Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
They strip the data of anything that might lead to a specific patient being identified (name, address, etc.). This is required for HIPAA law compliance, and is standard in big data research.
Anders - He indicated he was hoping it would begin relatively soon, but no firm date was stated. I suspect, given that pharma companies are involved in the study despite the concurrent NIH funding, they want the companies involved to provide at least some of the resources. My experience is that contracts between academic institutions and industry to do studies like this sometimes take a long time to get approved, e.g., 6+ months due to lawyers. Institutions want to make sure they retain rights for any resulting novel discoveries. They may just be waiting on the contract to be approved, and are assuming it will be done by Jan 1.
Anders - One of the UCLA investigators confirmed to me that the combo delay is due to them working out funding issues.
Covariance - I agree with your interpretation. Regarding there still not being an "interim blinded publication" formally announced via PR or 8K, I do not see this apparent delay as indicating anything nefarious or bad. I think many posters do not realize the inherent slowness of the publication process. Writing a paper with a large group of investigators who all have to sign off on it can be slow by itself. Even after submission to a journal, it usually takes 3-4 weeks MINIMUM to get the results of the initial reviews, then the paper has to be revised with all investigators again having to sign off on the revision (usually several weeks or more). Then you wait another 4 weeks for reviews of the revised version. If the initial reviews are negative, they would have to start the process over at another journal. Only after the paper is officially accepted by a journal is any announcement of the imminent publication of a paper even possible, and it is possible (even likely) that an announcement would not be made until the paper is actually released as a preprint online (another 4-6 weeks after acceptance, sometimes longer, as this does not occur until after the journal sends it to and it is processed by the publisher). If it were to appear in certain high impact journals, an announcement of the paper could not be made until an embargo date is hit. I take what Bosch said about the paper being "finalized" in late summer as being accurate - the fact that there is still not any follow-up announcement of the publication at 3+ months after it "finalization" was mentioned is totally consistent with normal publication timelines.
I agree that this publication is a stopgap to address the delays in locking the dataset for formal analysis. Because it only addresses blinded data, which I think is somewhat unusual, I do not expect this to appear in a high impact journal. That does not mean it won't generate interest, but I think keeping reasonable expectations on where and when this might appear is important. That said, I am as anxious as others on the board to see the final UNblinded results.
Absolutely positive for AMRN. The longer it takes to finish the better.
JL - Just putting the straw man out there
Raggie - From the title on the AMRN website, looks like the RWE poster is mainly showing the CV harm of high TGs in Kaiser patients, rather than the value of V. But, I would be surprised if they don't find a way to slip in a plug for V in context of TGs being bad for you
Kev - May not be an accident. AMRN may have specifically targeted doing a study with Kaiser to convince them that V is worthwhile covering to reduce TGs. We will see in 2018. Getting a large HMO to do a study that argues for use of your companies drug is a great way to increase scripts (brilliant).
Raf - They locked database later than originally stated AND it was still only 76%. Guidance now for study stop by "end of 1st quarter." Originally was end of 2017, then early in 1st quarter 2018. The 76% just confirms and puts a number on what we knew - events are clearly slowing (even more than we thought). COULD be due to events dropping in both groups (Kiwi's argument), but my opinion is that it seems much more likely this drop is largely attributable to drops in MACE events in the V group, particularly given that events tracked very closely to "as expected" through the 60% interim. Has CV care really changed that much between the 60-80% IAs?
North - That would be fascinating
North - I did not realize you had a science background, but I knew about your patent law history. I am an academic scientist by day. The patent work I have done is on the side as an expert witness, but I have "enjoyed" learning patent lingo - it is a different language.
Kiwi - I have been involved in drug patent challenges. I beleive that whether or not generic high purity EPA products violate the V patents depends on the wording AMRN used in the patents. The patents I have seen list a range of doses rather than a single dose (to cover themselves broadly). So, if the ARMN patents for V said "EPA between 80 - 100% pure, with less than 10% DHA", all high purity EPA currently sold as DS would violate the patents. I have not read the patents, so do not know if this is the case though.
For What It's Worth - Per an e-mail I received from Dr. Prins in response to a question I asked, the combo study (DCVAX-L + checkpoint inhibitor) that is one of the projects in the new UCLA NIH SPORE grant will be targeting recurrent GMB only, will take place only at UCLA, and has NOT yet started recruiting patients. He said they are waiting on company funding, and given the way NIH funded trials work (NIH covers study space, nursing expenses, MD effort, and most direct study costs), this is most likely related to a contract to pay for study drugs (contracts for studies like these even when they appear straightforward can take a while in my experience once lawyers get involved in editing every word). He did say he hoped it would be starting soon though.
