A saliva-based COVID-19 diagnostic that is being used to test thousands of students every day at the University of Illinois at Urbana-Champaign is now covered by an emergency use authorization (EUA) from the US Food and Drug Administration.
That clearance, announced on Aug. 19, makes it easier for other labs across the country to use the method and marks a significant milestone in using saliva samples for widespread screening of asymptomatic people returning to universities and workplaces.
The University of Illinois test, called I-COVID, relies on a sample-processing method developed by researchers there (bioRxiv 2020, DOI: 10.1101/2020.06.18.159434) and joins a growing list of saliva-based tests that promise to bypass some of the bottlenecks that are holding back testing efforts. The study describing the test was published earlier this year on a preprint server, and like the other preprint papers described in this story, it has not yet been peer reviewed.
The most common way to diagnose COVID-19 relies on an uncomfortable nasopharyngeal (NP) swab that is pushed deep into someone’s nostril to gather fluid from the back of their nose and throat. These samples are typically run through diagnostic devices that use reverse transcriptase polymerase chain reaction (RT-PCR) to detect the presence of RNA from SARS-CoV-2, the virus that causes COVID-19.
But this approach has several drawbacks. Gathering samples is laborious—the tests are most reliable when samples are taken by trained medical staff, who need to wear protective equipment to avoid exposure to the virus. The procedure is also unpleasant for the person being tested, possibly deterring people from taking repeated tests. Public health systems have also struggled with shortages of NP swabs, along with the reagents used to prepare the sample for RT-PCR.
Saliva tests offer a way around these issues. “The nice thing about saliva is that you don’t need a swab or any fancy equipment; you just spit in a tube,” says Laura E. Lamb at Beaumont Health, Michigan’s largest health-care system, who has developed a saliva-based COVID-19 test.
At the University of Illinois, students and staff are required to regularly stop off at one of 40 collection stations on campus. They spit into a labeled tube, which joins a rack of samples that is collected every hour and sent to a dedicated on-site lab for RT-PCR analysis. The system has been running since July and is now completing 15,000 tests per day—enough to test every asymptomatic person on campus twice a week, at a cost of roughly $10 per test.
“Commercial labs are taking 5 days, sometimes worse, to get results back to people. On campus this summer we’ve been getting results back to people in 5 hours,” says the university’s Paul J. Hergenrother, part of the team that developed the saliva-processing method. Speedy turnaround is key: those with positive tests are alerted on the same day so that they can immediately isolate themselves while a contact tracing team tries to warn everyone the infected person has met.
Recent outbreaks at the University of Notre Dame and the University of North Carolina at Chapel Hill show just how much is at stake. As undergraduates flood into schools and universities explore various testing regimes, the coming weeks will serve as a real-time experiment on whether frequent saliva sampling can curb COVID-19 spikes and help campuses stay open.
Saliva is a messy fluid, stuffed with enzymes and inhibitors that can stymie the biochemistry needed to run RT-PCR. Back in May, diagnostics company Helix reported that saliva-based RT-PCR testing was far less accurate than tests based on NP swabs, with an average sensitivity of 69.2% and 98.9%, respectively (medRxiv 2020, DOI: 10.1101/2020.05.11.20092338). Sensitivity indicates the rate of true positive test results. Helix declined an interview request for this story, but in May the company’s chief scientific officer James Lu said, “Our view is that saliva will not be a viable sample type for either community testing or for determining back-to-work status.”
Other studies have come to a different conclusion. A team led by Anne L. Wyllie at Yale School of Public Health found that saliva from COVID-19 patients could actually offer more accurate diagnostic results than NP swabs (medRxiv 2020, DOI: 10.1101/2020.04.16.20067835). And on Aug. 14, researchers at Hokkaido University reported screening results from about 2,000 asymptomatic people, which showed that saliva samples run through RT-PCR offered a sensitivity of 92%, slightly better than from NP swabs (medRxiv 2020, DOI: 10.1101/2020.08.13.20174078).
Wyllie suggests that different saliva-processing methods can have a profound effect on the success or failure of RT-PCR. “We’re seeing more and more studies in favor of saliva because people are figuring out methods to make it work” to avoid the problems created by saliva’s complex mixture of biomolecules, Wyllie says.
For Illinois, sample preparation involves heating saliva at 95 ºC for 30 min and then mixing it with a combination of buffer and detergent. This step kills the virus, making for safer sample handling, and exposes the virus’s RNA. It likely also deactivates enzymes and other molecules in saliva that would interfere with RT-PCR. After that, the test uses standard RT-PCR reagents and instruments. Hergenrother says it took a big commitment from the university to set up the logistics for the system, which involves about 200 people at sampling stations and the laboratory. On Aug 10, the university created a company called Shield T3 to set up similar testing systems at other sites across the country.
The Illinois method is very similar to another protocol called SalivaDirect that Wyllie helped to develop. It uses heat and a dash of proteinase K, which makes the saliva less viscous and easier to pipette, to prepare saliva samples for RT-PCR (medRxiv 2020, DOI: 10.1101/2020.08.03.20167791). Wyllie says the test has a sensitivity of 88–90% in asymptomatic individuals, and although that means the test could miss one in 10 infected people, frequent retesting can help to improve those odds. On Aug. 15, the FDA granted the protocol an EUA, enabling certified labs to use the method once they have registered with the SalivaDirect team.
