In a speech on Nov 12, 2014 at the 10th annual Personalized Medicine Conference, FDA head Dr. Margaret Hamburg provided some data to help give perspective on how much progress has been made in advancing personalized medicine.

The first point she made was that “two decades ago, only 5 percent of the new molecular entities (NMEs) approved were targeted therapies. In the last two years, targeted therapies accounted for more than 25 percent of the NMEs approved during that period.”

She also pointed out that “more than 100 drugs” now include pharmacogenomic information in their labeling, and highlighted that one third of those labels include information on “drug metabolism and dosing issues.”

Dr. Hamburg also discussed at some length the issues around FDA’s proposed regulation of LDTs, and stated that faulty LDTs had the potential to “jeopardize the advancement of personalized medicine.”

In statements meant to bolster opinion of the Agency’s ability to review and approve advanced technologies, Dr. Hamburg described the recent approval of Illumina’s MiSeqDx platform and cystic fibrosis (CF) assays.

FDA assessed the clinical validity of the two CF assays by utilizing a well-curated, shared database of mutations, rather than requiring the sponsor to independently generate data to support each mutation’s association with the disease. FDA’s flexibility reduced the burden on the sponsor and the review time, but did not compromise FDA’s ability to review the test’s safety and effectiveness.

Presumably the database utilized was ClinVar? Has FDA ever made any kind of statement about the database(s) it uses for validation of disease-associated mutations?

In a world of rapidly expanding clinical NGS it it critical that test developers know how their tests will be evaluated, and in fact if an agreed database is being used, test-by-test review seems unnecessary.

If a CLIA lab is using a sequencing platform that has been analytically validated by FDA, and is analyzing sequences that come from an FDA-approved database, does it really need each of its tests independently validated?