Zykadia - A High Level Summary of the Companion Diagnostic Story

(First published on LinkedIn)

Last year, I was riding a rented Triumph Thruxton in Utah, when I received the message that the data from the Novartis ASCEND-4 trial looked good with progression-free survival (PFS) significantly improved when compared to chemotherapy. Yesterday, the press release came out that FDA has granted first-line approval. Having left Novartis around the same time, here is a summary of the events I recalled while riding through Utah's beautiful landscape.

In 2013, when I was asked to suggest a prospective patient selection strategy for the phase III trial the term precision medicine had not reached today's popularity. We simply discussed selecting the right patient by choosing a companion diagnostic test that would be technically and commercially viable, would minimize the diagnostic risk component of the ASCEND-4 trial and lead to a successful approval of Zykadia in first-line patients. Call it a pragmatic approach. The risk component is an important one in this kind of company-internal discussion. Drug development is expensive, and you do not want to fail because of inaccurate patient selection. Unless you use a test that has already obtained FDA approval for another drug and the same indication, you cannot avoid risk entirely. On the other hand, you also do not want to use technology that is not sufficiently validated yet. Furthermore, once implemented, the prospectively validated test will be used for the duration of the trial (3 years) and should not be changed in the middle of the clinical investigation to avoid confounding the efficacy results.

When we decided to use the Ventana IHC test in 2013, it had not received FDA approval yet, and public domain validation data was still relatively sparse. Some people were pushing for a nucleic acid based test; others were undecided and looked for more fancy technological solutions (NGS) based on potential future needs. The biology supported the use of an IHC test in this case. First, background ALK protein expression in lung tissue appeared negligible. Second, there was a strong correlation between ALK DNA rearrangement and the expression of the ALK protein. This is an ideal biomarker situation because the genetic mechanism is the causative driver of de novo ALK protein expression.

The internal debate was a classic one, with the conservative 'forces' pushing for an already FDA-approved test based on DNA detection to minimize risk. The 'progressive' forces pushing for a more complex (and therefore riskier) NGS approach assuming a future need to check for other genes as well (gene panel). This approach would have significantly increased the risk to the project as other genes would have needed to be validated as well. Pharma uses technology but is not typically in the business of developing it. It is not a core competency. Getting a drug that works to the patient as quickly as possible should remain the main focus. In the end, management listened to the internal debate and supported the right technical and partnering choice. Yesterday, it paid off for the patients, Novartis, and the excellent diagnostic partner Ventana.