Multicenter Clinical Evaluation of the Luminex Aries Flu A/B & RSV Assay for Pediatric and Adult Respiratory Tract Specimens

ABSTRACT Influenza A and B viruses and respiratory syncytial virus (RSV) are three common viruses implicated in seasonal respiratory tract infections and are a major cause of morbidity and mortality in adults and children worldwide. In recent years, an increasing number of commercial molecular tests have become available to diagnose respiratory viral infections. The Luminex Aries Flu A/B & RSV assay is a fully automated sample-to-answer molecular diagnostic assay for the detection of influenza A, influenza B, and RSV. The clinical performance of the Aries Flu A/B & RSV assay was prospectively evaluated in comparison to that of the Luminex xTAG respiratory viral panel (RVP) at four North American clinical institutions over a 2-year period. Of the 2,479 eligible nasopharyngeal swab specimens included in the prospective study, 2,371 gave concordant results between the assays. One hundred eight specimens generated results that were discordant with those from the xTAG RVP and were further analyzed by bidirectional sequencing. Final clinical sensitivity values of the Aries Flu A/B & RSV assay were 98.1% for influenza A virus, 98.0% for influenza B virus, and 97.7% for RSV. Final clinical specificities for all three pathogens ranged from 98.6% to 99.8%. Due to the low prevalence of influenza B, an additional 40 banked influenza B-positive specimens were tested at the participating clinical laboratories and were all accurately detected by the Aries Flu A/B & RSV assay. This study demonstrates that the Aries Flu A/B & RSV assay is a suitable method for rapid and accurate identification of these causative pathogens in respiratory infections.

Respiratory viruses such as influenza virus and respiratory syncytial virus (RSV) are the most common cause of acute illness and physician visits in the United States (7). The frequency of respiratory viral infections is highest in children under 4 years of age. School-age children average 5 to 8 respiratory virus infections per year, and adults average 2 to 4 infections per year (8,9). In the United States, an estimated 200,000 people are hospitalized annually due to influenza alone, with more than 36,000, typically the elderly and immunocompromised, dying from the disease each year (10).
Traditional laboratory tests, such as culture, serology, and direct immunofluorescence, have been used to identify the infectious pathogens. However, the slow results from culture and the poor reliability of rapid immunoassay-based respiratory tests present challenges to clinicians and hospitals (1). Molecular diagnostic tests such as PCR methods offer high sensitivity, speed, and cost advantages over traditional methods, such as culture or direct immunofluorescence (DFA), and have been the focus of clinical laboratories for more than a decade. Rapid and early diagnosis of the causative pathogen in respiratory illness aids in patient diagnosis, treatment management, and the avoidance of overprescribing antibiotics.
The Aries Flu A/B & RSV assay (Luminex Corporation) is a real-time PCR-based in vitro diagnostic test for the identification of influenza A virus, influenza B virus, and respiratory syncytial virus (RSV) in nasopharyngeal swabs (NPS) from patients with suspected respiratory tract infections. The assay is used with the Aries instrument, a clinical multiplex test system that automates nucleic acid preparation from a clinical sample, performs real-time PCR detection, and reports multiple test results from the assay. The NPS specimen is introduced into the Aries Flu A/B & RSV assay cassette by pipetting, and the cassette is placed into an Aries system module via a cassette magazine. Within the assay cassette, the specimen is lysed and the nucleic acids are extracted along with the sample processing control (SPC) present in the cassette. The extracted nucleic acids are transferred to lyophilized PCR reagents contained within the PCR tube attached to the cassette.
The assay uses target-specific fluorescently labeled primer pairs to amplify nucleic acid sequences from influenza A/B virus and RSV found in the sample, in addition to amplifying assay internal control sequences. The Aries assay chemistry is based on an expanded genetic alphabet technology, using the synthetic DNA base pair 2=-deoxy-5-methyl-isocytidine (iC) · 2=-deoxyisoguanosine (iG). The isobases (iC and iG) pair specifically with each other and not with natural nucleotides and are efficiently incorporated during PCR (11,12). During PCR amplification, a quenchermodified iGTP is incorporated by the polymerase opposite to an iC and a fluorophore reporter attached to the PCR primer. If the target is present and is amplified, the assay fluorescence decreases with every cycle as the amplification product accumulates. The decrease in assay fluorescence is monitored in real time on the Aries instrument.
Reported herein are the results of a multicenter clinical performance study that evaluated the Aries Flu A/B & RSV assay over a 2-year period on 2,479 pediatric and adult subjects with suspected respiratory infections presenting at four different North American clinical institutions. The clinical performance of the Aries Flu A/B & RSV assay for detecting influenza A virus, influenza B virus, and RSV was determined in comparison to that of the FDA-approved xTAG respiratory viral panel (RVP).

