key: cord-308921-lpcjhxvg
authors: Stellrecht, Kathleen A.; Cimino, Jesse L.; Wilson, Lisa I.; Maceira, Vincente P.; Butt, Shafiq A.
title: Panther Fusion® Respiratory Virus Assays for the detection of influenza and other respiratory viruses()
date: 2019-11-09
journal: J Clin Virol
DOI: 10.1016/j.jcv.2019.104204
sha: 
doc_id: 308921
cord_uid: lpcjhxvg

BACKGROUND: Nucleic acid amplification tests (NAATs), such as PCR, are preferred for respiratory virus testing, due to superior diagnostic accuracy and faster turnaround time. Panther Fusion® Respiratory Assays (Fusion), which includes FluA/B/RSV (FFABR), Paraflu and AdV/hMPV/RV, offers a modular approach to syndromic testing on a fully automated platform while improving gene targets and expanding the test menu. OBJECTIVES AND STUDY DESIGN: We evaluated Fusion using 275 consecutive nasopharyngeal specimens previously used in an analysis of five PCRs, as well as 225 archived specimens. RESULTS: Of the combined 500 specimens, 134 were positive for influenza A (FluA), 54 for FluB, 65 for RSV, 64 for parainfluenza (PIV), 24 for adenovirus (AdV), 21 for humanmetapneumovirus (hMPV), and 40 for rhinovirus (RV) with Fusion. Of the positive samples Fusion correlated with historical results for all but one, despite multiple freeze-thaws cycles of this collection. Fusion was positive for an additional 33 samples, including 11 FluAs, 7 RSVs, 3 PIV3s, 3 AdV, 6 hMPV and 3 RVs. These samples were retested with corresponding Prodesse (Pro) assays using quadruple sample volume. This resolver test confirmed Fusion results for an additional 4 FluAs, 4 RSVs, 1 PIV3 and 3 AdVs. The sensitivity and specificity ranges of Fusion were 99–100% and 98–100%. Limit of detection (LOD) analyses were performed on a variety of Flu isolates. The LODs ranged from 2.69 to 2.99 log copies/ml and demonstrated superior LOD as compared to previously published data for some assays or to concurrent analyses with two new commercial tests.

Respiratory tract infections (RTI) are common and are associated with significant health burden. The major viral agents of RTI include FluA/B, RSV, hMPV, PIV1-4, AdV, and RV. The spectrum of diseases associated with viral infection of the upper and lower respiratory tract include the common cold, otitis media, influenza-like illness, croup, bronchiolitis and pneumonia; all of which can be caused by any one of these viruses, leading to diagnostic limitations based on symptoms alone.

Rapid identification is important for both therapeutic and infection control purposes. Traditional rapid diagnostics, such as immunoassays, produce quick results and are simple to perform but have sub-optimal sensitivity (Reviewed in [1] ). NAATs, which are rapid and have enhanced sensitivity, are considered the method of choice by many and are recommended by IDSA Guidelines [2] [3] [4] . However, performance differences have been observed among commercial NAATs, particularly after 2014 when sequence divergence in the matrix gene of A(H3N2) viruses emerged [5] [6] [7] . Especially problematic was a C163 T mutation that was first observed among 3C.2a clades of A(H3N2) [8] [9] [10] . Commercial assays with significantly reduced sensitivities after the C163 T mutation emerged, included Prodesse ProFlu+ (PFlu) and Xpert® Flu (Xpt).

The Panther Fusion® Respiratory Virus assays on the fully-automated Panther Fusion® system include the Flu A/B/RSV, Paraflu, and AdV/hMPV/RV (Table 1 ). This new system from Hologic has redesigned amplification reactions as compared to Pro. The FluA component still targets the matrix gene, but it uses a dual target approach with multiple probes for added redundancy to help safeguard against genetic drift. Both FluB and RSV now target the matrix gene and the AdV hexon gene target is designed to detect all AdV genotypes. The gene targets for hMPV and PIV 4 are the nucleocapsid genes, while those for PIV [1] [2] [3] and RV are the hemagglutinin-neuraminidase region and the 5′ UTR, respectively. This test system also expands the menu of virus detected with the inclusion of a RV and PIV4.

This study looks at the performance of the Panther Fusion® Respiratory Virus panels on a collection of samples that were previously analyzed with five other respiratory virus assays [7] , to effectively enable a comparative analysis. This population set was also highly representative of the A(H3N2) clade (3C.2a), truly challenging any FluA assay.

