Amarin Corp Plc (AMRN)

[HDGabor]

sts- (and to all)

One thing - maybe the most important regarding early stop - came to my attention recently only (thanks to AVI-)

later learn that some key secondary endpoint that was trending positively but didn't reach the high statistical standard for success that is associated with an interim look but that might have proven successful had the trial continued

- additional number of SE(s) will / could be higher than additional number of PE, since (a) as a 3rd (or later) event for CV Risk Stratum 1 (please note: they had one prior the study) or (b) 2nd (or later) event for CV Risk Stratum 2 will be counted as SE(s) ... some SE will have higher total number of events than PE / 1,612 (e.g. key SE – Composite of CV death, nonfatal MI,4 or nonfatal stroke – easily could be higher with 2nd, 3rd, etc. event)

- Based on Crestor label (JUPITER result), I think P<0.002 is enough for label (indications and usage) ... meanwhile CVD on label would be admirable, statistically significant (p<0.0422) „only” will be a great result also (Please note: all SE could be / will be on label - inc. non stat. sign. SE(s), but not as "indications and usage" (IU) but as "Clinical Studies";see Crestor / UA, HR=0.59, p=0.093) ... any result for labeling (IU) will be the cream on the cake .., I stand corrected, but not to many drug showed stat. sign. RRR of CV Death (e.g. Crestor / Jupiter did not, CVD p=0.315)

- I do not think that Amarin would like to get label (or get stat. sign. result) for any SE definitely, it isn't a requirement. Based on their guidance above I think they would like to see as many as possible SE to be stat. significant, (but it could be nil also). It could be the key SE only (1st SE … assuming that none of the SE was stat. sign. (0.0022) at 2nd IA and only this one could be at final (p<0.0422)) or Fatal or nonfatal stroke (7th SE … assuming that 1-6 was stat. sign. (p<0.0022) or trending positively and – at least this or the 8th, etc. also - could be stat. sign. at final (p<0.0422)).

Since the required RRR of PE - to hit P=0.0022 at 2nd IA - was low, but enough for label, the continuation "should be" due to:
(a) SEs weren't robust and / or subgroups result wasn't consistent
or
(b) SEs were robust and subgroups result was consistent, none (or some) of the SE / subgroups were (weren't) stat. significant (p<0.0022), but based on the 1st IA and / or 2nd IA analysis any (these) could be stat. sign. (p<0.0422) at final.

Since - e.g. for key SE - the required bar for p is lower (0.0422 vs 0.0022) and event number is more at final than at 2nd IA (and number of events of key SE will / could be higher than of PE) the required RRR of key SE will be lower than the required RRR at 2nd IA or of PE at final.

I think (b) is the case.

Best,
G

ps.: to avoid any meaningless, neighborhood based post ... this post is about label, stat. sign. and continuation not about the interpretation of final result by the medical community.