NEW YORK – Three decentralized tests for SARS-CoV-2, the virus that causes COVID-19, have been granted Emergency Use Authorization from the US Food and Drug Administration in the past 10 days. But whether deploying testing closer to patients will fundamentally change how the ongoing outbreak ultimately plays out may hinge on how many tests can be quickly deployed and how clinically accurate they prove to be.
Decentralized molecular tests are meant to bring lab-quality test results nearer to patients and provide results within the timeframe of an office or emergency department visit so that they can guide treatment decisions. These tests, and any accompanying equipment, need to be small or portable, require minimum sample processing, and little training to use correctly in low-complexity testing environments, typically those authorized to run CLIA-waived testing.
The newly authorized rapid, decentralized tests from Cepheid, Mesa Biotech, and Abbott Laboratories use instruments that have already been FDA cleared and CLIA waived for other indications, such as influenza testing, although FDA has noted it does not CLIA categorize EUA tests. These three EUA SARS-CoV-2 tests differ in terms of the sample types required, time-to-results, and workflows. Importantly, they also differ in their current instrument installed base.
This week Andrew Cuomo, the governor of New York state, said the COVID-19 epidemic might be resolved with "an inexpensive home test or point-of-care test that can be brought to volume," while former FDA Commissioner Scott Gottlieb declared the authorization of the Abbott ID Now test to be a "game changer" on Twitter.
Experts in clinical virology, however, are somewhat more cautious.
Richard Hodinka, a professor of microbiology at the University of South Carolina School of Medicine in Greenville and former director of the clinical virology laboratory and the Children's Hospital of Philadelphia, is an authority on molecular testing at the point of care.
"Having these systems available for SARS CoV-2 testing may be extremely impactful and offer a number of distinct advantages," Hodinka said in an email. However, "companies need time to ramp up their manufacturing capacity and product inventory in order to supply material to any and all customers desiring to use their products," he said.
Frederick Nolte, director of clinical laboratories at the Medical University of South Carolina in Charleston, said his lab went live a week ago with the Abbott m2000 lab-based high-throughput system and is now testing 400 patients a day, adding staff, and hoping to max out at about 1,000 tests a day, but the lab has not yet been able to get any decentralized testing up and running.
Alex Greninger, assistant director of the University of Washington clinical virology laboratory, said, "every new test is helpful" since they can build capacity and relieve the existing strain on the supply chain of other testing. "What’s interesting about these [decentralized] platforms is that there's no particular logistics for the sample, so you don't have to drive or fly them to a reference lab," he said.
Cepheid's test was the first to be authorized, and it accepts two sample types; nasopharyngeal swabs and nasal aspirates. It can be run in labs using the GeneXpert Dx or GeneXpert Infinity systems, or in decentralized settings using the GeneXpert Xpress system to run a single test or the firm's GeneXpert Xpress IV to four tests per instrument.
Mesa's test is authorized for use on throat and nasal swabs. It can be run on the firm's handheld Accula system, which is also exclusively distributed in North America by Sekisui under the Silaris brand.
Abbott's test is authorized for nasal, nasopharyngeal, and throat swabs, and runs on the firm's ID Now system, which was formerly known as the Alere i.
Manufacturing and distribution
To make a dent in the ongoing COVID-19 outbreak, the tests – and the test instruments – need to be distributed in high enough numbers.
Among the three decentralized tests, Abbott has the biggest installed footprint and highest current manufacturing capacity.
Specifically, Abbott has approximately 18,000 instruments out in the world in the US and will now begin ramping up to make 50,000 tests per day, according to Norman Moore, director of scientific affairs for infectious diseases at Abbott Diagnostics. The firm expects to ultimately produce more than a million tests each month, he said.
In contrast, Cepheid has 5,000 systems placed in the US, and a representative said that the firm is currently ramping test production. Mesa Biotech has previously said it has a few hundred users, mostly in physician office labs, and the firm can make 10,000 tests per week. It is also "actively adding personnel, additional shifts, and fast tracking an already planned facility expansion" according to Laura Dullanty, the firm's senior manager for marketing and commercial services.
