The need to understand and track the spread of SARS-CoV-2, the virus that causes the disease COVID-19, has highlighted the importance of diagnostic testing. Widespread diagnostic testing is needed not only to identify who is sick right now, but also who is at risk, and critically, who is immune due to previous exposure. Discussion of diagnostic testing for SARS-CoV-2 is omnipresent, and as you have likely gleaned from the news, not all tests are equally effective or rapid and there are huge issues around availability.
But what kinds of tests are needed, and what do these tests actually test for?
First and most acutely, we need tests to determine if a person is actively infected with the virus. This allows for infected people to be isolated and appropriately treated. Tests of this nature are called viral tests, meaning that they test for the presence of the virus in a patient sample.
The second type of test needed determines if a person is immune to the virus due to previous exposure. These tests will be critical for knowing whether a person can return to public life and help reignite the economy. Known as serologic tests, these tests identify if a person has antibodies that can recognize and respond to the virus, preventing future infections in that person.
Another way to categorize diagnostic tests is whether they are performed in a laboratory or at point-of-care; laboratory tests will always take longer since they require clinicians to ship samples to a centralized laboratory, but point-of-care tests take longer to design and deploy since they need to be robust against user errors.
One of the most important trade offs made when designing or selecting diagnostic tests is between the rate of false positives (telling someone they are sick when they’re not) and false negatives (telling someone they are healthy when they’re not). Obviously, neither is good, but which false result is more or less damaging depends largely on the context. In the case of the COVID-19 pandemic, false negatives are far more dangerous than false positives due to the highly contagious nature of the virus.
Here are the major test types used for viruses, some of which are currently in use or under development for SARS-CoV-2. You can find more detailed information at the FDA’s website.
Quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) tests are front and center as they were the first to be developed by the CDC and others. Since SARS-CoV-2 has a genome made of RNA, the first step is to use a reverse transcriptase enzyme to translate RNA into DNA in a patient sample.
To determine whether the virus’s genome is present, the test executes a “find and copy” (to use an analogy from computing) function on that DNA. Carefully designed molecular probes seek out a specific sequence that is known to be unique to the virus. If they find the sequence, they copy it many times generating large quantities of double-stranded DNA (more on how that works in this episode). Fluorescent molecules tag the copies of double-stranded DNA, so fluorescence measurements can be used to figure out roughly how much virus was present in the original sample.
To do this test, you need to know what sequence you’re looking for, which is why it was so crucial that scientists published the SARS-CoV-2 genome sequence within weeks of the first few cases. Thanks to the code-like nature of DNA, these tests can be developed very quickly. Scientists can take off-the-shelf qRT-PCR tests and simply modify the “search” sequence in their molecular probes. However, for clinical use these new sets of probes need to be tested and approved by the FDA, or in the case of an emergency, issued “Emergency Use Authorization” (EUA), which still takes some time. A list of all EUA can be found here.
Rapid RT-PCR tests are RT-PCR tests that can be done at hospital triage or at the bedside in less than 30 mins, without needing a laboratory.
Companies that develop these tests optimize the qRT-PCR protocol to speed up the DNA copy step described above and automate the entire process. The downside is that the tests have to be run on proprietary instruments, so it’s only available at hospitals that have invested in those instruments. The most common example is the rapid flu test. But what you gain in speed, you lose in accuracy — multiple studies (Chu et al, Chartrand et al., CDC guidelines) report that rapid flu tests miss a large fraction of cases due to low test sensitivity — errors that would be really dangerous in the current COVID-19 pandemic.
The FDA has issued EUA for two rapid qRT-PCR tests, the first is called Xpert Xpress SARS-CoV-2 from Cepheid and the second is called ID Now COVID-19 from Abbott. Cepheid’s GeneXpert readers and Abbott’s ID Now (f.k.a. Alere) readers are available in many, but not all, hospitals and clinics in the U.S, so while these tests will be widely available, they won’t be everywhere.
Viral antigen tests look for any molecules that are only found in the virus and that can be recognized by a type of specialized type of protein called an antibody. To detect these viral molecules — called antigens — scientists use a test called an immunoassay.
