As next-gen sequencing-based diagnostic tests are making their way into the clinic on multiple fronts, the US Food and Drug Administration is involved in developing processes to evaluate them, with the intent to regulate NGS-based tests eventually.
At the Advances in Genome Biology and Technology conference in Marco Island, Fla., last week, Zivana Tezak from the FDA's Office of In Vitro Diagnostics and Radiological Health offered some insights into the agency's initiatives and intentions around ultra high-throughput sequencing.
As of today, the FDA has not approved any next-gen sequencing diagnostic tests and its thinking around them is still evolving. "Nothing is finalized on how we are going to regulate these tests," Tezak said.
However, regulation will come as NGS tests are moving from a few pioneering labs to a growing number of smaller laboratories, a trend that became clear at last fall's Association for Molecular Pathology Annual Meeting, she said.
In the meantime, a growing number of academic and private laboratories are launching laboratory-developed NGS tests in CLIA-certified laboratories, aimed at diagnosing rare inherited diseases, detecting tumor mutations that can be targeted by drugs, or diagnosing chromosomal aneuploidies prenatally. Those tests range from gene panels to exome and whole-genome analyses.
Tezak explained that the FDA usually regulates diagnostic tests according to the risk of their intended use. For example, if doctors decide to administer chemotherapy or to conduct surgery based on a test's result, it would require the highest-level review, whereas a genetic carrier screening test for a disease like cystic fibrosis would be considered lower risk. For that reason, the FDA avoids categorizing tests by technology type, for example lumping all NGS-based tests together.
Developers submitting a test to the FDA for approval generally have to demonstrate its analytical and clinical performance, and to validate the instrumentation and software.
For next-gen sequencing-based assays, however – both gene panels and whole-genome tests – it will be much harder to do this because they analyze many more variants than existing tests that cover, for example, just a few dozen mutations.
To come to grips with these issues, the FDA almost two years ago sponsored a public workshop on approaches to assess the analytical validity of ultra high-throughput sequencing for clinical diagnostic applications, where it sought input from academic, government, and industry researchers as well as other stakeholders, both on the evaluation of the technical performance of a next-gen sequencer and on the bioinformatic data analysis (CSN 6/29/2011).
"What we got back was new questions and almost no answers to the questions we posed," Tezak said. "Since then, the answers to these questions are slowly falling into place."
One suggestion made at that meeting was to decouple a test from the platform it runs on. Usually, Tezak said, the FDA reviews a test in its entirety, from the clinical samples to the platform to the results reported to clinicians. However, meeting participants proposed that the agency should evaluate a sequencing platform based on specific characteristics – determining whether it is reliable, can be manufactured consistently, and how it performs, for example – and then separately evaluate the tests to be run on these platforms. "That is something that we are seriously considering, and have been considering since then, and we have been talking to people about how to implement that," Tezak said.
Indeed, the FDA has already begun assessing NGS-based tests using this strategy. Last month, Illumina said that it had submitted its MiSeqDx system, a version of its MiSeq sequencer, for approval by the agency, while separately submitting a cystic fibrosis diagnostics assay and a carrier screening version of the CF assay, which both run on the MiSeqDx system (CSN 1/9/2013).
"This is a first submission like that," said Greg Heath, senior vice president and general manager of Illumina's Diagnostic unit, at the time.
While the preanalytical steps of NGS-based tests as well as the data analysis methods keep changing rapidly, Tezak said at AGBT, the sequencing platforms "have maybe reached a little more stability."
Regarding the validation of the analytical performance of a next-gen sequencer – whether it measures what it is supposed to accurately and reliably, and how precisely – Tezak said that it will be impossible to assess the platform's ability to detect every single variant, but noted that a good strategy might be to look at the accuracy for a representative set of markers. It will be challenging, though, to define such a set of markers, she said, and to determine what percentage of them a platform needs to cover.
Any such set should include analytically challenging markers in genes of known clinical relevance, such as genes with high or low GC content or with homopolymers. "You want to know what the platform does for all these," she said.
Another open question is whether positive results need to be confirmed by Sanger sequencing, as most labs offering NGS-based diagnostic tests do today. Not only may Sanger sequencing not always be a "gold standard", she said, but also, verifying every result by Sanger will get "prohibitively expensive."
Reference materials will play an important role in validating the performance of sequencing platforms, and the FDA has been collaborating in various efforts to develop such standards, such as the National Institute of Standards and Technology's "Genome in a Bottle" project (CSN 9/5/2012). The agency is also involved in developing microbial reference materials and is helping with a Centers for Disease Control and Prevention-led initiative to address the standardization of file formats for clinical use.
Regarding the clinical performance of NGS-based tests, the FDA will not be asking for clinical studies unless the tests contain entirely new markers. As with existing genetic tests, the agency will accept evidence from the scientific literature establishing the clinical validity of the markers or guidelines from professional societies.
However, there is need for a unified resource for clinically relevant variants to interpret the data, Tezak said. So far, these have been kept in "silos of knowledge" at different institutions offering genetic testing, both academic labs and companies.
A number of groups – including the National Human Genome Research Institute, the National Center for Biotechnology Information, the American College of Medical Genetics, and the International Collaboration for Clinical Genomics – are interested in setting up such a clinical-grade database, which manufacturers could use to validate their systems, and the FDA is "interested in helping these efforts out," she said.