This one is going to take some explaining, but it is relevant for the search to figure out what level of antibodies is protective.
The study measured antibody levels over time in dental health care workers in the UK. It turns out that dentists and dental hygienists are at higher risk of catching covid than are their receptionists. This makes sense, given how close dental professionals have to get to someone's mouth.
But the really interesting thing is that they followed people, taking blood samples at three and six months, and asking about second infections in those that already had prior infections. They found seven people who had reinfections. All were symptomatic. When they looked at their blood samples from before the reinfection, they found that their antibodies were below the level of detection of their antibody test.
Even more interesting is their comparison of their antibody test to a set of international standards for measuring antibody activity. I wasn't aware that these existed until I saw this paper. This means that we will hopefully start seeing researchers using IU/ml (international units per milliliter). There are already international standards for some other antibodies, for example to measles. This means that you can get your blood tested for antibodies to measles and get an answer back in international units. You can then compare that number to the number known to be protective.
These researchers compared their antibody test to two international standards. This means that they ordered antibodies with a standardized amount of activity and ran them through their test. They used two standards and came up with similar but not identical results. The 20-136 standard is the first one set up by WHO, and the 20-162 standard is a blend from people with high amounts of antibodies but low neutralizing titers, which is typical in those recently recovered from mild covid or a those that had a single shot of RNA vaccine. Using the 20-136 standard, the lower limit of detection of their antibody test is somewhere between 20 and 30 IU/ml.
This is where it gets even more interesting. The documentation for the Roche test (used by Labcorp) says that it is calibrated against their own internal standard, which is a mixture of two different monoclonal antibodies (so they can get as much as they want forever) that bind to the spike RBD. Specifically, 1 nM of these antibodies equals 20 U/mL in the Roche test. So they are not calibrating to an international standard.
However, the UK-based National Institute for Biological Standards and Control (which keeps samples of the standards and sends them out to researchers) has run the 20-136 samples in the Roche / labcorp test. They ran a series of dilutions, and from them we can estimate that a sample that is 1000 IU/ml is about 780 U/ml on the labcorp test. So to convert from Roche/Labcorp's U/ml to IU/ml, you multiply by 1.3, or to convert from IU/mL to Roche U/ml, you divide by 1.3. This calculation only works for Roche results over 10 U/ml.
So if we go back to the 20-30 IU/mL, we can convert that to Roche 15-23 U/ml. I had previously estimated 20-50 using an entirely different set of studies. This is a really good sign that we have a reasonable idea of where the threshold of protection is -- so far it appears to be about 20. This study only had seven reinfections, so I'd really like to see a larger study before anyone makes it official. But this estimate is helpful for people trying to make decisions now.
I've heard from a lot of people who are nervous because their antibody levels are lower than those of other people. Just because theirs are higher doesn't mean that yours aren't protective.
I still think the point where I would get a booster (barring B.1.351 in the US) is still about 50, just because I'm only checking every two months and it can decline over time.
If the true threshold of protection is 20, and not 50 on the labcorp test, that also means that the threshold for the B.1.351 would also likely be lower than my previously estimated 200-500, which is a good thing.
Clearly we need more data here, but I will take whatever I can get. And the establishment of an international standard is a very good thing.
Resources / Sources / Links for further reading
Study on dentists:
https://journals.sagepub.com/doi/pdf/10.1177/00220345211020270
Numbers for dilutions of international standard 20-136 run on the Roche / Labcorp test:
https://static1.squarespace.com/static/5739222a27d4bd28d98e3ce9/t/60669eb5c894556d34c420ee/1617338039275/20210326_NIBSC_Serology_Slides.pdf
WHO document on the establishment of the original international standard:
https://cdn.who.int/media/docs/default-source/biologicals/ecbs/bs-2020-2403-sars-cov-2-ab-ik-17-nov-2020_4ef4fdae-e1ce-4ba7-b21a-d725c68b152b.pdf
Info sheet on the Roche / Labcorp test:
https://www.fda.gov/media/144037/download
Info sheet on international standard 20-136:
https://www.nibsc.org/documents/ifu/20-136.pdf