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.
No, FDA is not that simple. FDA review will look at numerous things and raise many questions. Ones I can imagine are related to: dropouts, titration, no dose response (if 30mg shows no benefit) and this a small trial (160 in treatment), etc. FDA will have many more questions that we can’t even think of.
You should read a public briefing docs — which is typically 100-300 pages long, and you’ll see the extreme extent to which FDA analyzes/scrutinizes.
I have been very very impressed by FDA for that reason. They deserve all the praise (even though they are NOT flawlessly competent).
SAP specification doesn’t mean too much — SAP may have 10s of different analysis methods. Odds might have been there. Doesn’t mean much.
I don’t think FDA ever approved anything about the trial.
It was implicitly clear that the trials endpoints were mean-change in Cog and ADL wrt placebo. That’s the norm, that’s what everyone does, and that’s what was in the clinicaltrials and company presentations (implicitly). No one has been heard if odds-ratio and the used thresholds before. Why? It’s all cherry-picking (from SAP) and/or made-up.
Fine. But 50mg needs to show the mean-changes in ADL and Cog being met. Yes, if that happens, AVXL may have a case.
The below numbers are pretty “convincing” that the drug “works”. However, that’s not how drugs are approved (and rightly so). One needs to show the (pre-determined) primary endpoints being met with statistical significance.
Group | total n | Num>.5Cog | Num>-3.5ADL
Placebo | 163 | 20 | 5
Treated | 308 | 63 | 24
30mg | 160 | 21 | 8
50mg | 148 | 42 | 16
Thanks for a great SUMMARY.
The only thing that will get the drug approved is “statistically significant” (p < 0.05) mean-change in Cog and/or ADL and/or CBR wrt Placebo for 30mg/50mg (pooled analysts is unconvincing).
Odds can only be supporting evidence (not core), IMO.
Of course, if we see awesome biomarker results (Ie, change compared to placebo) then the evidence becomes more convincing.
Overall, I don’t think approval chances are zero — but minimal is the best-case scenario.
So, what’s your guess on what’s going on? If CM is not dumb — is he lying? Is he a fraud? Doesn’t know what WS would think about his changes of endpoints over and over again?
Done venting for today. I pray some sense comes into CM’s senses, and we are saved from torture of months of waiting at low levels.
Nope, not convincing. Intelligence is way more than a Munich degree and corporate titles. Yes, Kellog is difficult to get into — but it’s still an MBA school, where intelligence is not the first thing they look for.
I’m just venting my frustration. No, I’m not here to save you guys — I can’t even save myself from this bloodbath.
I do not believe Missling is a fraud or a liar. I really do not. I just think he is dumb..really really dumb.
Missling's reply in CC to the question why the data did not include details of the decline in ADCS-ADL
====
That's very important to understand. So the CDR sum of the boxes, the ADAS-Cog and the ADCS-ADL co-primary endpoints of odds ratio were prespecified specifically in the SAP, in the statistical analysis plan. And the reason for that is because we are aiming for showing an improvement for patients, and that's something which we knew from previous Phase IIa trials and the Parkinson's dementia trial that we should go after that also for differentiation of the drug and for labeling reasons. That is really the key to understand. We are aiming for that, and we prespecified explicitly that analysis of improvement over a threshold of clinically meaningful improvement even in the clinical study -- statistical analysis plan.
====
I mean really?! If, for whatever stupid reason, odd-ratio was the original intent -- why did he fail to mention that/those-words in the clinicaltrials.gov and numerous company presentations! And, what gives him the sense to use odds-ratio anyway in a PIVOTAL trial, without FDA's explicit permission. The bottom line is -- he is either really really stupid (to have picked odd-ratio in a pivotal trial) or really really stupid (to think that people will people his above answer)!
I'm still extremely mystified as to what is going on in CEO's mind. I really am. He can't possibly be really really stupid. But, I just can't come to any other conclusion .. however hard I try!!!
Is the situation fixable? I believe it is -- Missling could come out and do the perfect PR/CC next week. Sure, we may miss ADCS (that's not fatal, IMO). I'm sure the data already revealed is accurate -- he just needs to convince the world of its accuracy by giving additional details. If it is not accurate, he should be honest and state the inaccuracies.
If the above happens --- in the worst case, if he backtracks, the stock may go to $5-6, but there will be hope of future. In the best case, the stock may shoot up to $12-15.
If I were the CEO, I would think thats a no-brainer. What is he thinking? How can he be so dumb? Is it in the German culture -- to believe that you can make a fool of the outside world?! I mean really -- his answer to Sumit Roy in CC about ADCS-ADL data, is really shocking (he really thinks everyone is just a fool to believe that bs?!).
Regarding Rett: Yes, the data will be good. But, what if Missling delays it by 1-2 years by this stupid moves (e.g., CRO taking 5 months to give data, Missling taking 5 months to meet with FDA, and then 6 months to submit an NDA) ..... such delays can kill the stock. Of course, in the meanwhile, he will bungle up the PR completely! I doubt he gets it still.
Not sure what the confusion is. I meant “Missing p by a whisker is not a failure. Eg, P = 0.051 is not a failure by itself.”
He needed 3 years to design/think.
