Amid questions about testing accuracy, the Food and Drug Administration (FDA) continues to warn doctors and the public against using antibody tests to definitively diagnose a past COVID-19 case.

The FDA on Tuesday morning e-mailed a MedWatch safety alert saying that while the agency will continue reviewing data from test-makers, it doesn't expect that an antibody test alone "can be shown to definitively diagnose or exclude" the viral infection. Such test results need to be judged in the context of patient history, including the timing of symptoms and exposures and the results of other tests.

COVID-19 is a viral illness that causes the body's immune system to produce special proteins called antibodies to fight off infection from the SARS-CoV-2 virus discovered in Wuhan, China, late last year.

An accurate antibody test could be a major tool to end stay-at-home orders like the one in place in Minnesota, because it would theoretically identify people who have immunity to the novel coronavirus. Researchers have not yet figured out how long COVID-19 antibodies protect those who have them, nor how accurate the tests to detect them are.

"Determining the next steps in our response to COVID-19 is partly dependent on an accurate assessment of our national efforts thus far, and the quality of data for making this decision is dependent on accurate testing products," FDA Commissioner Dr. Stephen Hahn said in a statement.

The FDA said it is collaborating with the National Institutes of Health and the Centers for Disease Control and Prevention to establish performance assessments of antibody tests at the NIH, including tests already on the market and others still under development.

Few details were available as of Tuesday. The NIH confirmed that the Frederick National Laboratory for Cancer Research, which has expertise in antibody testing because of past work with the human papillomavirus, is validating COVID-19 antibody tests with the FDA.

The FDA said the results of the NIH's performance assessments may help determine whether tests should receive emergency-use authorizations in the U.S. However, the FDA has also loosened its normal restrictions on how tests get to market, as it tries to balance the need for more testing against the risk of using tests backed by scant evidence of accuracy.

FDA guidelines published in March say companies can sell COVID-19 antibody tests without any authorization or review from the FDA. But the test-maker must do several things first: validate the test internally; tell the FDA about the test; and include disclaimers saying the test is not FDA-approved and cannot be used to definitively rule out COVID-19 infection.

Diagnosing the novel disease is difficult, partly because of its confusing symptom presentation. Many people with COVID-19 experience fever, coughing and shortness of breath, but some don't, even 14 days after exposure.

In general, the virus seems to cause mild or no symptoms in roughly 80% of cases, while 5% of cases may require critical care in a hospital. But those estimates are based on confirmed cases. Some researchers think the rate of symptomatic cases may be much lower, but it's impossible to tell without widespread, accurate antibody testing.

There are two major types of tests for COVID-19, molecular and serological. Molecular tests were the first to be developed in the U.S., and they are designed to react to the presence of viral genetic code. They can detect the virus earlier in the course of disease, doctors say, but may be vulnerable to missing cases because of imperfections with nasal or throat specimens.

Serological tests look for antibodies in blood samples. COVID-19 antibodies aren't thought to be detectable in the blood until 13 or 14 days after onset of symptoms, which typically don't appear until after an initial incubation period of two to seven days from initial exposure, Mayo Clinic researcher Elitza Theel said.

There's strong commercial interest in developing "point-of-care" antibody tests capable of delivering a result while the patient waits. At least 90 are under development and some have been submitted to the FDA to get emergency-use authorization based on initial validation testing done by the manufacturers.

But researchers like Theel say there's a risk that rapid antibody tests could be used too soon in patient care, which could cause negative readings in people who have actually been exposed to the virus.

"The vast majority of individuals are going to be negative if you test them too soon with a serological assay," Theel said.

Joe Carlson • 612-673-4779