Saliva-Dry LAMP: A Rapid Near-Patient Detection System for SARS-CoV-2.

The highly infectious nature of SARS-CoV-2 necessitates the use of widespread testing to control the spread of the virus. Presently, the standard molecular testing method (reverse transcriptase-polymerase chain reaction, RT-PCR) is restricted to the laboratory, time-consuming, and costly. This increases the turnaround time for getting test results. The study sought to develop a rapid, near-patient saliva-based test for COVID-19 with similar accuracy to that of standard RT-PCR tests. A lyophilized dual-target reverse transcription-loop-mediated isothermal amplication (RT-LAMP) test with uorometric detection by the naked eye. The assay relies on dry reagents that are room temperature stable. A device containing a centrifuge, heat block, and blue LED light system was manufactured to reduce the cost of performing the assay. This test has a limit of detection of 1 copy/µL and achieved positive percent agreement of 100% [95% CI 88.43% to 100.0%] and negative percent agreement of 96.7% [95% CI 82.78% to 99.92%] on saliva. Saliva-Dry LAMP can be completed in 105 minutes. Precision, cross-reactivity, and interfering substances analysis met international regulatory standards. The combination of ease of sample collection, dry reagents, visual detection, low capital equipment cost, and excellent analytical sensitivity make Saliva-Dry LAMP particularly useful for resource-limited settings. immediate development of rapid, near-patient tests for COVID-19 9,10 . Near-patient tests must yield straight-forward results which are easy to interpret. For remote settings, these tests should not be reliant on cold-chains and sophisticated equipment. To meet this immediate need, we developed Saliva-Dry LAMP, a rapid, near-patient saliva test for COVID-19 that uses lyophilized dual-target reverse transcriptase-loop-mediated isothermal amplication (RT-LAMP) with uorometric detection by the naked-eye 11 . This test can be performed on a portable and low-cost device that we manufactured.


Introduction
Due to the highly infectious nature of SARS-CoV-2 and its ability to be transmitted by asymptomatic individuals 1 , widespread testing for COVID-19 is critically important to preventing the spread of the virus 1 .
The COVID-19 pandemic has put immense demands on molecular testing infrastructure 2,3 . Presently, the standard method for COVID-19 testing is real-time polymerase chain reaction (RT-PCR) [4][5][6] . This method cannot be deployed outside of a laboratory. However, for the sake of contact tracing and self-isolation, the utility of a test relates to how quickly one can receive the results of the test after the sample is obtained 7,8 . Tests which require transporting samples to a centralized laboratory increases this time. Not surprisingly, governments have pushed for the immediate development of rapid, near-patient tests for COVID-19 9,10 . Near-patient tests must yield straight-forward results which are easy to interpret. For remote settings, these tests should not be reliant on cold-chains and sophisticated equipment. To meet this immediate need, we developed Saliva-Dry LAMP, a rapid, near-patient saliva test for COVID-19 that uses lyophilized dual-target reverse transcriptase-loop-mediated isothermal ampli cation (RT-LAMP) with uorometric detection by the naked-eye 11 . This test can be performed on a portable and low-cost device that we manufactured.

Patient Samples and Ethics:
Clinical specimens used in this study were anonymized saliva from individuals in Alberta collected between May and September 2020. No clinical information was obtained. Saliva was collected in UTM®-RT (COPAN Diagnostics Inc., Murrieta, USA) for ease of use 12 . The research involves human participants and was performed in accordance with relevant guidelines/regulations. Informed consent was obtained from all participants and/or their legal guardians, and was approved by Conjoint Health Research Ethics Board (CHREB) at the University of Calgary (REB20-0402/0444).

RNA Extraction:
Saliva diluted in universal transport media (140 µL total) was mixed with 560 µL of a concentrated preparation of lysis buffer and spiked with 2 µL of 50,000 pfu/µL MS2 bacteriophage (Zeptometrix, Buffalo, NY). Buffers used are described previously by Zainabadi et al. 13 . This lysate was hand shaken, then incubated at 61 ºC for 5 minutes. The lysate was applied to a spin column (Omega Bio-Tek Inc., Norcross, USA) and spun in a mySPIN™ 12 (Thermo Fisher Scienti c Inc., Waltham, USA) for 110 seconds at a peak speed of 11,300 RPM. The ow-through was discarded and 500 µL of wash 1 was applied to the column. The column was centrifuged again (110 seconds, 11,300 RPM) and the ow-through discarded. Next, 500 µL of wash 2 from was applied to the column before centrifugation for 170 seconds at a peak speed of 11,300 RPM. Columns were then transferred to new collection tubes and 50 µL of elution buffer was added. RNA was eluted with a nal spin (110 seconds, 11,300 RPM).
Lyophilized RT-LAMP Reactions("Dry LAMP"): Reference RT-PCR: The US Centres for Disease Control and Prevention N1/N2-gene RT-PCR was performed according to CDC-006-00019, Revision: 01 16 on the corresponding nasopharyngeal swab collected concomitantly with the saliva sample on which LAMP was performed.