RESULTS
A total of 2,504 NPS specimens from subjects suspected of having respiratory tract infections were collected in the prospective study. Twenty-five specimens were excluded based on inclusion/exclusion criterion violation or protocol deviation, leaving a total of 2,479 eligible unique specimens for subsequent data analysis. Of these, 1,017 were collected from January to March 2015, while the remaining 1,462 specimens were obtained between November 2015 and February 2016. Table 1 summarizes the general demographic information from the subjects that were included in the data analysis. Among the 2,479 specimens included in the analysis, 47.1% were from male patients (n ϭ 1,168) and 52.9% (n ϭ 1,311) were from female patients. Most of the specimens were collected from individuals who either were hospitalized or presented to the institutions' emergency departments (52.4% and 28.9%, respectively). Outpatients represented 18.7% of the study population. A total of 1,577 (63.6%) subjects were adults 21 years of age or older, while the remaining 902 (36.4%) subjects were pediatric patients Ͻ21 years of age.
Clinical runs using the Aries Flu A/B & RSV assay were conducted between December and March 2016 on specimens that were either stored frozen at Ϫ80°C (n ϭ 1,316 [53.1%]) or kept refrigerated at 4°C to 8°C (n ϭ 1,163 [46.9%]) prior to testing. The vast majority of the frozen samples (77.3% [n ϭ 1,017]) was kept stored for 3 to 12 months, while 23.6% of samples were collected 1 to 3 months before testing with the Aries Flu A/B & RSV assay. A difference in sensitivity with regard to the age of frozen specimens could not be detected. Similarly, when comparing storage conditions, no significant difference between the frozen and refrigerated sample results could be seen (data not shown).
Valid Aries Flu A/B & RSV assay results (i.e., positive or negative) were obtained on the first attempt from 2,458 specimens (  It is important to note that predictive values should not be considered intrinsic to any test, as they vary depending on the disease prevalence. Although the PPV of the Aries assay for influenza B virus was lower than those for influenza A virus and RSV, the prevalence of this virus (1.9%) was also observed to be lower in the study population than for influenza A virus (12.6%) and RSV (11.2%). This had a significant impact on PPV despite the fact that the false positivity rate of the assay for influenza B virus was only 0.6% in this evaluation (14/2,479).
Twenty-four specimens were identified as positive by the comparator method but negative by the Aries Flu A/B & RSV assay (i.e., false negative). Of these, 10 specimens (41.7%) were confirmed as negative by bidirectional sequencing analysis. These included 7 for influenza A virus, 2 for influenza B virus, and 1 for RSV. There were also 84 specimens that were identified as negative by the reference method but positive by the Aries Flu A/B & RSV assay (i.e., false positive). Of these, 39 specimens (46.2%) were confirmed as positive by sequencing analysis. These included 4 for influenza A virus, 3 for influenza B virus, and 32 for RSV.
Because the prevalence of influenza B virus was lower than those for influenza A virus and RSV in the study population, the prospective sample set was supplemented with archived specimens. A total of 40 unique preselected specimens positive for influenza B virus were included in this supplemental evaluation. The specimens were tested in a blind

DISCUSSION
A rapid detection of viral pathogens causing a respiratory infection provides physicians a number of advantages in treating patients and managing outbreaks. In addition to confirming an infection's viral basis and avoiding unnecessary antibiotics, differential diagnosis of respiratory infections enables the isolation of hospital patients, a reduction in the length of stay for minor infections, better control of outbreaks and hospital transmission rates, and more cost-effective clinical outcomes (13,14).
Traditional laboratory tests have limitations in the hospital setting, including a slow time-to-result, poor assay performance, and high cost (1). Without rapid test results, hospitalized patients either are not immediately isolated or are unnecessarily isolated as a precaution, potentially creating inefficiencies and increased health care costs. A poor sensitivity or specificity of an assay results in either inaccurate diagnoses or a lack of confidence in results. High assay costs make certain laboratory tests unfeasible and shift the financial burden to other aspects of clinical care, such as the decision to risk a potential outbreak or to isolate patients merely as a precaution (15).
The advent of rapid molecular diagnostic tests for respiratory viruses allows clinicians to gain diagnostic insight early enough in treatment to immediately and effectively impact clinical outcomes. However, such tests are only as effective as their proven assay performance, providing confidence and reliability for making the best possible clinical care decisions. This multicenter clinical study demonstrated robust performance of the Aries Flu A/B & RSV assay in subjects suspected of having respiratory tract infections. The primary metrics addressed by this study are the clinical sensitivity and specificity compared with those from existing molecular diagnostics for these pathogens. Other assay performance characteristics such as sample-to-answer time, ease of use, and a reduction of errors were evaluated in terms of their impact on the usage of the assay in the clinical setting.
