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.
There is growing sentiment based on communications from AMRN that "robustness" of the SEs are the dealbreaker as to whether the study will be stopped at the 80% IA. Two obvious key SEs to me are subgroup analyses in the diabetes population (about 70% of the total sample) and within gender subgroups. So, I took the total sample (n=8,175) and the final desired number of targeted MACE events (1,612) to examine scenarios for what the stat significance would be for these two subgroups at the 80% IA (assuming equal sample sizes across the two study groups, and equal gender distribution within each study group). The 80% IA targeted MACE event total would be 1,290. Results are interesting:
Assuming a 19% RRR as in the JELIS overall sample (and in line with my and Kiwi's prediction in the contest), the significance of the PE would be p<.0001, significance in the diabetic subsample would be p=.0005, and significance in men only or women only would be p=.0032.
Assuming there is a 15% RRR like the study was designed to detect in the overall sample, the significance of the PE would be p=.0013, significance in the diabetic subsample would be p=.0066, and significance in men only or women only would be p=.0232.
The company IR presentations have estimated that approx 50% of the total sample would have High TGs and low HDL-C, another key SE subgroup. Results would be identical to that for sex above (i.e., significant).
While there is no way to model what pattern the individual MACE components will show, analyses of all of these would be in the full sample and should have plenty of statistical power. This fact combined with the modeled results above tells me that it still would not be at all surprising to get a stop at the 80% IA. Obviously, this would be even more likely if the overall RRR is larger than in the models above as has been suggested by numerous people on the board. Maybe (as has been suggested recently) the company is underpromising and hoping to overdeliver???
Tas - Or, event rates are actually increasing substantially (more so in placebo) over time (as in JELIS), and this was their prediction all along. I agree that the longer we go without hitting the 80% trigger, the less plausible this seems to be as an explanation.
Bhatt is blinded to the actual study results also - Only the DMC members can see the results.
I am fairly certain based on my experience that the FDA never suggests any interim analyses (except in situations where futility and patient harm might be an issue). That was totally AMRN.
STS - Thanks for the comments about your situation. That fits what I would expect. I have always wondered about how V might work for "real" chronic pain. Since starting EPA (sadly, no V prescription yet), I no longer need to take naproxen on a near daily basis, which has got to be a good thing.
Pyr - The SE issue is not just re: components of the MACE, all of which would reflect the full sample size. Any SEs that reference subgroups (e.g., diabetics, men only, etc) will have a much smaller sample size, so even with an RRR of 18% (or whatever) they might not be significant even if the same RRR is significant in the full sample. Increasing the sample size for subgroup analyses is a primary reason to continue even if the PE meets significance criterion.
JL - I agree - You are absolutely correct.
Raf - There is nothing stopping AMRN from providing the DMC with new guidelines for stopping (including very specific SE patterns required) based on the updated SPA. In fact, I am sure they did this. I would make no sense not to.
Noelan -
The new SPA with the FDA includes the stop rules for stopping at the 80% interim, and the FDA has signed off on them. If the DMC recommends stop and the company agrees, the FDA is not going to have any problem with this. I am also sure, given their prior ADCOM experience, that the company in setting the stop rules made them convincing enough that any ADCOM would agree the drug works.
Kiwi -
Yes they are (maybe even more often than every 3 months, I am not sure what the schedule is). Hawthorne effect applies to both groups, and will not alter hard objective physiological outcomes (e.g., MI, re-stenting). Keep in mind the whole reason for including a placebo control group is to address any influence on results simply of being observed repeatedly, taking part in a trial, expectations, etc.
Pitt -
Yes - This has happened (posted previously). In one case, the results were significant at the interim and no longer significant at the end of the study. This unusual situation is most likely with smaller studies where a few unusual patients can swing results dramatically. Not going to happen in R-it - too many patients.
Kiwi - Clinical trials always provide free drug on study visits. Their regular pharmacy and insurance has no involvement.
CBB - For inflammatory chronic pain conditions (including low level chronic musculoskeletal back pain which affects about 30% of U.S. adults), V should help reduce pain levels somewhat. Not going to be a cure all based on my experience, but might provide the same benefit as NSAIDs (motrin, alleve), without the risks and side effects. Not sure about how V would affect opioid requirements, but assume it could permit lower dosages potentially. Definitely not harmful and probably helpful.
