Bfw,

Thanks for setting me straight. As I had written, "All just my opinion. I may be wrong here." I'm no pro in deciphering clinical trial designs. I do see in the BETonMACE ClinicalTrials.gov listing that atorvastatin and rosuvastatin are listed as separate interventions. The primary outcome is clearly stated as the time to first occurrence of adjudication-confirmed narrowly defined MACE (single composite endpoint of CV death or non-fatal MI or stroke). However, in various presentations Resverlogix and/or members of the clinical steering committee have shown a slide titled "BETonMACE subgroups" that states: 

Pre-specified subgroup analyses for the primary endpoint include:

– Rosuvastatin/Atorvastatin

– < 30 days/> 30 days post-acute coronary syndrome

– LDL/HDL/TG’s above and below median

– HbA1c above and below median

– eGFR >60 mL/min and < 60 mL/min

• Also change in eGFR for all patients with eGFR <60 mL/min

So does the above mean that ALL of the above pre-specified subgroup analyses of the primary outcome are valid even if the total population fails to achieve statistical significance. I find this hard to believe. That is why I expressed confusion/skepticism in my prior post.

With all of the secondary outcomes, pre-specified subgroup analyses for the primary outcome, exploratory outcomes, as well as the CKD and cognition sub-studies....it gets really confusing to keep things straight (see here for a previous post that reviews all of this). 

To further add to this, slide 18 of the current Corporate Presentation doesn't discuss the pre-specified subgroup analyses of the primary outcome at all. We do know that there have been some trial protocol amendments/changes to BETonMACE (i.e allowing dosing past 104 weeks, skipping the futility analysis at 125 events, skipping the sample size re-estimation analysis at 75% of events). Is it possible that they dropped these pre-specified subgroup analyses of the primary outcome too? I really, really wish Resverlogix and the steering committee would put out a BETonMACE rationale, design and baseline measures manuscript that would lay all of this out in detail!

There is often a formal, fixed testing sequence for pre-specified outcomes. I think this is referred to as hierarchical testing. Endpoints/outcomes are tested in a fixed sequence until the first non-significance is arrived at. At this point, subsequent statistical tests on outcomes/endpoints further down the line are only considered exploratory even if they show a difference. The review below on type 2 diabetes CVOTs discusses this hierarchical testing issue.

Cardiovascular Outcomes Trials in Type 2 Diabetes: Where Do We Go From Here? Reflections From a Diabetes Care Editors’ Expert Forum

http://care.diabetesjournals.org/content/41/1/14

As for the comments on ASSURE plaque regression from Tada and Koo, the observation of the responder population for plaque reduction was a post-hoc analysis (not pre-specified) of a Phase 2 trial that failed to achieve its pre-specified primary outcome. So although post-hoc data would suggest plaque reduction as a mechanism, this could not be used for product labeling or marketing without a follow up clinical trial to prove via a pre-specified outcome analysis that apabetalone reduces plaque volume. Yes, for the total population the 0.40% reduction in percent atheroma volume (PAV) in the apabetalone group versus baseline failed to achieved statistical significance (p=0.08) or meet the goal of a 0.6% reduction. More concerning in ASSURE was the fact that the placebo group (0.30% reduction in PAV versus baseline, p=0.26) nearly had the same benefit as the apabetalone group. The between group difference in PAV was P=0.81. Similar result for total atherome volume (TAV), as you can read in this ASSURE abstract. So even IF the within group change in PAV from baseline (primary outcome) had achieved statistical significance, there is the elephant in the room that apabetalone had pretty much zero benefit when compared to placebo! This unexpected placebo response is another reason why I urge caution to assuming too much about the expected BETonMACE placebo MACE event rate based upon prior trials such as EXAMINE and ELIXA.

BearDownAZ