Approved with 40% experiencing liver injury?<p>Check out the statistical evaluation sections on the study outcomes:<p><pre><code> https://www.fda.gov/vaccines-blood-biologics/tissue-tissue-products/elevidys
</code></pre>
I could see approval override when evidence is marginally positive but impossible to improve (e.g., for lack of study subjects - here ~150 with a rare genetic disease) when the benefit is clear and there is no safety risk (adverse effects - AE).<p>The primary outcome didn't fail by much in the small sample. It's an assessment scale of physical activity which can be hard to get right and perhaps prone to false negatives for slow responders. So, maybe.<p>But the liver injury (lots of bad labs) is a significant side effect, as is nausea/vomiting (~50%), fever (~30%), and thrombocytopenia (8%). Even with that, there was no discontinuation d/t AE, because the patients (parents) are so motivated. I'd like to see how quickly patients recovered esp. to normal liver labs, and also understand how whether other gene therapies have shown similar reactions to be transient and benign.<p>There's probably a lot of variability in immune systems at that age, with possible long-term effects from adverse training. The variability could result in severe problems in a few cases, and the adverse immune training could result in lifelong auto-immune diseases, etc. Which means there's reason for caution.<p>As for corruption...<p>Fatalism about corruption is exactly what corrupt regimes nurture, because it passivates objectors and activates collaborators. Please let's not go there.<p>The FDA process situation now is OK: both staff and advisors are on the record, so any inconsistent top-line decision is clearly discoverable. As far as I can tell, staff and advisory board selection is mostly per scientific skill, albeit with industry representation, and also on the record so that influence can be traced. Above all, reasons are justified and traced to studies on the record, and most marginal approvals (like this one) include post-marketing monitoring to claw back the approval as needed. (My primary objection now is that study data is mostly non-public, and AFAICT there's no requirement to publish negative data).