There is a nice graph posted on ST by Short Fish Fry - Trend for total weekly retail sales growth. Seems to fit a quadratic trend quite nicely. If it holds, we will be at 25,000 total scripts weekly by around August 1 of this year. I would love to see R-it halted at the same time we hit 25,000 scripts - would psychologically magnify the effect on the PPS in my view.
Once R-It succeeds, cost to the company to get the FDA to approve it is virtually nothing. Almost all costs are entailed in doing the trial.
Nothing has a real RRR of 80%, not even V
No knowledge of combo. The research summary I saw many years ago (in a medical journal) about aspirin's benefits indicated that it had a very small effect size. The recommendation to change guidelines to include low dose aspirin was driven by the fact that it had no real side effects, was cheap, and could be used by virtually everyone, so the population level benefits even with a small effect were substantial. Not sure if anything changed since then. Based on this, I don't think ubiquitous low dose aspirin use (which should be roughly equal across groups in R-It) will really change the results by much.
Kiwi - You happen to be in one of the best organized healthcare organizations there is (even if they don't want to cover V). I don't believe that Kaiser is typical of all cardiology care, but I understand your point. Sadly, my cardio friend is stuck in the old way of thinking and is only willing to consider V if the trial succeeds (which obviously is too late as an investment opportunity).
and the p-value for stoppage will be higher than p<0.022, and more like p<0.035.
Pyr - You are showing your ignorance here. A higher P value is EASIER to achieve (need a smaller effect), not harder to achieve as you imply. The p value at 60% was a very high hurdle, but the p<.02 at 80% IA is much easier to achieve even with a relatively small effect size of RRR = 15%.
Kiwi - Hawthorne effect is real in short term situations, but there will also likely be regression to the mean over a 3-5 year study. Behavior change in the absence of an immediate payoff is hard to maintain. I will admit though, some may change and feel better enough to maintain the gains (like you did). I agree with JL that you oversimplify the ease of making significant behavior changes. I have a cardiologist friend who has repeatedly described the self-induced nature of serious events in her patients - they are unable to make even the simplest lifestyle changes to benefit their health. So, they rely on meds and roll the dice.
JFM - Well said.
Kiwi - I get your point and we will see. To get in the study, stented patients in the study had to be stented before they came in, and enrollment ended a while ago. Most were enrolled before 2016. If they had a problem with their "old style stents," then they would show up as events in R-it. Other issues are not everyone in R-it is stented, and that risk reduction due to this type of stent may have nothing to do with the mechanisms of V, which could still leave much room for reducing residual risk. These changes may have an effect but I do not believe they in any way eliminate the possibility of a nice RRR. By the way, I noticed that you and I both have 2 of the lowest RRR values listed for our contest guesses (both around 19%). Mine is based on the full group JELIS RRR (= 19%), with design and population differences between R-It and JELIS that increase versus decrease likely efficacy offsetting each other (e.g., Increase RRR = more men, higher EPA dosage, higher risk population; versus decrease RRR = stronger statins, improved CV care [your stents], less subjective MACE definition). I still would be totally happy with RRR = 19% and think this would be a very important and clinically meaningful outcome.
Kiwi - Interesting and true, but practice changes slowly. R-It has been going on now for 5 years. Surely there are some US patients in R-it who may be getting these newer interventions, but overseas patients most likely are not. Also, there is typically a requirement to not make other treatment changes during study participation. So, I don't think we will have massive numbers of study patients getting miraculously more effective therapies. Some patients may ignore recommendations not to change their care, but I suspect that final R-It results may indicate the level of reduced risk based on how cardiology was practiced 5 years ago.
