Emerging COVID-19 serology tests aim for neutralizing antibodies
New COVID-19 serology tests focus on the antibodies believed most likely to block viral entry
The next wave of serology tests from big diagnostic players could get closer to telling people whether they are protected from COVID-19 by homing in on the antibodies most likely to block viral entry into cells. The new entrants highlight how far industry remains from defining ideal tests for COVID-19 exposure and immunity.
FDA has granted 15 developers Emergency Use Authorization (EUA) for COVID-19 serology tests, which detect antibodies against SARS-CoV-2 in a patient’s blood, a sign that he or she has encountered the virus and mounted some degree of immune response.
The difficulty of creating tests sensitive enough to account for immune variability in the population, but specific enough to not cross-react with other coronaviruses, has been compounded by at least three factors: the need to produce the tests as quickly as possible, the limited availability of patient samples for validation studies, and the low prevalence of COVID-19 exposure in population, which reduces confidence in positive results (see “Good Test Hunting”).
So far, antibody tests from Roche (SIX:ROG; OTCQX:RHHBY) and Abbott Laboratories (NYSE:ABT) have led the FDA-authorized tests in terms of production capacity and performance specs, with the caveat of uncertainty in sensitivity metrics due to the dearth of patient samples (see “Roche Unveils Test Specs”).
Roche and Abbott are among the six companies with FDA-authorized tests for antibodies against the SARS-CoV-2 nucleocapsid protein.
But two other diagnostic giants, Siemens Healthineers AG (Xetra:SHL) and the Beckman Coulter unit of Danaher Corp. (NYSE:DHR), think their new tests will be more predictive of protective immunity because they detect antibodies against the receptor binding domain (RBD) of the virus’ spike protein, which is thought to be key to viral entry into cells (see “Insights from Spike Protein Structure”).
Siemens received EUA from FDA for two versions of its test for anti-spike RBD antibodies on May 29; Beckman Coulter has submitted its test for EUA and is awaiting FDA’s decision.
A third, smaller company, Hangzhou Biotest Biotech Co. Ltd., also is focusing on the RBD of the spike protein. Hangzhou Biotest received EUA for its antibody test on June 4; the company did not return requests for comment in time for publication.
Eight other companies with FDA-authorized tests detect antibodies against the full-length spike protein or a broader set of domains within the spike protein.
“Just because you’re binding to the spike doesn’t mean that you have neutralization.”
Virus neutralization assays are the gold standard for determining whether a patient’s antibodies block infection. Yet because these tests take days to perform, use live viruses and cells, and require specialized facilities and personnel, bona fide detection of neutralizing antibodies isn’t practical for routine population screening.
Siemens and Beckman Coulter believe their lab-based tests -- optimized for machines already widely used around the world -- could serve as high-throughput proxies for neutralizing assays.
“Antibodies that block the RBD of the spike protein have a high chance of being neutralizing, and that’s why they’re the target of many of the vaccines under development,” said Beckman Coulter President Julie Sawyer Montgomery.
Detecting antibodies that are likely to be neutralizing could help identify potent convalescent plasma donors and monitor the effectiveness of vaccines over time, she added.
Reagent company GenScript Biotech Corp. (HKEX:1548) aims to take that idea a step further.
The company has developed a functional assay to detect antibodies that block the interaction between recombinant versions of the SARS-CoV-2 spike protein RBD and its receptor on host cells, ACE2. GenScript submitted its functional assay to FDA for EUA on June 3; the test already has provisional authorization for clinical use from Singapore’s Health Science Authority (HSA) and CE marking in Europe.
“Just because you’re binding to the spike doesn’t mean that you have neutralization,” said David Martz, GenScript’s VP of new product management in life science. “This is probably one of the highest levels of stringency you can put on that sort of a test, short of using a live virus assay.”
The choice of target antigen is one of the earliest and most fundamental decisions in serological test design, said Deepak Nath, Siemens’ president of laboratory diagnostics.
Serological tests use recombinant versions of target antigens as bait to trap antigen-specific antibodies in patient sera. The presence of trapped antibodies is then reported via a color- or fluorescence-based readout in a laboratory immunoassay analyzer device, or in a point-of-care dipstick cartridge (see “COVID-19 Diagnostic Tech Tableau”).
Nath and Beckman Coulter’s Sawyer Montgomery both said their companies based their decisions to target the RBD of the spike protein on the emerging literature on neutralizing antibodies, which they say continues to support that call.
Examples include May studies in Science and Nature Communications describing RBD-targeting neutralizing antibodies derived from convalescent COVID-19 patients and immunized transgenic mice, respectively.
Sawyer Montgomery said sticking with the RBD was worth delaying the path to market.
“If we had, for example, chosen a nucleocapsid for development, that would have helped us get to market faster, because there were a number of nucleocapsids commercially available very early, but we thought the receptor binding domain of the spike protein was the right choice,” she said. “Getting the test right is certainly worth a few weeks of extra time.
“Getting the test right is certainly worth a few weeks of extra time.””
