Replies to post #58958 on Amarin Corp Plc (AMRN)

[Sam81]

JL, I totally agree with you. in fact in yesterday's conference, JT clearly mentioned that their base case scenario is for 2017 and should there be overwhelming results at the 60% interim expected during year 2016, a halt and stoppage of the trial is an option.

[HDGabor]

JL-

We had a same discussion one week ago, so I have to assume:
- you do not want or could not understand what I am saying or
- a post from HDGabor makes you see red and do not think before post

So,
- nobody said it belongs to the FDA, BUT the SPA is in place
- yes, "The company can do anything wants", but what a "surprise" they did not increase the TG enrollment criteria without SPA modification (-> FDA was involved)
- nobody said "Amarin does to change the trial design". Any analysis AFTER full enrollment isn't a change of the design and could not affect it since ALL patients were enrolled at the time of the analysis

So the best policy is to stick to the pre-trial protocol which is designed to produce meaningful data, and not open the presents before Christmas.

The protocol includes an analysis @100% and 60%. BOTH could produce meaningful data! Without full details:
- final: eff. 15%, p =< 0.046
- 60% interim: app. 25-27%, p =< 0.0076

I do not see any reason, why additional interim analysis @80% or @50% (AFTER FULL ENROLLMENT) could not produce meaningful data. ie:
- @80%: eff. 20%, p =< 0.025
- @50%: eff. 35%, p =< 0.005

All interim could provide a statistically meaningful and valid result.

To be more precise: the @50% will be a little bit higher and the more exact definition (independently from event number) is "interim after full enrollment)

Since time is money, both additional interim look (but again: if any) is could for Amarin (@"50" - 4 months earlier than @60 / in November, @80 - 11 months / halfway between @60 and @100)

Best,
G

[HDGabor]

JL-

I forgot to mention one thing (addition to #58998):

Meanwhile you are fine with an analysis based on guess / "ethical" reason (actual event# is less than expected) without protocol, you are saying that additional predefined look with protocol is dangerous / risky.

Err... WTF?

Best,
G