A SARS-CoV-2 antigen rapid diagnostic test for resource limited settings

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is the causative agent of COVID-19 disease. RT-qPCR has been the primary method of diagnosis; however, the required infrastructure is lacking in many developing countries and the virus has remained a global challenge. More inexpensive and immediate test methods are required to facilitate local, regional, and national management strategies to re-open world economies. Here we have developed a SARS-CoV-2 antigen test in an inexpensive lateral flow format to generate a chromatographic result identifying the presence of the SARS-CoV-2 antigen, and thus an active infection, within a patient anterior nares swab sample. Our 15-min test requires no equipment or laboratory infrastructure to administer with a limit of detection of 2.0 × 102 TCID50/mL and 87.5% sensitivity, 100% specificity when tested against 40 known positive and 40 known negative patient samples established by a validated RT-qPCR test.


SARS-CoV-2 antigen RDT device design and manufacture.
Our SARS-CoV-2 antigen RDT utilizes lateral flow immunoassay technology. Each test strip consists of a plastic backing card, sample pad, conjugate pad with the detection particles, nitrocellulose detection membrane with immobilized antibodies, and wicking pad. The test strip design, assay principle, and visual result interpretation are illustrated in Fig. 1. Nitrocellulose membrane cards are prepared by adhering 25 mm × 301 mm wide nitrocellulose membrane (Sartorius Stedim Biotech) to 60 mm × 301 mm vinyl backing cards (DCN Diagnostics) using a Matrix 2210 Universal Laminator (Kinematic Automation). A test line consisting of mouse anti-nucleocapsid protein antibody solution (Meridian Bioscience), and control line consisting of goat anti-chicken IgY solution (Bio-Techne), are printed onto the nitrocellulose membrane using low contact pressure nozzles on an IsoFlow Reagent Dispenser (Imagene Technology). Printed membrane cards are dried at 37 °C and stored in sealed foil pouches with silica desiccant until further processing.
To prepare the conjugate pad, mouse anti-nucleocapsid protein antibody (Meridian Bioscience) is conjugated to 150 nm gold carboxyl nanoparticles and chicken IgY (Bio-Techne) is conjugated to 40 nm gold carboxyl nanoparticles as control as previously described 13 . Both conjugation protocols utilize standard EDC chemistry as recommended by the particle manufacturer (nanoComposix). Test and control particles are combined in a drying buffer containing sugars, stabilizing proteins, and surfactant and sprayed onto 8 mm × 300 mm glass fiber conjugate pads (Millipore) using an atomizer nozzle on an IsoFlow Reagent Dispenser. Sprayed conjugate pads are dried at 37 °C and stored in sealed foil pouches with silica desiccant at ambient room temperature until further processing.
Printed nitrocellulose membrane cards are assembled with sprayed and dried conjugate pads, 17 mm × 300 mm cellulose fiber sample pads (Millipore), and 20 mm × 300 mm cellulose fiber wicking pads (Millipore) using a Matrix 2210 Universal Laminator. Pads are overlapped as depicted in Fig. 1A to allow for capillary flow from the sample pad, through the conjugate pad, up the nitrocellulose membrane, and finally into the wicking pad.
Fully assembled cards are cut into 3.8 mm test strips using a Matrix 2360 Programmable Shear (Kinematic Automation) and assembled into plastic test cassettes (Venus-lab) using a Closure-1 press (A-point Technologies). Individual cassettes and 1 g silica desiccant (ULINE) are sealed into foil pouches (Labels Inc.) to prevent moisture from degrading antibody components.
The current Test Kit contains 25 test cassettes sealed in foil pouches with desiccant, 25 sterile nasal swabs, 25 sample extraction tubes with 400 µL of sample extraction buffer each, 1 positive and 1 negative control swab, a package insert, and a quick reference instruction card. Intended use will allow for testing and result interpretation at the point of care, where a patient nasal swab sample is collected and immediately extracted into the buffer of the sample extraction tube. The tube is capped with the attached dropper tip and 3 drops are applied to the RDT cassette; signal is interpreted visually after 15 min as depicted in Fig. 2 (consent to publish was obtained from the person whose image is included in Fig. 2). The sample extraction buffer consists of a phosphate buffered saline solution with casein and mouse IgG as blocking agents, Brij 35 as surfactant and lysis agent, and sodium azide as preservative.
