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.
Raf - I think they are trying to do an amended ANCHOR application to expand to lower TG levels, and since the original application proposed a TG-based cutoff, they have to stick to the same thing in the amended application. They are giving the FDA the study they asked for (R-It) to address data deficiencies in the original application.
Knowing what I have heard about Chinese business practices, whether a trial is needed or not may come down to whether the right palms were greased with the right amount of "incentive."
Thanks JL (and HD). The fact that the company's updated 5.9% projected placebo rate happened to agree with what the later FOURIER results showed was just coincidence, but it is nice supportive information (Kiwi disagrees though).
Laura - No - Not possible.
James - The boss can say no stop, and expect no stop, but the stop is dependent on whether the results meet the criteria the company has given in advance to the DMC. If they had truly wanted no stop, they would not have given the DMC stopping rules with specific p values that are relatively easy to meet with a reasonable RRR. If there is a 25% RRR, p<.00001 for the PE, and all SEs are consistent, there is no advantage to going to completion.
Kiwi -
Regarding your question, the problem is that there is no guarantee that the RRR and p value will get better at the end of the study - weird outlier data points do occur. There are cases where an effect observed at an interim analysis disappears when run to completion. I don't think this would happen, but it is a risk for the company, and it is possible that the RRR would not change at all between 80% IA and end of the study (but obviously the p value would increase). Given the example numbers you proposed, and assuming SEs were generally consistent so results would not be questioned, I would undoubtedly go for 18% RRR with p<.01 now rather than a slightly larger RRR and better p value later solely for purposes of getting an expanded label a year sooner. This RRR would beat FOURIER and it is significant. I know the p value is often misused, but in truth it is an a priori dichotomous decision. Significant is significant as long as it meets the criterion, no matter how small the p value is. BTW - An RRR of 18% would be more like p<.0001. All of what I have said is obviously just my opinion, but I am trying to think through a good rationale for my argument that the study might reasonably be stopped at the 80% IA.
In a recent post I said I thought the odds of stop at 80% IA were >60% (probably a 2/3 chance), and I wanted to explain my thinking.
1) The company has repeatedly said that "events are tracking to projections." We know they increased the sample size by 1,000 patients before starting to address the possibility that the placebo rate was only 5.2% (which is exactly what FOURIER in a similar population recently found). We can therefore assume that their projections use this lower placebo rate assumption, and that they assumed at least a 15% RRR with V as in the power analysis (although it is certainly possible they used something higher in their projections - they are not tied to the power analysis assumption). It is always pointed out on the board that "tracking to projections" for overall events across groups could be due to the placebo rate being much lower than expected. This is possible, but not highly likely, as it is inconsistent with FOURIER and all the prior trials they used in coming up with their projected placebo rate in the first place (and I am sure a lot of thought went into this given the implications). What no one seems to mention is that "tracking to projections" could just as easily reflect a higher placebo rate (maybe 5.9% as in the original designs) and consequently a much larger RRR for V. My take on all this is that "tracking to projections" most likely reflects at least a 15% RRR for the PE.
2) I have done calculations using the actual total event numbers at the 80% IA, and if we see a 15% RRR or more, we meet (and well exceed) the required p value to stop.
3) The company has repeatedly said that the "key SEs must be robust." They have NOT said they need to be statistically significant, and in fact, I think at one time they specifically said they did not have to be significant, just robust. What does robust mean and what are the key SEs? The recent trial design paper made very clear that the key SEs are the individual MACE components that comprise the PE. So if the individual MACE components do not have to be significant, but need to be robust, I take that to mean that they should show all show similar direction of effects and roughly similar RRR to the PE (although they would not be significant because there are fewer events for each individual component). JELIS is the only information they have to go by. In Matsuzaki et al (2009) describing the secondary prevention subgroup most like the R-It population, the hazard ratio for the overall MACE event PE was 0.77, and the hazard ratio for each individual component and combined endpoints was 0.79 or less (except for sudden cardiac death; HR = 0.93). So there was high consistency across almost all MACE components, and there is no reason to think that R-It would not show the same thing. My take is that if the PE shows >15% RRR, most (if not all) of the "key SEs" will show something similar, so it should be robust as required to stop at the 80% IA.
4) At the 80% IA, they have enough info to see trajectories, and whether there is growth in the EPA effect over time. Looking at the change from 60% IA to the 80% IA should be enough to tell them that the effect is solid. My guess based on JELIS is that they would see acceleration of the divergence between groups at the 80% IA compared to the 60% IA. If not, there is no advantage to going to the end, as results will not change (just the p values). If there is increasing divergence, the question is whether this alone would justify continuing the trial in order to accrue an additional 8-9 months of data. As has been pointed out previously, this will mostly serve to shrink the confidence interval around the outcomes which makes the p value lower, but it really is almost impossible to change the RRR by a dramatic amount in this short period. Would they give instructions to the the DMC that in effect would delay label expansion by a year in order to move from an RRR of .18 to .20, or move the p value from p<.001 to p<.00001? I don't think they would. It would not be in their business interest if the observed RRR in R-It beats Fourier (i.e., > 15% RRR) and the p value for the PE well exceeds the stop criterion.
