California Research Team Analyzes Performance of 12 Serology Tests

Consistent method used to evaluate different serology tests

CEO SUMMARY: Researchers with the COVID-19 Testing Project used a multidisciplinary effort to analyze and compare the performance of 12 serological tests. One finding is that, 16 to 20 days or more after a confirmed infection with a molecular test, many of the 12 serological tests analyzed were 80% positive. Using donated blood samples, the researchers assessed the volume and number of antibodies the SARS-CoV-2 produced in infected patients.

FROM MID-MARCH UNTIL MAY 4, manufacturers launched more than 200 serological tests to identify the antibodies patients produce in response to infection from the new coronavirus.

The performance of these tests was mostly¬†unknown because the federal Food and¬†Drug Administration (FDA)‚ÄĒunder the¬†rules it issued on March 16‚ÄĒallowed almost¬†all of these 200 tests on the market without¬†the usual review. Without performance data¬†except from the manufacturers themselves,¬†clinical laboratories followed the caveat emptor¬†principal of buyer beware when choosing¬†among these tests for clinical use.

The FDA reviewed the performance of only a select few of the tests that applied for the FDA’s emergency use authorizations (EUAs) before May 4, when the FDA issued new guidance for serological tests for SARSCoV-2, the virus that causes the COVID-19 illness.

Seeking to fill the gap in performance review, a team of researchers analyzed the performance of 12 of the serology tests in a significant effort to do a comparative review to benefit clinical laboratories, public health officials, patients, and clinicians.

The researchers provided a significant¬†contribution, because the FDA‚Äôs March 16¬†policy for serological tests provided what¬†the agency called ‚Äúregulatory flexibility¬†for developers offering such tests without¬†FDA review and without an EUA.‚ÄĚ Test¬†manufacturers were required to simply¬†notify the agency that they had validated¬†their tests, and they needed to provide disclaimers¬†about the limitations of the tests.

Additionally, the federal agency added that,¬†‚ÄúThe FDA does not review the validation,¬†or accuracy, of data for these tests unless an¬†EUA is submitted.‚ÄĚ

Emergency Use Authorization

As of May 7, the FDA listed the performance data for only 12 serological assays that had received EUAs and yet, according to published reports, the agency’s website listed 200 or more serological assays that were approved for sale in the United States.

The regulations on March 16 were intended to make it faster and easier for test manufacturers to bring tests to market to track the SARS-CoV-2 pandemic. As a result of requiring little in terms of supporting performance data, test manufacturers flooded the market with tests, some of which performed poorly, produced inaccurate results, or were fraudulent.

In response, on May 4 the FDA issued¬†new rules requiring all companies seeking¬†to sell serological tests in the United States¬†to apply for EUAs. According to a review¬†of FDA data on May 8, 116 were no longer¬†authorized and 12 had received EUAs. The¬†FDA site made no mention of what became¬†of the other antibody tests, suggesting that¬†the manufacturers may have withdrawn them¬†from the market. (See ‚ÄúFDA Replaces March¬†16 Serology COVID-19 Test Rules‚ÄĚ in this issue.)

Researchers Step Up

In the seven weeks between the FDA’s March 16 decision and the May 4 revised policy on antibody tests, there was no accepted, industrywide standard for a high-quality COVID-19 serological assay for labs and pathologists to use for comparison. Therefore, the team of researchers from two universities and two bioengineering companies in California decided to develop a way to compare the performance of 12 serological tests on offer.

In April, the researchers published their findings online. The lack of performance data had created a pressing need for an independent body to produce a consistent method to evaluate the performance of serological tests for the SARS-CoV-2 infection, said Patrick D. Hsu, PhD, an Assistant Professor and Faculty Fellow at the University of California, Berkeley. A bioengineer and geneticist who normally works in genome editing, Hsu is one of the researchers leading the multidisciplinary COVID-19 Testing Project (

The researchers involved in the project represent UC Berkeley, the University of California San Francisco, the Chan Zuckerberg Biohub, and the Innovative Genomics Institute.

Given the urgent need for comparative data on the assays, the researchers performed head-to-head comparisons of commercially-available lateral flow assays (also known as rapid serology tests) and enzyme-linked immunosorbent assays (ELISAs), the researchers wrote on the web.

‚ÄúWe do not have a gold standard for¬†evaluating serological tests, and that‚Äôs¬†something we look for in modern medicine,‚Ä̬†Hsu commented in an interview¬†with The Dark Report. ‚ÄúTherefore,‚Ä̬†he continued, ‚Äúwe conducted this study¬†by taking in case samples from PCR-positive¬†patients that have been seen in¬†San Francisco hospitals.‚ÄĚ

Hospital Patients’ Specimens

Upon admission, those patients were given the reverse transcription-polymerase chain reaction (RT-PCR) test. Over the course of their hospital stay, blood from these patients was drawn multiple times for follow-up testing and the researchers used those samples in their analysis.