Recombinant Sf 21 (baculovirus)-derived SARS-CoV-2 nucleocapsid protein Met1-Ala419 (Bio-Techne) is used as quality control throughout the test manufacturing process. The SARS-CoV-2 antigen rapid test is designed to be interpreted visually; however, for this study we also employed a hand-held RDS-2500 reader The test cassette can also be read using a cassette reader for quantitative evaluation.

Sample preparation for clinical performance testing.
For clinical performance studies, as samples were not collected directly into our RDT kit sample extraction buffer tube, a 10X formulation of the sample extraction buffer was prepared and added directly into the saline sample matrix to mimic our intended sampleextraction buffer concentration ratio. De-identified samples were removed from the − 80 °C freezer and allowed www.nature.com/scientificreports/ to thaw and equilibrate to room temperature in a sample processing hood. Once thawed, the samples were briefly vortexed and centrifuged for 5 min at 4000 rpm. For each sample, 10 µL of 10X extraction buffer was added to 90 µL of nasal swab saline sample, pipetting up and down to mix. 85 µL of this resulting sample mix was dispensed via pipette into the sample well of the RDT cassette. Additional nasal swab samples collected into saline were mixed 1:1 directly into the kit sample extraction tubes (400 µL of saline collected sample into 400 µL 1X extraction buffer) to better mimic the RDT intended use per the test kit instructions. Three drops (approximately 85 µL) were added to the sample port of the RDT cassette.
In each case a timer was started upon sample addition and results were interpreted after 15 min.

SARS-CoV-2 antigen RDT device interpretation.
After the 15-min incubation, the cassette was visually assessed and determined to be positive or negative. Positive results were indicated by the presence of a blue test line of any intensity at the "T" location and the presence of a red control line at the "C" location. Negative results were indicated by a lack of blue test line of any intensity at the "T" location and the presence of a red control line at the "C" location. An invalid test would have no red control line at the "C" location; there were no invalid RDT cassettes observed out of > 200 cassettes tested, therefore in the studies reported here, the invalid test rate was 0%. Immediately after visual inspection cassettes were inserted into the RDS-2500 reader and read to measure the signal intensity at both the test and control line locations.  14 . Heat inactivated SARS-CoV-2 virus (BEI Resources, isolate USA0WA 1/2020) was diluted into a pool of 10 individual RT-qPCR SARS-CoV-2 negative nasal samples collected into saline. Linearity was evaluated by testing serial dilutions of inactivated SARS-CoV-2 virus in pooled negative nasal swab samples and the intensity of the test line was evaluated by visual inspection (Table 1) as well as quantified using the RDS-2500 reader (Fig. 4).

SARS-CoV
Following visual inspection, RDT cassettes were read in the RDS-2500 reader to quantify line intensity. Test line signal counts were correlated to viral copy input. Figure 4 demonstrates a linear response between the amount of viral material applied to the cassette and the amount of specific SARS-CoV-2 signal detected at the test line. The correlation coefficient for the resultant best fit line demonstrated an R 2 of 0.9684, indicating an acceptable dose-response for the N-antigen detection assay.  Table 2 below and the assay image of each RDT cassette can be found in Supplemental Fig. 1. The LOD was verified using inactivated virus spiked into pooled negative nasal swab samples and found to be 3.59 × 10 4 genomic viral copies or 200 TCID 50 /mL per RDT cassette (Table 2).
To quantitatively assess the sensitivity of the SARS-CoV-2 antigen RDT, each of the fifty cassettes used in the determination of the assay LOD were read using the RDS-2500 reader. Figure 5 demonstrates the range of signals detected at the test line and the control line locations.