Clinical specimens included 275 consecutive nasopharyngeal swab specimens, in 3 ml of viral transport medium, received into the laboratory for the detection of respiratory viruses during a 2-week period in the winter of 2015 (age range 22 d to 93 yr., median 25 yr., 45% pediatric cases, Table S1 ). These specimens were previously used in a prospective analysis of PFlu and PFAST (Hologic, San Diego, CA), FilmArray Respiratory Panel 1.7 (RP, BioFire, Salt Lake City, UT), and cobas® Influenza A/B test (cIAB, Roche Diagnostics, Indianapolis, IN) specifically for the detection of FluA. Subsequently, Xpt (Cepheid, Carlsbad, CA) was analyzed retrospectively. Specimens were stored at -80°C, after original testing with PFlu/PFAST (49%) or RP (51%), thawed for cIAB and RP or PFAST testing, frozen at -80°C, thawed and frozen for Xpt testing. Some were frozen and thawed additional times for previous discrepancy analysis. They were again thawed for testing by Fusion. Selected archived samples positive for respiratory viruses (n-225), previously tested by various combinations of NAATs, were analyzed.

Clinical specimens containing low Ct values on Fusion (presumably high titers) of RSV, PIV1, PIV3, hMPV, AdV, and RV were serially diluted for endpoint comparison of Fusion with the appropriate Pro assays (PFLU, PFAST, ProParaflu+, ProAdeno+, or Pro_hMPV+, Hologic) except for the RV sample which was compared with RP2. 

Fusion Assays (Hologic) were performed on the fully automated Panther Fusion® with continuous, random access. This instrument utilizes universal nucleic acid extraction and PCR chemistry. The assay specific reagents are available in ready to use reagent cartridges. Initially, 500 μL of specimen was added to a Panther Fusion Specimen Lysis Tube containing 750 μL buffer and 360 μL of the mixture is used for an extraction. The nucleic acid was subsequently eluted into 50 μL and 5 μL were amplified for an effective sample volume tested of 14.4 μL.

LOD studies were expanded to newer tests kits, RP2 and Cepheid Xpert® Xpress Flu (Xpress), both of which have been marketed as having improved sensitivities among currently circulating strains of influenza. Testing was performed in accordance with the manufacturer's package insert.

Specimens were considered true positive (TP) using a composite reference standard (CRS) defined as positive with previously published results [11] . Samples equivocal for FluA with RP were considered positive by that test method. The resolver test involved a modification of Pro assays to enhance the analytical sensitivity. Specifically, viral RNA was extracted from 0.4 ml of specimen (twice the normal volume) using the easyMAG extractor (bioMérieux, Durham, NC) and eluted to a volume of 25 μL. RNA extracts, 5 μL (effective sample volume tested of 80 μL), were amplified with the appropriate Pro assay on SmartCyclers (Cepheid). Sensitivities, specificities, and confidence intervals (CI) were determined using Microsoft Excel 2016 (Redmond, WA) [11] . Probit analyses for the limit of detection with a 95% probability of detection were performed using SPSS version 8.0 (IBM, Armonk, NY). 

The performance of Fusion was first evaluated using 275 consecutive nasopharyngeal specimens collected during the peak of the influenza season in January-February 2015. This sample set was previously used in a prospective analysis for cIAB, PFlu/PFAST, RP and Xpt. The incidence of FluA in this population was previously considered to be 24%, exclusively A(H3N2), with 66 true positive cases. Even though this collection of specimens has been used in multiple studies and has gone through multiple freeze-thaws cycles, FFABR was FluA positive for 65 of the TP cases ( Table 2 ). The one false negative sample was previously positive by RP only. FFABR was positive in an additional 10 samples which were previously negative by other methods.

The 10 additionally positive samples were analyzed with a resolver test which involved analyzing quadruple the initial specimen volume for testing by PFAST, resulting in an additional 4 T P specimens. As a result of reclassification after resolver testing, the incidence of FluA in the sample set was now 25%. The sensitivity and specificity of FFABR for the detection of A(H3N2) circulating in 2015 were 98.7% and 97.1%.

This sample set was also positive for multiple other viruses, including RSV (55), PIV2 (1), PIV3 (10), AdV (7), hMPV (8) (17), hMPV (7), and RV/EV (20) , as well as 25 samples negative for all Fusion detectable viruses but positive for either coronaviruses or atypical bacteria (Table 3) Fusion correlated with all historical positive results except 6 samples negative with Fusion RV but positive for EV/RV by RP. These six samples were tested by our lab developed assay for EV and were positive, excluding them as false negative cases. Fusion had 29 positive test results not detected by another method, including 1 more FluA, 7 RSV, 3 PIV3, 3 AdV, 6 hMPV and 3 RV. These samples, except the 3 positive for RV, were analyzed by the appropriate resolver tests, which confirmed Fusion results for 4 RSV, 1 PIV3 and 3 AdV. The sensitivity of Fusion ranged from 99 to 100% for the various viruses and the specificity ranged from 98% to 100% (Table 4 ).