Rosie - I stand corrected. I was going by information I found several years ago when someone suggested I try red yeast extract.
My understanding is that the only red yeast rice still sold today has had the statin-like chemical removed due to the issues noted. Not clear what's left does anything at all to lipids.
UFO - Go look at time between 60-80% IA. Almost guaranteed to do the same thing this time. There will be about a $0.40 trading range, and it will repeatedly cycle up and down within this for no apparent reason until about 2-3 months before the final readout.
V day guess:
Onset: Feb 7, 2018
Top Line Announcement: June 5, 2018
PE RRR%: 19%
JL - Let's hope you are psychic.
Raf - You are correct. I am sure the total events they analyzed were a little over 80%. However, if they had stopped the study, they could have included all people eventing since March and data lock for the 80% IA, which would have brought the total closer to 90% of the 1,612 targeted at the end of the study. Irrelevant now though.
I was wrong in my guess of high probability of stop...oh well. End game is the same as always. Data reported by the company on Patient years to date and known rate of PY accumulation, plus known number of events to stop the study, means all that matters is the placebo event rate. If you believe it will be 5.2% or higher, then you can be pretty safe in assuming the trial will "succeed" on the primary outcome (15% RRR was targeted). All the results on SEs and results in subgroups are unknowns at this point that may add gravy to the PE RRR. To get a 30% RRR at the end, we will need a placebo rate of about 5.6%, which is not impossible but certainly not guaranteed (recent trials were lower). I am going to hold on to this stock, but with expectations for the final RRR in the 15-20% range. If the trial fails (<15% RRR), I believe strongly that there will be nice results in at least some clinically-relevant subgroups that could be beneficial to V sales over time. While still a positive, I think this would take a while to be reflected in the stock price.
Avi - I totally agree with your statistical and labeling comments. I do recall though in one of the calls in the past year that a statement (possibly during Q/A) was made that stated directly that "no, the SEs do not have to be statistically significant - they have to be robust." I have no idea where exactly this appeared though.
Avi - I know there is some ambiguity on this. Obviously, they hope to show this - maybe they mean that they are hoping that when analyzed at 80%, CV death pops out as significant and they can include this on the label. Whether they predicate the entire study continuation decision on documenting this for an FDA label is unclear. Prior to the last CC, there were no statements I am aware of that would lead me to believe they want a specific event type listed on the FDA labeling (but would be great if this works out). I keep going back to their specific statement last spring that the SEs did NOT have to be significant to be robust.
Kiwi - All I can figure is the costs of ramping up sales. But really have no idea. This does not fit with the increasing sales of V over time.
Re: SE "robustness" issue. Discussions on the board recently have commented on the order of statistical testing for SEs (components of MACE) described in the design paper and the p values required for significance of these. JL and others have noted that R-It might continue to the end of the trial simply to maximize numbers of events for more rare SEs (like CV death) so that they achieve significance. Makes sense, but...
An FDA guidance I posted quotes from a while back made very clear that the only reason you would need statistical significance for SEs using ordered testing on the MACE components was IF FDA labeling was being sought for those specific components ("V reduces CV deaths"). It would obviously be terrific to be able to say this.
The company, on the other hand, in a conference call when they revealed the revised SPA said quite specifically that statistical significance was not required for the SEs, but rather they just needed to be "robust." They know the FDA rules, so this says to me that they are simply going for expanded labeling to a wider TG range, rather than worrying about (i.e., instructing the DMC via rules provided) getting labeling for "death" or "MI" specifically. If it all works out perfectly using sequential testing, I am sure they will go for this more detailed labeling. But if not, I think they will be quite happy with just an expanded TG indication.
Bottom line is that if we can trust what the company said, there is no reason they have to delay results for a year on the chance that individual MACE components might become significant solely for the purposes of getting a very specific labeling that is more "gravy" (death) than meat (TGs).