The Aries system is a standalone sample-to-answer platform that does not require additional equipment or computing power, minimizing the system's footprint on the laboratory bench. The system is capable of running cassette-based assays, including both in vitro diagnostic (IVD) assays and user-defined protocols. With onboard nucleic acid extraction, internal controls, and barcode scanning, the instrument is designed to minimize laboratory errors. The Flu A/B & RSV assay requires less than 10 min of hands-on time, and results are available in approximately 2 h of run time for up to 12 samples per instrument, allowing assay results for all 3 pathogens to be reported for up to 48 patient samples per 8-hour shift.
Reliability is an essential component when choosing the right assay for any clinical setting. Unwanted repeat runs due to invalid results not only disrupt the laboratory workflow but can also lead to extended turnaround times or even cancellation of testing if insufficient sample material is available for reruns. This is especially important during seasonally driven high volumes and for decision making for triaging. The Aries Flu A/B & RSV assay demonstrated a remarkably low value of 0.8% (21/2,479) of invalid results. All of those 21 results needed to be repeated, but provided reliable results upon repeat. Similar or inferior reliability data are seen with comparable assays; GenMark Diagnostics eSensor RVP showed a 1.5% invalid rate, the Verigene RVϩ Nanosphere showed 9.7%, FilmArray RP (BioFire Diagnostics) showed 3% and 1%, and Prodesse Pro Fluϩ showed 5.7% (16)(17)(18). xTAG RVP (Luminex) performed with a 1.1% invalid rate in this study and 2% elsewhere (19). Even the point-of-care approved assay Alere i showed invalid rates of 4.6% and 5.1% in two studies, with the inability of reruns due to an insufficient quantity of specimen (20,21 (RUO) to that of laboratory-developed reverse transcriptase PCR (RT-PCR) assays (LDA), immunochromatographic assays, and a direct fluorescent antibody. Clinical sensitivities were 93.3 to 98.6% and clinical specificities were 100% compared with those of LDA (23).
Limitations with this study include the chosen time for specimen collection that led to the inferior amount of influenza B specimens, as they had just begun to appear in each season and subsequently demanded a further collection process. Furthermore, due to the inability to definitively identify influenza A H1N1 pdm09 strains by both assays, namely, the Aries Flu A/B & RSV and the xTAG RVP, definitive conclusions about the specific performance in the identification of different influenza A strains that are predominant during the seasons of testing cannot be drawn.
In conclusion, the Aries system and the Aries Flu A/B & RSV assay provide a rapid reliable diagnostic tool for hospitals and clinical labs needing to identify viral pathogens earlier in the treatment continuum. The integrated sample-to-answer process enables clinicians to make confident decisions about patient isolation, care, and treatment that could potentially improve clinical outcomes and reduce health care costs.

MATERIALS AND METHODS
Institutional and ethics reviews. The present study was conducted under institutional review board (IRB) or research ethics board (REB) approval at all participating sites.
Prospective specimen collection, processing, and testing. Specimens included in this clinical study included leftover nasopharyngeal swabs (NPS) prospectively collected during the 2014 to 2015 and 2015 to 2016 influenza seasons. Clinical specimens were collected from pediatric or adult patients suspected of having respiratory tract infections who presented at four geographically diverse clinical sites located in North America. Clinical sites were located in the eastern United States (Northwell Health Laboratories, Lake Success, NY), central United States (Department of Pediatrics, Washington University, St. Louis, MO), southwestern United States (Baylor Scott and White Health, College of Medicine, Temple, TX), and Canada (Department of Pathology and Molecular Medicine, St. Joseph's Healthcare, Hamilton, Ontario, Canada). Respiratory sample collection was performed with each institution's standardized NPS collection kit and in accordance with their individual procedures. All clinical specimens were submitted fresh to the institutions' clinical laboratories and were processed as per their standard procedures and as ordered by the referring physician. Specimens that were not properly collected, labeled, transported, stored, or received timely and in good condition were excluded from the study.
Upon receipt, any leftover specimen that met these study inclusion/exclusion criteria was provided in a blind manner to an individual at the site who was not directly involved in the study. Each eligible specimen was then divided into multiple aliquots. One aliquot of the specimen was kept at the clinical sites for Aries Flu A/B & RSV assay testing. Another aliquot of the specimen was shipped to a centralized testing facility (Luminex Molecular Diagnostic, Toronto, Ontario, Canada) for comparator xTAG RVP testing. Two additional specimen aliquots were generated and stored frozen at Ϫ80°C for either repeat testing or discordance analysis.
