The comment period for the FDA draft guidance addressing the use of public gene-variant databases to support clinical validity for next-generation sequencing (NGS) systems is closed, but at least one stakeholder sees a major point of ambiguity that could stymie regulatory review of these high-volume DNA tests. San Diego-based Illumina Inc. said in its comments to the docket that the draft fails to spell out what sort of role these databases will play, speculating that they could serve as anything from industry-wide standards to special controls for specific tests, a difference that could have important implications for premarket review.

The FDA released the public gene variant database/NGS draft in July, at the same time as the agency released a draft guidance for the use of standards for NGS systems for detection of germline disease, two documents the agency has labored over for some time. The FDA conducted a webinar in July to answer questions regarding both drafts, during which the agency suggested it would conduct inspections or audits to ensure the gene databases are well managed for curation, interpretation, and other matters. (See Medical Device Daily, July 8, 2016.)

The draft guidance spells out details such as the documents the agency would need to see in order to approve a database for recognition, along with procedures required to maintain recognition. The draft further pointed to the possibility of third-party recognition procedures and entities, citing the third-party review program for 510(k) applications as an existing use of such regulatory instruments.

The agency stated that data from these recognized databases would "generally constitute valid scientific evidence" that a sponsor could employ in support of a premarket filing, data that could be used, "at least in part," to support claims of clinical validity. Depending on the agency's views of the sufficiency of the evidence supporting the genotype-phenotype relationships as characterized in that database, the agency said it could waive any need for scientific evidence other than that found in the database.

Illumina noted in its comments to the docket that the agency had cleared two of the company's assays based in part on a database for genetic variants, and said there are few industry standards for genomic database formats. However, the company said there are at present "no empirically validated methods to adjudicate differences" between experts on the point of interpretations of gene variations.

In addition to the question of what sort of regulatory role the databases will serve, Illumina asked whether FDA review of applications invoking genetic database information would be subject to "established timelines," possibly a reference to the target review times for 510(k)s and PMAs as provided for in user fee agreements.

The Advanced Medical Technology Association's (Advamed) diagnostic division suggested the agency consider "a wider application of this proposed approach" to include other technologies, such as tests using multiplexed polymerase chain reaction methods. Advamed also urged the FDA to clarify whether the final guidance would pertain in its entirety to proprietary databases as well as public databases.

The association's letter recommended the agency be more descriptive about the meaning of the term "qualified expert" as applied to the draft's language regarding interpretation of the presence or absence of a gene variant. However, Advamed also took exception to the FDA's position that summary literature is necessarily inferior to data made available for independent evaluation, arguing that such a categorical declaration fails to account for the possibility that the literature "can play a vastly important role, particularly in a rapidly moving area such as sequencing."

Another point of interest for Advamed's diagnostic division was the agency's use of the term "validated methods" in the context of the scientific evidence generated by a database. The association pointed out that one interpretation of the term could be validation by means of the CLIA regulations, but also that the methods of validation are likely to evolve, a predicament that would fly against a highly prescriptive definition of the term.

One area of the draft that would seem to suggest a conventional good manufacturing practices approach is the discussion of version control for database standard operating procedures used to govern functions such as aggregation, curation and interpretation. Advamed said this portion of the draft lacks definition, but nonetheless suggested that the version control paradigm be applied to the addition of new evidence regarding the variants housed in the database.

The association further suggested the FDA recommend that operators of the databases adopt a standard database format in an effort to avoid interoperability issues. The group said any guidance the agency could offer regarding the nomenclature used to describe information, such as genetic variants and genomic coordinates, might make such information more readily usable.

In its comment to the docket, the Association for Molecular Pathology (AMP) expressed skepticism as to whether the FDA is statutorily authorized to govern lab-developed testing services, but said also that the agency should work with CMS "to facilitate streamlined coverage determinations and higher payments for tests" that a lab posts in a publicly available database.

AMP said the FDA certification of a test may convey the notion that non-certified databases are less reliable, which the association said would impinge on the practice of laboratory medicine. AMP added that any requirement that a participating database conform to the draft's transparency requirements "may deter some owners of proprietary databases from participating in the program," which could in turn impede the use of high-quality databases as a source for establishing the clinical validity of an NGS test.