Celebration Thread! Biogen is going to the FDA with Aducan.

[J11]

Correlation.GIF

Dose Curve 3.gif

Aducan Weighted 2.gif

[J11]

Correlation 3.gif

[J11]

Sorry everyone I know that it must be rough with all of these posts, yet here is another.

What I was thinking about was pages 220 and 223 from the Briefing document.
From one of my earlier posts I was quite impressed by how strongly the phase 1 results fitted a linear line through the origin
when considering the aggregate dosing arms on the plane defined by SUVR change and CDR-sb change. I thought why not extrapolate this line out further with the help of page 223 which shows how much amyloid removal is thought achievable. The phase 1 and phase 3 trials achieved ~0.25 removal (from 1.4 --> ~1.15). Yet, the expected maximal removal is closer to ~0.36. How much more CDR-sb benefit could be achieved by going the maximum? I extrapolated the line from the phase 1 trial and found the CDR-sb value of 1.29. You can also plug this into the formula in the regression above. One problem is that I did not force the (0,0) point to (0,0). We know by definition that the origin is exactly where the placebo should be. I could have simply added in a great many points at the origin and forced the line through strong weighting to include 0,0. As I did not do this the regression line is actually 0.056 CDR at x=0. So the line above is slightly high and the slope is also somewhat too steep.

Nevertheless I think the fact that the aducan trials left cognitive benefit unreported is a useful point to mention. It is not clear if this calculation is entirely accurate, though it does help to give some estimate of this undeclared benefit. Page 217 of the Briefing Document give the phase 1 10 mg/kg CDR-sb benefit as -1.08. Extending out to the full amyloid benefit gives ~ -0.21 additional benefit by the extrapolation. The phase 3 regression appears to be flatter, so the benefit there would be expected to be less (though this probably relates to the demographic of the phase 3s trial composition which had many younger patients with minimal decline. The more advanced patients in age etc. might also be expected to capture more of the cognitive boost as in the above extrapolation.
Obviously we would all like to see the longer term dosing results for aducan, however unfortunately the futility analysis prevented a large uninterrupted dataset from accruing over the long term.

In a future post I would like to discuss the idea that these FDA Briefing Documents would benefit from being "peer reviewed". I find it surprising that almost everything else that is published is peer reviewed yet the actual Briefing Document that is of public record for the approval of a pharmaceutical oddly is not. This opens the potential to bewilder the non-technical reader more than enlighten them. In parallel peer reviewed articles of the clinical trials from the sponsor during the approval process would also be informative.

I have done my best to respond to the document, though clearly those with expert level knowledge would be not overly amused with my efforts. Then why not have posted rebuttals from these experts? It is important to get this deliberation right and part of that deliberation is the public discussion about the Briefing Document. Clearly some of the deep thought needs to be reserved for the FDA supercomputers, yet there are obvious points of disagreement that have been exposed in the document especially the statistical addendum that could profitably entertain a rebuttal. (I would be interested in how a high level factor analysis. I am still very unclear why this issue was raised by the statistical report, though now that it has been broached a reply would be welcome. Extracting and analyzing what I previously referred to as the Alzheimer Factor (gothic A) might produce lower p-values than were reported in the Briefing Document.

[J11]

After all of these posts what I have concluded?

It is fairly clear to me that aducan is effective.

At a statistical significance level of p<0.01, I would say that it does help with Alzheimer's dementia.
I don't think that predicting down to super-low p-values is realistic because there were some confusing aspects
to the trials that still seem somewhat murky: The thousands of patients that are in the long term
extension will help to clarify these issues.

Approval of aducan will be such an enormous win for the AD community. This will begin the era of controlling
Alzheimer's dementia. I am very excited to see what will happen when reduced methylene blue is added in (assuming
that this will provide additive (or better benefit).) More broadly this will be the start of a new era of a super-aging society.

Even while our loved one was coping with profound loss of cognitive ability, I always had the impression that given the choice that they'd be up running around and having fun. This is an insight that I would never have expected from the typical mainstream portrayal of dementia. Even extremely severe dementia did not stop our loved one from enjoying life. A world beyond Alzheimer's will be what Dicken's described as Recalled to Life. The adventure is about to begin again.

