Skip to main content
Premium Trial:

Request an Annual Quote

Saliva Samples Slow to Catch On for Large-Scale Coronavirus Testing


NEW YORK – When the US Food Administration granted the first Emergency Use Authorization for a saliva-based SARS-CoV-2 PCR test to the Rutgers Clinical Genomics Laboratory at RUCDR Infinite Biologics in April, many expected that other assays using saliva samples would quickly follow and scale to large numbers.

After all, what could be a more convenient way to provide a sample for testing than spitting into a tube? Not only could individuals collect their own sample — even at home, after the FDA amended Rutgers' EUA in May — to help reduce the use of personal protective equipment by healthcare workers, but saliva would also be less invasive than nasopharyngeal (NP) swabs and relieve swab shortages.

But three months after the Rutgers EUA was issued, only a few other labs have developed and deployed saliva-based COVID-19 tests, most at a relatively small scale. They include Phosphorus Diagnostics, which received EUA in June; P23 Labs, which obtained EUA for a test that can use saliva and other samples this month; Fluidigm, which has submitted a saliva test for EUA; Soft Cell Laboratories; Sports Medicine Research Testing Laboratory; and academic groups at Yale University, Columbia University, the University of Chicago, the University of Illinois at Urbana-Champaign, and the University of Southampton in the UK.

Meanwhile, several labs that provide large-scale COVID-19 diagnostic testing — including two with extensive experience in saliva-based genetic testing — have explored saliva for their SARS-CoV-2 tests but abandoned it, for reasons that include lack of sensitivity, challenges with sample collection and processing, and the cost of collection devices. Instead, they have found nasal swabs, which can also be self-collected, to be more suitable for their operations.

In general, saliva may be less appealing to labs developing a COVID-19 diagnostic test than other sample types because the US Centers for Disease Control and Prevention does not include saliva among the recommended specimens for testing. Also, the FDA still requires each saliva-based test to be validated against recommended sample types for authorization. 

A likely reason for the agencies' reluctance to recommend saliva across the board is that published studies on saliva-based SARS-CoV-2 testing have reported inconsistent results. A recent meta-analysis of saliva-based testing studies, published as a preprint in May, found that saliva tests, overall, were 91 percent sensitive, compared to 98 percent for NP swab tests, but individual studies varied in their results. A study involving more than 600 individuals published in April by researchers in Australia, for example, found that saliva detected about 85 percent of the cases that NP swabs did.

"Some studies that have been published show saliva is an adequate sample type for SARS-CoV-2 testing, and there are others studies that have shown that it's actually not a very good sample type," said Alicia Zhou, CSO of genetic testing company Color, which has developed a loop-mediated isothermal amplification (LAMP) COVID-19 test that received EUA in May. Color serves a number of community testing sites in the San Francisco area and elsewhere and currently has a capacity of about 10,000 tests per day.

Why saliva behaves so differently across studies could have a number of reasons. "It could be that the sample itself is not a good sample, although that seems unlikely, given that some people have shown that it works," Zhou said. More likely are differences in the sample collection devices and the chemistry they use to stabilize the saliva sample, she said. Some groups have found, for example, that the chemistry can interfere with downstream processes such as RNA extraction, so modifications are needed for the test to work properly.

James Lu, cofounder and CSO of genetic testing company Helix, agreed that "the literature on saliva-based testing is pretty mixed," unlike that for other sample types.

Helix, which has analyzed DNA from tens of thousands of saliva samples for its population genomics projects and consumer genomics customers, conducted its own evaluation study of saliva-based SARS-CoV-2 testing, published as a preprint in May, and found saliva to be 30 percent less sensitive than NP swabs for detecting COVID-19 in samples from a community setting, a result that was unacceptable for its goals. "We had very high hopes that saliva would work," Lu said. "We were disappointed that we could not get it to work."

For their study, the Helix researchers deliberately chose to test samples from a community setting rather than a hospital, and they expected these to include individuals who were only mildly symptomatic or even asymptomatic. They also analyzed samples from a group of convalescent COVID-19 patients who needed a negative test to return to work. In both cases, they saw a significant reduction in sensitivity when using saliva samples compared to NP swabs.

It is unclear whether the reduced sensitivity was due to their saliva testing method or because the saliva samples did not contain enough viral material. "We acknowledge that people have different chemistries and different collection methods, so they may be able to detect what we cannot," Lu said.