Banked specimen collection and testing. Banked specimens were stored frozen at Ϫ80°C over a period of 12 months prior to testing. Due to the lower prevalence of influenza B virus observed in the prospective study, the sample set was supplemented with banked (preselected) influenza B-positive specimens derived from both seasons, namely, 2014 to 2015 and 2015 to 2016. These specimens were collected at a single clinical laboratory (collection site) located in Canada (Department of Pathology and Molecular Medicine, St. Joseph's Healthcare, Hamilton, Ontario, Canada). The presence of the expected pathogen (influenza B virus) in each of the preselected specimens was confirmed by xTAG RVP testing. To minimize bias, preselected positive specimens (n ϭ 40) were distributed and tested along with an equal number of unique negative specimens in a randomized double-blind fashion at the four testing sites that were involved in the prospective study. Both positive and negative results were included in the final analysis.
Comparator testing with the xTAG RVP assay. An aliquot of each specimen was extracted using the NucliSENS easyMAG method (bioMérieux, Inc.), and the total nucleic acid isolates were tested by the xTAG RVP assay (Luminex Molecular Diagnostics, Toronto, Ontario, Canada) at a centralized testing facility. xTAG RVP testing was performed by trained personnel in accordance with the manufacturer's instructions provided in the kit package insert and within sample stability claims.
Aries Flu A/B & RSV assay testing. All specimens were assessed by the Aries Flu A/B & RSV assay (Luminex Corporation, Austin, TX) by one or two trained operators at each of the clinical sites. These operators were blind to the routine clinical results to minimize potential bias. Assay runs and allowable reruns were performed in accordance with the manufacturer's instructions. Briefly, assay cassettes were removed from the pouch and capped and the foil seal was removed. Sample tubes and assay cassettes were placed in the Sample Prep tray. The samples were vortexed briefly and 200 l from each was pipetted into the assay cassettes. The cassette caps were closed and the loaded cassettes were placed into the cassette magazine. The loaded magazine was inserted into one of two modules in the Aries instrument and the run was started. In the first phase of the prospective study (2014 to 2015 influenza season), specimens were collected and then stored frozen at Ϫ80°C prior to being tested with the Aries Flu A/B & RSV assay. Most clinical specimens collected during the second phase of the prospective study (2015 to 2016 influenza season) were tested by the Aries Flu A/B & RSV assay at the clinical sites after being kept refrigerated at 4 to 8°C for up to 96 h after collection.
Discordant analysis with bidirectional sequencing. Any discordant specimens where the Aries Flu A/B & RSV assay results were different from the comparator xTAG RVP assay results were assessed by bidirectional sequencing using analytically validated primers that were directed against genomic regions different from those for the Aries Flu A/B & RSV assay. Discordant specimens were subjected to bidirectional sequencing using M13 forward and reverse primers and the Sanger dideoxy sequencing method to retrieve DNA sequences. Following extraction by the bioMérieux NucliSENS easyMAG extraction method, specimens were subjected to RT-PCR using the OneStep Qiagen reverse transcriptase PCR kit (Qiagen, Hilden, Germany). Amplicons were then treated with exonuclease I and shrimp alkaline phosphatase enzymes to remove unincorporated primers and deoxynucleoside triphosphates (dNTPs) left over from the PCRs. Dye-labeled terminator cycle sequencing reactions were performed using the BigDye Terminator v3.1 cycle sequencing kit (Thermo Fisher, Waltham, MA). Any unincorporated dye was removed using the BigDye Xterminator purification kit (Thermo Fisher). Sample electrophoresis and sequencing analysis were performed on the 3730xl analyzer (Thermo Fisher) using the 3730xl data collection software (v 3.1.1) and sequencing analysis software (v 5.4). Sequences that (i) were at least 200 bases in length, (ii) had a PHRED score greater than or equal to 20 for at least 90% of the bases, and (iii) contained fewer than 5% ambiguous base calls were considered for further analysis using BLAST (NCBI). Acceptable matches to BLAST reference sequences were those with greater than 95% query coverage and identity with an expected value (E value) less than 10 Ϫ30 compared with the reference sequence.
Data collection and analysis. The performance of the Aries Flu A/B & RSV assay for detecting influenza A virus, influenza B virus, and RSV was compared with that of the xTAG RVP assay. Accuracy determinations for each viral pathogen were based on the fraction of positive (or negative) results by the comparator method, which were also positive (or negative) by the Aries Flu A/B & RSV assay. An Aries Flu A/B & RSV assay result was considered to be a true-positive (TP) or true-negative (TN) result if it agreed with the comparator method result for the pathogen in question. The exact (Clopper-Pearson) method was used to calculate 95% confidence intervals. Estimates of sensitivity (or positive percent agreement [PPA]), specificity (or negative percent agreement [NPA]), the positive predictive value (PPV), and the negative predictive value (NPV) for each pathogen were calculated based on two-by-two tables (reference method result versus result from Aries Flu A/B & RSV) for the entire prospective data set.