The FDA is becoming more amenable to screening and surveillance tests for the COVID-19 pandemic, although the distinction between test uses is not always clear. Toby Lowe, the associate director of the Office of In Vitro Diagnostics and Radiological Health (OIR), said on the agency’s weekly town hall that the difference between surveillance and screening tests is whether the individuals who are screened can act on the information thus derived.
Lowe addressed the difference between surveillance and screening when using pooled samples on the agency’s March 24 town hall, offering a scenario in which swabs taken from 10 individuals are pooled. This would constitute screening if the intent is to collect additional specimens from each individual if the pool test is positive, and Lowe said the logic is that screening is invoked if the intent is to determine each individual’s status such that the individual would be able to act on the information. Conversely, if the information is strictly a matter of public health surveillance with no opportunity for individuals to receive confirmatory testing, the test by definition functions as a surveillance test.
For sponsors seeking serial screening claims for an existing emergency use authorization (EUA) for qualitative detection of the virus, Lowe said the molecular testing template already spells out the agency’s expectations for the addition of claims for individual screening. For a serial screening claim, the sponsor might not need any additional data prior to filing a supplemental request, but that supplement should include information on the proposed post-EUA validation study and the proposed testing interval.
No predicate does not mean no comparator
Lowe clarified that most test types cannot recite a predicate device with the exception of a test that uses the Biofire de novo as a predicate. However, the agency may ask the sponsor to use a comparator test during clinical evaluations. The agency recommends that an over-the-counter test offer 90% positive percent agreement (PPA) with an index test, but the PPA for an over-the-counter test can be as low as 80% for serial testing claims. Stenzel clarified that these metrics are applicable to tests authorized for use with symptomatic individuals.
The FDA’s grant of the Biofire de novo does not affect any EUAs other than the company’s own EUA that was granted for the same assay, Lowe said, nor does it affect EUAs obtained by other tests. Lowe said the agency is amenable to the use of archived specimens for validating multi-analyte assays that include influenza pathogens, a necessity in some instances due to the exceedingly low prevalence of influenza over what is ordinarily a busy flu season. This type of test might use the Biofire de novo as a predicate as the Biofire tests for several species each of coronavirus and influenza, but Lowe said that such a 510(k) might need a prospective clinical study. However, this can be a single-site study, assuming the sponsor uses both positive and negative samples.
The sponsor should select positive samples that present a range of viral loads, including loads that are low enough to approach the test’s cut-off for sensitivity. However, the agency may require that such an application be followed by a study of the test using three collection and testing sites.
Regarding prescription vs. non-prescription use lateral flow antigen tests, Lowe said the clinical validation should be conducted at the type of site where the label indicates the appropriate site for use. Such a test can be authorized as a serial test based solely on validation in a symptomatic population. The agency will want to see usability data for a test to be used by lay users when the test is prescribed by a physician, but for over-the-counter, lay user tests, the sponsor should present user comprehension data with the EUA filing, Lowe said.
EUAs filed for over-the-counter tests should probably include an asymptomatic claim and related validation data inasmuch as a health care professional is typically not available to direct test usage. The FDA may allow the sponsor of such a test for use in a serial screening scenario of asymptomatic individuals to delay gathering all the validation data until the test has been granted an EUA. This is contingent, however, on the test having achieved a PPA of 80% when used on symptomatic individuals.