Illumina jumping into cancer is hardly surprising; they've made a number of announcements of increasing interest in the field. Plus Illumina keeps pushing sequencing outside research into consumer (Helix) and medical (Verinata for non-invasive prenatal testing). In preparing my pre-JPM piece, it did cross my mind to ponder out loud if Illumina might move further into the clinic -- or even start evolving away from a focus on sequencing technology to application of sequencing technology - -but they never made it into the blog.
Illumina's announcement is a bit of a rude, if not entirely unexpected, jolt for the existing players in the field, such as Pathway Genomics and Foundation Medicine, which are already using Illumina's technology for cancer diagnostics and liquid biopsies (or planned to go there). Given Illumina reputed large margins, Grail might be offered sizable discounts which would enable Grail to undercut the market. Sequenom ran into a similar issue in NIPT when Illumina acquired Verinata, and appears to be losing: Sequenom just announced the shuttering of one of their clinical labs and a 20% layoff. Grail might be initially shooting for different markets, particularly than Foundation, which is focused on supporting clinical care not on screening, but once the assays are developed it would be very easy to move them into the clinical monitoring market -- and very tempting if pioneering companies have established that payors will cover such tests.
The promise of Grail is that healthy adults could be screened routinely for cancer, detecting ominous mutations from just simple blood draws. Catch the tumor early, and treatment could cure early and frequently.
As with any screening, both false positives and false negatives are a worry. I haven't kept up with the liquid biopsy field, but it would seem certain that the biggest longitudinal datasets are probably multiple orders of magnitude less than what even modest acceptance of Grail would generate. Suppose, for example, that 1% of Americans decided to get the test: that's over 3 million tests. What will the positive rate be? Can the health care system process all the secondary tests, let alone pay for them?
More importantly, if you get a positive, what then? Cancer is probably the disease most in danger of scaring patients with the treatments more than they are scared by the disease. Modern targeted therapies and immunotherapies are huge improvements on radiation therapy or cytotoxic chemotherapy, but are hardly benign. Immunotherapies are amazing, often resulting in cures, but torquing up the immune system can also induce autoimmune disorders. So do you treat healthy individuals with these powerful agents just because they seem to be shedding EGFR mutant DNA/ Do you send the patient for PET imaging? If they are shedding KRAS mutant DNA, for which no treatment exists at the moment (though clever minds are working to change that), what do you do? If the liquid biopsy screams "cancer" but it can't be found by other approaches, does an otherwise healthy patient go on chemotherapy? Or do you just watch-and-wait?
On the other hand, it isn't clear which dangerous tumors don't shed DNA (or at least, what fraction of patients have such tumors), and even many that shed may be hard to spot. For example, pancreatic adenocarcinoma is notorious for nearly always being KRAS-positive, and having lost a dear relation to the disease I would heartily applaud finding it early. But, what about breast cancer? Specific point mutations are far less common in breast cancer -- so would a liquid biopsy screen be relatively insensitive for breast cancer? That wouldn't be a show-stopper, but establishing such blind spots and publicizing them appropriately will be critical tasks for Grail and others in this field, with the hope (perhaps slim) of avoiding the overselling which has happened with mammograms and PSA.
This has been the challenge of cancer screening in the past, whether it is mammograms or PSA: how do you establish the predictive value of these tests and design appropriate follow-up? In the worst case, screening tests can drive large degrees of expensive invasive testing and potentially harmful treatments, when some of the tumors detected might be under the thumb of the individual's immune system and unlikely to ever pose a lethal threat. Really establishing the value of these early detections and early interventions will require huge, long clinical studies.
Those are all important caveats, but I'm glad Illumina is plunging forward -- and bringing some very good minds to bear on the problem. For example, former NCI head Richard Klausner is leaving his post at Illumina, but will be a director at Grail, and Klausner is very sharp and very on top of the changing tides of oncology. Grail will also have the resources to sequence a lot of archival samples to start exploring the population characteristics of circulating mutations. So I applaud Illumina for launching Grail and hope they succeed -- but retain a healthy skepticism as to whether it will improve health and patient for when we might know an answer to that key question. Finally, as a society we'll be faced with uncertainty around the value much earlier than any certainty, and like it or not payors and medical providers will have to deal with it.