Originally published May, 1998
IVD Technology News
FDA joint panel weighs uses of gold standards
Gene amplification technologies are beginning to challenge the traditional paradigm of culture as the gold standard for the diagnosis of infectious diseases. But shifting paradigms isn't an easy matter.
Although many researchers believe that nucleic acid amplification (NAA) methods will eventually supplant culture, FDA wants to ensure that the new tests are actually an improvement and that any discrepant results can be handled appropriately. In February, the agency convened a joint meeting of four medical specialty panels related to clinical diagnostics to discuss these issues.
When comparing assays, sensitivity and specificity are the most important yardsticks. Although culture has long been considered to have perfect specificity, new data suggest that the method is less than 100% sensitive. This means that a positive result from culture truly indicates disease, but that a negative result could be a missed diagnosis. Specimens that are negative with culture but positive with NAA methods have generally been ruled false positives. Many researchers believe that such discrepant results are truly positive, reflecting the greater sensitivity of NAA assays.
Determining valid methods for data analysis and retesting was a focal point of the joint panel discussions. Speaking at the meeting, Tim Green, a statistician at the Centers for Disease Control and Prevention (CDC), noted that "even a slight deviation from perfect specificity greatly affects the bias of sensitivity estimates obtained via discrepant analysis." Biases can arise because retest specimens have been selected based on their culture results. When trying to convince FDA that NAA assays are as good as the culture gold standard, said Green, it is the responsibility of manufacturers and clinicians to avoid such biases and to use sound statistical analyses.
But according to Linda Ivor, a clinical scientist at Gen-Probe, Inc. (San Diego), "Any possible bias created by selecting false-positive specimens for discrepancy analysis appears to be negligible when the test's true sensitivity is greater than 90%." At the panel meeting, Ivor urged that alternative tests of similar sensitivity be used for resolving such discrepant results. For its amplified chlamydia test, Gen-Probe has implemented a research "alternative amplification" test that FDA has agreed to integrate into discrepancy testing. But the company has yet to convince FDA that such alternative tests should be used in final calculations of the assay's performance. For clinical trials that involve diseases with definite diagnosis, such as tuberculosis, Ivor suggested that physician diagnosis be considered the gold standard.
Although formal guidance on the use of gold standards may be a long time coming, the joint panel made several recommendations to standardize the process of accepting NAA testing methods. In selecting a gold standard, the panel recommended that the relationship between analyte and disease state be clearly defined and supported with data. To resolve discrepant results, retesting should include both discrepant and concordant specimens. The panel also cautioned against the use of alternative amplification tests using the same molecular technology, but with target primers that have not been clearly linked to disease. —G.W.
