Thursday, August 22, 2019

OPKO, 4KScore, PMA Status of "Risk Tests," VALID, and Regulatory Uncertainty

Let me show that the title of this blog is more than a word salad.

Opko 4KScore

On August 7, 2019, OPKO released its Q2-2019 results, and including remarks about its adventure with the FDA and its 4KScore test, a multi biomarker MAAA test that helps evaluate the risk of men with intermediate PSA levels (above normal but single digits).   The 4KScore test has suffered some published negative LCDs, for several years, but a favorable LCD is currently pending.

New News re 4KScore and FDA

Here's the transcript.

OPKO announced in June it had submitted 4KScore to FDA as a de novo 510(k) test.  Here.   That makes sense.  The FDA has reviewed breast cancer prognostic risk tests under 510(k) - see the Mammaprint test (K070675) and the similar Nanostring Prosigna test (K130010).    See the FDA's 2007 guidance document for these MAAA risk tests here.

Not gonna work for 4KScore.   OPKO informs us, 4KScore has to be PMA:
Based on comments received from the FDA, in July, we withdrew that submission and plan to resubmit the application as a PMA, after [...] having to meet the PMA submission guidelines. We anticipate the resubmission will occur in the next few weeks...There is a different test already approved by FDA as a PMA route application. So, in this case they wouldn't allow us to do it as a de novo....We expect to submit the PMA later this month and we have a typical review timeline for PMA, it’s somewhere around nine months to a year.
Risk Categories: Experts Have Low Concordance

Recently, the trade journal 360Dx had a long interview with ARUP regulatory expert Jonathan Genzen (here), which tracks back to his 2019 article on FDA/LDT regulatory issues (here and open access) and his 2017 article (here). 

Genzen also was co-author on a very interesting 2019 article on how noisy "risk categorization" is, how much disagreement there is on an expert-to-expert basis (in other words, there is low concordance).  See Mohlman, Genzen, Weiss, Schmidt, 2019, here

4KScore this summer seems to be a case study of this -  similar prognostic MAAA's are variably classified as 510(k) or PMA based on relatively small clinical scenario differences.

DAIA, VALID and Risk Categories

New efforts at Congressional FDA diagnostics reform hinge importantly on separation of regulatory workload by risk category.   E.g., see articles on VALID here, here.  (See also a little-known 18 page HHS comment on VALID from March 2019, here.)

But:  If regulatory "risk" categories are intrinsically noisy and inconsistent - asserted with data by Mohlman et al - we should all be aware of that in forming views about what it will be likely putting LDTs under the umbrella of VALID.



The FDA-reviewed Mammaprint and Prosigna tests cited above are 510(k) tests that help physicians make major decisions in breast cancer, such as giving or not giving adjuvant chemotherapy.   Tests that refine or hope to refine decision-making in regards to prostate cancer/elevated PSA (like 4KScore) are, however, PMA tests.  See e.g. the free-PSA test (P060005), the Hologic Progensa test (P100033), the Beckman PHI test (P090026).

While there are a bonanza of specialty tests to handle prostate risk, recent articles and letters in Arch Path Lab Med tout the value of simply reflexing an intermediate PSA (4-10) into free-PSA testing.  See Ferraro here,  Schifman here; original lead article here.