But, many of the rapid MDx instruments typically run one, or a few tests at a time, so users need multiple instruments to increase throughput. "It is not a solution for high-volume testing, it is a niche solution for those situations where you need results quickly," Nolte said.
On the ID Now, a strong positive COVID-19 result – which presumably is due to a patient having a high viral load – can be detected in five minutes. But a negative result requires 13 minutes, translating to fewer than five negative tests per instrument in one hour.
"Most patients in most locations will be negative," Nolte noted.
The Cepheid and Mesa tests take significantly longer to run each test – 45 minutes and 30 minutes, respectively, although using a four-module Cepheid instrument could potentially increase the throughput. A representative from Cepheid also noted that the firm's COVID-19 assay has been granted EUA to run on the entire GeneXpert family of instruments, which also includes two-, 16-, 48-, or 80-module configurations. Testing on these instruments under EUA had been designated for labs authorized to perform high and moderate complexity tests, but the representative noted that some hospitals do use the high-throughput systems nearer to patients, essentially in decentralized lab spaces.
Nolte said his hospital system in South Carolina has a large telehealth presence, and it started a drive-through sample collection tent early on in the outbreak.
"There are four stations patients can drive up to, and I was thinking, what if we had an ID Now instrument at each station so while they are waiting in their car, they can get the results," he said. "But that is difficult to scale up, because you need lots of instruments."
The Cepheid test is also on Nolte's short list for decentralized testing, as is a BioFire test – which was granted EUA but for not for use in low-complexity settings – because his hospital system uses a high volume of the FilmArray Respiratory Panels.
Overall, Nolte said decentralized testing is not going to help much with the thousands of tests that need to be done daily. "It is going to help in situations where you need to know quickly, instead of waiting 24 hours," he said, such as the 20 percent of tests that his lab processes for patients in the emergency department who could likely benefit from stat processing in order to make admitting or treatment decisions.
Greninger concurred that in the current EUA decentralized tests can likely absorb tens of thousands of tests each week, "but not the hundreds of thousands or millions that we need to guarantee widespread testing."
In contrast, just one lab-based Roche Cobas 8800 instrument, assuming it can be properly supplied, can run 20,000 tests per week, Greninger said.
Indeed, Marrio Torres, head of Roche Molecular Diagnostics, commented on the firm's COVID-19 strategies. "We strongly believe that we can most effectively address this global health crisis by focusing our efforts on ensuring the availability and capacity of our respiratory diseases across our platforms," he said. "We've seen a tremendous increase in demand for influenza, RSV, and Strep A testing … [and] we are committed to ensuring that needs are met for these three important respiratory tests at the point of care and that the urgent demand for high-volume testing is met with the Cobas SARS COV-2 Test on the Cobas 6800/8800 Systems at this time."
In terms of rule-out testing, Nolte's lab had originally provided routine BioFire RP results for every patient that was tested for SARS-CoV-2.
"When that volume got to be hundreds per day, we were getting the COVID-19 results out before we could complete all the BioFire runs," because that instrument also runs a single test per instrument, with each test taking one hour, and the lab still needs to run other diagnostic tests on that instrument. The lab also found that the results of the RP panel, particularly for patients who had samples collected at the drive-through clinic, were not impacting patient care.
The three tests can reach different points distal to the lab. The Cepheid system is often dubbed "near-patient" because it is more of a bench-top instrument. The Abbott system is the next smallest, and also sits on a surface, while the Mesa system is handheld, so could theoretically be used at the bedside.
But although point-of-care tests are being touted as empowering small clinics and rural areas that don't have access to testing, Nolte emphasized that low-throughput testing could become a bottleneck in itself. "Once volume scales up, it is no longer a workable solution."
The tests are also currently being distributed selectively to areas hardest hit by the outbreak.
"We're working with the [Trump] administration to deploy the tests to areas where they can have the greatest impact," said Abbott's Moore.