The reader can interpret signal outside of the actual band, which increases the perceived background signal (as shown by the overlapping error bars of the 1.79 × 10 4 viral genomic copies and the 3.59 × 10 4 viral copies Table 1. Linearity study summary. The dilution in bold (3.59 x 104 viral copies/RDT, 2.0 x 102 TCID50/mL) was the lowest detectable dilution in the series. Visually interpreted test results of heat inactivated reference SARS-CoV-2 virus dilutions into pooled negative nasal swab sample (n = 2 per dilution). All samples with 1.79 × 10 4 viral copies or less per RDT cassette as well as the zero-virus negative control did not generate a visible test line and were interpreted as negative. All test cassettes have a visible red control line, indicating all tests were valid.    Fig. 5A). Visual images of representative RDTs are therefore shown in Fig. 6 to highlight the intensity level of the band for each level of inactivated virus-spiked negative nasal swab samples alongside a RT-qPCR positive swab sample. Weak positive blue test lines were visible in the "T" location of cassettes 1 and 2, both at and above 3.59 × 10 4 viral copies/RDT. No test line was visible on cassette 3, below 3.59 × 10 4 viral copies/RDT. For a comparison, the RT-qPCR positive swab sample demonstrates a clearly visible test line on cassette 4. Control lines were visible at location "C" for all cassettes 1-4.
Clinical sample performance testing. Clinical sample performance testing as outlined by the FDA EUA recommendation is a minimum of 30 positive specimens and 30 negative specimens collected either retrospectively or prospectively. Tests should demonstrate a minimum sensitivity of ≥ 80% for sample types tested 14 . Retrospective nasal samples collected into saline were mixed with 10X extraction buffer to mimic our intended sample-extraction buffer concentration and run on RDT cassettes; all samples were tested for SARS-CoV-2 by RT-qPCR using the ExoDx COVID-19 RT-qPCR Test (a CLIA-validated FDA EUA-authorized CDC protocol; this assay has an LOD of ≤ 0.85 copies/µL). Samples are considered positive by RT-qPCR if both the N1 and N2  A total of 35 out of 40 nasal swab samples positive by RT-qPCR were shown positive by RDT, demonstrating 87.5% sensitivity, and 40 out of 40 patient nasal swab samples negative by RT-qPCR were negative by RDT, demonstrating 100% specificity. This performance from a preliminary evaluation is comparable to SARS-CoV-2 RDT devices that have received FDA EUA approval 15 . However, it is important to note that the reported clinical sensitivity of these assays is dependent on the viral load of the patients in the clinical cohort. The viral load varies widely between patients and is dependent on when and how the sample was taken during the course of an infection, making it challenging to compare across different studies. The concordance between RDT and RT-qPCR testing of patient nasal swabs collected into saline is summarized in Table 3 and Fig. 7; data for each individual test cassette is listed in Supplemental Table 1 and individual cassette images are shown in Supplemental Fig. 3.
SARS-CoV-2 RT-qPCR-positive samples that had Ct values of 25 or lower were universally positive in the SARS-CoV-2 antigen RDT. RT-qPCR positive samples that were not detected as positive by RDT had Ct values of 25.3, 25.5, 25.7, 26.5 and 33.5, respectively, for the N1 assay. This indicates that the LOD for the SARS-CoV-2 antigen RDT assay correlates with patient samples that have Ct values around 25 for this CLIA-validated FDA EUA-authorized CDC test protocol.