Pyr - Have to admit you are generally correct here
Kiwi -
Based on what JT has said in the past, I think it means that the composite rate is where they projected it to be at this point, given their assumed placebo event rate and assumed efficacy of V when they started the study. That doesn't prove that V is working as intended, but at least where we are now is consistent with V working as planned. The only ways to get to the composite rate where we are now that is consistent with projections that does not involve V working as intended is: 1) both groups are having fewer events than in past similar studies in similar populations on statins, or 2) V makes people worse and the placebo group is doing great. #1 seems unlikely given that the sample is apparently about 70% diabetic (someone posted that yesterday), has elevated Trigs, and 50% also have low HDL), and #2 would be inconsistent with all prior studies of EPA. I don't buy Pyr's thesis of a high % of dropouts in the placebo group - Moderate trig elevations as in R-it are not a treatment target for cardiologists, even if patients see them and discover their trigs have not lowered. There is also little incentive to drop out and start using V outside of the study because it is not yet a proven treatment. Why would they spend money on V when no one knows yet whether it does anything to reduce risk? My suspicion is that many of the study patients are using the planned study visits as free ongoing cardiac care and may no be seeing their regular cardiologists intensively during the study.
Kiwi -
Expectation of 5.2% in placebo arm means the trial was designed and powered a priori to test an assumed rate of 5.2% in placebo with 15% RRR in the V arm.
HD - You are correct. Nothing can be added in the middle of a trial. The SE changes were just a new plan to analyze data already being collected. Banking blood at the study outset for analyses not planned initially might be complicated, as different assays require different processing (and more money). DNA banking for future genetic analyses is common in academic research these days, but I am not sure they do this in industry-sponsored trials regularly.
JF - Not exactly. What I recall posting was about what can happen when trials stop early vs. not, with an example of a trial that could have stopped early with significance ending up being continued and being nonsignificant on completion. It obviously is more common to be nonsignifcant at early interim but significant later due to increased event numbers. My take on the ethics of an early stop of R-it is that this is an unusual case because V is already approved and available to anyone who wants it (potentially). Ethicists could argue that V as an effective treatment to reduce CV risk is not being withheld since it is already available, and therefore the ethical need to stop early for overwhelming efficacy is not as compelling, and must be balanced by the benefits to other patients (from greater strength of the secondary outcomes) if the trial is continued.
Mr. Main - Keep in mind that any numbers we get on event rates in the study until it is unblinded are ONLY referring to overall rates across both groups. You cannot infer anything from this information about what is happening in the placebo group. If the overall rate is lower than originally projected, the most likely explanation is that V is working very well, so fewer events are occurring.
If JT said that in conference today, that is great news. This seems likely to increase the RRR based on the science that has been posted, and also means we will have a large sub-population of diabetics even at the 80% IA (makes this key secondary endpoint more likely to be significant).
Zum - If that was what was said, he must have misspoken. The only thing that could be referring to is overall event rate - No one but the DMC knows how many are in either group until study is unblinded.
Fishy - Thanks for the perspective. Makes sense.
CBB - I love that indication on the label: "Vascepa is indicated for people at risk of old age"
Seems to me you both make sense - doesn't high volume necessarily mean larger numbers on both sides of the transaction are present? Can't have buyers without sellers and vice versa. Seems to me that the imbalance between them is what drives PPS changes - if there is high volume and PPS is increasing, this means more buyers than sellers, right? Or, is my supply/demand take on this too simplistic?
Pyr -
CPXX was completely written off by the market. Insty ownership was nil.
Pyr - Re: your #3, statins do not work via the same mechanism as EPA or aspirin (aspirin and EPA are not really the same either). These distinct beneficial mechanisms are why they might have additive effects rather than duplicate effects. EPA also addresses two distinct mechanisms which might be complementary in terms of reducing risk (Trig lowering and anti-inflammatory). R-It will see whether that is the case.
I think I understand your short argument, that you make less money but with less risk. I disagree with your thesis that chances of R-It success are as low as you postulate. Based on this, I am willing to take JL's asymmetric bet.
Zum - Thanks. Interesting stuff. V could treat not only symptoms, but the underlying disease. Good argument for preventative V prescribing in pre-diabetics.
HD - That is an excellent and obvious point.
Bio - Interesting - I won't disagree if that's what studies show, but I find that a bit surprising. There is a large body of evidence that in general, women report higher pain levels than men to the same stimuli (at least in lab studies). Regardless, sex clearly cannot be ignored in conducting and interpreting studies.