NEW YORK – A study published Thursday found that a coronavirus serology from Abbott Laboratories was more accurate than PerkinElmer business Euroimmun's coronavirus serology test.
In spite of that, the study – conducted by researchers at Washington University in St. Louis and Barnes Jewish Hospital department of laboratories – noted that the Abbott test still did not perform as well as the company has said it can.
The study was published online Thursday in Clinical Chemistry and looked at Abbott's SARS-CoV-2 IgG assay and Euroimmun's Anti-SARS-CoV-2 ELISA assay, both of which have received Emergency Use Authorization from the US Food and Drug Administration.
Looking at 256 samples – comprising samples from 103 from patients with COVID-19 detected by PCR testing and 153 samples from patients without the disease – they found that Abbott's test, which runs on the company's Architect platform, had a specificity of 99.4 percent and a sensitivity of 93.8 percent 14 days or more after symptoms began.
Meanwhile, Euroimmun's assay, running on Inova Diagnostics' Quanta-Lyser 240 instrument, had a specificity of 94.8 percent and a sensitivity of 85.4 percent 14 days after the onset of symptoms. For Euroimmun's test, borderline results were primarily reported as positive results.
Further, the current prevalence of people in the US with antibodies to SARS-CoV-2 is unknown, but assuming a 5 percent prevalence of people with SARS-CoV-2 antibodies – the threshold by which the FDA predicts the performance of companies' assay – the positive predictive value of Abbott's test is 89 percent and of Euroimmun's test is 46 percent, "highlighting the importance of a high specificity," the study authors said.
They called the difference between the sensitivity and specificity of Abbott and Euroimmun's tests "statistically insignificant" and said more studies would be necessary to confirm that Abbott's test had higher sensitivity and specificity. Masoud Touloue, PerkinElmer's vice president and general manager, diagnostics, also noted the statistical insignificance, saying in an email "Overall, it is a misrepresentation to conclude that the Abbott SARS-CoV-2 assay demonstrated higher sensitivity and specificity than the Euroimmun SARS-CoV-2 assay in this particular study."
Touloue added that because Abbott's test was performed on its own instrumentation, while Euroimmun's was performed on Inova's, the study "does not result in a true comparison."
Meantime, a spokesperson from Abbott called the study "consistent with other studies that have demonstrated the high reliability of Abbott's SARS-CoV-2 IgG test."
Both firms also noted that other studies have demonstrated higher accuracy with their tests than that shown in the WUSTL publication.
Touloue noted that the company has performed studies with larger sample sizes, such as one with the University Medical Center Hamburg-Eppendorf that tested 1,122 samples and detected specificity of nearly 100 percent.
An Abbott spokesperson also cited a study from the University of Washington, which tested 1,020 samples and found 100 percent sensitivity and nearly 100 percent specificity at 17 days or more after the onset of symptoms.
The WUSTL researchers said that although Abbott's test demonstrated a higher sensitivity than Euroimmun's, it is still below Abbott's stated 100 percent sensitivity for the test. One explanation, they noted, is that Abbott's test subjects were likely patients who had recovered fully from the virus, while the Washington University study included patients who were still hospitalized.
The researchers also said there were four discordant results between the serological tests and PCR tests, with two of the three patients on chemotherapy for leukemia and the other patient diagnosed with a previous enzyme deficiency, which may have contributed to the lack of antibody responses. Touloue added that "PCR testing can sometimes result in missing true positives," and said the three positive samples detected by Euroimmun but not by PCR were "likely not false positives, but, rather, missed by PCR."
Diagnostic sensitivity for both tests was low during the first 14 days of symptoms, which is not unexpected as an individual may not produce enough antibodies to SARS-CoV-2 during that time for robust detection. The Abbott spokesperson said the company's test was only for use after two weeks of symptoms, similar to other antibody tests being marketed for COVID-19.
This low sensitivity "argues that serologic status should not be assessed until 14 days post-symptom onset and confirms that molecular assays should remain the primary method for COVID-19 diagnosis," the study said.
In a statement, Chris Farnsworth an instructor at Washington University School of Medicine at St. Louis and the corresponding author for the study, said that their work "demonstrates the importance of testing patients who have fully recovered from infection and hospitalized patients with multiple comorbidities including immunodeficiencies."
Additionally, the study results suggest that "given the relaxed regulatory requirements by the FDA on SARS-CoV-2 serology tests, it is crucial for clinical laboratories to uphold the rigor of assay validation and to determine if assay performances, as reported in package inserts, are accurate."
In addition to the lower Abbott sensitivity, relative to what the company says it can achieve, Farnsworth and his colleagues said that for the Euroimmun test, only nine specimens were used to determine the diagnostic sensitivity cited in the test's package insert, when the Clinical Laboratory and Standards Institute recommends using at least 50 specimens.
The WUSTL study comes amidst a rapid surge in the number of coronavirus antibody tests hitting the US market, accompanied by concerns about the performance of some of the tests.
There has been much conversation surrounding potential false positive results from serology tests and debates about whether the FDA was too lax with its validation requirements.
In response to the concerns, the agency recently revised its guidance and said that coronavirus antibody test makers would need to prove that their tests meet certain performance thresholds.