The FDA recently approved a new type of coronavirus test, an ANTIGEN TEST. Do NOT confuse this with an ANTIBODY TEST! Finding a viral antigen (viral “piece”) in someone tells you the virus is in that someone – now. It’s like catching a robber red-handed. But finding an anti-viral antibody in someone is more like finding a fingerprint the robber left on his way out – it tells you someone was infected in the past (or is currently recovering).
Biochemically, antigen tests works similarly to an antibody test, BUT in REVERSE. Antibody tests are used to detect little proteins called antibodies that your body makes to specifically bind to viral parts to help fight off infections and prevent re-infection. Antigen tests look for those viral parts, which are called antigens and are usually parts of viral proteins. So, these tests give you very different information. Antibody tests tell you if someone “has been” infected – in the past or currently (though antibodies don’t show up until later in the infection). Antigen tests tell you if someone is currently infected – they detect the virus itself which is only there when you’re actively infected, so these tests don’t tell you anything about past infections. Antigen tests are thus another form of diagnostic test, joining the group led by the gold standard RT-PCR tests, but not performing as well, missing a lot of infections. Here’s the science cuz, biochemistry!
Format-wise, they’re “lateral flow assays” – strips in a cassette similar to pregnancy tests, but instead of peeing on them, you “snot on them”. It’s not quite as simple of just blowing your nose (but wouldn’t it be awesome if in the future they had Kleenex that changed color in response to specific infections?…). Instead, how it works is that they stick a swab in your nose (one of those giant not-q-tips), they swirl it around a bit to get a good sample, and then they swirl that around in some “magic liquid” that has chemicals that break open the viral particles (lyse them).
These “viral particles” consist of the virus’ genetic blueprint (genome), which is a single strand of RNA, and is what the RT-PCR tests look for. The RNA is wrapped up a coating of a “nucleocapsid” protein (N), and enclosed in a protein-studded lipid membrane. One of the proteins studding this oily membrane is the spike protein (S), which is the one that juts out “crown-like.” Because this S protein is responsible for binding to cell receptors and allowing the virus to get into our cells, it’s been getting most of the attention. But the most abundant viral protein is actually that N protein, and it’s this N protein that the new antigen test looks for.
The exact contents of the magic liquid are proprietary, but it contains detergents – these look similar to the lipids making up the membrane, so they’re able to wedge their way in between those lipids and break up the membrane, allowing the viral contents to spill out. (this is also why soaps and detergents (which are just synthetic soaps) are able to kill the virus: https://bit.ly/soapsanddetergents )
Now that N’s out, the test needs to capture it and show it to us.
This capturing is actually done with anti-N antibodies. But these aren’t antibodies that come from the person being tested. Instead, they’re antibodies that have been made in a lab, attached to a fluorophore (a chemical group that lets of light of a specific wavelength when you shine light of a different specific wavelength on it) and put onto that strip of paper in the cassette.
These tests use antibodies for the same reason your body does – they’re good at specifically binding viral pieces, helping the body do things like flag infected cells for destruction, call for backup, and keep watch after the virus has been conquered in case it tries to return (which is why antibody tests can detect past infection).
At the molecular level, antibodies (aka ImmunoGlobulins) are “just” little proteins which have generic adapter parts (constant regions) as well as unique parts (variable regions) that allow them to bind specifically to different viral parts (antigens). “Antigen” is just a shorter way of saying “that specific thing that the antibody binds to” and it typically is part of a viral protein. Terminology-wise, we typically specify the antigen after “anti-“ So, an anti-N antibody is an antibody whose antigen is the N protein. And these are the ones the test uses – in a couple different ways.
Note: There are different layout strategies for doing these lateral flow tests & I don’t know exactly how this specific one’s organized inside the cassette since they keep those details inside the company as far as I can find. So I’m just going to describe one of the common schemes that is used and what I think some of this company’s other tests use… https://bit.ly/3cp4udG
When the clinician sticks that patient sample of spilled-out viral “guts” onto one end of the strip, the viral gut parts get a chance to interact with fluorescent beads coated with anti-N antibodies. If there is N protein present, the anti-N antibodies will latch on. And then, thanks to capillary action, the mix gets wicked through the paper strip to the other end, similarly to how water spreads across a paper towel. (capillary action seems like magic but it’s really just because water’s super sticky but also likes to spread out, so it pulls itself through spaces in the paper’s fibers, dragging its friends with it).