Combining everything above, I conclude based on rational grounds only (no fancy models) that a stop is more likely than not at the 80% IA. If it does not stop, results are probably a more modest RRR (close to the 15% RRR design target) which could still show very nice effects in certain subgroups, so not a disaster.
Raf - Tocopherol is vitamin E and is there to prevent oxidation. Gelatin is the capsule. The others are sugars but not sure why they are there, but probably related to preservation.
Zum - Translation - EPA helps turn on production of multiple proteins that do good things for our arteries.
Jl - Ditto. Well said.
Rosie - Your observation seems spot on. I think what the EPA-relevant research evidence landscape shows is kind of like a Rorschach test - you can make anything of it. IMO - greed makes us interpret this ambiguous evidence as indicating high RRR and $$ can be expected, whereas fear makes us expect a total failure of R-It. The underlying evidence is the same regardless. The problem is that existing evidence is incomplete. No one (except JELIS) has tested pure EPA in a large CV outcomes trial, and no one including JELIS has tested a high EPA dose in such a trial. When I think back over the evidence posted here and on PubMed, I focus on:
1) JELIS and its sub-studies - these are an overestimate of what is likely in secondary prevention in R-It due to outcome differences, but they do suggest at least 15-20% RRR is likely with pure EPA over multiple years.
2) The January meta-analysis focused on DHA+EPA trials and showing >15% RRR in those with TGs >150 in controlled trials of high risk patients, and similar effects in uncontrolled cohort studies. This meta-analysis also showed a somewhat larger effect in diabetics, who are quite common in R-It
3) The CHERRY study which showed actual plaque regression when EPA and statins are combined.
4) The well-understood anti-inflammatory MOA of EPA, and known importance of inflammation in CV disease and other diseases.
5) Other human and animal studies (Zum's huge list) showing multiple MOAs of EPA that could plausibly reduce plaques and CV risk via both TG and non-TG actions
The flip side is multiple studies, almost exclusively low dose and combining DHA and EPA (i.e., fish oil and fish consumption studies) showing a lack of much if any beneficial clinical effect, and meta-analyses including studies of non-high risk populations concluding that there are small effects of DHA+EPA if any.
IMO - How we weigh this conflicting evidence is very biased by what we want to see, and our emotional state at the time (fear vs. greed). Trying to put on my dispassionate scientist hat, I believe the positive evidence is more reflective of the R-It design and population, and is more compelling as a hint of what R-It might show. I do my best to stay focused on these facts and to have confidence in all this positive suggestive evidence, whichever way the emotional winds are blowing )i.e., I am long). My best guess given what I know of statistics is that there is a >60% chance of stop at 80% IA even with an RRR much smaller than many talk about here (RRR = 15% would clearly meet the PE stop criterion, and most key SEs [MACE components] in JELIS showed the same pattern of effects as the PE). If it goes to completion, even if the RRR for the PE is <15%, I think the positive evidence above suggests that it is very likely that at least one key subgroup would show nice effects of V (High TG/low HDL, diabetics, men only, etc). This information could be used to promote V in these groups even if the label is not expanded due to the first amendment win - the PPS would in my view therefore not collapse long term (but might drop for a while the recover). This company is on the verge of making money even in its current very restricted market. So I am choosing to go with greed rather than fear...
Chas - Nice summary. Thanks.
BB- My guess is recent low volume and high stability reflect absence of speculation and PPS manipulation (wash-dry-repeat cycles) that had been occurring in recent months. Too risky now with no one knowing when R-It results will be released or what they will show. That plus good sales (i.e., fundamental support) may be keeping it right where it is.
Zum - That can't be good. Seems like it would ramp up immune-mediated inflammation. JL want to chime in?
Thanks Sam - Nice to see the refill numbers showing the highest % increase for once. It always has seemed like these lag behind where I would expect them to be.
BB - Maybe they just wanted to get it out of the way before the chaos of the 80% IA ensues?
FFS - IMO - Too early for leaks to occur. They will not even have the data for the 60-80% cleaned and entered until at least a month from now. This requires a lot of work. If we get to early June, and you see PPS support fail, that might reflect actual knowledge of negative results (or at least knowledge that the trial will not stop at 80%).
Chas - Good points.
When an early stop depends on sufficiently high estimates of multiple endpoints, the odds change dramatically. With three endpoints, for example, the odds of the estimate of at least one being significantly low go from 40% for a single endpoint to about 80% (100 x (1 - .6^3) = 78.4). That is, from probably no problem, to very likely a problem.
Raf - Since we never heard a statement regarding intentional over-enrollment, I was just assuming that this slight over-enrollment reflected excess patients enrolled because it took a while to shut down enrollment at all sites once the target was hit.
Zum -
...if the number of such events appears less than, and inconsistent with projections, the Sponsor will consider (under blinded conditions) re-calculating the number of patients needed to achieve the target number of events within the desired timeline or extend the follow-up period.
