Photo courtesy of Alex Proimos via Flickr
In the world of NGS a sea-change is occurring, and that is the shift to clinical applications of this technology to personalized medicine. It seems like a long time ago and it was only in 2005 when the Roche / 454 GS-20 first started appearing in genomics laboratories, and 2007 when the Solexa 1G first appeared. I was involved with some of the earliest adopters of NGS into clinical genetics laboratories way going back go 2008 and 2009 – these groups were quick to realize the potential of the technology, undertaking the risk of a very fast-changing technology, with regular changes to systems, reagents, protocols and software, all of which militate against adoption into a regulated laboratory-developed test environment.
With the advent of the Ion Torrent™ PGM in early 2010, and rapid development within a 3-year timeframe, this system has matured to the point of an expected regulatory submission in 2013. By rapid development, the readlengths have gone from about 100 base-pairs to the release this month of 400 bp reads; the densities along with this readlength have increased the throughput some four or five-fold; automation of the template prep (the OneTouch™ System) was rolled out; and the accuracy has improved via software improvements. (Chad Nusbaum makes this point at his Ion World 2012 talk – available here on YouTube if you haven’t seen it already.) Here’s a link to the December newsletter – page 2 has a nice graphic that illustrates the progress from 2010 to late 2012. (The link requires a free registration to the Ion Community, where news, information, discussion and protocols are available.)
There are market reports (like this one from DeciBio) that show higher than the overall market growth (DeciBio has a 5 year annualized growth rate of 17%, with the applied segment at 26%); I’ve seen projections for particular applied segments to be projected at twice that (over 50% per annum). On top of this, I recently came across an investor research piece (complete with interviews of thought-leaders) that indicated that the clinical uptake was ‘faster than even we projected six months ago’.
The stories in the popular press continue the drumbeat. For example, an online women’s magazine More had this piece entitled ‘Cancer, a Huge Leap Forward’ with the subtitle, “An underreported revolution in cancer treatment is giving hope to people who have this illness now—or fear they’ll get it in the future”. The institutions in this article where personalized medicine for cancer is taking hold is across the U.S. shows where this leadership is coming from: Massachusetts General in Boston (no surprise), to Memorial Sloan-Kettering Cancer Center in New York (no surprise), the University of California Los Angeles and the Oregon Health and Science University. The piece ends with the quotation, “If you looked at World War II prior to D-Day, you’d have said, ‘We’re never going to crack this.’ My view is that at least we’ve arrived on the beaches.”
I tell friends and acquaintances that should I be so unfortunate to have cancer, that I would go to one of three centers to get the latest in personalized medicine: MassGen, MSKCC and Baylor. Certainly the NCI maintains list of designated cancer centers, some 67 in all. However, these three have the benefit of many, many clinical trials going on simultaneously, looking at scores and scores of new medications and treatment protocols for cancer; on top of this they are on the leading edge of applying NGS into the clinical research realm, by sequencing a panel of cancer genes and cancer gene pathways from individual tumor specimens and comparing variants to the individual’s germline sequence data.
Of course no method is 100% efficient, and the data may not be 100% complete due to variation in sequence coverage, stochastic and systematic error, and limitations of sample input (FFPE treatment of samples make isolation of intact DNA a challenge). However to the oncologist making patient treatment decisions, the presence of a particular tumor mutation in a particular gene pathway implicated for that particular tumor type can open up the opportunity to make real difference in treatment – the mechanism of action (MOA) for the array of potential treatments is well-known, and with the mutation data the right medication can be given to the right patient. If the data is only 90% complete (i.e. 90% of the genes have sufficient sequence coverage given the limitations cited), there is a very good chance that a given patient’s mutation will show up.
Steve Job’s pancreatic tumor was sequenced via whole-genome sequencing, and it was early in the timeline of NGS development. Apparently he was one of only 20 individuals who had their genomes sequenced, which puts the timeline at about 2008, at a cost of $100,000. This effort was a collaboration across several top institutions (Stanford, Hopkins, the Broad Institute, Harvard), and is remarkable in that this was a failure of the most advanced technique for cancer treatment, and yet made public. It isn’t very often that a non-diagnosis of an NGS approach is publicized. The reasons for this failure could be straightforward (the underlying driver mutation had incomplete coverage and thus was missed, a false-negative result), or due to the devilishly difficult nature of cancer itself (a fast mutating metastasis or a yet-to-be-discovered driver mutation for pancreatic cancer).
So there is disruptive innovation taking place on the frontiers of cancer treatment, and this trend isn’t going away, it is only getting larger and more prevalent. Increasing numbers of patients, informed of these advances, are more often insisting upon getting this kind of personalized treatment. And with the majority of the US population served by community hospitals, the trend toward pushing NGS closer to the patient will only grow.