Because previous studies demonstrated FluA assay performance variations can be strain associated, we performed LOD analyses with six isolates of Flu. We also included RP2 and Xpress in this analysis as these assays were new to market and considered to have improved viral strain coverage. Fusion demonstrated excellent analytical sensitivity with low LODs ranging from 2.69 to 2.99 log copies/ml ( Table 5 ). The LODs were highly consistent across the strains and clades of FluA. The assay was highly reproducible with coefficient of variances ranging from 0.5 to 4.2% across the dilutions (data not shown). RP2 and Xpress demonstrated improved analytical sensitivity and consistency as compared to the manufacturer's previous test systems. However, the LODs were still relatively higher, particularly with newly circulating A(H3N2) subclades.

To assess differences in analytical sensitivity for the other Fusion targets versus Pro or RP, clinical specimens containing RSV, PIV1, PIV3, hMPV, AdV, and RV were serially diluted for endpoint comparison. Fusion was positive for an additional 10-fold dilution for RSV, PIV3, and AdV and two 10-fold dilutions for hMPV, while the endpoint positivities were similar between Fusion and Pro or RP for PIV1 and RV (data not shown).

Since their introduction, there have been numerous studies regarding the performance of NAATs for respiratory virus detection, whether multianalyte panels or Flu or Flu/RSV specific, as various systems came on the market to fill individual niches and needs (reviewed in [12] ). Prodesse assays had always been thought of as a necessary comparator system, if not the gold standard [13] [14] [15] [16] [17] [18] . But FluA target failures became common with A(H3N2) subclades that emerged in 2013 [7, 9] . In fact, since 2014 most all circulating A(H3N2) viruses had M1 gene mutations which significantly affected the sensitivities of PFlu and Xpt. As a result, many manufacturers modified their systems to enable expanded clade coverage. Indeed, Hologic uses influenza redundancy in the design of the Panther Fusion FluA/B/RSV assay.

Our sample populations offered a unique opportunity to challenge the performance of Fusion with A(H3N2) samples containing the highly problematic C163 T mutation in the M1 target region [6, 7] , as well as evaluate the performance among historic clades and strains of virus. Furthermore, many of these samples have been used in previous analyses of other systems, effectively enabling a multisystem analysis. In a recent study by Banerjee et al. [19] , FFABR was compared to 5 FluA/B PCRs, including cIAB and RP. However, this study involved selected pediatric specimens collected over five respiratory virus seasons, as opposed to all specimens received within a period of high incidence, which is a better challenge of test sensitivity. In addition, pediatric cases are known to have higher viral titers, which also poses less of challenge with regards to test sensitivity. As a result, the sensitivities of all assays with this contrived population were comparable.

Interestingly, in our study, Fusion was positive for all but one virus previously detected in these samples, even though this collection has been used in multiple studies and has gone through multiple freezethaw cycles. There were 12 additional Fusion positive samples, which were confirmed true positive by analysis on Pro using higher sample input volumes. Indeed, the need for more sensitive methods for discrepancy analysis for Fusion has been demonstrated by others [20] . This leaves 21 samples not confirmed by a different method. It is not clear if this phenomenon is due to the greater analytic sensitivity of Fusion or due to true false viral detection. The superior analytical sensitivity was demonstrated in the LOD and endpoint dilution studies, in addition to the fact that the test performed exceptionally well with highly compromised samples in terms of specimen handling for maintaining viral RNA titers. Indeed, it appears the FP samples had much lower virus titers, as demonstrated by Ct values (Supplemental figure  1) ; hence, a system with superior analytical sensitivity would demonstrate better performance with such specimens. Furthermore, the additional positive viruses were only observed in samples collected during periods of high viral prevalence, for example during the 2-week period of 25% FluA incidence. False positive detection or signal would be a random event and be equally evident at periods of low viral incidence. It is also important to point out that 3 F P samples were due to RV and unfortunately, we only had one other system available in our lab to test for RV. Hence, there was no true referee of the difference in the two systems.

Although this manuscript was not a reanalysis of cIAB, Pro, RP, and Xpt, the reclassification of some samples as TP would result in changes in the sensitivities and specificities of these other test systems. In particular, for the consecutive samples from 2015, the specificities of these other assays would improve to 100% from our previous publication [7] . However, the sensitives would decrease by 4 to 10%. It is also important to point out the choice of reference method in the absence of a gold standard has a significant impact on test performance. If this study involved submission of a diagnostic test to the Food and Drug Administration (FDA), a majority criteria would be required for a non-reference standard and a resolver test would not be allowed for the determination of positive and negative agreement, so as to prevent the introduction of bias in favor of the new test [21] . However, it is also known that this approach is biased against a new test with superior analytical sensitivity and in such situations a CRS is recommended [11] .

