Thursday, December 8, 2016

Discovery and Profitability in Molecular Diagnostics: Patents vs Time Constants of R&D

Recently, Genomeweb ran two articles on patents and molecular diagnostics.   The first article, on November 15, covered discussions at the annual AMP meeting.  Roger Klein MD JD, of Case Western, debated with Timothy Stenzel, COO of Invivoscribe.   In a second article, December 5, the rate of patent applications was assessed by Duke's Arti Rai and Santa Clara University's Colleen Chien.  Rai and Chien studied 27,000 patent applications from 2002 to 2014.  The data was presented at a Utah conference and is not yet published.  (For an earlier Rai publication, here.)

Patents provide a barrier to entry, which can encourage investment by allowing a window of protected time for investment recoupment.   This could be more important in situations where there is a large upfront investment required, but a relatively low ongoing marginal cost, and, where the product could otherwise be easily observed and replicated.

We can draw a line between these articles and a December 8 article, also in Genomeweb, that Interpace Diagnostics (Parsippany, NJ) has obtained coverage for 200M patients, including major payers like Aetna and Medicare and some BCBS plans for its ThyraMir thyroid nodule classifier test.  According to the article, " Molecular testing using ThyGenX and ThyraMir has been shown to reduce the rate of unnecessary surgeries in indeterminate cases."   More after the break.

Besides patents another barrier to entry in molecular diagnostics is the nature of the data required.  For example, for tests like Oncotype DX Breast, development required large historical archives of paraffin block, at least 5-10 years old, with highly accurate longitudinal followup.  Finding such archives is a limiting resource, and while a number of such international archives have been available for breast cancer prognostics, a very common cancer, similar archives in less common cancer will be rare.

With a thyroid nodule classifier, the time constant is much shorter.   Patients are relatively plentiful, and the time constant between an FNA and a surgical result is fairly short.   The accuracy of the test (defined as the concordance between the molecular FNA prediction and the results of a surgery) can be logged fairly quickly.   This allows more impressive data to be accumulated relatively quickly, and allows enough data to be published to win with technology assessments at payers.  On the other hand, it allows new entrants to join the market more quickly.

The new entrants may have one less barrier to entry.  A new test requires validation in practice; if physicians and patients have a molecular diagnostic that is (for example only) 80% accurate, will it impact the care pathway?   That has to be shown.  But once it is shown once or twice in large cohorts of real-world doctors and patients, it seems likely that a new (second or third or fourth) test that is also 80% accurate would be used the same way, just like a Siemens and a GE MRI scanner with similar images (looking the same and both 80% accurate) would be expected to have the same impact on healthcare outcomes.  And, to extend the analogy, neither Siemens or GI will spend $100M to study the impact of its MRI on colon cancer outcomes, because everyone would assume the usefulness of the other brand MRI would be the same and there would be no advantage to the investor.

Investors are usually aware of this - what is the patent landscape barrier to entry, and what is the real-world barrier to entry in terms of patients and data.   These issues are not so much on the radar of health technology assessors.


Anecdotal footnote.

When I was in graduate school in the 1980s, I had the chance to learn from Lloyd Fricker PhD, now a professor of molecular pharmacologist at Einstein.  For his PhD at Johns Hopkins under the legendary Sol Snyder, he isolated an endorphin cleaving enzyme.

This type of work had previously involved very tedious radioimmoassays. Fricker developed a ligand with a radiolabeled polar molecule on one end.  He set up racks of tubes with aqueous solution below and a polar supernatant.  In the tube that the enzyme was being purified into, the polar half the of molecule would be cleaved free, and the radiolabel would pop up into the polar supernatant.  The enzyme activity could be isolated just by testing which tube had radiolabeled polar supernatant.  This enormously sped up the time constant of discovery, leading the group to victory in being the first to isolate the enzyme.  If labs are racing to discover something, and your lab can make the methodology work 10X faster, you'll win.