Yesterday’s failure of the Impala study, which sent Savara’s stock down 75% this morning, was put down by the company to an unexpectedly high placebo response. Investors should not buy this highly unoriginal excuse: placebo patients actually did as Savara had expected.
At best this points to Savara running a trial that was too small and insufficiently powered. At worst the premise behind the asset in question, Molgradex, might be flawed, which does not bode well when the company seems determined to press on and take the failed data to regulators.
On Savara’s side is that the lung disease Molgradex targets, autoimmune pulmonary alveolar proteinosis (APAP), is rare, so the US FDA might be willing to give the group a break. And the standard of care, whole lung lavage, is extremely burdensome, while Molgradex, an inhaled GM-CSF, could offer an important convenience advantage.
Unfortunately, no statistical clinical benefit has emerged, according to the results of Impala. The 138-patient trial’s primary endpoint measured change in alveolar-arterial oxygen difference (AADO2) at 24 weeks, and found an average 12.1mmHg improvement for Molgradex versus 8.8mmHg for control, yielding a non-statistically significant p value of 0.17.
The numerical benefit fell in the range around which Impala had been designed: in April Savara had told Vantage that it expected 10-12mmHg for Molgradex (Impala had been powered to show a 10mmHg reduction), and said 6-8mmHg was possible with whole lung lavage (Savara hopes to take a leaf out of Insmed’s book, April 16, 2019).
Nevertheless, on an analyst call yesterday the company insisted: “The study did not meet its primary endpoint evidently due to a much larger than anticipated placebo effect.”
In April Savara also pointed to three key secondary measures: change in patient-reported St George’s respiratory questionnaire and the six-minute walk test, and time to whole lung lavage. Only the first yielded a change against control that was nominally significant, and the company slipped in the bizarre revelation that it had yet to “fix some statistical analysis errors”.
Since Molgradex is numerically, if not statistically, better on all the key endpoints, it could be argued that the project is doing something. And the respiratory questionnaire is where it should excel, given that this relates to quality-of-life measures.
The problem is that Savara seems to have run completely the wrong study. In hindsight it might have been better to try demonstrating Molgradex’s non-inferiority to lung lavage on a relevant efficacy measure – ethically it could not have denied control subjects this standard of care – and superiority on quality of life.
Whether the US FDA would have deemed this sufficient is another question. Savara will find out soon enough, as it now intends to meet US and EU regulators to discuss whether the failed study can back approval or whether it needs to run additional trials.
If the study conducted was too small Savara might argue that it was too difficult to run anything bigger, given the rare nature of APAP, but if Molgradex cannot actually beat standard of care in a clinical setting it is difficult to see the logic behind it.
Evercore ISI analysts wrote that the signals in Impala were “strong and clinically relevant, and the patient population is small, so perhaps [Savara] can file”. However, they put back launch expectations for Molgradex from 2020/21 to 2024.