The composite PE is made up of the 5 MACE components. To have a good RRR for the composite measure, these components largely have to agree at least in their trends ("be robust"). I predict a stop at 80%, and if there isn't one, I think the final RRR is going to be (hopefully) close the originally targeted 15% RRR. They would be continuing in this case because the PE is not strong enough to stop and they are seeing trends that it may be improving a bit (e.g., slowing of events hence the new guidance on final event in 2018, changes in slope of events between 60% and 80%). Just my opinion obviously.
FFS - Don't know about their having unblinded info, but not impossible - I worked with a private pharma company once who had a blinded trial in the DMC phase, and a scientist who was supposed to still be blind told me off the record that the results for primary outcomes didn't look good. Whether or not this is occurring with Amarin, the company has every incentive to slow walk a stop at 80%. The longer they wait, the more events they have in the final locked dataset, and the closer to 100% they will be. I really doubt a stop at what amounts to 90% of planned events really will be seen with a jaundiced eye by anyone. Questions raised with prior trials halted early have been with trials stopped much earlier in the study.
Yes. But I do research work including trials.
Raf - That is why a 30% "real RRR" might only be 19% on actual analysis.
Fits with their statement at 60% IA that "we are continuing to give the trial time for results to become more robust" (paraphrased). But, that is a W..A.. theory
Marz - Re: Timing - I think it might look bad for the company to have a CC saying the interim analysis process is ongoing, then only a few days later announce the DMC decision. Might look like they were sitting on information during the CC without disclosing it. I suspect that we will see a delay of at least another 7-10 days before any announcement. Just my opinion though.
One additional comment - It has not really been discussed on the board, but by the end of the full trial (if it goes to completion), I believe that an RRR < 15% for the primary MACE could still yield an effect reaching statistical significance using their stated criterion (p<.0422).
Kiwi - Normally in a clinical trial, patients have to agree to stay on the same treatment regimen (at least CV related in a study like this) that they were on at study outset throughout their trial participation - this is done because of potential confounds over time for exactly the reason you noted. If people want to try other new treatments, they would normally have to drop out of the study. Some may do so outside of teh study but I do not think this would affect a large proportion of patients, especially those in countries where newer drugs may not be available, approved, or affordable.
Kiwi - I just ran models using 33,000 (low end) to 34,500 PYs (realistic high end), 1612 MACE events, and with a range of placebo values. Under the low end scenario (which I think is unlikely), a placebo rate of 5.3 gives an RRR of 15.7%, so we would need a placebo rate of at least about 5.28 or so to hit the designed target RRR of 15%. At 34,000 PYs, a placebo rate of 5.1 gives an RRR of 14.1 and a placebo rate of 5.2 gives an RRR of 17.6. So the sweet spot in this model is a placebo rate of about 5.13%. Under an optimistic model (PY = 34,500), a 15% RRR would be hit at a placebo rate of about 5.08.
Going with the more realistic middle scenario (34,000 PYs):
Placebo
Rate RRR
5.3 21.1
5.4 24.4
5.5 27.6
5.6 30.1
5.7 33.6
5.8 36.5
5.9 39.3
6.0 42.0
Raf - Totally agree that this change in language is a very good indicator of eventual success. Supports JL's argument about time in trial being critical. I agree on your PY estimates (about 500 PYs per month X 10 months between March 15 2017 and January 15, 2018). Given everything we know (including that they were slower to hit the 80% trigger than expected), I still see a stop at 80% as being > 50/50, just not sure by how much.
HR - Not to mention that the DCF models seem to have to make assumptions about likelihood of R-It success. This would seem to be a hugely influential variable that they essentially seem to make up a number for.
Zum - You may be correct up to a point - I know they can analyze subgroups and report in scientific papers, which opens up 1st amendment win promotion on this basis. However, I think you are correct that they may not be able to get labeling for subgroups unless their prespecified order of analyses are met.
JL - All true. Good summary.
Thanks Kiwi - You too. I have been posting less because the board seems to be very anxious and irritable these days, due to pending results. Still check it regularly though. I keep coming back to the patient years we know as of March, and the number of events at that time. Knowing these, it really only comes down to what the placebo rate is. At 5.4% or higher placebo rate, all data available say we are good. Less than this, and it moves to a subgroup focus as noted in your other post.
Kiwi - Appreciate the reminder that even if R-it initially looks like a failure (<15% RRR), some type of positive subgroup results are almost certain, which could still expand the market.
Raf - There are statistical mediation tests that can tease this out (google "Preacher and Hayes and mediation" if interested).
Zip -
market has apparently accepted the fact that R-IT will be continued