So, now you have your N protein bound to an anti-N antibody that’s bound to a fluorophore. All that’s left is to catch it. And we need to catch it separately from any antibody that is unbound, so we don’t confuse them. This is where the second set of antibodies comes in. A line of stuck-on anti-N antibodies is waiting to greet them when they reach that part of the strip. These bind to the (now-labeled) N protein – if it’s present. If the N protein isn’t present, the labeled antibody will keep flowing through and get captured by a line of secondary antibodies that recognize the initial antibodies. So the fluorescently-labeled antibody will get stuck in one place if N is present, and in a different place if it isn’t. Now you just need to look to see where it is.
Unfortunately, unlike the colored readout from your pregnancy tests, a fluorescent label requires some special equipment to see because fluorophores only shine light when you first shine a specific wavelength of light on them. http://bit.ly/jablonskidiagrams
It’s less special equipment than is needed for RT-PCR tests, but you still need a reader machine.
The first antigen test to get authorized in the US, is the Quidel company’s Sofia 2 SARS Antigen FIA, which got the “yellow light” (Emergency Use Authorization (EUA) only, not full approval) on 05/10/20. FYI, FIA stands for Fluorescent ImmunoAssay. The machine it uses is called a Sofia 2. It’s a little bench top machine. A lot of doctors & clinics might already have them machines because similar tests are used to test for other diseases, like strep throat, flu, or RSV.
The test takes about 15 minutes. The test has to develop (it takes some time for all those molecules to stick, wick, and restick) so there’s a wait step. And the machine has modes where you can do the wait step in the machine so you can just walk away, or you can do the wait step on lab bench so you don’t hold up the machine. Then, after the 15 minute wait, you stick it in the machine and get the result in a minute. This allows you to process more samples, though you still have to stick them in one at a time.
The test kits also come with control samples to make sure the test is working: a positive control (a swab coated with non-infectious SARS proteins) to make sure the test can detect what it’s supposed to and a negative control (a swab coated with non-infectious strep protein) to make sure the test doesn’t just always give you a positive readout.
The test is super specific – it didn’t cross-react with other seasonal coronaviruses, like the milder ones that cause some common colds, so you’re unlikely to get a false positive. BUT the test only has 80% sensitivity – which means that, if you were to test 100 people who had the disease, 20 of them would be told they’re fine. For this reason, according to their brochure, “Negative results should be treated as presumptive and confirmed with a molecular assay, if necessary for patient management.” https://www.quidel.com/sites/default/files/product/documents/EF1438900EN00_0.pdf
More on how to interpret test accuracy: https://bit.ly/testterminology
I didn’t learn the details I wanted from their brochure, but here’s another factoid I picked up: because the N protein of SARS-CoV (the original) & SARS-CoV-2 (the current coronavirus) is so similar, the test can detect either. It can’t tell you which one you have but it’s super super super unlikely it’s the “plain” SARS-CoV
This test is currently only approved for doctor/clinical lab use, but I’ve heard hype about companies thinking they can test all their employees everyday before work and prevent infection from getting in. They need to know that they’ll only be able to keep out 80% of infected people with this screening method. And once the virus gets in… So, in “screens” these tests can be useful for keeping the infection rate lower, but not for preventing it entirely. And, if you’re a person who gets a negative result, it’s not exactly comforting to know that your test is technically only “presumptive” until confirmed by PCR.
Another downside of the only currently approved antigen test is that it requires that machine to process. So, although the tests themselves are cheap (~$5) clinics/labs, etc. that want to run them also have to buy this machine if they don’t have it. Not great for developing nations or areas with limited resources…
I’m hoping that antigen tests will be developed that are colorimetric – they give a colored readout you can see with your eyes (think pregnancy test) instead of a fluorescent one you have to scan in a machine. This would allow them to be truly field-deployable. But a benefit of fluorescence is that it’s more sensitive. The machine is better than our eyes – and even with this increased sensitivity, the test still misses a lot of infections.
I’m hopeful about cheap, rapid tests that detect the virus’ genetic information (the thing the RT-PCR tests look for). Some of these use CRISPR technology and I will hopefully do a post on them.
more coronavirus resources (including details on how the other tests work) here: https://bit.ly/covid19bbresources