From Triplex to Quadruplex: Enhancing CDC’s Respiratory qPCR Assay with RSV Detection on Panther Fusion® Open Access™

1. Introduction

The coronavirus disease 2019 pandemic (COVID-19) has dramatically altered the epidemiology of other respiratory viruses. Between 2020 and 2021, the traditionally endemic influenza A (IAV), influenza B (IBV), and respiratory syncytial virus (RSV) were virtually absent due to public health measures aimed at containing the spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2; SC2) [1,2]. In the post-pandemic period, as previously anticipated [3], influenza, SC2, and RSV infections represent a global public health threat due to their high transmissibility, morbidity, recurrent seasonal waves, and economic burden [4,5].

The overlapping symptoms of respiratory infections pose an additional challenge to healthcare systems by increasing the risk of complications and hindering therapeutic strategies [6,7]. These findings underscore the need to strengthen surveillance, diagnosis, and prevention systems, particularly for vulnerable populations, including young children, older adults, and individuals with comorbidities [8]. In this regard, the rapid, specific, and simultaneous detection of respiratory pathogens is a key strategy to guide clinical management and support control efforts [9]. Improved timeliness of diagnosis and multiplex detectability can also contribute to reducing antibiotic use [10], limiting the need for additional laboratory tests [11], and optimizing cost-effectiveness [12].

Although multiple diagnostic approaches exist, nucleic acid amplification tests (NAATs)—particularly multiplex RT-qPCR—are considered the clinical gold standard due to their sensitivity and specificity and their ability to detect multiple targets in a single reaction [13,14,15,16,17,18]. In 2021, the U.S. Centers for Disease Control and Prevention (CDC) introduced the CDC Flu-SC2 assay for simultaneous detection of influenza A, influenza B, and SC2 [19]; however, it does not include RSV, and its semi-automated workflow requires manual handling. This study aimed (i) to develop an RSV A/B RT-qPCR assay (LDT RSV) compatible with CDC Flu-SC2 chemistry, and (ii) to validate a fully automated laboratory-developed respiratory assay (LDRA) on the Panther Fusion® Open Access™ system for simultaneous detection of influenza A, influenza B, SC2, and RSV in nasopharyngeal swabs from patients with suspected respiratory infection.

2. Materials and Methods

2.1. Clinical Samples, Ethical Aspects, and Eligibility

This single-center study was conducted by the Research and Development (R&D) Group of the Molecular Biology Department at Referencia Laboratorio Clínico (Santo Domingo Oeste, Dominican Republic). We used residual material from 405 nasopharyngeal swabs collected in Universal Transport Medium (UTM; Copan, CA, USA) from patients with clinician-ordered molecular testing for respiratory infection by IAV, IBV, SC2, and/or RSV. Specimens were randomly selected between October 2023 and September 2025. In accordance with laboratory confidentiality and informed-consent policies, all patients consented to research use of residual specimens and publication of results with privacy safeguards; accordingly, all samples were de-identified prior to analysis.

Pre-analytical acceptance/rejection criteria were applied according to the laboratory standard operating procedures (SOPs) for respiratory NAAT testing. Specimens were rejected if they: (i) were mislabeled or had mismatched identifiers; (ii) were collected in an inappropriate tube/transport medium; (iii) were in a leaking or broken container; (iv) had insufficient specimen volume; (v) were stored or transported outside SOP specifications; (vi) had gross specimen quality issues likely to compromise NAAT performance (e.g., obvious contamination); or (vii) had no or invalid test prescription. All residual clinical samples included in this evaluation satisfied the laboratory’s acceptance criteria. No demographic eligibility criteria were applied. Residual specimens were stored at −70 °C immediately after routine testing.

2.2. Assay Design and Optimization

2.2.1. Oligo Design

Oligonucleotides for LDT RSV were designed as previously described [20,21]. In brief, 8,661 complete RSV sequences (types A and B) were retrieved from the NCBI Virus database (https://www.ncbi.nlm.nih.gov/labs/virus/, accessed on September 2025) and aligned with MAFFT algorithm in Unipro UGENE v48.0 (Unipro, Novosibirsk, Russia). After alignment validation, redundancy removal, and reduction to 8,100 sequences, a conserved region within the RSV Matrix (M) gene was selected by sequence-entropy analysis (Supplementary Figures S1 and S2, based on Supplementary Files S1–S4). Using the alignment-derived consensus sequence as reference, three primers and one TaqMan probe were designed employing Oligo v7.60 (Molecular Biology Insights Inc., CO, USA), Benchling™ (Benchling, CA, USA), and OligoAnalyzer™ (Integrated DNA Technologies, IA, USA) (Figure 1). The probe was labeled at the 5′ end with cyanine 5 (Cy5) fluorophore, phosphorylated and labeled at the 3′ end with the Iowa Black® RQ dark quencher (IAbRQSp), and contained an internal TAO™ quencher at the ninth nucleotide (5′→3′). The LDT RSV oligos were designed to minimize heterodimer formation with the CDC Flu-SC2 oligos [19]. To improve resilience to mutations in circulating SC2 variants, the CDC SARS-CoV-2 detection probe was modified. All oligos comprising the quadruplex LDRA assay (CDC Flu-SC2 + LDT RSV) are provided in Table 1.

All oligonucleotides were synthesized and purified by HPLC by Integrated DNA Technologies (IDT, IA, USA). Optimal purity was defined as ≥85%. All oligonucleotides were shipped lyophilized.