Nath said Siemens consulted vaccine developers on the choice of target antigen, though whether the best vaccine antigen will make the best target antigen for serological tests is unclear.
“The timescale over which these relationships are established are typically measured in years,” he said. “But it’d be fair to say based on past evidence that when you design a test, you want to try and get as close as possible to the vaccine targets.”
According to BioCentury’s COVID-19 Resource Center, the spike protein is disclosed as a target antigen for at least seven out of 14 clinical vaccine candidates and 50 out of 133 preclinical vaccine candidates. All seven clinical candidates target the full-length spike protein; at least nine of the preclinical candidates specifically target the RBD.
One clinical and four preclinical vaccine programs list the SARS-CoV-2 nucleocapsid protein as a target antigen.
Roche spokesperson Patrick Barth said targeting the nucleocapsid protein yielded the most accurate assays in the context of the pharma’s platform. “We assessed all different possibilities when coming up with our most promising antigen candidate. In our experience and how our test is set up, our modified recombinant nucleocapsid antigen showed the best performance.”
Big league stats
The push to make COVID-19 serology testing widely available and accurate enough to guide public health decisions has put a spotlight on test sensitivity, specificity and production capacity.
The limited availability of COVID-19 patient samples compared with non-infected controls for validation studies has led to greater uncertainty around sensitivity compared with specificity. That uncertainty is reflected in wider 95% confidence intervals, which represent the range of values over which one can be 95% certain contains the test characteristics. Wider intervals mean the result is known with less precision.
Siemens’ test, which captures all antibody isotypes, has a specificity of 99.8% based on 1,091 negative control samples, with a 95% confidence interval of 99.34%-99.98%. Its sensitivity is 100%, but that number is based on 42 patient samples collected more than 14 days after virus detection; the 95% confidence interval is 91.59%-100.00%.
Siemens expects to produce more than 25 million tests per month starting in June; according to Nath, the company has the largest “installed base” of immunoassay analyzers in the U.S.
Beckman Coulter’s test captures only IgG, an antibody isotype that emerges late in immune responses. The company declined to disclose performance metrics, but said its test’s sensitivity and specificity are comparable to other tests that have received EUA.
Beckman Coulter plans to deliver 30 million tests per month by end of June, and has more than 16,000 immunoassay analyzers worldwide, 3,500 of them in the U.S.
Roche and Abbott reported performance specs similar to Siemens’ in their Information for Use (IFU) documents; independent studies by Public Health England (PHE) confirmed the companies’ specificity metrics but reported lower sensitivities (“Hampered Sensitivity in Independent Studies”).
In April, Roche said it would produce “high double-digit” millions of tests per month by June, while Abbott said it would distribute 20 million per month starting in June (see “Heavy Hitters to Scale Uo Serological Tests”).
Nath expects independent validation results from PHE and others will soon be available for the Siemens test, and said the company is updating its own performance evaluations as new samples become available, with the goal of moving beyond EUA to full FDA approval.
“We’re not done,” he said “Our goal is to pursue the 510(k) and all the rigor that process involves.”
While most serology tests detect antibodies by looking for the presence of a color or fluorescent signal, GenScript’s test looks for its absence.
The lab-based immunoassay involves coating the bottom of a plate with ACE2 protein, flowing through patient sera, and then exposing the plate to a labeled version of the spike protein RBD. If a patient’s antibodies blocked the interaction between ACE2 and the RBD, the label will not be detected.
The test can be performed in manual, automated or high-throughput immunoassay formats, and reads out within an hour.
Beyond getting closer to finding neutralization antibodies, GenScript said its interference assay has the added advantage of being antibody isotype- and species-independent, meaning it can be used for zoonotic COVID-19 surveillance.
The test was developed in the lab of Wang Lifa, director of the Programme in Emerging Infectious Diseases at Duke-NUS Medical School in Singapore. GenScript licensed the technology from the university, and partnered with diagnostic developer Corgenix Medical Corp. to validate the assays.
GenScript reports the test has a sensitivity of 93.8% based on 166 patient samples collected at least eight days after viral detection or symptom onset and 27 samples collected at seven days or less after viral detection or symptom onset, and a specificity of 99.4% based on 480 negative control samples.
Like other COVID-19 serological test developers, GenScript has reported sensitivity and specificity as positive percent agreement (PPA) and negative percent agreement (NPA) with polymerase chain reaction (PCR)-based diagnosis of infection. But the company is also conducting studies that compare test results to virus neutralization assays; according to Martz, initial concordance studies using a small set of samples are promising.
Martz declined to disclose the scale at which GenScript plans to produce its tests, but said the company is in negotiations to scale up manufacturing.
He said the question of when to administer this functional test, and how to use the information it yields, will be up to clinicians.
“We’re not trying replace an IgG test. Have you been exposed and have you raised antibodies is one question,” said Martz. “A follow-on question is are the antibodies neutralizing, and the last question would be, are you immune?”
ACE2 - Angiotensin-converting enzyme 2
SARS-CoV-2 N - SARS-CoV-2 nucleocapsid protein
SARS-CoV-2 S - SARS-CoV-2 spike protein