9

these endpoints in the post-hoc analysis permit a reliable assessment of cardiovascular

10

safety and given the limitations of the data, it seems like they might. Actually it’s

11

unknown because there is no real adjudication. All of us have participated in clinical

12

trials where you may comment on an individual patient but adjudication, which is more

13

strict, may or may not be accurate. So the real question is, how accurate is this post-hoc

14

adjudication? Is it enough to then have these custom MACE and broad MACE to base

15

the decision on at this point in time? Any comments? Dr. Veltri?

2

suggestions on how to analyze the data. There is one more way to handle data that has a

3

small number of events and that is, instead of conditioning on the number of events, the

4

total number of events across both groups, you don’t do that. You say, is there any

5

common event probabilities? Suppose the event probability is the same in the two

6

groups. Is there any probability that would be consistent with the data shown? So there

7

is just one more test, Statistic Bernard’s test, which might be done in addition.

17

done a wonderful job in looking at this custom MACE concept. If you have a database

18

where these CV death and mild stroke have been adjudicated, then you have that same

19

database where you have investigative reports looking at specific MedDRA preferred

20

terms, you can actually possibly compare the actual CEC adjudicated in these clinical

21

trials versus the customized approach. I suspect that is going to be very close. I think

22

that’s one way of trying to validate this when you don’t have the CEC adjudication.

1

DR. PROSCHAN: Yes, there is just one more thing in terms of

8

DR. BURMAN: Thank you. One statement is, please discuss whether

141

16

DR. VELTRI: It’s the same comment I had before. I think the FDA has

23

DR. BURMAN: Thank you. Other comments?

Scribes, LLC Toll Free 1-800-675-8846 www.scribesllc.com