2.2.2. In Silico Specificity Evaluation

In silico specificity of the RSV primers and probe was assessed as previously described [21] by evaluating homology, exclusivity (cross-reactivity), and potential amplification of human background DNA. Homology was assessed by open NCBI BLAST tool searches (https://blast.ncbi.nlm.nih.gov, accessed on September 2025). For 100% coverage, high-homology sequences were defined as ≥95% identity and an expectation value (E-value) ≤10^–2 [22]. Exclusivity was assessed by closed searches against sequences from ≥68 clinically and biologically relevant respiratory microorganisms (Supplementary Table S1); low homology (and thus low cross-reactivity likelihood) was defined as E-value >10^–2 irrespective of coverage or identity [21].

Potential amplification of human DNA was evaluated with MFEprimer version 3.1 [23] (https://mfeprimer3.igenetech.com/ or https://m4.igenetech.com/, accessed on September 2025). The following parameters were set: “Background Database”, “UCSC - Homo sapiens - hg38 – Genome”; “Allow mismatch at the 3’ end”, “No” (default); “Tm min”, “48 °C” (minimum Tm required for binding stability between the primer and its binding sites: T_a – 10 °C); “Concentration of divalent cations (usually MgCl₂)”, “4.0 mM”; “Annealing oligo concentration”, “800 nM” (optimized primer concentration); “Product size”, from “0” to “1,000” bp.

Inclusivity was estimated by design-coverage analysis using SCREENED v1.0 tool [24] (Sciensano Galaxy External; https://galaxy.sciensano.be/, accessed on September 2025) with all sequences available in the NCBI Virus (https://www.ncbi.nlm.nih.gov/labs/virus/, accessed on September 2025) and GISAID EpiRSV™ (https://gisaid.org/, accessed on September 2025): 3,705 RSV A and 3,178 RSV B sequences from NCBI Virus, and 11,752 RSV A and 13,910 RSV B sequences from GISAID. Sequences with missing or ambiguous information were removed.

2.2.3. Synthetic Controls: Ultramer™ Duplex

Synthetic Ultramer™ duplex oligonucleotides (≤200 bp) containing the target regions for IAV, IBV, SC2, RSV A, and RSV B were used as amplification controls during assay optimization (Supplementary Table S2). Controls were synthesized and shipped lyophilized (4 nmol) by Integrated DNA Technologies (IDT, CA, USA) after standard desalting.

2.2.4. Ultramer™ Duplex Control Preparation: Stock Solution

Ultramer™ duplex controls were reconstituted and stock solutions prepared as previously described [21]. Briefly, each 4 nmol control was reconstituted in 1 mL IDTE buffer (10 mM Tris, 0.1 mM EDTA), pH 8.0 (IDT, IA, USA). Thirteen additional 1:10 serial dilutions were prepared in the same buffer, yielding 14 concentration levels per control (stock solution panel: 2.41 × 10¹⁵ to 2.41 × 10² copies/mL).

2.2.5. Assay Optimization

Physicochemical conditions for each monoplex assay composing the quadruplex LDRA were optimized as previously described [20,21]. Salt concentrations (KCl and MgCl₂) were optimized first, followed by primers/probe concentrations. PCR-grade betaine (5 M; Sigma-Aldrich, MO, USA) and annealing temperature were then optimized using a 10 °C gradient (54–64 °C: 54 °C, 54.7 °C, 56 °C, 57.9 °C, 60.4 °C, 62.3 °C, 63.5 °C and 64 °C). All other cycling parameters were kept at the default Panther Fusion® Open Access™ RNA amplification standard protocol so that the critical amplification time (55 min) was not exceeded.

Primer and probe reconstitution solutions (PPR) for optimization experiments were prepared in 8 mM Tris-HCl buffer. Optimization runs used the off-label Panther Fusion® Open Access™ RNA/DNA Enzyme Cartridge (Hologic, CA, USA) protocol on a CFX96-IVD thermal cycler (Bio-Rad, CA, USA) and Ultramer™ duplex controls at a nominal concentration of 2,41 × 10⁵ copies/mL (Ct ≈ 30). The optimized thermocycling program was: 1 cycle at 46 °C for 8 min, 1 cycle at 95 °C for 2 min, and 45 cycles of 95 °C for 5 s and 58 °C for 21 s (fluorescence read). MyAccess™ Software v2.1.2.1 (Hologic, CA, USA) was used to implement the LDT protocol on the Panther Fusion® system; the final Open Access™ protocol is provided in Supplementary File S5.

Supplementary Table S3 summarizes the optimized PPR composition for the quadruplex LDRA.

2.2.6. Amplification Efficiency and Multiplex Compatibility

Amplification efficiency (E) and multiplex compatibility were evaluated post-optimization as previously described [21]. Ten-fold serial dilutions (2,41 × 10⁹ to 2,41 × 10² copies/mL) of each synthetic Ultramer™ duplex control were tested in triplicate. The fluorescence threshold was set at approximately the midpoint of the linear phase of the amplification curves. Standard curves were generated by plotting cycle threshold (Ct) values versus log₁₀(input concentration; C_i) using Microsoft® Excel (Microsoft Corp., WA, USA), and E and its 95% confidence intervals (95% CI) were calculated from the regression slope (m) using the following equation.

E = (10^(-1/m) - 1) × 100%.

(1)

E was considered acceptable for values between 90% and 110% with goodness of fit (R²) ≥ 0.98.