Or perhaps even more accurately, a surge in the centenarian work force!
Our house acted as something of a global refuge as friends from all over the world would stop by and complain about the current generation had no interest in working and only wanted to be promoted into management. Perhaps the future will be the centenarians will do the actual work and the young'uns will do all the managing.

[roxanne]

Thank you J11 for all you hard work. It is so amazing that you were able to elucidate for me and all the people in the forum how Aducan can have a positive impact on our loved ones suffering from Alzheimer's. So beautiful, the adventure is about to begin again....I'm feeling hopeful

Roxanne

[floramaria]

After all of these posts what I have concluded?
It is fairly clear to me that aducan is effective..

I also thank you, J11, for the tremendous job you have done of looking through all of the information from the trials on Aducanumab and posting what you learned. You have have dedicated so much time and energy to this. I would not have even tried to understand it on my own, and am very grateful for your contribution.

[J11]

Thank you roxanne.

I wanted a nice clean read-out on my best understanding of the Briefing Document.
The thread might have gotten somewhat distracted (as a consequence of my wanting to make sure that I
made some attempt to comment line by line on the main ideas), though my conclusion is at least concise.

Admittedly after all of these years it is still surprising that we are this close to winding down Alzheimer's.
The underlying seemingly reasonable assumption that the brain (more specifically the Alzheimer brain)
is just so complicated that it cannot be repaired no longer seems tautological. There are a wave of approaching
treatments. These treatments will have a transformative effect for so many of those on forum and beyond.
Once the scientific method is allowed to feedback upon the results turbocharged by the enormous Alzheimer's
economy, we might finally see a level of support that will push us well past the goal line.

[J11]

Thank you floramaria.

During high school, I would be given some assignment and told to write it up.
I really hated those assignments (I just did not know what to do with them).

The problem that I had was I wanted to do what I did with the Briefing Document; basically
lock myself up for a few months and see what I could think up. But they never allowed you to
do that. It was always, "Oh, well maybe spend a weekend or two on it and then call it quits."
I never understood that.

If you ask someone a question, don't expect to see them for a few months (perhaps years), that's
what makes sense to me. I might not have been as comprehensive as I would have liked with the
Briefing Document, though I was not sure how much more others on forum could take. (So this is a somewhat
abridged version.)

[J11]

Correlation 1a.gif

[J11]

The strong correlation of the 103 study does not seem to have received the attention that it deserved.
The statistical appendix of the Briefing Document found all sorts of technical faults with the study, yet what I find especially persuasive
as seen in the figure above (bottom purple line) is the strong linear correlation. When you plug in r= 0.9559 and n=6, p-value is 0.0029.
This supports strong intuition that there was indeed a strong relation between dose and response in the 103 study. Questioning whether the 10 mg/kg dose actually hit significance then has less importance: What is important is that the totality of all the points is supporting the efficacy claim. I realize that this was a dosing study and it really might be too much to ask for everything, though having a hold back sample would add yet more powerful evidence. Basically, draw your regression line with the dosing build-up and then with the hold back sample (and the determined amount of amyloid difference from placebo) predict where the point should fall given the regression formula and confirm the prediction. If it were me I think that would be something that I would want to add into all phase 1,2 and 3 clinical trials. This might actually be something that is done with clinical data sets.

The reason why I included the orange line pivoting from 0,0 is that i wanted to show how a different population sample might change the regression relationship. What I can imagine happening is that when you add in more patients with earlier illness, it would pivot up from the later patients in 103. Both the orange and purple lines would actually be "correct"; they would just show the regression for different cohorts.

With the 103 study everything fits nicely along the line and it makes a great deal of sense. With the 301 and 302 studies (blue dots) life is fairly complex. The low dose is not actually defined by a single dose; some patients were dosed at 3 mg/kg and others at 6 mg/kg. Same with the high dose, different patients received different doses. As can also be seen the 301 study is way up there and its abscissa does not align with the 302 study. On this figure the phase 3 studies really do seem confusing, though clarification is added in the Briefing Document. I added in the admittedly somewhat implausible regression that the data suggest for the combined phase 3s. As a request it would be highly appreciated if such confusing clinical trials could be avoided
(if possible) in the future.