Thanks to Bionug for planting this seed with an earlier post. It begs the question, what would the BETonMACE primary MACE results have looked like if BETonMACE had a higher proportion of CKD patients?

In that post from Bionug, it was noted that on page 8 of the latest MD&A: "...The proportion of CKD patients [in BETonMACE] was 11%, lower than anticipated (possibly because of competing trials) which compares to 25-29% from similar DM post-ACS populations."

This "25-29%" seems about right, since EXAMINE trial had ~29% and ELIXA had ~23%. In contrast, there were only 288 patients in the CKD subgroup (eGFR<60) in BETonMACE out of the ~2425 patients in the trial ( see slide 19 of the post-AHA slides). There were 124 treated w/ apabetalone and 164 treated w/ placebo. This equates to 11.9% of the BETonMACE population. Note, the CKD sub-study poster from May 2018 indicated 259 patients in the CKD sub-study, so there must have been a few more patients recruited between the abstract submission/poster preparation and the end of BETonMACE recruitment.

Back to slide 19, the 3-point MACE primary endpoint data was broken down like this:

eGFR<60: Apabetalone 13/124 (10.4%), Placebo 35/164 (21.3%), Hazard ratio 0.50

eGFR>60: Apabetalone 112/1084 (10.3%), Placebo 114/1041 (11.0%), Hazard ratio 0.94

Total: Apabetalone 125/1212 (10.3%), Placebo 149/1206 (12.4%), Hazard ratio 0.82.

As you can see, the biggest apabetalone benefit was in those w/ eGFR<60. 

Looking back at the Corporate Update transcript (thanks imtesty) prior to start of enrollment, the plan was:

"As a subset with the BETonMACE, we estimate based on our historical data, that we will have somewhere in the neighborhood of about fifty percent-ish [BDAZ note: I think this is type and should have read fifteen percent-ish] CKD patients. That’s the bottom left. So in BETonMACE we could have up to three, four or five hundred CKD patients."

So they had 288 CKD patients, below the 300-500 target. 

Time for hypotheticals....what if BETonMACE had twice as many CKD patients as the 288? This would put in in line with the patient composition in ELIXA and EXAMINE. For this practice exercise, I will keep the total number of patients the same (~2400), but make the CKD group be 23.8% of total patients instead of 11.9%. I will also even out the placebo vs. apabetalone patient split to make the number of each patients in each subgroup the same, but keep the event rates the same for each subgroup as they were for the real data. By doing this, the total number of events increased to 287 from 274.

eGFR<60: Apabetalone 30/288 (10.4%), Placebo 61/288 (21.3%), Hazard ratio 0.50

eGFR>60: Apabetalone 95/918 (10.3%), Placebo 101/919 (11.0%), Hazard ratio 0.94

Total: Apabetalone 125/1206 (10.36%), Placebo 162/1207 (13.42%), Hazard ratio 0.772

The real study produced a 0.82 hazard ratio (18% RRR) with a p-value of p=0.11. Would this hypothetical 0.772 hazard ratio (22.8% RRR) with more CKD patients have produced a statistically significant reduction for the total population? I don't know, but it sure would have been close.

BDAZ