Another limitation with this study is the fact that the resolver method does not test another target. However, our need was to have a resolver that more closely matches the LOD of the Fusion. Having an assay with a low LOD is not an easy feat; hence, it's not easily accomplished with another gene target for the sake of a less biased resolver. Pro was chosen as the resolver because 1) it had LOD lower than any of the other commercial assays [10] and 2) it can be modified to increase the effective testing volume. Likewise, Sanger sequencing which is often used as a resolver, it doesn't have a low LOD. For example, HIV genotyping can only be performed on patients with viral loads of 1000 copies/ml. Indeed, our lab was only able sequence the influenza matrix gene for samples with low Ct values [10] .

Besides the exquisite sensitivity seen with Fusion, the system has numerous other benefits (Table 6 ). It is a fully automated, high throughput system with on-demand testing capabilities. The strategy used by Hologic is for a modular approach to syndromic testing, with 3 respiratory assays for 10 targets. However, it is lacking subtyping for FluA strains. Another plus is that the reagents are stable on-board for up to 60 days; however, refrigeration is needed for assay cartridge storage. In summary, Fusion offers exquisite sensitivity for the detection of respiratory viruses on a fully automated platform. There are many options for respiratory virus testing, each with their own niche. Fusion appears to be the leader with regards to sensitivity, protections against genomic drift, while maintaining high throughput and minimal hands on time.

Reagents were provided by Hologic, Inc. (USA). The previously published study used for comparison was sponsored in part by Roche Diagnostics (USA).

Studies were performed in accordance with IRB requirements at Albany Medical Center. 

The authors' institution received several research grants from Abbott Molecular, BD, BioMerieux, Cepheid, Hologic and Roche Diagnostics. Dr. Stellrecht received reimbursement of travel expenses for attending meetings and conferences from Abbott Molecular, Hologic, Quidel and Roche Diagnostics. Lacks subtyping for influenza A strains Modular approach to syndromic testing, the same extract can be used for 1 to 3 respiratory assays

The storage capacity for the reaction tubes does not meet the needs of the maximum specimen capacity Highly automated with little hands on time Controls cannot be tested prior to specimen testing; at least one specimen or a blank must be tested concurrently On demand testing Process up to 335 respiratory tests in 8 hrs. Multiple PCRs and TMAs on board; can run up to 12 different protocols at one time Reagents stable on-board for up to 60 days. Reagent and consumable tracking with advance warning if more is needed and when it is needed

Rapid antigen tests for influenza: rationale and significance of the FDA reclassification

Update on influenza diagnostics: lessons from the novel H1N1 influenza A pandemic

Diagnostic accuracy of novel and traditional rapid tests for influenza infection compared with reverse transcriptase polymerase chain reaction: a systematic review and meta-analysis

Clinical practice guidelines by the infectious diseases society of america: 2018 update on diagnosis, treatment, Chemoprophylaxis, and institutional outbreak management of seasonal influenza

Influenza A virus drift variants reduced the detection sensitivity of a commercial multiplex nucleic acid amplification assay in the season

Severe sensitivity loss in an influenza A molecular assay due to antigenic drift variants during the 2014/15 influenza season

Effect of genomic drift of influenza PCR tests

Incidence of matrix genes mutations affecting PCR tests among influenza H3N2 clades circulating during the 2014/15 season

History of matrix genes mutations within PCR target regions among circulating influenza H3N2 clades over ten-plus-years

The drift in molecular testing for influenza: mutations affecting assay performance

Methods and recommendations for evaluating and reporting a new diagnostic test

Molecular testing for respiratory viruses

Evaluation of three influenza A and B real-time reverse transcription-PCR assays and a new 2009 H1N1 assay for detection of influenza viruses

Comparison of the FilmArray Respiratory Panel and Prodesse real-time PCR assays for detection of respiratory pathogens

Evaluation of the Cepheid Xpert Flu Assay for rapid identification and differentiation of influenza A, influenza A 2009 H1N1, and influenza B viruses

Clinical accuracy of a PLEX-ID flu device for simultaneous detection and identification of influenza viruses A and B

Influenza and respiratory syncytial virus detection in clinical specimens without nucleic acid extraction using FOCUS direct disc assay is substantially equivalent to the traditional methods and the FOCUS nucleic acid extraction-dependent RT-PCR assay

Comparison of three commercial RT-PCR systems for the detection of respiratory viruses

Comparison of six sampleto-Answer influenza A/B and respiratory syncytial virus nucleic acid amplification assays using respiratory specimens from children

Evaluation of performance characteristics of panther fusion assays for detection of respiratory viruses from nasopharyngeal and lower respiratory tract specimens

FDA, Statistical Guidance on Reporting Results From Studies Evaluating Diagnostic Tests

We thank the Molecular Diagnostics staff for their help with collecting respiratory specimens and performing original testing. This study was presented in part at the 2018ASM Clinical Virology Symposium.