COVID: antibody testing, and concerns
a recent summary reviewed the status of SARS-CoV-2 antibody testing (see https://www.covid19compendium.com/1486-2/ ), brought to my attention (yet again) by jon pincus, with comments incorporated below by him, anna wald, and julita mir
sorry in advance. this is a really complex issue, the data are a bit all over the place and sometimes not completely detailed, the tests themselves are of very questionable quality, and i have included lots of studies. so hopefully the conclusions are not obscured by all of the details. but accurate antibody testing is a really important issue and i thought it made sense to delve into some of the complexities. a much easier blog tomorrow n hydroxychloroquine...
Details:
--there are >90 FDA-nonapproved tests on the market (ie, validated by the manufacturer, but allowed on the market without FDA approval)
--there are 3 tests authorized by the FDA based on performance data submitted to the FDA: cellex, ortho clinical diagnostic, and chembio diagnostic systems (the latter requires a small reader device); see hyperlinks in the document for details. Abbot labs is also releasing a new antibody test
--of note, there are problems (perhaps some quality concerns, perhaps many geopolitical: the current swipes between trump and china) leading to stopping shipments of cellex from china to new york, the curent epicenter of covid-19 (see https://abcnews.go.com/International/crucial-coronavirus-antibody-tests-destined-york-city-caught/story?id=70199489 , from 4/17)
--NIH is beginning a study this month to assess 10,000 adults without documented Covid-19 or current symptoms consistent with Covid-19 who have positive antibodies (including people who think they recovered from covid-19 but were never tested): blood samples will be collected either at the NIH or at home after a virtual clinic visit. unclear what test they will be using, to my perusal. But, at least this is being done .... and, hopefully with a reliable test (see below)
--a Chinese study found seroconversion of 173 patients (using combined IgM/IgG assay) was 11 days after symptoms, with antibody better than PCR afer 8 days (serocoversion at median 5-14 days), though the RNA remained positive for the full study (> 21 days). not clear what test was used
--these results were consistent with my oft-cited Munich study http://gmodestmedblogs.blogspot.com/2020/03/covid-19-update-31120.html
--and do raise the issue, as with pertussis, of ordering different tests at different times from symptom onset: PCR soon after symptoms start, then comb with antibody, then just antibody later
but, there are real concerns about antibody testing, including:
--there is no clear, rigorous documentation about the accuracy of these tests
--a report from Spain suggested 30% sensitivity; one from UK found that none of the 3.4 million kits from China worked well. see https://www.businessinsider.com/coronavirus-spain-says-rapid-tests-sent-from-china-missing-cases-2020-3
--we definitely need negative controls from patients with an array of symptoms from an array of different non-SARS-CoV-2 viruses
--there are other very real concerns about the actual clinical benefit of antibody testing:
--there is lots of variability between the duration of antibody responses after prior coronaviruses:
--MERS for up to 3 years, and SARS up to 2 years (though, as per below, lack of measurable titers in the lab do not necessarily translate to lack of immunity)
--and the protection from common coronaviruses does not seem to last: people keep getting infected with them every winter
--though there does not seem to be many cases of reinfection with SARS-CoV-2 documented (there was one early case i read about in china, but hard to know for sure, and our limited testing makes it even harder to document here). this does suggest that there is at least short-term protection from an actual infection, but:
--i have never seen any rigorous data showing that people who develop antibodies after a real infection and have no evidence of viable virus left (eg neg PCRs or tissue cultures) do not get reinfected with very mild or asymptomatic cases but can then carry and potentially transmit the virus
--as per prior blog: 4 health care professionals developed Covid-19 in Wuhan, had subsequent 2 PCRs that were negative, but 5-13 days later had positive PCR (and repeated with a kit from another manufacturer: see covid pos PCR after 2 negs jama2020 in dropbox, or doi:10.1001/jama.2020.2783). they all remained asymptomatic and no family members were infected. but does this secondarily positive PCR reflect suppression then re-emergence of a live virus? Or are these actually asymptomatic reinfections??? are these people able to transmit virus to others??? unclear. would have been good to have viral cultures in these people when the PCR is positive.
--in terms of vaccine-induced immunity: some antibodies are more protective than others (ie future vaccines need to be people-tested and we should not assume that a good IgG response is sufficient)
--another recent study found a reliable test from Mt Sinai Hosp in New York, testing 50 banked human serum samples from patients with and without confirmed viral infections (hantavirus, dengue, coronavirus NL63) and 4 samples of plasma/serum from 3 COVID-19 survivors (see https://www.medrxiv.org/content/10.1101/2020.03.17.20037713v2.full.pdf ). they did find neutralizing antibodies in these subjects. but quite a small study.... and limited data on specificity of test from other viruses. is this small study generalizable???