However, his team also tested two saliva collection devices, including the one authorized in the Rutgers EUA, using contrived samples and found the detection limit to be similar to that for swabs. "That argues that [the lack of clinical sensitivity] may be driven more by biological reasons than a technical reason," Lu said. It may be, for example, that saliva works great for highly symptomatic individuals but less well for mildly sick ones. "We have not seen a definitive study showing great correlation [between sample types] in a community setting," he said. "At least in our hands, with the chemistry that we were trying, we could not get it to be reproducible in the kind of fashion that we thought was reliable for community testing."

As a result of the study results, Helix developed its own SARS-CoV-2 test, which received EUA last week, for use with swabs, with a focus on lower nasal swabs. "We now know that anterior nares or nasal swabs are a perfectly viable self-collection methodology that's easy for people to use," Lu said. Swabs have not been a bottleneck recently, he said, but the company has seen shortages of collection tubes and devices and expects reagents to become "a bit more tight" as the demand for testing grows, though it has still been able to maintain next-day turnaround times for the vast majority of samples.

Helix is also still working on a sequencing-based SARS-CoV-2 test, for which it has already submitted an EUA application. This test, Lu said, would offer a differentiated supply chain and be more sensitive than its PCR assay.

The Broad Institute, which set up large-scale PCR-based COVID-19 testing in April, also considered saliva and ultimately decided against it. Stacey Gabriel, senior director of the institute's genomics platform, said one issue was the collection of saliva from patients in hospitals and residents of nursing homes. "In those settings, saliva is really difficult because the collection devices require a fair amount of saliva and patients sometimes have a hard time generating that much," she said.

In addition, saliva "presented some challenges" to the liquid handling robots that enable a high degree of automation for the Broad's test, because the samples tended to be heterogeneous and viscous, she said.

"We did try it in a limited way," she said, but found the results to be worse than for swabs, likely because there wasn't enough sample or the automation system didn't handle it well. "It was not great in our hands," she said. "I'm not doubting what Rutgers can do, but for us, it didn't seem to be the answer."

Instead, the Broad went ahead with nasal swabs. "While I would totally agree that the widespread feeling is that the NP swabs are uncomfortable and hard to accommodate, I would say that the anterior nasal swabs gave really good results and were much easier to tolerate, so we put our efforts into the anterior nasals," she said.

For Color, which has been collaborating closely with the Broad Institute on SARS-CoV-2 testing, another argument against saliva was the cost of the collection devices. Commercially available devices can easily cost more than $10 each, Zhou said, "and if you're talking about trying to do massive testing … that cost attached to the collection kit itself feels untenable when you're trying to really lower barriers of access to this type of testing."

The company's goal was to create an infrastructure that enabled tens of thousands of tests per day, at low cost and with no reagent supply constraints. Neither saliva, the sample type it was most familiar with from its clinical genetics tests, nor next-generation sequencing, the analysis technology it had the most experience with, served that goal, according to CEO Othman Laraki.

Another issue with saliva that people tend to underestimate, he said, is that it requires no eating, drinking, or smoking for half an hour or so before testing. This can become a problem at high-throughput testing sites that try to get people in and out within minutes, he said.

"When we took the full picture in mind, even though there's been a lot of excitement around saliva — and we were actually also excited about it at one point — we decided that it was not the ideal set of tradeoffs," Laraki said. "We actually did run some experiments and we have no doubt that saliva can work, but the variability of peoples' results does indicate a more variable process."

Color opted to adopt dry nasal swabs for its LAMP assay instead, which have almost completely replaced the NP swabs it was initially using. "We did some work with the Gates Foundation and a number of others and found that it is the right approach to dramatically driving down the costs," Laraki said. "A nasal swab is basically a fancy Q-tip, and the tube, it's under $1 of raw costs for the collection device." It also doesn't require a stabilization solution that would add extra plastic and create more safety concerns. Dry nasal swabs, in addition, appear to lend themselves more readily to pooled testing than other collection devices, he said, which is something Color is interested in.

Madhuri Hegde, CSO of PerkinElmer Genomics, said that her company has "not given up on saliva" for SARS-CoV-2 testing but has not seen great results so far. Initially, it tried to validate saliva testing in a drive-through testing setting, she said, and found that swabs worked just fine but saliva samples varied in quality because people did not adhere to the rules of not eating, drinking, or smoking. "I can't even describe how the tubes looked," she said. "They had air in it, food particles, sputum."

Now, PerkinElmer is testing saliva samples collected by collaborators in a clinical setting, from patients that are expected to have a high viral load, which it plans to analyze shortly. "I think we will eventually get to a place where we are also validated on saliva, but I am not certain if it is the right sample type to use on asymptomatic individuals because these individuals may not have a high viral load and we might be issuing false-negative results," Hegde said. "That is my biggest concern."