For multiplex compatibility, Ct values and E between monoplex and multiplex conditions were computed. Multiplex compatibility was considered acceptable when each monoplex–multiplex pair: i) the |ΔCt| ≤ 0.5; ii) the E acceptability criterion described above was satisfied; and iii) no indications of significantly compromised sensitivity in the multiplex reaction (e.g., failure or inefficient amplification) were observed.

For these purposes, the off-label protocol of the Panther Fusion® Open Access™ RNA/DNA Enzyme Cartridge was used on the CFX96-IVD analyzer.

The E per target was verified under the definitive conditions of the quadruplex LDRA assay on the Panther Fusion® Open Access™ system.

2.2.7. Panther Fusion® Open Access™ RNA/DNA Enzyme Cartridge: Off-Label Protocol

Briefly, the wells of the Panther Fusion® Open Access™ RNA/DNA Enzyme Cartridge containing the enzyme aliquots were punctured using a nuclease-free punching tool. The lyophilized enzyme was reconstituted with 20 μL of PPR, incubated at room temperature (18–25 °C) for 1–5 min, and mixed by vortexing in short, repeated pulses to minimize foaming. Amplification and detection were performed on the CFX96-IVD analyzer using 96-well PCR plates in a 25 μL reaction volume (20 μL master mix (MMX) + 5 μL DNA template).

2.3. Analytical Performance

2.3.1. Analytical Sensitivity (Limit of Detection)

The limit of detection (LoD) was estimated using dilution panels prepared from the five synthetic Ultramer™ duplex controls. Starting from stock solutions at 2.41 × 10⁷ copies/mL, each control was first diluted to a working concentration of 6.4 × 10³ copies/mL in a UTM/Panther Fusion® Specimen Lysis Tube (SLM) medium matrix. UTM from negative nasopharyngeal swabs was mixed with SLM (Hologic Inc., CA, USA) at the manufacturer-recommended proportion (0.5 mL UTM per 0.71 mL SLM) [25] to generate the UTM/SLM.

Each LoD panel consisted of seven additional 1:2 serial dilutions prepared from the working solution in UTM/SLM (final volume, 1,000 µL), yielding eight 1:2 dilution levels ranging from 6.4 × 10³ copies/mL (2.3 × 10² copies/reaction) to 5.0 × 10¹ copies/mL (1.8 × 10⁰ copies/reaction). These concentrations correspond to 1.5 × 10⁴ to 11.2 × 10² copies/mL in the primary UTM specimen (prior to mixing with SLM).

Copies/mL in UTM/SLM were converted to copies/reaction using the equation shown, assuming: 360 µL sample aspirated, 50 µL elution volume, 5 µL eluate per amplification reaction, 100% extraction recovery, and 1000 as the conversion factor from mL to µL.

$$\begin{gathered} \mathrm{C}\left(\mathrm{PCR}\right) [\mathrm{copies}/\mathrm{reac}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}]=\mathrm{C}(\mathrm{UTM}/\mathrm{SLM}) [\mathrm{cop}\mathrm{ie}\mathrm{s}/\mathrm{mL}] / 1000 \times 360 / 50 \times 5 \\ \mathrm{C}\left(\mathrm{PCR}\right) [\mathrm{copies}/\mathrm{reac}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}]=\mathrm{C}(\mathrm{UTM}/\mathrm{SLM}) [\mathrm{copies}/\mathrm{mL}] \times 0.036 \end{gathered}$$

(2)

Where, C(PCR) [copies/reaction] is the number of copies of the target sequence in the PCR reaction; C(UTM/SLM) [copies/mL] is the concentration of the target sequence, in copies/mL, in the UTM/SLM.

Equivalent concentrations in the primary UTM specimen were calculated by volumetry using the stated mixture volumes (V(UTM/SLM) = 1.21 mL; V(UTM) = 0.5 mL).

$$\begin{gathered} \mathrm{C}\left(\mathrm{UTM}\right) [\mathrm{copies}/\mathrm{mL}]=\mathrm{C}(\mathrm{UTM}/\mathrm{SLM}) [\mathrm{copies}/\mathrm{mL}] \times \mathrm{V}(\mathrm{UTM}/\mathrm{SLM}) \left[\mathrm{mL}\right] / \mathrm{V}\left(\mathrm{UTM}\right) \left[\mathrm{mL}\right] \\ \mathrm{C}\left(\mathrm{UTM}\right) [\mathrm{copies}/\mathrm{mL}]=\mathrm{C}(\mathrm{UTM}/\mathrm{SLM}) [\mathrm{copies}/\mathrm{mL}] \times 1.21 \mathrm{mL} / 0.5 \mathrm{mL} \\ \mathrm{C}\left(\mathrm{UTM}\right) [\mathrm{copies}/\mathrm{mL}]=\mathrm{C}(\mathrm{UTM}/\mathrm{SLM}) [\mathrm{copies}/\mathrm{mL}] \times 2.42 \end{gathered}$$

(3)

Where, C(UTM) [copies/mL] is the concentration, in copies/mL, of the target sequence in the primary UTM specimen; C(UTM/SLM) [copies/mL] is the concentration of the target sequence, in copies/mL, in the UTM/SLM.

Using a single lot of specific (PPR) and generic reagents, a single instrument, and one operator, 20 replicates of each member of the evaluation panels were analyzed, and the frequency of positive results was recorded at each concentration level. C₉₅ LoD and their 95% CI estimates were obtained by probit regression of detection probability as a function of concentration. The cutoff was set at the maximum possible threshold cycle of the employed amplification and detection protocol (Ct = 45).