Droplet Digital-PCR:
A high titre positive sample was quanti ed using a Bio-Rad QX200™ Droplet Digital™ (dd) PCR system relying on the Bio-Rad SARS-CoV-2 ddPCR Kit (Bio-Rad Laboratories, Hercules, CA) 17 . The ddPCR master mix consisted of (per sample) 2.5 µL One-Step RT-ddPCR reverse transcriptase, 6.25 µL One-Step RT-ddPCR Supermix, 1 µL 300 mmol/L dithiothreitol, 1 µL of each forward and reverse primers, 0.5 µL probe (20 µM primers and 10 µM probe), 7.5 µL RNase-free water, and 5 µL of extracted RNA. A 20 µL aliquot of each template mastermix was added to the sample well of the droplet generation cartridge, with 70 µL of droplet generation oil for probes. Thermocycling was done with the Bio-Rad C1000 Touch™ Thermal Cycler before measurement with the QX200™. Cycling conditions were 50 °C for 1 hour, 95 °C for 10 min, 40 cycles of 95 °C for 30 sec and 60 °C for 60 sec, then 98 °C for 10 min. Ramp rates were 2 °C/sec.

Limit of Detection Studies:
The limit of detection using commercially-available instruments and the Biobox was determined using a patient sample (nasopharygeal swab diluted in 25% saliva, 75% UTM 18 ) which was quanti ed by the Bio-Rad SARS-CoV-2 ddPCR Kit (Bio-Rad Laboratories, Hercules, CA). This sample was serially diluted to achieve a range from 1 to 0.25 copies/µL.

Cross-reactivity and interfering substance studies:
Potentially cross-reactive respiratory pathogens were tested with Saliva-Dry LAMP using inactivated stocks from Zeptometrix (Buffalo, USA) (Table S2). For interference testing (Table S3), negative samples and samples contrived to 9X LOD were spiked at the indicated concentrations with substances expected to be commonly found in saliva.

Clinical Validation:
Saliva is not collected routinely for COVID-19 diagnosis in Alberta. Given the low prevalence of COVID-19 in Alberta during this study, saliva and corresponding NP swab samples had to be collected from individuals who previously tested positive by RT-PCR. Clinical saliva samples were selected to re ect the natural distribution of viral loads in the population during early infection (See Figure S1). Plots, and 95% con dence intervals (Clopper-Pearson) were performed using MATLAB R2020b (The Mathworks Inc., Natick, USA).

Biobox fabrication:
A custom-made device, termed "Biobox" (Fig. 3), was developed to execute the sequence of steps for Saliva-Dry LAMP -centrifugation, isothermal incubation and naked-eye uorescent detection respectively. The Biobox comprised of three components -centrifuge, heating block and transilluminator (470 nm light emitting diode, LED, arrays). The design was prepared using Solidworks™ 2020 (Dassault Systems, Waltham, USA). All housing parts/ xtures were fabricated using a fused deposition modeling (FDM) 3D printer (Anycubic C, Commerce, USA) with poly-lactic acid (PLA) lament unless speci ed. The centrifuge rotor was fabricated using polycarbonate lament. The transilluminator consists of two LED arrays -a 6 x 8 LED array mounted inside the Biobox and a pair of 2 x 8 LED arrays mounted on the sides of the cap to provide illumination from the sides. A second cap was placed on the transilluminator with acrylic sheet window to block the wavelengths emitted by the LED's but not the intercalating dye. The aluminum heating block was machined to house both 2 mL and 1.5 mL microcentrifuge tubes. The temperature of the heating block was maintained at 61 o C using three heating elements and three thermocouple sensors. The centrifuge was made with a direct current (DC) powered brushless motor (Tmotor F40 Pro3 2600Kv, Nanchang, P.R.C.) mounted on an aluminum bracket. The centrifuge rotor was mounted on the brushless motor and achieved 8000 RCF. All components were controlled by an ESP32 microprocessor. The device is operated through the user interface using an LCD display and pushbuttons.
A DC power supply of 21-23 V was used to power the device.

Saliva-dry LAMP performed on the Biobox:
The lyophilized RT-LAMP reagents for the ampli cation of SARS-CoV-2 on the Biobox were obtained from Illucidx Inc. (Calgary, Canada). Lyophilized reactions consisted of the master mix described previously 14 but employed the dye mix described above. A proprietary excipient mix was also added. For extractions on the Biobox, conditions were identical as those on the commercially-available instruments with the exception that centrifugation times were 50 seconds shorter (due to faster ramping rates). Lyophilized reactions were resuspended with 25 µL of extracted RNA, mixed, then 30 µL of mineral oil was added on top. LAMP was run for 45 minutes at 61 0 C and visualized with the Biobox LED transilluminator.