PerkinElmer has a close relationship with Everlywell, Hegde said, which received an EUA for a nasal swab home collection kit in May. PerkinElmer is working on validating a home collection SARS-CoV-2 kit at the moment that will use Everlywell's nasal swabs, which are working well, she said. "I don't think we will launch saliva, unless our next validation round, which is in hospital patients, is to our satisfaction" and has similar high sensitivity as PerkinElmer's current assay, she said.

Quest Diagnostics, likewise, is not considering saliva-based testing at the moment. "Right now, current saliva collection technologies do not meet our standard for use with COVID-19 diagnostic testing," a spokesperson said in an email, adding that Quest continues to evaluate new technologies and services and may consider saliva testing in the future if appropriate.

In the meantime, a handful of labs continue to bet on saliva testing. Phosphorus Diagnostics, which received EUA for a saliva test in June, including at-home sample collection, published a preprint of its validation study last month, which found saliva to have about 97 percent positive and negative agreement with NP swabs. The company currently has capacity to run about 700 tests per day, almost all of which is being used.

According to Gabriel Lazarin, VP of medical affairs and marketing at Phosphorus, differences in performance between various saliva studies are likely due to differences in methods. "That is, for example, one reason we looked at multiple extraction methods in our study," he said in an email. "Some yielded poor performance, and thus this is an important thing to report in data that categorically states saliva is worse."

Overall, he said, saliva offers a convenient and noninvasive option for large-scale testing that reduces the risk to healthcare workers. "There is sufficient data across EUAs and publications indicating at minimum equivalence to NP swabs in individuals presenting with symptoms, which gives it an advantage for testing both in the clinic and at home," he said.

Fluidigm has also developed a saliva-based SARS-CoV-2 RT-PCR assay, for which it has submitted an EUA application. Clinical validation of the test, which is now commercially available, showed 100 percent concordance with NP swabs, according to CSO Andrew Quong.

"There are many reasons saliva has not been used in the past; many are due to perception regarding the inhibitors in the samples that may confound the assay or that the number of viral particles is lower in saliva than other samples," Quong said in an email. However, he noted, the company's Advanta assay has overcome the issues around inhibition with a two-step process that doesn't require RNA extraction. "The clinical data supports the hypothesis that there is sufficient viral load in saliva and there are other published reports suggesting the same," he said.

Another testing lab, Soft Cell Laboratories, has validated a saliva test that uses the Spectrum Solutions collection device that is part of the Rutgers EUA. After submitting its data to the FDA, the company was told that it had conducted a bridging study and, as a high-complexity lab, did not require EUA, according to John Maclennan, VP of business development and IP at Soft Cell Biological Research. The firm has been offering the test for almost three months now.

Academic labs have also continued to develop saliva-based COVID-19 tests. A group at the Yale School of Public Health, for example, said last month that it has launched a study with the National Basketball Association to further validate its saliva test, which now requires no RNA extraction. At the time, the Yale team said it was planning to submit the test to the FDA for EUA.

Also last month, a group at the University of Illinois at Urbana-Champaign published a preprint describing a saliva-based SARS-CoV-2 RT-PCR assay that also requires no RNA extraction.

In addition, a group at Columbia University last month described a rapid saliva-based RT-LAMP assay in a preprint. The test, which also requires no RNA extraction, uses technology that was adapted from an assay used in preimplantation genetic testing at Columbia's fertility center. On Wednesday, Sorrento Therapeutics said it licensed the Columbia test and plans to submit it to the FDA for EUA, calling it a "key asset in our diagnostic solutions."

Yet another group, at the University of Chicago, said last month that it has developed a droplet digital PCR-based saliva test and compared it to nasal swabs, though it has not published its results yet.

Finally, researchers at the University of Southampton in the UK said last month that they are exploring a saliva test, based on Optigene's COVID-19 LAMP assay, in collaboration with the National Health Service and the Southampton City Council, testing it in a large number of physicians, essential workers, university staff and students, and households.

Meanwhile, Rutgers RUCDR Infinite Biologics, seemingly the only laboratory conducting saliva-based SARS-CoV-2 diagnostic testing at large scale so far, has teamed up with others to distribute its test more widely. At least two companies — Vault Health and Vitagene — allow individuals to order the Rutgers test online, and a third firm, IxLayer, offers a platform to facilitate patient-initiated or physician-ordered testing.