The results of the research could¬†inform healthcare providers and public¬†health and government officials conducting¬†serological testing for the new coronavirus,¬†Hsu explained.¬†‚ÄúTo power our analysis adequately,¬†we used different time intervals and then¬†placed all the samples into five-day bins,‚Ä̬†Hsu explained. ‚ÄúThis is a question of biology,¬†because we have patients who were¬†seroconverted when their antibodies were¬†expected to rise.

‚ÄúWe know from past research that¬†the IgM antibody is generally thought to¬†rise first, whereas IgG antibodies can take¬†more time while the immune system revs¬†up,‚ÄĚ noted Hsu. ‚ÄúBut without making any¬†assumptions about what type of antibodies¬†would come first and which type of¬†tests might be more sensitive, we wanted¬†to compare all of the assays head-to-head¬†in a very systematic fashion.

“To do that for each of the time-point intervals for every specimen, we tested the performance of each one against the lateral flow assays (LFAs) and against two ELISAs.

LFA versus ELISA Bake-Off

‚ÄúFor each specimen, we wanted to evaluate¬†how the tests performed, collect that¬†data, and then compare the results,‚ÄĚ he¬†said. ‚ÄúThat way we could see what comes¬†from this bake-off.

‚ÄúGenerally what we found is that, at¬†16 to 20 days, or at more than 20 days,¬†many of these tests were over 80% positive,‚Ä̬†Hsu explained. ‚ÄúWhat we don‚Äôt¬†know, however, is whether that 80% positive¬†rate should be 100% positive. All¬†we know is that these samples were PCR¬†positive before we ran our analysis.

‚ÄúWe don‚Äôt necessarily know when¬†antibodies will rise and if the antibodies¬†should all have developed to show a¬†rate of 100% positive after three weeks,‚Ä̬†observed Hsu. ‚ÄúWe certainly expect samples¬†from certain patients not to have¬†seroconverted because some might be¬†immunosuppressed, for example. When¬†we reviewed the patient data, we decided¬†that assumption might be correct.

‚ÄúAlso, we had a relatively small sample¬†size of patients who were still hospitalized¬†after 20 days,‚ÄĚ he recalled. ‚ÄúThat¬†makes sense intuitively, because after that¬†many days, these patients would likely¬†have been released because they were¬†from an ambulatory population.‚ÄĚ

Adding Assays to the Study

During the interview, Hsu said he and his colleagues were considering expanding the sample set beyond the initial 12 COVID-19 serological assays analyzed. At press time, it was not known if the researchers would continue after the FDA issued new rules on May 4. After the initial interview, Hsu did not respond to questions from The Dark Report.

‚ÄúWe released our findings as a preprint¬†so that we could report our preliminary¬†results in advance of a formal¬†peer review,‚ÄĚ he said during the interview¬†on May 2. ‚ÄúWe did that because¬†we saw the urgency of the situation¬†and the number of tests that were¬†fraudulent but on the market.‚ÄĚ If they¬†expand the sample set, the researchers¬†may want to add data from later time¬†points.

Although antibody tests have an important role to play during the pandemic, treating physicians may want to confirm a positive antibody test with another antibody test, he suggested.

‚ÄúYou could confirm results with a¬†different type of serology test that might¬†look at a different antibody isotype or a¬†different type of antigen,‚ÄĚ Hsu explained.¬†‚ÄúYou could look at IgM verses IgG versus¬†IgA. Or you could look for reactivity¬†against proteins in the virus. That type of¬†confirmatory testing could help improve¬†significantly the specificity of a result by¬†looking across multiple different antigens¬†or antibody isotypes.‚ÄĚ

From the pre-print manuscript of their findings, the researchers have posted the results of the performance analysis of the 12 assays on the web, and they are interested in adding to the data as they evaluate more tests, Hsu said.

Evaluating More Tests

Some lab companies that make COVID-19 serological tests were interested in¬†adding their tests to the dataset, he said.¬†‚ÄúThere has been an outpouring of interest¬†from many manufacturers or distributors¬†that want their tests evaluated,‚ÄĚ Hsu¬†commented. ‚ÄúWe‚Äôre trying to figure out¬†the best way to prioritize those requests¬†because there are many more tests that we¬†could evaluate, but we would need more¬†blood samples from infected patients.

‚ÄúAs researchers in this new field,¬†we also want to understand the correspondence¬†between venipuncture¬†serum‚ÄĒwhich is how our study was conducted‚ÄĒ¬†with blood from fingerstick sample¬†collection,‚ÄĚ commented Hsu. ‚ÄúPlus,¬†we want to increase the number of samples¬†that we have in the later time points¬†from symptom onset.