SalivaDirect was the fifth saliva-based COVID-19 test to be approved by the agency, and the Illinois team effectively piggybacked on that authorization by showing that I-COVID operates in the same way and offers equivalent performance. In a surprise intervention, the US Department of Health and Human Services declared on Aug. 19 that certified COVID-19 laboratories would no longer require an EUA from the FDA to run such “laboratory developed tests.” However, tests lacking that authorization do not benefit from a legal immunity to liability suits from patients, as provided by the Public Readiness and Emergency Preparedness Act, a factor that may deter some labs from using unauthorized tests.
The first saliva-based COVID-19 diagnostic won FDA authorization back in April. That test was developed at RUCDR Infinite Biologics, a biorepository at Rutgers University, which has partnered with several companies to sell saliva collection kits online. These samples are mixed with a preservation agent and go through RT-PCR testing at the Rutgers lab, which can process 50,000 samples per day at a cost of about $100 per test, and offers results that match the accuracy of NP swabs. On Aug. 19, the Rutgers lab relaunched as an independent company, Infinity BiologiX, a move it hopes will help to ramp up its testing capacity.
Several universities have been using the Rutgers test to screen students before they return to campus. But there are a wide variety of approaches toward testing across the sector: some universities just test symptomatic students, while others offer voluntary testing for asymptomatic students and staff. The University of California, Berkeley, for example, is running a pilot study called Free Asymptomatic Saliva Testing (FAST) that has completed more than 4,000 tests so far, which could lead to a routine screening system.
Many of the researchers behind these saliva-based tests are already working on improvements to expand testing capacity and speed. Automated sample processing would help; so too could loop-mediated isothermal amplification (LAMP), an alternative to PCR.
Unlike PCR, which cycles between different temperatures to amplify DNA, LAMP uses reagents that can operate at a constant temperature of about 65 ºC. This makes for a cheaper and simpler diagnostic process, Lamb says. Her team has developed its own saliva-based RT-LAMP system that produces similar results to NP swabs and is seeking funding to further develop the test (PLOS One 2020, DOI: 10.1371/journal.pone.0234682). Researchers at Columbia University invented a similar RT-LAMP system for saliva testing that indicates the presence of SARS-CoV-2 with a simple color change (medRxiv 2020, DOI: 10.1101/2020.06.13.20129841).
Still, testing protocols like I-COVID inevitably rely on a great deal of sweat and toil to provide enough screening capacity for an entire university, and automated high-throughput systems should be far more efficient. For example, Fluidigm’s Advanta Dx SARS-CoV-2 RT-PCR assay can process 6,000 saliva samples per day on a single machine. After some saliva-processing steps, each sample is dropped into one of 192 wells on a microfluidic cartridge that plugs into the company’s Biomark device. Tiny amounts of each sample are drawn into nanowells at the heart of the cartridge, where they undergo PCR. Working at this microfluidic scale cuts down on reagents, and results still match those from NP swabs. Fluidigm was awarded $36.8 million by the US National Institutes of Health’s Rapid Acceleration of Diagnostics (RADx) initiative to scale up production of the test, and on Aug. 25, the FDA granted the test an EUA. “I believe our technology is more than sufficient to be the principal workhorse for surveillance-based monitoring of populations,” says Chris Linthwaite, Fluidigm’s CEO.
Fluidigm also hopes to tag the nucleic acids in each sample with short strands of DNA that act like bar codes to identify which patient provided the sample. Several samples could then be mingled into a single well because the bar code would allow a positive result to be traced back to the right person. By running multiple samples in one well, the company could increase testing capacity to up to 48,000 samples per machine per day, Linthwaite says. Meanwhile, Ginkgo Bioworks aims to use a similar bar-coding approach to pool thousands of saliva samples for next-generation sequencing. Whereas PCR works by selectively amplifying a specific nucleic acid sequence from the virus, next-generation sequencing can read the code of millions of strands of nucleic acids in parallel. The RADx initiative has awarded Ginkgo $40.7 million to scale up its COVID-19 diagnostics.
Nevertheless, these methods still require centralized testing facilities, which introduces a delay between collecting a sample and returning a result. “What we haven’t done away with yet is the requirement for a large lab,” says Stephen A. Rawlings at the University of California, San Diego, who studies COVID-19 antibodies in saliva. “That’s why point-of-care tests will be a game changer.”
Such tests could deliver a result on the spot, without a technician running the procedure in a lab. For example, diagnostics company Todos Medical has developed a 10 min saliva test that detects 3C protease, an enzyme that SARS-CoV-2 produces to enable viral replication. The test uses a vial coated with a protein that binds to this protease and triggers a fluorescence response to indicate a positive result. The method is currently in clinical trials in Israel.
In the UK, Avacta Life Sciences is collaborating with Cytiva on a saliva-based lateral flow test that uses antibody-mimicking proteins called affimers to detect the SARS-CoV-2 spike protein and indicates results like a pregnancy test. The test, which takes just a few minutes to deliver results, is now undergoing clinical validation studies.
For now, a recent modeling study lends support to the University of Illinois’s testing strategy. Researchers at Yale and Harvard calculated that screening students a few times a week, in combination with measures such as frequent hand washing and mask wearing, could be enough to prevent major outbreaks on campus (JAMA Network Open 2020, DOI: 10.1001/jamanetworkopen.2020.16818). Hergenrother is optimistic that will be borne out at Illinois. “Frequent, repeat testing is really the key to controlling the pandemic,” he says.
This story was updated on Aug. 31, 2020, to mention that Fluidigm's test received an emergency use authorization from the US Food and Drug Administration on Aug. 25, 2020.