Amarin Corp Plc (AMRN)

[AVII77]

The sequence of events is:

* Data Freeze
* Data Scrubbing
* Database Lock
* Data analysis
* DSMB meeting

Data freeze is the date of the "snapshot" of the trial outcomes. Say, for example June 1. The interim will analyze all events that occurred up till June 1st and none after that date. (the DSMB may additionally review events after that date but they should have confidence that the data as of "data Freeze" is very accurate)

Data Scrubbing is ongoing in the trial but has a particular focus at an interim. Prior to database lock, they need to be sure they capture all relevant data. That includes outcomes (have them fully adjudicated) and it also includes making sure there are no details missing on any of the items they are supposed to capture (is someone missing a BP measurement? why?, is there an inconsistency on a patient report? why? etc.). This is the time to make corrections and catch errors. Also, missing data is bad. If it turns out that there is an imbalance of missing data between arms, it could be very bad ("missingness" stymied the recent IMPROVE-IT trial and zetia didn't get the label expansion even though the results were Stat Sig (there were other issues besides missing data)).

Once the Database is "scrubbed" it can be "locked". No changes/corrections/additions are allowed after DB lock. If the interim is successful, the analysis based on this database is the one that will (usually) make it on the label. It is important that it is as complete and accurate as possible.

Once locked it is immediately available for analysis by the stat guys (working for the DSMB). They will prepare the reports for the DSMB to analyze. The templates for these reports are usually pre-specified and it's just a matter of making the calculations and graphs relative to the actual data (often they have previously completed trial runs with fake data to look for potential problems).

The interim results are forwarded to the DSMB a few days prior to the DSMB meeting so they can review.

The DSMB meets, first in open session with the sponsor (w/o discussing unblinded data) and then in closed session w/o the sponsor where they discuss the unblinded data.

After the DSMB meeting they issue their recommendation to the Steering Committee. To guard against unintentional signaling of results, they often have a pre-printed form they fill out with 3 or 4 boxes.

___Stop for Futility
___Continue as planned
___Continue with the following changes (usually safety related recs)
___Stop for efficacy


In the case of an efficacy stop, they unblind the SC to the results so they can decide whether or not they want to accept the DSMB rec. Sometimes just the Chairman of the SC is initially unblinded (to prevent bias in case the decision is to continue anyway).

They may or may not approach the FDA with the successful interim results. The FDA is not going to say "yeah, stop, you're good". They will remind the sponsor of the risks of stopping early. At best, they would say something like " the interim results appear to support the filing of the sNDA but the ultimate determination for approval will be a review issue". If it were a new drug, the FDA might encourage submittal of an expanded access IND but that wouldn't happen here. With a SPA here for REDUCE-IT, it is likely they know full well what the FDA wants the data to show before stopping the trial.

Anyway, I'm rambling. I guess the important point I wanted to make was that it is likely we have not yet hit "data freeze". Their reference to patients coming in over the "next several months" suggests the results of those visits will be included in the interim. (Patients continually come in for visits, why mention it in the PR if it were not relevant to the interim).

It is entirely possible that, once adjudicated, they will have more (or less) than 60% of events. Whatever it is, they can easily adjust the interim Lan DeMets alpha spending to the appropriate value.

But for those looking for a date of event 967, I would use something like June 1, maybe even July 1. It's pretty clear that is when they expect "data freeze". My comments in red below.

REDUCE-IT patients are in the process of completing a study visit over the next several months, after which (Data Freeze) additional time is required by the contract research organizations to (data scrub) finish collecting and preparing data for transfer to and analysis (Data lock) by the DMC.