances of the Prodesse ProFLU+, FilmArray RP, and Verigene RV+ assays for the detection of influenza A virus were calculated. These characteristics were calculated considering each assay as the gold standard (Table 5). When we considered the Prodesse ProFLU+ assay as the gold standard, 74 samples were true positives (TP) by the FilmArray RP, with only 61 TP samples by Verigene RV+ and 81 TP samples by qRT–PCR, resulting in a sensitivity of 90.24% for the FilmArray RP, a sensitivity of only 84.72% for the Verigene RV+ assay, and a sensitivity of 95.29% for qRT–PCR. When we considered the FilmArray RP assay as the gold standard, 74 samples were TP by the Prodesse ProFLU+ assay and by qRT–PCR, and only 60 TP samples were obtained using Verigene RV+, resulting in sensitivities of 100% for the Prodesse ProFLU+ assay and qRT–PCR and 92.31% for the Verigene RV+ assay. When we considered the Verigene RV+ assay as the gold standard, 61 samples were TP by the Prodesse ProFLU+ assay and qRT–PCR and there were 60 TP samples obtained by the FilmArray RP assay, resulting in a sensitivity of 98% for all three assays. When we considered qRT–PCR as the gold standard, a sensitivities of 88.04, 85.06, and 76.25% were found for the Prodesse ProFLU+, FilmArray RP, and Verigene RV+ assays, respectively.
We describe here a comparative study of the FilmArray RP, Verigene RV+, and Prodesse ProFLU+/FAST+ multiplex platforms for the detection of respiratory viruses in 171 clinical samples from the 2011–2012 influenza season in Belgium. Since 31 samples had an invalid result for one or more assays, the detection rate of influenza virus and influenza A virus subtyping was evaluated on 140 samples. A total of 75 (53.57%) samples were influenza A virus positive for at least two assays and/or positive for an influenza A virus HA or NA sequence, and 10 samples (7.14%) were found to be concordant influenza B virus positive for at least two diagnostic assays. Influenza surveillance data obtained by the Belgian Scientific Institute of Public Health (WIV–ISP) indicated that throughout the 2011–2012 influenza season, 59% of the tested samples were influenza virus positive, of which 56% were influenza A virus positive and 3%of which were influenza B virus positive (http://influenza.wiv–isp.be/). The influenza A viruses were almost all subtyped as A/H3N2. Compared to former seasons, the intensity of the epidemic was moderate. This distribution of influenza virus–positive samples is comparable to the positivity rates found in our study (60.71% versus 59%). We also detected coronavirus, human metapneumovirus, rhinovirus/enterovirus, and parainfluenza 1 virus. Only two positive RSV samples were observed, a finding that was in the line of our expectations since the epidemiological peak of RSV fell in the beginning of December 2011 in Belgium. Adenovirus, parainfluenza virus types 2 to 4, and coronavirus HKU1 were not found. In 44 samples, none of the respiratory viruses included in the assays were detected. This can be explained by a viral load below the detection limit or by infection of these patients by respiratory pathogens that are not covered in one of the assays, e.g., bocavirus, coronavirus 229E or OC43, Bordetella pertussis, Chlamydophila pneumoniae, Mycoplasma pneumoniae, and Streptococcus pneumoniae (21).
To our knowledge, this is the first study in which four FDA-cleared assays for the detection of respiratory pathogens were compared. Previously conducted studies have compared the performances of a single FDA-cleared test (e.g., the Prodesse assays, FilmArray RP, or Verigene RV+) to either conventional cell culture methods, in-house real-time PCRs, Luminex xTAG RVP assays, or other molecular assays (1, 2, 7, 11–13, 15), but none of these studies compared more than two FDA-cleared assays to each other. By keeping the freeze-thaw cycles of the samples identical for all of the assays, we ruled out specimen storage effects, facilitating a fair comparison. This comparative study has shown that the FilmArray RP and the Prodesse ProFLU+/FAST+ assays have better sensitivities for the detection of influenza A virus compared to the Verigene RV+ assay (Table 4). The lower detection limit of these assays is in the range of 3.5 to 4.5 log RNA copies/ml. Verigene RV+ failed to detect influenza virus in two samples with viral loads of >5 log RNA copies/ml (Table 4) and demonstrated a high percentage of invalid results (mainly due to the failure of an internal control). As described earlier, many factors should be taken into account in selecting the best molecular diagnostic assay for the detection of respiratory pathogens (Table 3). For the single detection of influenza viruses, the combination of the ProFLU+ and ProFAST+ assays may be the best choice for a high-throughput laboratory analyzing samples in batches. The FilmArray system may be more convenient when a relatively small set of samples per day must be analyzed and/or when the lab is not fully equipped for RNA extraction and/or real-time PCR detection. However, the cost of the FilmArray RP test for 1 sample is rather high. As recently described (22), the implementation of the FilmArray RP in a core laboratory improved the testing turnaround time for respiratory viruses. The early detection of respiratory pathogens has several implications for patient quarantine strategy and treatment. High-risk patients populations (for example, patients in intensive care units with respiratory distress and immunocompromised patients) could greatly benefit from a broad and rapid screening of different respiratory pathogens. In addition, the early discrimination between cluster I and II influenza viruses could influence treatment decisions for the use of broadly neutralizing antibodies such as CR6261 or CR8020. With the rise of antiviral influenza virus interventions with specific targets (e.g., neutralizing antibodies CR6261 and CR8020), the early discrimination between cluster I and II influenza viruses will become increasingly important, supporting the need for an early detection and a highly specific intervention strategy.
We thank the physicians and all of the volunteers that participated in this study. We also thank Janssen Biobank and the Lab Operations and Medical Affairs department from Janssen Diagnostics for their logistic support and Hans De Wolf for helpful scientific discussions.
Published ahead of print 3 July 2013