2.3.2. Analytical Specificity: Cross-Reactivity (In Vitro Exclusivity) and In Vitro Inclusivity

Analytical specificity was evaluated using pathogen panels assembled from commercial quality-control materials and external quality assessment (EQA) schemes. Commercial controls/panels included NATtrol Respiratory Verification Panel 2.1 (Zeptometrix, NY, USA), Respiratory Multiplex (1 to 5) Q Control (Qnostics, Scotland, UK), Amplirun® SARS-CoV-2 RNA Controls Amplirun® SARS-CoV-2 B.1.1.7 (Alpha) RNA Control, Amplirun® SARS-CoV-2 B.1.351 (Beta) RNA Control, Amplirun® SARS-CoV-2 B.1.617.2 (Delta) RNA Control, Amplirun® SARS-CoV-2 P.1 (Gamma) RNA Control, Amplirun® SARS-CoV-2 BA.1 (Omicron) RNA Control. (Vircell, Granada, Spain), Amplirun® Total SARS-CoV-2/FluA/FluB/RSV Control (Swab) (Vircell, Granada, Spain), and AccuPlex™ H5N1 Influenza Reference Material Kit (LGC Clinical Diagnostics Inc., MA, USA). EQA materials were from CAP ID3 and IDR (CAP: College of American Pathologists; 2023 and 2024) and QCMD INFTP24 (QCMD: Quality Control for Molecular Diagnostics; 2024). Each panel member was tested in triplicate with the LDRA assay.

For cross-reactivity assessment, 30 respiratory pathogens were tested at clinically relevant concentrations (per manufacturer specifications): Adenovirus A, Adenovirus 1, Adenovirus 3, Adenovirus 14, Adenovirus 31, Bordetella parapertussis, Bordetella pertussis, Chlamydia pneumoniae, Coronavirus 229E, Coronavirus HKU-1, Coronavirus NL63, Coronavirus OC43, Enterovirus A16, Enterovirus 68, Enterovirus A71, Influenza A, Influenza B, Mycoplasma pneumoniae, Metapneumovirus 8, Metapneumovirus A2, Metapneumovirus B2, Parainfluenza 1, Parainfluenza 2, Parainfluenza 3, Parainfluenza 4, Rhinovirus 1A, Rhinovirus 16, RSV A, SARS-CoV-2, and Legionella pneumophila.

For in vitro inclusivity, 32 distinct (no duplicate) variants/strains were tested: i) 17 IAV variants/strains, including seasonal H1N1 (H1N1sea), 2009 pandemic-derived H1N1 (H1N1pdm09), H3N2, H5N1 and H7N7 subtypes; ii) four IBV variants/strains spanning Victoria and Yamagata lineages; iii) seven SC2 variants/strains including Alpha, Beta, Gamma, Delta and Omicron lineages; and iv) four RSV variants/strains spanning subtypes A and B.

2.3.3. Precision

Within-laboratory precision of the quadruplex LDRA assay was assessed using a five-member panel prepared in UTM/SLM. The panel included one negative matrix member and four doubly positive members consisting of IAV + SC2 and IBV + RSV at two concentration levels: low (30 < Ct ≤ 35) and moderate (25 < Ct ≤ 30). Only low–moderate and moderate–low pairings were evaluated for each pathogen combination. Panel pathogens were IAV, A/NY/02/09 (H1N1pdm09); IBV, B/Florida/02/06 (Victoria); SC2, USA-WA1/2020; and RSV type A, all from NATtrol Respiratory Verification Panel 2.1.

Each panel member was tested in duplicate by a single operator in one daily run on two Panther Fusion® systems, using two PPR lots, over 14 consecutive days. Panel samples were stored at 2 °C to 8 °C throughout the evaluation, and a new PPR vial from each lot was thawed daily.

Agreement with expected results was calculated for each pathogen and panel member. Within-laboratory precision, and its components (within-lot, between-lot, between-instrument, within-run/day [repeatability], and between-run/day precision), were summarized as standard deviation (SD) and coefficient of variation (CV).

2.3.4. Method Comparison: Agreement Assessment

Agreement was assessed using 405 clinical samples tested with the quadruplex LDRA assay and two commercial comparators: Panther Fusion® SARS-CoV-2/Flu A/B/RSV Assay (PFRA; Hologic Inc., CA, USA) and Allplex™ SARS-CoV-2/FluA/FluB/RSV Assay (APRA; Seegene Inc., Seoul, South Korea). Discordant specimens were retested with all assays. Persistently discordant results were adjudicated using the consensus rules described below (see Diagnostic Performance section). Specimens yielding an “invalid” result were excluded.

2.3.4.1. Allplex™ SARS-CoV-2/FluA/FluB/RSV Assay (APRA)

Clinical samples were tested with APRA, a multiplex RT-qPCR assay for simultaneous detection of SC2, IAV, IBV and RSV A/B. Viral RNA was extracted with the MagNA Pure 96 DNA and Viral NA Large Volume kit (Roche Molecular Systems Inc., NJ, USA) on the MagNA Pure 96 instrument using the “Viral NA Universal LV 4.0” protocol (500 µL input; 50 µL elution), following the manufacturer’s instructions [26]. Amplification and detection were performed on the CFX96-IVD analyzer [27] using kit-provided enzyme mixes and oligonucleotides in a 20 µL reaction volume. The assay includes endogenous (Endo IC) and exogenous (Exo IC) internal controls.

