The FDA held a public meeting last week to solicit input from the scientific community, manufacturers and clinical labs about how regulations should be applied to Next Generation Sequencing (NGS) for diagnostics. One thing was clear; everyone believes NGS will be heading to the clinic faster than any new big technology to hit the diagnostics scene, ever. Highlights of the meeting were a technology overview presentation by Dr. Vincent Magrini from Washington University, a presentation about how to detect variation across genomes and the implication of these variances on pharmacogenomics and the bioinformatics tools that will be required to do these types of analyses by Dr. Russ Altman from Stanford and the panel discussions about technical performance and bioinformatics. The public comments were mostly thoughtful suggestions from lab groups and manufacturers about how some of the issues might be tackled, and one blatant infomercial.
The questions posed were interesting if not vexing. Predicted clinical applications discussed were resequencing, tumor/normal profiling, expression, gene panels, germ line tumor diagnostics, somatic mutation profiling and (OMG) whole-genome sequencing as a Dx. NGS is fundamentally different from any of the molecular technologies that have hit the clinic. At first glance, my thoughts were that back in the day of viral load, we thought getting PCR approvals would be hard and then were shocked and awed when micorarrays were approved for diagnostic use (with a 510(k), de novo clearance nevertheless!), so maybe NGS would be the same? Reagents, instrument, consumables, software and a little more data? Nope, nope and nope.
There are some fundamental differences at play. Namely, when PCR and microarrays entered the clinic there was one front running technology that set the standards for all others to follow. With NGS, there are several platforms utilized in research today and each has different chemistries that do not necessarily yield concordant results. How does one decide what is truth?
There are no reference materials to use for sequencing. And the sample prep is complex. And there is no way to define a “standard workflow”, as the number of sequencing passes to get the desired accuracy will change by application, as will the analysis. And the amount of data to be analyzed will be incredible and will most likely break every LIMS system in the clinical labs today. And there were questions about what reference method one would use to validate results during test development? Older sequencing methods like Sanger sequencing? Another NGS platform? Southern blots? (just kidding). Add to this the imperfect alignment software, random or systematic errors across a huge number of bases sequenced, (which if you are looking at the whole genome is substantial), and the fact that there are regions of the genome that won’t sequence very well (high GC content, centromeres and telomeres). The discussion about possible errors and how to handle them in a Dx setting made my head ache.
Without a doubt NGS will require a new regulatory paradigm. Perhaps instrument only approvals will become reality? The challenges are daunting, but the pace at which this technology is moving from research to the clinic is exciting. Some great suggestions came from the meeting, namely the NGS community should get together to form data standards a la microarrays (MIAME), the American College of Pathologists (AMP) communicated the formation of a working group to tackle NGS standards, and NIST is looking at the problem as well. There was an open invitation from FDA to make suggestions about a potential framework for evaluating clinical diagnostics based on NGS.
What wasn’t discussed that will impact the development of NGS Dx tests was the recently issued draft RUO guidance. Most of the reagents used today are RUO, so getting all of the reagents needed for a test cleared will make for an interesting race. The meeting was informative and collaborative. It was nice to see a real effort from the clinical community, the manufacturers and FDA to build the regulatory framework together to have a rational discussion about the intersection of regulation and technology, unlike some of the LDT regulation discussions. I am looking forward to see how this plays out; it certainly will not be boring! I give this meeting 5 stars and two thumbs up for all of the panels and presenters.
The replay of the meeting will be available on the FDA website in about two weeks.