‚ÄúOur goal is not to exhaustively evaluate¬†every test‚ÄĒin part because that might¬†not be possible,‚ÄĚ he added. ‚ÄúAnd, this¬†work is not what most of us do normally.¬†A lot of the scientists working on¬†this project are repurposing themselves¬†to add to the response among scientists¬†to address this new coronavirus. We want¬†to highlight this issue and perhaps inspire¬†others to scale up this work to do more¬†test-performance evaluations.

‚ÄúAt the least, we hope that some regulatory¬†bodies will review our data,‚ÄĚ concluded¬†Hsu. ‚ÄúThose would be very good¬†outcomes for our work.‚ÄĚ

At a time when large numbers of COVID-19 serological tests are coming into the market, but without the comparable range of data provided with diagnostic assays that are reviewed by the FDA and cleared for use, the work of the COVID-19 Testing Project is a valuable resource for clinical laboratories.

In Manuscript for Peer Review, Researchers Describe Evaluation of COVID-19 Serological Tests

IN A MANUSCRIPT PREPARED LAST MONTH FOR PUBLICATION IN A PEER-REVIEWED JOURNAL,  researchers from the COVID-19 Testing Project described the steps they followed to evaluate 12 serological assays.

The researchers collected 130 samples of blood from 80 individuals who had molecular tests that confirmed infections from the SARS-CoV-2 virus. They also had 108 blood samples from another group of patients and those specimens were pre-COVID-19 samples collected from American Red Cross blood donors.

The analysis shows that the 12 tests¬†performed relatively well in identifying IgM,¬†IgG, or both antibodies.¬†‚ÄúFor each test, we quantified detection¬†of IgM and/or IgG antibodies by¬†time period from onset of symptoms and¬†assessed specificity and cross-reactivity,‚Ä̬†the researchers wrote in the pre-publication¬†manuscript. For their analysis, the researchers¬†tested the patients‚Äô specimens over five time periods: one-to-five days, six-to-10¬†days, 11-to-15 days, 16-to-20 days, and¬†more than 20 days.

‚ÄúThis study also seeks to provide¬†feedback to manufacturers about areas¬†of success and necessary improvement,‚Ä̬†the researchers explained. That feedback¬†would be useful given that the extent and¬†time to development of antibodies are not¬†fully understood and may vary among all¬†patients, even those who have RT-PCR confirmed¬†cases, they added.

Rather than report the sensitivity of¬†each assay, the researchers compared the¬†percent positivity rate by time interval, in¬†part because the percent positivity rate rose¬†over time after the onset of symptoms.¬†One important finding was that high¬†rates of positive results were not reached¬†until at least two weeks after the onset of¬†clinical symptoms. Therefore, ‚Äúdiagnosis¬†at time of symptom onset thus remains¬†dependent on viral detection methods.‚ÄĚ

Another important result is that for patients with more severe levels of illness, the 12 assays showed a trend to higher positive rates within time intervals. But the researchers urged caution when interpreting this finding because they had limited data from ambulatory patients.

‚ÄúThe majority of samples evaluated after¬†20 days post-symptom onset had detectable¬†anti-SARS-CoV-2 antibodies, suggesting¬†good to excellent sensitivity for all¬†evaluated tests in hospitalized patients three¬†or more weeks into their disease course,‚Ä̬†they wrote.

More research is needed on ambulatory or asymptomatic patients to guide the appropriate use of serological testing and it’s important to note that researchers do not know the extent to which positive serology-test results reflect a protective immune response, they wrote.

To ensure consistent and meaningful results, Hsu and colleagues recognized that patients tend to respond differently when infected with the virus. Many are asymptomatic initially and may not produce antibodies for many days or several weeks, he said.

Research Highlights about COVID-19 Serology Tests

FOR MOST OF THE TESTS EVALUATED IN THE COVID-19 TESTING PROJECT, the research showed specificity greater than 95% for most of the tests and specificity of more than 99% for two lateral flow assays (LFAs) and for one of the two enzyme-linked immunosorbent assays (ELISAs), the researchers reported.

There were two ELISAs (Epitope and an in-house developed test) and 10 immunochromatographic lateral flow assays:

    • BioMedomics
    • Bioperfectus
    • DecomBio
    • Deep Blue
    • Innovita
    • Premier
    • Sure Check
    • UCP
    • VivaDiag
    • Wondfo

Researchers provided data on each assay showing the percentage of specificity for each antibody (IgM, IgG, or both) and the percentage of positive results for each antibody at each of the five time intervals (one-to-five days, six-to-10 days, 11-to-15 days, 16-to-20 days, and more than 20 days.

Contact Patrick Hsu, MD, at



You are reading premium content from The Dark Report, your primary resource for running an efficient and profitable laboratory.

Get Unlimited Access to The Dark Report absolutely FREE!

You have read 0 of 1 of your complimentary articles this month

Privacy Policy: We will never share your personal information.