APRA targets the SC2 N, RdRp, and S genes; the IAV M gene; the IBV NS2 gene; and the RSV M gene. Results were interpreted with Seegene Viewer v3.33 (Seegene Inc., Seoul, Korea) using channels FAM (SC2 S gene and RSV), HEX (SC2 RdRp gene and IBV), Cal Red 610 (SC2 N gene and IAV), and Quasar 670 (Endo and Exo IC). Total time from extraction to results was approximately 4 hours.

2.3.4.2. LDRA and Panther Fusion® SARS-CoV-2/Flu A/B/RSV Assay (PFRA)

All clinical samples and analytical-specificity panel specimens were stabilized in SLM according to the manufacturer’s instructions [25] and stored at 2–8 °C for ≤10 days. Stabilized specimens were tested with both the candidate LDRA and the comparator PFRA per manufacturer specifications [28].

PFRA is supplied as lyophilized, ready-to-use reagents in cartridges (up to 12 tests per cartridge; up to 96 tests per box), whereas LDRA uses a user-prepared, target-specific PPR together with the generic Panther Fusion® Open Access™ RNA/DNA Enzyme Cartridge. By default, the Panther Fusion® extraction protocol aspirates 360 µL of stabilized specimen and elutes nucleic acids in 50 µL. RT-qPCR amplification/detection for both assays occurs in the Fusion® module with a 25 µL reaction volume (20 µL MMX + 5 µL eluate). Both assays used Panther Fusion® Extraction Reagent-S and Internal Control-S (IC-S; Hologic Inc., CA, USA), although IC-S is reported only by PFRA.

PFRA targets the IAV, IBV and RSV M genes and the SC2 Orf1ab gene. LDRA targets the IAV and RSV M genes, the IBV NS2 gene and the SC2 N gene. The Panther Fusion® system interprets fluorescence in channels FAM (IAV in both assays), HEX (IBV in LDRA; RSV in PFRA), ROX (SC2 in both assays), Quasar 670 (RSV in LDRA; IBV in PFRA), and Quasar 705 (internal control (IC): human RNase P for LDRA). PFRA interpretation is proprietary; LDRA was configured as: “positive” when the sigmoidal fluorescence curve crossed the predefined threshold (200 Relative Fluorescence Units (RFU) for IAV and IBV; 100 RFU for SC2; 50 RFU for RSV; 100 RFU for RNase P), “negative” when the threshold was not crossed and the IC was positive, and “invalid” when all channels (including the IC) were negative.

Turnaround time from extraction to results on the Panther Fusion® system was approximately 2.5 hours for both assays.

2.4. Diagnostic Performance

Diagnostic sensitivity (dSens), diagnostic specificity (dSpec), positive predictive value (PPV), and negative predictive value (NPV) for LDRA, PFRA and APRA were calculated using 2 × 2 contingency tables against an expected result defined as the per-target consensus across the three assays. Consensus rules were: A) “positive” if at least two of the three assays were positive for the same target; and B) “negative” if at least two of the three assays were negative.

2.5. Mixed Infection Detection Capability

Agreement for mixed-infection detection was assessed using only samples positive for any of IAV, IBV, SC2, or RSV (n = 335). A consensus reference was defined as: A) “mixed infection” (Mix) if at least two of three assays detected the same ≥2 pathogens in the same specimen; and B) “single infection” (Sin) if at least two of three assays detected the same single pathogen only.

2.6. Statistical Analysis

Slopes (Ct vs log₁₀(C_i)) and amplification efficiencies (per target) for LDRA were estimated by linear regression in Microsoft® Excel; 95% CI were computed from the standard error of the slope.

For precision, Grubbs’ test was applied to identify outlier Ct values. After outlier exclusion, a three-way ANOVA was used to estimate within-laboratory precision components (within-lot, between-lot, between-instrument, between-run/day, and within-run/day precision), summarized as SD and CV.

Agreement for target detection and mixed-infection detection capability among LDRA, PFRA, and APRA was assessed using 2 × 2 contingency tables to compute positive/mixed-infection percent agreement (PPA/MixPA), negative/single-infection percent agreement (NPA/SinPA), overall percent agreement (OPA), and Cohen’s kappa (κ-index). Excellent agreement was defined as OPA ≥ 95% and κ > 0.8[29,30].

Wilson score intervals were used to compute 95% CI for agreement estimates (PPA/MixPA, NPA/SinPA, OPA), positivity rates (PR), and diagnostic performance metrics (dSens, dSpec, PPV, NPV). The 95% CI for κ-index and SD were calculated using Wald for binomial proportions and Exact/MLS methods, respectively [31].

The McNemar-Mosteller nonparametric test was used to assess the statistical significance of differences in PR, dSens, and dSpec between the candidate and comparator assays (LDRA, PFRA, and APRA).

Analyses were performed with Analyse-it® for Microsoft® Excel, Ultimate Edition v6.16.2 (Analyse-it Software Ltd, Leeds, UK). A statistical significance level of 5% (p < 0.05) was considered.

3. Results

3.1. Design Evaluation and Assay Optimization

3.1.1. In Silico Specificity Evaluation

BLAST analysis of the RSV primers/probe showed that sequences meeting the predefined high-homology criteria (coverage and percent identity ≥95% with E-value ≤10^–2) were observed only for RSV (Supplementary File S6). In the closed-search exclusivity panel, all 68 respiratory-related microorganisms yielded coverage/identity <95% and E-value >10^–2, corresponding to 100% in silico exclusivity (Supplementary File S7). MFEprimer v3.1 predicted negligible amplification of human background DNA by the RSV oligos (Supplementary File S8).

