Read on Twitter
1. First the thanks: @tyler_ripperger @JenniferUhrlaub, Rachel Wong, Makiko Watanabe, Ryan Sprissler, Janko Nikolich-Zugich, @UAZHealth, @UAZBIO5, @ImmunobiologyUA, and many others not on Twitter. Huge amt of work. https://medrxiv.org/cgi/content/short/2020.08.14.20174490v1
2. Back in April we decided (were voluntold?) to set up a serological assay to be used across AZ, before the worst had hit. The mandate was for high-risk jobs (e.g. HCW, etc.). We began by copying @florian_krammer with RBD ELISAs. But low seroprevalence makes things tricky...
3. RBD worked great to pick out positives, but there were still a few duds that got in. Of ~6000 people, we saw 73 positives based on RBD signal. Problem is that 13 did not have neutralizing Abs. Could be that they cleared the virus w/o, but more likely to be false positives...
4. That is only a 0.2% false positive rate, which is really good. But because we had so few true positives, it meant that ~20% of the positives would be wrong. Since neuts were not done in CLIA, those results are not returnable. We needed an orthogonal antigenically distinct test
5. Now briefly on to the next topic du jour. We also followed Ab titers over time both across subjects and for some of them longitudinally. Bottom line is that this looks like a pretty conventional Ab response. Early rise, a partial decline, and a more stable 2nd phase that lasts
6. ...for at least a few months. This is similar to work by @florian_krammer , @JenGommerman , @PepperMarion , @ForrestKJones. Reports of the death of antibodies have been greatly exaggerated.
7. Final thanks again to @TheBcellArtist and Daved Fremont at WashU and @florian_krammer and the Path lab at Sinai for getting us started.