The FDA promotions refer to off-label indications which the company has data in peer-reviewed published form to support. ANCHOR fits this, but DES does not.
The AMRN website has a section for medical professionals that (carefully) lists information supporting V for off-label uses, with language (IIRC) referencing permission to do this because of the IA win (see below). Maybe the company has gotten approval for certain promotions but it has just been under the radar?
See this link: https://www.vascepahcp.com/anchor/background.html
Amarin may now disclose truthful, non-misleading information not included in the VASCEPA Prescribing Information to healthcare professionals pursuant to a federal court order issued on March 8, 2016 in Amarin Pharma Inc. v. United States Food and Drug Administration, 119 F. Supp. 3d. 196 (S.D.N.Y. 2015).
That sounds great, but are you technically BUI?
Kiwi - That's true, but raw % reductions for V in JELIS were similar between angina, MI, and other hard outcomes. It was just a sample size issue because of too few events for the non-angina outcomes. I think restenting was the only outcome not showing a similar effect.
Thanks Zum.
Kiwi -
All true...also JELIS was highly female. I have listed everything you noted in previous posts, and agree that they are in some cases reasons to expect a smaller RRR (that is why my official RRR guess was only 19%). However, some evidence from other trials I have seen posted suggests having more males In R-It might improve the RRR, as would higher TGs in R-It (we are testing the JELIS high risk subgroup in R-It).
BTW - For long-term trials of drugs without notable psychoactive effects (i.e., patients cannot tell obviously that they are on it) that are focused on objective outcomes ("blinded endpoint analysis" in JELIS), I am not really worried about placebo effects driving MACE outcomes. What are they going to do, think themselves in MIs, death, and re-stenting
AVI - Good catch - I had never noticed that. Certainly suggests that we might expect a particularly strong RRR for R-It patients (who are sicker than in JELIS) in the High Trig/Low HDL group. This will be about 50% of the R-It sample per the company. However, we clearly cannot expect a similar 53% RRR in R-It given the strong impact of subjective angina on JELIS findings (a MACE component that has been tightened up in R-It)
HD and Raf - For the question being posed, the issue is not whether there is accelerating divergence between V and placebo over the whole trial (which does seem to be the case). The issue is whether there is significantly accelerating divergence between V and placebo arms between 3.5 years (80% IA) versus 4 years (100%). My opinion is there may be an increase, but likely small. I think AVI's recent point about narrowing of the confidence intervals with more events is the best argument for continuation. However, if there is a clinically meaningful and significant effect (and I think there will be), we should be seeing it by the 80% IA.
Isaeed - Not necessarily - The biostats people on the DMC, even if the RRR looked good at the 60% IA, might have elected to continue to the 80% IA to insure that the effect persisted (that it was not a fluke, and that there was a real trajectory difference between the V and the placebo groups). I don't think we can assume that no stop at the 60% IA means anything one way or the other.
AVI - FWIW - The meta-analysis form January you recently reposted found that length of trial did not influence the magnitude of the omega-3 effect on CV outcomes. Agrees with what you said in your post. Maybe it is more an issue of the trial being long enough for events to occur rather than being long enough for EPA to work.
They certainly could.
Rosie - So old men, who tend to get gray hair, have more risk of heart disease. Very surprising
STS - In this scenario, AZN would own AMRN's patents so there would be no one to challenge the AMRN patents (it would make no sense to sue themselves for patent violation although they could in principle). They also could in theory launch Epanova and take V off the market in this scenario (dumb idea obviously).
If AZN buys AMRN, the patent ownership would be part of the deal as well.
BB - high-intensity statin therapy, ezetimibe, a beta-blocker, and aspirin
Map - FWIW - Added DHA in milk has nothing to do with TGs or cholesterol. It is there solely because of evidence that children need it for healthy brain growth, particularly if they are not breast fed.
Zip - Agreed - Based on JELIS subgroup analyses, the SEs (MACE components) should be fairly similar in pattern to the overall composite PE.
Rosie - Statistically it is slightly easier at 100%, but not enough to be really meaningful. A 15% RRR on the PE will be significant (despite a slightly more stringent requirement) at 80% IA. At 100%, an RRR of even 10% (approx) would be significant.
HR - If R-It succeeds, the issue of which jury it is won't matter. It would be exactly what the FDA asked for, and there will be no approval issues. Past negative evidence regarding importance of TGs don't apply to V entirely. Its mechanisms of action is anti-inflammatory as well as via TG lowering, so other drugs are not comparable. I also think there has been accumulating evidence of TG importance since ADCOM (e.g., genetic studies).
HR = Very simple. Evidence that TGs are a valid biomarker reflecting actual cardiac risk was deemed not strong enough to warrant approval. This violated the original SPA with FDA but was an accurate summary of the science at that time (FDA probably should never have approved the SPA in the first place). The FDA wanted to hold off approving the expanded label until AMRN provided evidence that V actually reduces risk of CV events, not just TGs.