--a recent Chinese study of 175 patients found that neutralizing antibodies for SARS-CoV-2 were detected 10-15 days after disease onset, but higher levels in those who were older, had higher CRP levels and lower lymphocyte counts. but 30% failed to develop high titers of neutralizing antibodies (from the article, it seems to be in the 1:1024 range or higher), and they were not detectable at all in 10. also none had detectable viral RNA (see https://www.medrxiv.org/content/10.1101/2020.03.30.20047365v1.full.pdf ). so, we cannot necessarily assume that prior infection (and perhaps especially milder ones, or ??asymptomatic ones, will lead to likely clinically significant increases in neutralizing antibodies)
--these neutralizing antibodies did not cross-react with the old SARS-CoV virus. (ie, we should not be surprised if neutralizing antibodies against SARS-CoV-2 do not work against future coronavirus infections, which are bound to come; or even if there is a significant genetic shift in the current one. so far the genetic shifts that we have found seem to be minor ones.)
overall:
-it is unclear that the antibodies we are measuring are necessarily the ones leading to human protection from infection (eg for vaccine development). the reasonable assumption is that neutralizing antibodies are likely to be the protective ones. and there are some suggestions that a titer > 1:640 is protective (and that is the one used for transfusion therapy: see http://gmodestmedblogs.blogspot.com/2020/04/covid-convalescent-plasma-seems-to-help.html). and, we do not know how long these neutralizing antibodies last in a titer that is protective
--and looking at titers of just IgG itself may not correlate with protection: antibody titers can wane to being below our arbitrary cutpoint and still be effective in preventing further infection (as with hepatitis B), eg undetectable antibodies may be stimulated from memory cells and lead to a protective anamnestic response). and high titers may be of antibodies that are not protective
--also, of note: a study of 64 of 200 healthy Chelsea residents in Massachusetts, our local hotspot, found that 1/3 had positive antibodies. about 1/2 of them had at least one potential covid-19 symptom, though those with known Covid-19 infections were excluded. would have been great if they also checked viral loads to see if these antibody-positive people still might have had transmissable virus. they used the BioMedomics test (see https://www.biomedomics.com/products/infectious-disease/covid-19-rt/ ), a 10-15 minute test, claiming a sensitivity of 89% and specificity of 91% (unclear what these numbers are based upon). see https://linkprotect.cudasvc.com/url?a=https%3a%2f%2fedition.pagesuite.com%2fpopovers%2fdynamic_article_popover.aspx%3fartguid%3d945fc329-63a4-4b4f-829d-93bdfef6d813%26appid%3d1165&c=E,1,nKn_Ur7L-aOpfMQCCwSRQDEl8_4wzVxKOWSlhEGDVBXo9uTNyfOhyWNyw5gJHkAOkAPLk02CWMr3quJNmNgdurf8bbIuUk1xjbuT069meIXK4g,,&typo=1
--i should reinforce that sensitivity and specificity are characteristics of the tests, that the numbers of false positives and negatives are dependent on the prevalence of disease in the community (or, also, the pretest probability). For example the above test specificity of 91%, which sounds pretty good, in a setting where there is a 2% prevalence, actually translates to 2 true positives in testing 100 people, but 9 false positives!! (ie, 4.5 as many patients per 100 having a positive test result really have not had infection for every person correctly identified….)
--and, even in Chelsea with a 33% prevalence, there woud be 33 true positives and 6 false positives per 100 people (ie, about 1 in 5 chance someone was incorrectly told they had the infection and were immune). the point here is that there really needs to be a highly specific test, in the 99% range...
so, reliable antibody testing is really, really important for many reasons:
--when people can go back to work, limiting self-isolation etc
--when health care workers or others can safely take care of Covid patients/friends/family
--improving our understanding of the epidemiology, and specifically the incidence of immunity in a community (which might show enough herd immunity to be protective overall in the community, now pegged at about 65%)
--the ability for antibody-positive patients with sufficient neutralizing antibodies to donate plasma for transfusion to sick Covid patients
--but we really do need to make sure that the antibody test is accurate (very high sensitivity and especially specificity), easy, rapid and available enough to test very large numbers of people
also there was an interesting scientific american article on immunity last week: see https://www.scientificamerican.com/article/what-immunity-to-covid-19-really-means/
geoff
If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@uphams.org
For access to the dropbox, go to link: https://www.dropbox.com/sh/0bmvtita8mzms11/XDTwHySFFg
Then go to "clinic", then to either "clinical stuff" for articles, or "powerpt presentations" for the powerpoint presentations
to get access to all of the blogs:
2. click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category
3. or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them
please feel free to circulate this to others. also, if you send me their emails, i can add them to the list
Comments
Post a Comment
if you would like to receive the near-daily emails regularly, please email me at gmodest@uphams.org