It’s TOO OBVIOUS.
Missling must get it now — that his handling of the TLD has not been well received (he didn’t seem to have gotten it before — with previous trial results).
What can he do now? Does he want to?
It’s very safe to assume the following:
1. ADCS-ADL delta failed.
2. P-values given have questionable accuracy (likely they are one-sided tails).
With that in mind, what can he do? I think he’ll just wait (a few weeks!) to reveal additional data (without saying much about already revealed data metrics).
Yes, but in 2014-2015, it might have been silly to spend much money on a 2a
That’s what I said.:)
It’s way beyond Missling and our grievances now. It’s all about how FDA will look at the data presented to them (more complete and explained, than what has been presented to us). We may remain in dark, but that’s likely irrelevant.
Missling doesn’t seem to be in the business of being transparent. Or he has sufficient reasons to hide things.
Is the data analysis done by CRO/third-party, or AVXL? Or that’s unclear?
Thanks for your analysis.
I believe there were (or were very close to) 25 Blarcamesine and 5 Placebo to achieve +3.5 on ADCS-ADL
Yes, it’s not a failure. FDA doesn’t consider 0.05 to be hard cut-off.
No, a company is not obligated to report all aspects of data.
Your #1 and #3 though valid concerns, I doubt they pass the bar for legal prosecution. I could be wrong, but will be surprised.
In either case, it’s really very sleazy and unethical for AVXL to make a fool of the world by claiming that they meant to use responder-analysis all along for their endpoints!
Regarding #2, there is a difference between “co-primary” and “multiple primary” endpoints.
For co-primary, the trial is a success if both are met. In this case, p needs to be only less than 0.05. Makes sense.
For multiple primary endpoints, the trial is considered a success if any one of them is met. In this case, p needs to be less than 0.025 for the one met. Makes sense.
Before you refute any of the above by quoting some link etc, please use your own mind too. It doesn’t make sense for p requirement to change for co-primary.
AVXL endpoints were co-primary.
Yes, Missling couldn’t have lied. But it’s easy to tell the half-truth.
This, I do believe the numbers in PR are accurate, but: (1) n value in unclear (Ie including if excluding dropouts), (2) p-value could be one-sided tails.
And, legally, it will be difficult to claim that he lied about ADCS-ADL endpoint wasn’t net (he just used a different analysis, mentioned in SAP).
Had Missling been completely honest and elaborate, the stock would be way higher (even with the worst-case of unseen data).
Thanks. I’m quite sure the delta-RSBQ numbers are solid too — except that Missling is literally stupid to not have disclosed those. His justification for using RSBQ-AUC (tied to CBR) is solid … except that he goofed up by making it sound like he changed the endpoints.
The stock has crashed hard partly/largely because of Cantor’s questioning the entire platform (and thus Rett too) now due to the side effects in AD trial. But that they may an overreaction.
IMO, Rett is solid — but execution can still continue to be flawed.
Regarding Rett trials:
Has the TLD been completely released on the two completed trials? Other than the RSBQ-AUC controversy (I can look past it), is there any other controversy or issue with the Rett trials’ TLD?
In my understanding, Rett TLD is solid except for the AUC issue. Just confirming without spending time gleaning through the past PRs/presentations.
Not many want to buy it in low 8s.
Must be manipulation. Or fraud perpetrated by Bp, sec, WS.
They are all stopping the eager buyers from buying the stock. How? Well, I can’t imagine … but perhaps threatening them with criminal consequences?
Oops, forgot one. Stopping buyers from buying more by instilling fear in them! Buyers are weak (easy to scare), and sellers .. well .. fearless and manipulative!
(Sarcasm — in case you didn’t get it.)
I don’t think anyone doubts that the result as are good/reasonable — just likely not enough for approval. But I think good enough for a buyout.
If your ADCS calculations are anywhere close, no wonder Missling didn’t share the delta ADCS change … it would be a miserable fail.
Can you share your calculations?
The only hope I have now — is a buyout.
Which can certainly happen. Perhaps, for $15. That will also solve all Anavex’s problems (Missling won’t be screwing it up!)
My criticism was on his write up on data (which really is meaningless at this point, without the critical issue assessed). The interactions can’t be criticized.
1. Possibly the bio markers being evaluated by AVXL in the other endpoints. But unlikely .. since these were exploratory endpoints. But plausible.
2. IND is relevant to conducting trials (shipping and administering a drug) in US, but not approval.
Isn’t it possible that he doesn’t know it? He doesn’t get it? He’s the CEO only for the first time in his life … and is from Germany.. not US.
Why does it happen only with AVXL and always?? Are they cheap?
Mayo: Your write up is complete BS. You don’t address the most critical issues:
1. Why not disclose mean-change in ADCS-ADL?
2. How are the 1.85 and p-values accurate?
Those are really the issues. If you don’t address these, everything else is BS.
Do you realize that 187% and 167% values are absolutely meaningless without the absolute numbers (even if you ignore that these weren’t the endpoints at all),
What’s odds gotta do with regression? Are you trying to determine an optimal threshold? Else, given a threshold, why do you need anything like regression to determine the odds and p-value.
As an example, Exondys got AA, but was rejected by Europe.