The US reported its first confirmed case of COVID-19 on January 21st. Eight weeks later, there still aren’t enough tests for the virus available for everyone who needs them. “It is a failing,” said Anthony Fauci, director of the National Institutes of Allergy and Infectious Diseases, at a House briefing last week. “The system is not really geared to what we need right now.”
People who are sick or have been in contact with sick people are struggling to get tested. Until last week, the number of tests that could be run per day in the United States was limited to around 7,000. Labs are struggling to get the supplies they need to meet the demand.
At the center of all of this chaos is a relatively straightforward type of test called a polymerase chain reaction, or PCR, that’s been around for decades. To run it, a doctor swabs a patient’s nose or throat and sends the sample to a lab. The lab then looks for tiny snippets of the virus’s genetic material. PCR analysis is complicated but reliable.
“If the health system is working well, those tests should be good and help us manage this epidemic,” says Catherine Klapperich, director of the Laboratory for Diagnostics and Global Healthcare Technologies at Boston University. “It’s frustrating that the testing we thought we could rely on didn’t roll out the way we expected it to.”
PCR works. But in an outbreak situation, even if it’s working well, it’s still too slow. Ideally, physicians would be able to run tests in an office or right at a patient’s bedside. The technology to test that way exists. But there hasn’t been a large-scale investment made to commercialize it, so there’s no clear pathway to get it up and running during an emergency.
“Point of care testing is required for these outbreak situations. We’re just not quite there yet as a scientific community,” Klapperich says.
Current gold standards
A week after the genetic sequence for the novel coronavirus first became available in January, German researchers had designed a PCR test that could detect the virus in a patient sample. That test became the basis for the World Health Organization’s (WHO) test used in countries around the world, including South Korea, but which the Centers for Disease Control and Prevention (CDC) declined to use. PCR tests are, in theory, fairly simple to create: scientists pick snippets of the virus’s gene and use a series of chemicals to look for that gene snippet in the sample. If they find the snippet, it means the patient has the virus.
PCR is the gold-standard testing platform for viruses because it’s highly sensitive, says Paul Yager, a professor in the department of bioengineering at the University of Washington — it can detect even a tiny amount of virus in a patient sample and is less likely to incorrectly have a negative result.
In addition, it’s very well-known. It was first invented in the 1980s, and the patents on the systems needed to run it expired in 2005, leaving it cheap to do for any lab. “If you know the sequence of the bug you’re looking for and know what bodily fluid to look in, designing that test should be relatively straightforward,” Klapperich says.
It’s straightforward, but it can be slow. For weeks in the US, tests for the virus were being done by hand. That’s a time-consuming, technical process. Even though PCR tests are simple to create, they’re complicated to execute: a lab technician has to carefully mix a patient sample with tiny portions of chemicals in tiny tubes, and any bit of contamination can ruin the test. The samples also have to be repeatedly brought up to high heat and back down in a process called thermocycling, which is done on a specific machine. It takes a few hours to get results back.
“Not every lab can do PCR. It requires a very clean lab, and it’s hard to troubleshoot,” Klapperich says. “I think of my hometown hospital — there’s no way they’re doing PCR in that hospital.” Labs that can do PCR testing need special approval to run tests for patients, and getting permission can take months. Klapperich says her research lab in Boston could run around 80 tests per day if it had approval. The University of Washington virology lab, which has multiple machines and many skilled lab technicians, can do up to around 2,000.
Errors and barriers
The German test adopted and disseminated by the WHO scanned for three specific viral genes. Countries around the world were able to adopt and scale that test. South Korea, for example, quickly started testing around 10,000 people each day for the virus. So far, South Korea has tested around 250,000 people.
Instead of using the WHO protocol, the CDC decided to create its own test using three different genes from the German test. That’s not an entirely unusual decision — the agency also created its own test during the Ebola outbreak in West Africa in 2015. But when some state labs tried to validate the test, it appeared to cause false positive results. The agency had to redesign and remanufacture test kits, which contributed to the delays in getting widespread testing up and running.