To further assess the SARS-CoV-2 antigen RDT kit components and determine if a higher dilution of the samples could be used (using the 1X extraction buffer instead of 10X), 10 additional samples were evaluated by combining 400 µL of saline-collected swab directly into the kit sample tube containing 400 µL of extraction buffer, pipetting up and down to mix, and adding 3 drops (approximately 85 µL) to the sample port of the RDT cassette. As was observed previously, all known positive RT-qPCR samples with Ct values ≤ 25.0 resulted in a positive test line on the RDT; one sample at 26.5 Ct reported positive on RDT while one sample at 25.28 Ct was negative. Visual results are depicted in Table 4 and Fig. 8. Table 3. Concordance of the SARS-COV-2 antigen RDT assay and the ExoDx COVID-19 RT-qPCR assay. SARS-CoV-2 antigen RDT results achieved 87.5% sensitivity and 100% specificity in a clinical cohort of 40 negative samples and 40 positives that were confirmed by RT-qPCR. www.nature.com/scientificreports/

Discussion
We have described the development of a SARS-CoV-2 antigen RDT with preliminary performance data that meets acceptance criteria for sensitivity and specificity as outlined by the FDA in their EUA guidance documents and is among the most sensitive lateral flow assays that do not require a reader. A summary of other Emergency Use Authorization (EUA) issued antigen tests for SARS-CoV-2 collected in October of 2020 from the Food and Drug Administration (FDA) website demonstrated a range of LOD from 1.0 × 10 2 TCID 50 /mL to 4.5 × 10 5 TCID 50 / mL 15 . Our LOD of 3.59 × 10 4 viral copies per test is equivalent to 2.0 × 10 2 TCID 50 /mL and therefore is among the most sensitive SARS-CoV-2 RDT devices. For comparison, the BinaxNOW COVID-19 Ag card has shown a limit of detection of about 4.04 × 10 4 to 8.06 × 10 4 copies per swab 16 . Additional studies to examine cross-reactivity with other infectious viral and bacterial species, potential interference from common endogenous substances, and additional clinical performance testing with prospectively collected samples should be performed to fully establish the performance of the test. All the current SARS-CoV-2 antigen RDTs have sensitivity disadvantages compared to the gold standard RT-qPCR. However, the fact that RDTs can be more widely distributed, are cheaper, do not require an instrument or a reader, and can be used without specialized training is a big advantage. A test that is less sensitive but more readily available may identify more cases than a more sensitive test that is less available. To enable efficient surveillance of COVID-19 and reduce community transmission, quick and inexpensive testing is required. The longer turn-around time for RT-qPCR (in many cases several days) also makes it less useful. Patients with lower Table 4. Visual interpretation summary of test kit-based study. 400 µL of RT-qPCR known positive SARS-CoV-2 saline swab samples added to sample extraction buffer tube and evaluated on SARS-CoV-2 antigen RDT. The results agree with previous cutoff for RT-qPCR positives on RDT around 25 Ct for the N1 assay. www.nature.com/scientificreports/ virus levels not detected by RDT may be in a later phase of a waning infection, or they may be early in disease course and becoming more infectious over time. While RDTs have a lower analytical sensitivity than RT-qPCR, the ability to pick up community spread may be higher. RT-qPCR was estimated to detect at best 10% of infections in the US, and this number is likely far lower in countries with less access to these types of tests 8 . These RDTs enable immediate results, ease of use, accessibility, and no need for sophisticated equipment. They can also be easily mass-produced, and tests that utilize different supply chains and reagents are needed when global demand for testing occur as was seen during this pandemic. Furthermore, RT-qPCR tests are not immune to potential false negatives as new SARS-CoV-2 variants pose an evolving challenge to global testing. The variants B.1.1.7 first identified in the UK 17 , P.1 first identified in Brazil 18 , and variant B.1.351 first identified in South Africa 19 , have drawn concern that laboratories performing RT-qPCR tests targeting the gene for the viral spike protein alone may have decreased sensitivity 20 . Our SARS-CoV-2 antigen RDT, as well as many other RDTs, targets the nucleocapsid protein. As of early February 2021, the Foundation for Innovative New Diagnostics has examined 5 of the most widely used SARS-CoV-2 RDTs and determined these 3 viral variants had minimal anticipated impact on performance 21 . Regardless, all SARS-CoV-2 test manufacturers must monitor globally emerging mutations, including mutations in genes coding for the nucleocapsid protein, for all diagnostic applications 22 .
It is critical that diagnostic innovation reach all populations globally to enhance surveillance and monitor viral transmission-and RDTs can play an important role in quenching this SARS-CoV-2 pandemic when used appropriately to help alleviate testing inequality. Settings where centralized laboratory tests are limited can use RDTs as a complement to RT-qPCR, and high-risk congregate settings can use frequent serial screening tests to quickly identify SARS-CoV-2 infection and inform control measures to limit transmission, especially in settings where RT-PCR testing is not available. A pandemic of this size and scope that has reached all corners of the globe will require a diagnostic approach that matches its scale, and SARS-CoV-2 antigen rapid diagnostic tests are a crucial tool in the arsenal we must gather towards this effort.