Do commercial antibody tests indicate protective immunity to #COVID19?
An important question.
Answer: We don't know, but correlations w/ neutralizing titres (NT) are weak enough that one should NOT count on it.
Short thread.
doi.org/10.1101/2020.0…
An important question.
Answer: We don't know, but correlations w/ neutralizing titres (NT) are weak enough that one should NOT count on it.
Short thread.
doi.org/10.1101/2020.0…
Background:
There are many commercially available antibody tests for COVID19. Most give you a qualitative (Yes/No) answer to whether you have antibodies from exposure to SARS-CoV-2. We still don't know how protective immunity from previous exposure is, nor duration of immunity.
There are many commercially available antibody tests for COVID19. Most give you a qualitative (Yes/No) answer to whether you have antibodies from exposure to SARS-CoV-2. We still don't know how protective immunity from previous exposure is, nor duration of immunity.
But at some point we will know these things and people will want to know their status and will take a commercial test.
Setting aside issues w/ false positives (1-specificity), a fair question to ask is what the Yes/No answer people get back from test is means.
Setting aside issues w/ false positives (1-specificity), a fair question to ask is what the Yes/No answer people get back from test is means.
One surrogate measure of the strength of immunity is the neutralizing titre of someone's blood/plasma: take live virus & mix it with different dilutions (1/titre) of plasma and put it on monkey vero cells. If virus is not neutralized it infects cells & forms foci.
If plasma has strong "neutralizing" effect, it binds to virus and virus doesn't infect cells. This is an effort to model what might happen when virus gets into our bodies - will our antibodies bind virus and stop it from infecting our cells?
But it's a cell culture assay w/ just plasma and our actual immune system is much much more (see interesting paper on T-cell immunity w/out antibodies:
medrxiv.org/content/10.110…).
So take results w/ grain of salt. It's not a animal/human challenge study but it's a start.
medrxiv.org/content/10.110…).
So take results w/ grain of salt. It's not a animal/human challenge study but it's a start.
New paper correlates numerical values from 3 commercial assays (Roche, Abbott, EuroImmun - 3 big names & widely used tests) w/ neutralizing titres and compares results w/ cutoffs recommended by companies for calling plasma sample positive for antibodies.
Correlations b/w tests and neutralizing titres are unfortunately pretty weak, & commercial cutoffs call many samples with very low titres +.
Translation: a + test does NOT mean you have a strong neutralizing response.
Translation: a + test does NOT mean you have a strong neutralizing response.
Since we don't KNOW what titre is required for you to be "immune" to re-infection (preventing infection, infectiousness, disease or some combo), authors use FDA cutoff titre (Dilution: 160) for donating plasma that is now being used to treat patients.
In each panel note the fraction of all dots above the dashed line that are to left or right of where the 160 dilution line would be (b/w 128&256). Many (~50%) are to left which means ~half of the sera testing + have only weak neutralizing effect.
Conclusion: A + antibody test from these commercial tests does NOT guarantee a strong neutralizing titre.
How well it indicates actual protection, especially when combined with T-cell (which is partly correlated w/ Ab: cell.com/cell/fulltext/…) & other factors is not yet known.
How well it indicates actual protection, especially when combined with T-cell (which is partly correlated w/ Ab: cell.com/cell/fulltext/…) & other factors is not yet known.
Free Add-on
Original paper also shows that people w/ worse clinical outcomes have higher neutralizing titres but this pattern is not detectable by commercial assays. I could speculate but I'd just be BSing.
Original paper also shows that people w/ worse clinical outcomes have higher neutralizing titres but this pattern is not detectable by commercial assays. I could speculate but I'd just be BSing.