All immunoassays follow a similar principle: an antibody is designed to find the antigens in a patient sample and create a color or fluorescence signal if successful. These tests can be done over the course of a few hours in a laboratory. However, as with qRT-PCR, diagnostics companies have developed some rapid tests (less than an hour) that automate the process. Some of these rapid tests use “lateral flow” of samples across a specially designed test strip to determine if there is a reaction between antigens and antibodies.
Antigen tests take longer to design compared with qRT-PCR tests since antibody design isn’t as much of a plug-and-play process as designing DNA-based tools. Many groups are working on antigen tests for COVID-19, but none are available for clinical use.
Serologic antibody tests, as mentioned above, are the only one of these tests that can detect whether a patient has ever been infected, rather than whether or not they are currently infected. These tests search the immune system’s “memory” to look for evidence that the patient’s immune system has ever seen the virus.
Under the hood, these tests are basically the reverse of the “viral antigen immunoassays” described above: viral particles are added to patient blood samples; if antibodies from the patient’s immune system bind them, the test produces a colored or fluorescent signal. Like viral immunoassays, most serologic tests take several hours and must be done in a lab. Rapid paper-based serologic tests using lateral flow (more on this test type below) are also in the works.
As we learn more about molecular biology, diagnostic tests are evolving as well. At a16z, we’re tracking some of the newest diagnostics technologies that scientists in academia, companies, and startups are working on today. Here are some of the most promising technologies we see on the horizon:
— You already know of the most famous paper diagnostic: the home pregnancy test! The name refers to the form rather than to the molecular details of the test, but in general, paper diagnostics can typically be thought of as quick-and-cheap viral antigen tests. In these tests, a patient sample is put on a piece of paper that’s coated with reagents for doing a viral antigen test. These parts of the paper change color when the virus is present, similar to how a colored line appears on a positive pregnancy test. These tests aren’t always the most accurate option, but they are the holy grail for point-of-care testing in low-resource countries since they don’t require a laboratory or any special instruments. Today, however, they’re mostly developed as academic projects since there’s a significant risk of user error and the commercialization opportunity in these low-resource environments is challenging. In the context of COVID-19, however, these tests could be a major game changer, since they would be cheap to produce and could be taken at home, reducing the strain on centralized clinics or hospitals.
— CRISPR-based technologies have been making headlines in therapeutics for the past decade, but these tools can also be used for diagnostics. Two startups — Sherlock Biosciences in Boston and Mammoth Biosciences in the Bay Area — are developing tests that use proteins from the CRISPR toolbox to search for SARS-CoV-2. If they find a gene sequence that matches the novel coronavirus, they create a colorimetric signal that can be read out on a test strip.
— Metagenomic sequencing (also called meta next generation sequencing or mNGS) is the opposite of a targeted search — it’s the readout of all genetic sequences present in a patient sample. This approach is definitely overkill if you know exactly what you’re looking for, but it’s the most comprehensive method for discovering what a patient may be infected with in a single go. Rather than doing targeted tests for each potential culprit — first a flu test, then a COVID-19 test, then a rhinovirus test, etc — this single test can identify any pathogen infecting the patient, even if it’s a novel strain of virus or bacteria.
Today, attempts to use metagenomic sequencing in the COVID-19 epidemic are limited to research use only. For example, a startup called IDbyDNA is partnering with sequencing giant Illumina to use metagenomics for epidemiological studies. mNGS is rare in clinical settings because it is still too expensive and slow. As the cost of sequencing rapidly decreases and the pool of genomics data in the world rapidly increases, mNGS will increasingly become used to identify pathogens in clinical settings — first in complicated cases, but we predict that eventually it will be used simply to run one definitive test for each patient.
* * *
Diagnostics are essential weapons in our arsenal against emerging pathogenic threats. As the COVID-19 pandemic has shown us, rapid and widespread testing capabilities can determine the course of a pandemic. Technologies that enable us to develop novel tests quickly, that provide results accurately and efficiently, and to distribute these tests rapidly and cost-effectively, will all be needed to tackle this and future outbreaks. Watch this space.
Judy Savitskaya is the cofounder of a stealth startup.
Jorge Conde is a general partner on the Bio + Health team at Andreessen Horowitz, focused on therapeutics, diagnostics, life sciences tools, and software.