Normally, state public health labs and commercial labs would have been able to create their own PCR tests once it was clear the CDC test wasn’t working. But because the country is under a public health state of emergency, any new test had to be approved for emergency use by the Food and Drug Administration (FDA). That slow process also contributed to delays in testing. On February 29th, the agency relaxed that rule, and commercial labs started COVID-19 testing.
Machines are available that can run PCR tests automatically, which also speeds the process. However, it took companies weeks to develop tests for those automated systems and for the Food and Drug Administration to approve them. The pharmaceutical and diagnostics company Roche, for example, didn’t get approval for an automated test until March 13th.
Experts say the lag time and slow rollout was confusing. South Korea is working with the same technology as the US and was able to ramp up testing much more rapidly — although, notably, it doesn’t have the same regulatory barriers that the US does. But the US also has experience doing this well. “During H1N1, that test came out quickly and was distributed,” Klapperich says. “I’m not clear what went wrong here.”
Point of care
PCR tests are still the most reliable method of diagnosing a viral infection in a patient, even if the US is currently struggling to get up and running. But researchers around the globe are also accelerating work on other types of tests that can be done more quickly and closer to patients.
Speed is important during a pandemic, Yager says, because — among other things — it lets doctors figure out who has a virus quickly. “What you don’t want to do is bring someone with COVID-19 into a geriatric ward. You want to separate them. You want something that would give you a result in 10 to 20 minutes,” he says.
Doctors regularly run flu tests in their offices and can have results back in around 15 minutes. Those types of tests don’t look for influenza genes in a patient sample. Instead, they look for proteins on the surface of the virus. Those types of protein-based tests for the novel coronavirus are in development, Klapperich says. “It’s attractive because it’s a fast way of doing the test,” she says.
However, those tests are far less sensitive than gene-based tests and are more likely to generate false negative results — to say that the virus wasn’t in a sample when it actually was. Proteins on the outside of viruses also change more frequently than their genes do, so the test could stop working. “The tests are only as good as what you know about the protein,” she says. Gene-based tests like PCR tend to be more accurate.
PCR is hard to bring to a patient’s bedside, though, because the thermocycling needed to run the test needs a lot of electricity. But there’s another approach to gene-based testing, called isothermal amplification, which is simpler and can work at a fixed temperature. “There are many ways to make isothermal methods work. Each one has its own drawback, and none is quite as convenient as PCR. But the drawbacks can be overcome because they’re smaller, simpler, and more lightweight,” Yager says.
Yager’s lab created a small isothermal system that could be used outside of a lab in 2017, and other labs have created similar systems. But even though isothermal tests have been put together in labs, they haven’t been manufactured on a large scale. “We had a commercial partner who decided about three quarters of the way through that they did not want to commercialize our test,” he says. “It was the commercial market that actually caused the project — which was technically successful — to fail.”
Now, dozens of companies are developing point of care tests for the novel coronavirus and are in different stages of development. For example, doctors at the University of California, San Diego are evaluating an experimental system that they say could return results in one hour; Mammoth Biosciences is developing a bedside test that would work similarly to a pregnancy test; and the biotechnology company Cepheid is leveraging their flu detection technology to build a rapid novel coronavirus test.
Normally, PCR methods work well. But in an infectious disease outbreak, when diagnosing patients quickly is so important, having the infrastructure to develop point of care tests would improve the response — to this and future pandemics. “The tools are there, for engineers. We have to put them together, which requires a lot of systems to work together,” Klapperich says.
But a lack of sustained investment over the past few years means the technology these tests use isn’t as well known as standard PCR and is still largely experimental. There isn’t a pipeline in place to ramp up point of care testing and get it to health care facilities in the same way there is for PCR. It’s also still not clear how exactly these types of tests compare with tried-and-true PCR. “That’s not because we haven’t been yammering about this for a while,” Yager says. “Companies haven’t seen a real need. Until now.”