The diagnostic industry in the U.S. and elsewhere has scrambled to keep up with the COVID-19 pandemic, and one of the key developments will be a test that can be used at home without medical supervision. However, Tim Stenzel, director of the U.S. FDA’s Office of In Vitro Diagnostics and Radiological Health, said on the Sept. 16 diagnostic town hall that the agency is keen on authorizing such a test, but has yet to receive any emergency use authorization filings. “We want to see a home test submission, and we’re willing to be very flexible here,” Stenzel said.
One of the callers on the town hall noted that the FDA had posted a template for submissions for at-home test in July, and that the requirement for an antigen test used in this context is an accuracy of 90% relative to a standard lab test. However, the agency also indicated it would like to see some sort of mechanism for reporting results to a public health agency, which could prove complicated if the device offers nothing more than a display that reads out the test results.
Stenzel said the FDA had recommended a 90% positive percent agreement for a test acquired over the counter, while positive percent agreement/sensitivity for a prescribed test is 80%. That metric can be as low as 70% if the test is packaged as part of a serial testing program, which consists of ether two tests on consecutive days or on days 1 and 3. Stenzel said the agency is “very eager to authorize a home diagnostic test,” but said the FDA has not received “a single EUA application for a home test.” The flexibility is driven in part by a dire need to bring a test to market to fill this testing niche, which may prove crucial in the effort to reopen businesses and schools.
Stenzel explained that the comparator lab test would be a high sensitivity molecular test because a less sensitive benchmark might engender too much inaccuracy once the test is widely distributed. The agency is keeping a close eye on sensitivity for rapid antigen tests and point-of-care molecular tests that are administered in the first five to seven days after onset of symptoms because of the persistent perception that this is the peak time frame for transmissibility of the SARS-CoV-2 virus.
LOD less important for asymptomatic subjects
For validation of tests used for those who are asymptomatic, the agency is still focused on ensuring that samples are drawn from subjects with similar levels of viral load. Stenzel said the FDA is not particularly focused on a test’s limits of detection (LOD), in part due to lack of an international standard for this pathogen. Another factor is that there isn’t always a direct relationship between LOD and a test’s clinical performance. This is particularly the case when trying to identify those who are actively shedding virus, Stenzel said.
As for the read-out function for a home test, Stenzel said the FDA has authorized only one visual read-out test, but that there are others in development. The agency is hoping that one or more additional such tests will hit the market in the near term, but home tests that are processed outside a CLIA lab are not required to report results by law. Still, Stenzel said, “we do want to stimulate thought about how those test results can be reported.”
Stenzel said the FDA and the National Cancer Institute (NCI) have struggled to obtain a sufficient volume of samples for development of reference panels for antigen test development, but matching sample type and titer has also been labor-intensive even when those samples were available. The FDA has grown inactivated virus to overcome the difficulty, but availability continues to be an issue despite the FDA’s and NCI’s best efforts.
Stenzel said that false positives have not received as much attention as false negatives, but they are an important factor to be aware of, particularly in the context of recent reports of false positive results for tests used in nursing homes. As the U.S. returns to work and school, screening protocols may be conducted on populations with low rates of prevalence, and thus seemingly high rates of specificity and sensitivity might misfire for a significant number of cases.
Stenzel said that molecular and antigen tests that offer a 95% rate of sensitivity and a 99% rate of specificity might seem ideally robust, but if the percent positive on a given population is 0.2%, the positive predictive value might be as low as one in six, or roughly 16%. “We have seen that level of positivity in a number of screenings,” Stenzel said, and the problem here is often that someone who is falsely diagnosed as positive may be moved into a COVID ward at a hospital as a result, where “they may unfortunately contract the disease.”
The consequences could be disastrous if that patient is at risk due to age or comorbidities, and Stenzel said the FDA needs a clear picture of a test’s sensitivity and specificity, but also of the positive predictive value for the specific population setting.