A comparison with analogous oligo designs previously published can be found in Supplementary Figures S3 and S4.

Inclusivity analysis of the LDT RSV oligos yielded predicted true-positive rates of 99.73% and 99.93% across 3,360 RSV A and 2,834 RSV B NCBI sequences, respectively. Consistently, analysis of 11,653 RSV A and 13,907 RSV B sequences from GISAID estimated inclusivity of 99.06% and 99.89%, respectively(Supplementary Table S4, based on Supplementary Files S9–S12). When NCBI and GISAID datasets were pooled, overall inclusivity (pooled inclusivity) reached 99.21% for RSV A and 99.9% for RSV B (Supplementary Figure S4).

3.1.2. Amplification Efficiency and Multiplex Compatibility

Monoplex assays met the predefined amplification-efficiency acceptability range (90% ≤ E ≤ 110%), and on the Panther Fusion® Open Access™ system the LDRA achieved R² ≥ 0.99 with all targets remaining within the predefined efficiency limits (Table 2, Figure 2, and Supplementary Figure S5).

3.2. Analytical Performance

3.2.1. Analytical Sensitivity (Limit of Detection)

Absolute and relative hit frequencies by target and dilution level are shown in Supplementary Table S5.

According to the probit regression, the LDRA C₉₅ LoDs were: 37.8 copies/reaction (95% CI: 29.5–58.0) for IAV; 13.9 copies/reaction (95% CI: 11.0–21.1) for IBV; 9.6 copies/reaction (95% CI: 7.4–14.6) for SC2; 12.3 copies/reaction (95% CI: 9.6–18.0) for RSV A; and 25.1 copies/reaction (95% CI: 20.4–34.4) for RSV B (Table 3 and Supplementary Figure S6).

3.2.2. Analytical Specificity

3.2.2.1. Cross-Reactivity (In Vitro Exclusivity)

No cross-reactivity was observed with any pathogens included in the analytical-specificity panel. Sigmoidal amplification curves were observed only in the channels corresponding to IAV, IBV, SC2, and RSV (Table 4).

3.2.2.2. In Vitro Inclusivity

All subtypes and variants of the respiratory viruses tested with LDRA produced “positive” results in the expected detection channels for IAV, IBV, SC2, and RSV; no false-negative results were found in the in vitro inclusivity panel (Table 5).

3.2.3. Precision

Two outlier Ct values and two invalid results due to instrument errors were excluded. After exclusions, agreement with expected results exceeded 99%. Within-laboratory precision for the moderate level was 0.33 (95% CI: 0.28–0.48), 0.23 (0.21–0.29), 0.46 (0.40–0.59), and 0.42 (0.38–0.50) for IAV, IBV, SC2 and RSV, respectively; and for the low level was 1.00 (0.91–1.17), 0.42 (0.38–0.52), 0.51 (0.46–0.66), and 0.52 (0.48–0.64) for IAV, IBV, SC2 and RSV, respectively (Table 6). Figure 3 summarizes the main components contributing to within-laboratory precision of the LDRA assay.

3.2.4. Method Comparison: Agreement Assessment

No invalid results were found with any assay during method comparison; therefore, no specimens were excluded. Relative to PFRA, four SC2 specimens and one RSV specimen were PFRA-positive/LDRA-negative (0.99% and 0.25%, respectively; 1.23% overall). Relative to APRA, one specimen each for IAV, IBV, and RSV was LDRA-positive/APRA-negative (0.25% each; 0.74% overall). Discordant specimens were reanalyzed with all three RT-qPCR assays, and the same results were reproduced. One specimen was ultimately considered SC2-positive by all three assays after testing six replicates, with hit rates of 2/6 (LDRA), 3/6 (PFRA), and 6/6 (APRA); APRA reactivity was observed only for the SC2 S gene across all replicates. Across both comparator pairs, agreement metrics for detection of the four viruses were: NPA ≥ 99.7% (95% CI: 98.1–99.9%), PPA ≥ 95.7% (95% CI: 89.3–98.3%), OPA ≥ 99.0% (95% CI: 97.5–99.6%), and κ ≥ 0.971 (95% CI: 0.944–0.999) (Table 7).

Discordances between LDRA and PFRA were associated with high Ct values (SC2: 37.3–39.7; RSV: 38.8). Likewise, discordances between LDRA and APRA occurred at similarly high Ct values (IAV: 36.6; IBV: 37.6; RSV: 38.0). Consistent with these low-template patterns, the specimen that showed variable detection across six SC2 replicates yielded comparable mean Ct values across assays: LDRA, 37.8 (2/6 positive replicates); PFRA, 38.3 (3/6); and APRA, 34.2 (6/6).

3.3. Diagnostic Performance

According to the McNemar–Mosteller nonparametric test, positivity rates (PR) for IAV, IBV, SC2 and RSV did not differ significantly between LDRA and the comparator assays PFRA and APRA (p > 0.7322). Using the per-target consensus reference (majority agreement among the three assays) (see Section 2.4), LDRA showed 100% dSens and dSpec for all targets (Table 8). PFRA yielded four SC2 false-positive results and one RSV false-positive result, whereas APRA yielded one false-negative result each for IAV, IBV and RSV; however, no statistically significant differences in dSens or dSpec were observed among assays (p > 0.1250).