Results
Analytical study of Saliva-Dry LAMP: The saliva dry-LAMP kit's work ow, on the Biobox, is depicted in Fig. 1. The limit of detection was determined with commercially-available instruments and the Biobox using a dilution series of a quanti ed contrived sample which spanned 1-0.25 copies/µL ( Table 1). All replicates tested positive at 0.5 copies/µL when using commercially available centrifuge and dry bath, whereas 4/5 were positive on the Biobox. A limit of detection con rmation was conducted using 20 replicates at 0.5 copies/µL prepared in the same way as described previously. Limit of detection con rmation was then retried successfully at 1 copy/µL (19/20 positive) (Table S1). Considering that no gold standard method exists for saliva yet, positive percent agreement (PPA) and negative percent agreement (NPA) was calculated ( Table 2). The CDC reference RT-PCR were run on the corresponding NP swab as reference methods ( Figure S1).  (Table ). Assay Precision, Cross-reactivity, and Interference: None of the 11 potentially cross-reactive respiratory pathogens tested showed any cross reactivity with Saliva-Dry LAMP in vitro (Table S1). None of the 18 potentially interfering medicines/substances tested showed any interference with Saliva-Dry LAMP in vitro (Table S2). The assay precision was con rmed with two samples twice a day for 20 days (Table S3). Variation arising from equipment was determined adequate with ve samples per day on three different sets of instruments for ve days (Table S4).
Performance of Saliva-Dry LAMP on the Biobox: In order to demonstrate functionality of the instrument designed and manufactured by our group (Biobox, Fig. 3), a limit of detection study was performed with Saliva-Dry LAMP. The limit of detection was determined using a dilution series of a quanti ed contrived sample which spanned 1-0.25 copies/µL. Four out of ve replicates tested positive at 1.0 copies/µL.

Discussion
In this study, we have developed a rapid, near-patient, saliva test for COVID-19 using lyophilized LAMP reagents with uorometric detection by the naked-eye. Experiments were designed to satisfy regulatory standards. Saliva-Dry LAMP showed no cross-reactivity or interference from any tested respiratory pathogens or medicines, respectively (Table S1, S2). In silico analysis by Mohon et al. did not identify any primer cross-reactivity in 13 relevant respiratory pathogens either 14 . As an RT-LAMP test, this method uses different reagents than RT-PCR, thus averting some supply chain bottlenecks and export restrictions 2,3,19 . Saliva-Dry LAMP detects SARS-CoV-2 from saliva, instead of the specimens from the standard nasopharyngeal swab, as there is a higher likelihood of detecting virus in saliva than detecting virus in nasopharyngeal swab specimens during the early phase of infection when diagnostic testing is most useful 20 . Saliva can be collected without a healthcare worker 20,21 and self-collection does not induce coughing, sneezing or bleeding 21,22 . Therefore, saliva collection avoids depleting critical supplies of PPE and swabs while reducing healthcare worker demand and exposure 20,21 . Saliva is also favourable for testing children as NP swabs are invasive 21 .
An important area of ongoing development for point-of-care nucleic acid tests is rapid RNA extraction. Standard laboratory RNA extractions are very time-consuming; however, replacing an RNA puri cation step with a simple inactivation step can compromise assay sensitivity 14,23,24 . RNA puri cations result in the concentration of viral RNA and the removal of ampli cation inhibitors, both of which increase sensitivity. Some rapid RNA extraction methods exist, but many of them require a cold-chain [25][26][27][28] (Table  S5 and S6), while still achieving an excellent limit of detection. Saliva-Dry LAMP has a capital equipment cost an order of magnitude less than RT-PCR when using commercially-available instruments, and a capital equipment cost nearly two orders of magnitude less than RT-PCR when using the Biobox (S. Rudgar, personal communication). Either of these options enables the deployment of Saliva-Dry LAMP in resource-limited settings. This method has its limitations. Firstly, the positive controls used in this study are not room-temperature stable reagents and require cold chain. Second, while the sample-to-result time near the patient is useful, the time required to perform the test is in approximately 105 minutes with a minimal throughput of four samples per run. Finally, the equipment developed still requires electricity and further re nements to increase the portability. A second prototype of the Biobox relying on a lithium-ion battery is feasible and is currently being evaluated. Future studies will aim to port Saliva-Dry LAMP onto a micro uidic cartridge, improving speed, and point-of-care feasibility.

Declarations
Competing Interests: DRP is scienti c advisor to Illucidx Inc., a University of Calgary start-up company supported by Innovate Calgary, which holds patents related to LAMP technology. All other authors declare no con icts of interest.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SciRepSupplementalMaterialJan132021.pdf