Nolte's lab is running the Abbott m2000 test on two instruments they already had, plus one that the company supplied them with to facilitate COVID-19 testing. "We've used that system for years. It's a workhorse … [and] they guaranteed us a supply of reagents and we were the second lab in the country to go live with it," he said.
However, Nolte said he also wanted to set up additional avenues for rapid testing, and his colleagues have been asking him how the two Abbott assays compare. But when he reached out to ask for an ID Now instrument, the company said they will be allocating whatever they can produce first to the states that are currently hardest hit by the outbreak.
Nationally, the positivity rate for COVID-19 testing is approximately 8 percent, Nolte said, while New York City is nearing an average of 30 percent with some neighborhoods as high as 70 percent. Nolte's lab in South Carolina currently has a 6 percent positivity rate.
"We're not in the hot zone yet," he said.
From a high-level viewpoint, he pointed out that even if manufacturers can make enough tests, the sample collection materials, and competent staff, also need to be in place and at scale.
"It's not just looking at the inventories that the manufacturers are talking about, it is about how do you effectively use those," he said. "If the collection kits have disappeared, it doesn't matter how many tests Roche or Abbott produces."
To prevent positive cases from slipping through the cracks and continuing the chain of transmission, tests need to be accurate, which is a function of sensitivity, specificity, ease of use, and potential for contamination.
For lab-based testing, Nolte and his colleagues confirmed the analytical sensitivity of the m2000 test in house, using SARS-CoV-2 genomic RNA quantitated by droplet digital PCR. "It looks like close to what [the company claims], of 100 copies per milliliter," he said, adding that any better than that is probably not needed for diagnosis of symptomatic patients.
However, "When I tell people we launched our assay, the first question is, 'What is the sensitivity and specificity,' and I say, 'I can't tell you,'" he said. The test is analytically very sensitive, and based on the primer and probe selection, it should be specific for this coronavirus, but that is as much as is currently known.
The same holds true for the EUA decentralized tests, which have all been validated in the lab with contrived samples.
All three tests are explicitly authorized to be run by trained operators in patient care settings outside of the clinical laboratory environment, and all three accept raw samples.
According to the instructions for use of each test, the reported limit of detections differ slightly among these tests: Cepheid detects as few as 250 copies per milliliter, Mesa's limit is 200 copies per reaction, and the Abbott LOD is 125 genome equivalents per milliliter.
However, whether the LOD and contrived sample analysis will translate into actual sensitivity in real patient samples remains to be seen.
"Everyone wants to know the clinical sensitivity and specificity of all of the FDA EUA-available tests, but I do not think you will see this information until the dust starts to settle," said University of South Carolina School of Medicine's Hodinka.
Abbott's Moore also highlighted that "performance characteristics, including accuracy data, will continue to be collected in the field."
Nolte noted that even though the outbreak is challenging, there still needs to be stringency among lab professionals, and he is drafting a white paper with his colleagues at the Association of Molecular Pathology noting that the validation for emergency use that the FDA is now accepting is "the absolute minimum anybody would consider before launching a clinical assay."
This may be why all three decentralized tests also explicitly state, "Negative results do not preclude SARS-CoV-2 infection and should not be used as the sole basis for treatment or other patient management decisions."
Using high-sensitivity molecular tests outside of the controlled lab environment has historically been an issue, as laboratorians have worried that non-expert use might lead to contamination and false results. In a number of cases previously reported – such as at Geisinger Health, Henry Ford Health System in Michigan, and in some pediatric hospitals – decentralized MDx systems have been adopted but are run by designated trained users, housed in special containment boxes, or only used in the laboratory for stat testing.
In an ideal world, the EUA POC MDx tests would have been evaluated using clinical samples run by actual end users in the field to make sure the accuracy is the same as in the lab.
Unfortunately, a thorough trial of these technologies "is just not possible in the throes of the outbreak," Hodinka said. This is especially true "given all of the patient testing that laboratories are currently doing, and particularly because of the shortage of collection swabs, transport medium, test kits, instruments, reagents, [and] consumables," he said.
"Everyone is playing catch up right now and doing the best they can under very extenuating circumstances," Hodinka said.