3.4. Mixed Infection Detection Capability

Among 335 specimens positive for ≥1 target, seven (2.09%) showed consensus evidence of mixed infection: one SC2 + IBV, one IBV + RSV, one IAV + RSV, three SC2 + RSV, and one SC2 + IAV. The LDRA assay detected all seven co-infections with no disagreement versus the consensus results (OPA = 100% [95% CI: 98.9–100%]; κ = 1.000). LDRA detected all seven co-infections without disagreement versus the consensus (OPA = 100% [95% CI: 98.9–100%]; κ = 1.000). PFRA detected the seven consensus co-infections plus three additional discordant co-infections (one IAV + RSV and two SC2 + IAV; OPA = 99.1% [95% CI: 97.4–99.7%]; κ = 0.819 [95% CI: 0.619–1.000]). APRA detected five of the seven consensus co-infections; one specimen was called SC2 only (instead of SC2 + IBV) and another IAV only (instead of IAV + RSV) (OPA = 99.4% [95% CI: 97.8–99.8%]; κ = 0.830 [95% CI: 0.599–1.000]) (Table 9).

4. Discussion

This work designs an oligo set for detection of RSV A/B and validates a laboratory-developed respiratory assay (LDRA), resulting from combining the LDT RSV with the CDC Flu-SC2 [19]. The new quadruplex enables fully automated, multiplex RT-qPCR detection of IAV, IBV, SC2, and RSV on the Panther Fusion® Open Access™ platform. In the current post-pandemic setting—where influenza, RSV, and SC2 can co-circulate and produce clinically indistinguishable syndromes [2,32]—such syndromic molecular testing can improve time-to-diagnosis and strengthen infection control and surveillance workflows [12]. In our proposal, using of human RNase P as an IC instead of IC-S adds a safety element because it ensures adequate sample collection. Assays, that rely only on process controls, such as the PFRA comparator used in this study, introduce an additional risk of false negatives when clinical samples have not been collected effectively [33].

Although a commercial assay is available for the Panther Fusion® platform (PFRA), we validated our approach using the system’s open channel because of local operational availability and the need to implement a resource-optimized workflow while maintaining equivalent quality standards. Laboratory-developed tests (LDTs) have been recognized as an important tool for laboratories seeking to adapt diagnostic capacity to local needs [34,35,36]. In settings where health insurance coverage can be fragmented or insufficient, as is often the case in the Dominican Republic, LDTs may facilitate more equitable access to molecular diagnostics, which remain relatively costly.

From an operational perspective, the Panther Fusion® platform provides mid-to-high throughput with continuous loading and random access, which can support timely reporting during seasonal peaks. Increased automation across pre-analytical to post-analytical steps reduces manual handling and may decrease the likelihood of human error, thereby improving reproducibility and supporting scalability in routine clinical workflows.

Analytically, the quadruplex LDRA demonstrated consistent performance in multiplex format, with amplification efficiencies within predefined acceptance limits and no evidence of target interference. The RSV component also showed high predicted inclusivity in silico across large RSV A and B sequence datasets, supporting resilience to circulating genetic diversity, and in vitro specificity testing against a comprehensive respiratory pathogen panel showed no cross-reactivity. Likewise, the remaining oligo designs derived from CDC Flu-SC2 for IAV, IBV, and SC2 detection (including the SC2 probe modification described here) retained high analytical specificity in silico (Supplementary Table S6) and in vitro (Table 4 and Table 5).

A dedicated analytical sensitivity study was performed for each target using standardized replicate testing and probit analysis. The LDRA assay achieved C₉₅ LoDs in the low tens of copies per reaction, corresponding to approximately 10³–10⁴ copies/mL, depending on the target. Direct LoD comparisons across candidate (LDRA) and comparator (PFRA and APRA) assays are inherently constrained by differences in sample materials, matrices, and reporting units. However, compared with the CDC Flu-SC2 assay, the quadruplex configuration showed slightly higher LoDs for influenza and SC2 (9.6–37.8 copies/reaction vs 5 copies/reaction) [19]. This difference may reflect the lower effective input volume (approx. 15 µL) dictated by the platform workflow. Overall, the LoDs observed for all four targets align with analytical sensitivity benchmarks commonly reported for contemporary multiplex RT-qPCR assays, particularly when assays are capable of detecting ≤ 10² copies/reaction (≤ 10⁴ copies/mL) [3,5,19,37,38,39,40].

The main contributors to within-laboratory imprecision for the LDRA assay were run/day (within-run/day precision or repeatability) and PPR lot (within-lot precision). Instrument effects (between-instrument precision) contributed to a lesser extent, depending on the pathogen and concentration level. For RT-qPCR assays, CV < 10% and CV < 5% are generally considered good and excellent, respectively [41]. Accordingly, the per-target precision metrics reported here for LDRA (CV: 0.9–2.9%; SD: 0.23–1.00) indicate high precision and demonstrate consistency, even near the LoD. These findings are also consistent with Shu, et al. [19], who considered variance values > 2 (SD = 1.41) unacceptable during CDC Flu-SC2 production quality control. Prior reports are consistent with these criteria [7,10,39], including manufacturer-reported precision for the commercial comparator assays (PFRA and APRA). For PFRA, a maximum reported CV of 2.04% was reported for a low level of IAV [28]. For APRA, a CV of less than 10% was considered acceptable [27].

