In a Cepheid-sponsored workshop at the Clinical Virology Symposium here this week, Columbia University professor Daniel Green outlined the virtues and weaknesses of RIDTs, noting that the US Food and Drug Administration recently changed its classification of these tests in order to ensure they are as sensitive and specific as possible.
There are three main methods for detecting flu in a patient sample, Green said — viral culture, antigen detection, and nucleic acid tests. Viral cultures, though accurate, take far too long to be useful in the clinic. When immediate treatment can significantly impact a patient, waiting 48 hours for test results makes no sense. And although nucleic acid tests are gaining in popularity because of their speed and accuracy, they're still prohibitively expensive to be very widely used, and their complexity can sometimes require skilled lab personnel that not all hospitals have, Green said.
That leaves RIDTs, which take only a few minutes to return a result and are relatively inexpensive. A sample is applied to a test strip, which detect viral particles using antibodies embedded in the strip. The newer tests even come with automated readers to reduce the possibility of human error when determining results.
But these tests are also less sensitive than culture or nucleic acid testing, Green said. They have no ability to subtype viruses, and recent studies have shown their sensitivity to be somewhere between 50 percent and 70 percent. According to the Centers for Disease Control and Prevention, their negative predictive value is not high enough for negative results to be trusted without verification with further testing, Green said.
The sensitivity problem of RIDTs came to light during the 2009 H1N1 pandemic. A study at that time found that RIDT performance varies, depending on the level of virus in a given sample. At the lowest concentrations of viral load, the performance of these test is inadequate, Green said. Another study of H3N2 showed some assays can't even detect viral strains when they're present at high levels in a patient sample.
In order to solve this problem, the FDA convened an advisory panel on RIDTs in 2013. After a period of study and public comment, the agency reclassified them from Class I to Class II with special controls. Class I in vitro diagnostics are low risk devices for which general controls, such as registration of manufacturing facilities and quality control procedures, are sufficient, Green said. Class II devices are those that pose a higher risk to patients if their effectiveness is compromised in any way. These devices require special controls in addition to general controls — performance standards, postmarket surveillance, review by the FDA under 510(k) procedures, and so on.
The FDA determined that the special controls for RIDTs would include minimum performance standards, checking of results against reference methods, annual monitoring, and provisions for public health emergencies, according to Green.
To set minimum standards of sensitivity, the agency focused on tests that were cleared within the last five years because they tended to demonstrate somewhat higher sensitivity. It also determined that RIDT manufacturers would need to obtain 510(k) clearance to certify that any new tests meet minimum performance standards. According to the new regulations, old products must be withdrawn from the market by January 2018 or must be modified to meet the new criteria.
The tests must be compared against reference methods such as viral culture or nucleic acid amplification tests, Green said. If compared to viral culture, the RIDT must have a sensitivity of at least 90 percent for detecting Influenza A and 80 percent for detecting Influenza B. When compared to nucleic acid tests, the RIDTs must have sensitivity of at least 80 percent to detect both Influenza A and B.
Further, because of the reclassification, all manufacturers must conduct annual analytical testing of their RIDTs with contemporary strains of the flu, including vaccine strains, with panels provided by the CDC. The protocol involves testing panels of viruses in three different dilutions — in order to claim that an RIDT detects a given strain of the flu, the manufacturer must show that the test can detect at least one dilution of that strain. If it can't, then the labeling must reflect that limitation.
And finally, Green said, manufacturers must make provisions for public health emergencies by testing the reactivity of their RIDTs with novel viruses as soon as they become available. Companies must also provide the FDA with written protocols as to how they plan to deal with such emergencies using their tests.