First, I think we had real bad luck.  If you look carefully at the plots, they were at their maximum separation at around 24 months.  So if the trial had ended by say November 2018, I think we would have a statistical success.  Still, the fact that the curves started separating right away, was a good sign.  

Second, undetermined death hit us a little harder than the placebo...12 deaths vs 9 deaths for placebo.  Given what turned to be a too low a powered trial, this swings the confidence interval too much against us, IMHO.

Third, I think the only way the FDA allows us to go ahead speedily, is from the eGFR data for those with kidney problems.  The # of people with existing eGFR < 60 was 124 for Apabetalone and 164 for placebo.  I don't know if this is sufficient data to persuade the FDA. But do note that medical science does not even have a first drug to give you, if you end up with kidney disease.  The fact that eGFR for those above 60 was not statistically significant may turn out to be in RVX's favor (helps with kidney disease patients...no impact otherwise).  But will this be the magic bullet for RVX?  No idea.  If not, then we are stuck with a bolt on trial...but more like a big bolt on trial.  The curves did separate...but we may need an even bigger bazooka than at least what I had assumed, just eye balling the charts and numbers.  

Fourth, for us, the safety issue worked fine...no Hy's Law getting us into trouble.  Heck...if the FDA is fine with SGLT2 side effects, then we are way above that.

Fifth, the placebo rate turned to be 9.7% versus assumed 10.5% at the outset.  This could be because the patients in some of these countries were getting far better standard of care (because of the trial).  The other explanation could be that after a heart attack or such event, everyone naturally has a "Come to Jesus" moment and started doing a lot more things right (eat better, take pills etc.).  It would be useful to break out the data for centers with poor medical infrastructure (Bulgaria for example) and the ones with much better ones (Germany?).  The absence of the USA and Canada could have more illuminating.  

Sixth, I think people are way underappreciating the 18 to 21% RRR.  This is in a very sick patient group, so every RRR percentage is a lot more valuable (in $ terms) than in a healthier population.  I would like to know the RRR for statins and SGLT2.  For Amarin's Vasepa, it was 25%.  

Seventh, the final placebo MACE rate seems to be 5.72 vs us at 4.75.  So we delivered what was expected...but the placebo rate was far lower than assumed.  That is a real mystery...and adds to my sense of bad luck.  Somehow, our placebo patient group turned out to be far healthier than what we all assumed!

Eight, the P=0.06 for patients excluding deaths of undetermined is statistically acceptable...and yet comes by excluding "undetermined"!  Bad luck yet again.

Iconoclast