The LDRA assay showed very high agreement with both commercial comparators across a broad Ct range, with percent agreement and κ (kappa) coefficient approaching 100% and only a small number of discordant results. Recent studies likewise report substantial inter-assay commutability and agreement among multiplex RT-qPCR assays for simultaneous detection of IAV, IBV, SC2, and RSV, with percent agreement values between 95% and 100% [3,5,6,7,8,16,38,42]. Although comparative statistics in the literature may vary across study designs and specimen sets, RT-qPCR assays for diagnosing IAV, IBV, SC2, and RSV generally meet commonly accepted acceptability criteria [29,30]. Against the study’s consensus reference standard (majority agreement among the three assays), LDRA showed 100% dSens and dSpec for each target. Importantly, because the consensus is derived from the three assays, LDRA’s “100% accuracy” should be interpreted as complete agreement with the majority-rule consensus rather than as proof of infallibility against an independent gold standard. Nevertheless, the concordance pattern—together with repeat testing of discordant specimens—supports that LDRA’s diagnostic performance is comparable to that of established commercial systems. No statistically significant differences were detected across assays (p > 0.1250), consistent with previous studies reporting diagnostic sensitivities and specificities that typically range from 95% to 100% and from 96.8% to 100%, respectively [12,17,18,43].

Discordant results in the comparator assays clustered at high Ct values, consistent with low-titer specimens in which stochastic sampling near the LoD can yield intermittent detection and increase the likelihood of apparent false negatives on repeat aliquots. In addition, multiplex assays may differ in target selection, chemistry, and signal processing; minor cross-channel fluorescence bleed-through (“crosstalk”) or differences in thresholding algorithms could contribute to occasional false-positive calls on specific platforms. In this dataset, the pattern of PFRA false positives (notably for SC2 and RSV) and APRA false negatives (one each for IAV, IBV, and RSV) is consistent with rare, low-frequency events expected near decision thresholds rather than with systematic assay failure, supported by the reproducibility of discordant calls upon reanalysis and by the observation that at least one SC2 specimen showed variable hit rates across replicate testing. Notably, SC2 false positives with PFRA—the most frequent discordance in this study—have been previously reported [16]. These false positives (two coincided with IAV co-detection) have been at least partially associated with software version 7.2.7 of the Panther Fusion® platform or the crosstalk correction algorithm [44]. Although the manufacturer has recently implemented an adaptive crosstalk correction to mitigate inter-channel artifacts in high viral-load samples, this study was performed using software version 7.2.7 [45]. In the SC2-positive agreed specimen that showed variable detection across six replicates, APRA detected only the SC2 S-gene target; this pattern may reflect target-specific performance, as RdRp detection sensitivity for that assay has been reported to exceed S-gene sensitivity [42]. This observation supports the added robustness of multitarget detection strategies for high-mutation RNA respiratory viruses.

Co-infections involving IAV, IBV, SC2, and RSV have been previously described [46,47,48]. Although infrequent, their detection has prognostic value because it can support risk stratification for complications and mortality [49,50]. From an epidemiological standpoint, the identification of co-infections may also signal periods of intensified viral circulation and inform the need for more active surveillance [51]. In our study, the mixed-infection frequency (2.09%) falls within the post-pandemic range commonly reported in the literature (0.3–10%) [16,50,51,52,53,54]. In this study, the most frequent co-infection was SC2 + RSV, which is consistent with the temporal overlap of three outbreaks during the sample-selection period (Supplementary Figure S7). This distribution may explain why SC2 and RSV were the most frequently represented viruses among co-infection specimens (3 of 7), followed by IAV and IBV (2 of 7). Differences in co-infection detectability across the three RT-qPCR assays and the consensus reference standard are consistent with the false-negative and false-positive patterns discussed above. As recently described for a multiplex IAV typing assay [55], apparent false negatives in consensus-defined co-infected samples may reflect preferential amplification of the higher viral-load target, which can occur in multiplex settings and may suggest some degree of competitive effect among targets/oligos. Despite these differences, agreement for co-infection detection remained high across assays.

Finally, several limitations should be acknowledged. This was a single-center study, which may limit extrapolation to other regions and epidemiological contexts. Although samples were collected over an extended period, the overall cohort size was modest, and no demographic, epidemiological, or clinical data were available; therefore, potential associations with patient factors or outcomes could not be explored. Because specimens were de-identified residual samples stored retrospectively, the observed detection frequencies reflect only the analyzed cohort and should not be interpreted as population prevalence. Moreover, the analytical sensitivity estimates were generated using synthetic evaluation materials with target concentrations not traceable to a certified reference standard; thus, LoD values may not be directly comparable across studies and may not fully reflect performance in clinical matrices. Finally, discordant results were not adjudicated using an independent orthogonal method (e.g., sequencing or an additional reference assay), which limits definitive attribution of assay-specific errors. Despite these limitations, the study provides a technically relevant analytical and clinical performance assessment of the proposed quadruplex approach.

5. Conclusions

The quadruplex laboratory-developed respiratory assay (LDRA), based on the CDC Flu-SC2 chemistry and expanded with RSV detection, demonstrated strong analytical performance and diagnostic estimates comparable to those of commercial comparator assays. These findings support LDRA as a practical multiplex option for the detection of IAV, IBV, SC2, and RSV, with potential utility for diagnostic testing and respiratory virus surveillance. Implementation on the fully automated Panther Fusion® platform may facilitate timely reporting and scalable workflows in medium- to high-throughput clinical laboratories.