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

Alzheimer's, cardiovascular, and other chronic diseases; biomarkers, lifestyle, supplements, drugs, and health care.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

These are the numbers for the Clarity trial.

The bottom one is the CDR-sb.
We have 1.67 for placebo and 1.22 for treatment giving a 0.45 difference.
The trial had 1795 split equally into 898 per arm (though I did not include drop outs which can be significant).
Standard error given for the 10 mg/kg was 0.21 and n=84 --> SD= 84^0.5 * 0.21 = 1.95
Standard deviation using the same approach for placebo was 2.0.

We were given the p value as 0.00005 so working it backwards we have a z score of 3.9 and this yields a standard deviation of 2.45 which will be included for both the placebo and treatment arms.
That was the CDR-sb topline which aligns well with the phase 2 result in relation to the standard deviation.


Probably the big question left to answer is the ADAS-cog 14.
In the phase 2 lecanemab trial this result readout as a 47% reduction in decline for treatment.
This suggests the p-value involved will be truly profound.
How profound?

Top figure shows the same number crunch.
The t statistic with infinite degrees of freedom is equivalent to a z score equal to the t value. Here z= 6.8.
Hmm, the table doesn't go that high.

Will have to go with an approximation.
Best I can figure the p-value is 0. 000 000 000 010 7 (10 leading zeroes!).
Easiest to just go with 107 ten trillionths; probably better still to just go with 10 trillionths; even one hundred-billionth.

There has been something of a tendency for the ADAS-cog and the ADCS-ADL to switch around. for example with Emerge, it was the ADCS-ADL that read out with a 46% reduction while the ADAS-cog readout out with a 25% reduction. Choosing the iADRS is the simple safe way to average out the tests and largely expect a smallish p-value.

I stated before on thread that I wasn't that comfortable going super low with p-values; there is the statistical truth and then there is the truth of everyday life. It seems all too easy for the truth of everyday life to be a bigger factor than what might be expected in an idealized world; probably best to just go p<< 0.000 01 and then just call it.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

Leca diff 2 ADAS-cog 14 p2 b.PNG
Leca diff CDR-sb 2a.PNG
You do not have the required permissions to view the files attached to this post.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

Leca diff 2 ADAS-cog 14 small 2.PNG
Leca diff 2 ADAS-cog 14.PNG
Leca diff 1a.PNG
You do not have the required permissions to view the files attached to this post.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

OK, I did a redo on the number crunch.

The post two up, shows the calculation using the information as given in the phase 2 article.
The calculated result given is slightly different. This might be true because the numbers were from MMRM.
It is possible that with MMRM they are including more of the available information from the drop out patients and this cannot be recapitulated by performing the as indicated calculation. If anything, we would then expect the true result for the p-value to always be lower than our calculations.

The problem with the first round of number crunching given yesterday was that it did not properly account for drop outs. The compensation that was used was to have the higher non-drop out number of the full Clarity patient population number and then just inflate the standard deviation until we hit a 3.9 z score.

When I thought about, I realized that it would be better to enter all the numbers from the phase 2 Lecanemab study and then solve for n. This should give us the best estimate. The standard deviations seem to largely remain constant; the one uncertainty is how different the sizes of the placebo and treatment arms will be at final readout. I assumed that they would be the same. It is possible, though that somewhat more patients will drop out in the treatment arm.

The bottom result from the stats calculator above gives us n= 600 in the placebo and treatment arms. Roughly, 350 patients are expected to drop out of these arms. This n was chosen as it then gives us a t-statistic of 3.9. This is the t-statistic previously found to yield a p-value = 0.00005 -- the reported p-value for CDR-sb given in the Clarity topline press release.

What I then thought was we could then pivot off this n = 600 into the ADAS-cog to give us a better p-value for it. We know how many patients are in the treatment and placebo arms from the CDR-sb calculation so we can include this in the ADAS-cog calculation. That's what the middle figure in the post above gives us. We have a t-statistic of 5.525. This is equal to a z score of 5.525 because the t-statistic has an infinite number of degrees of freedom.

A z score of 5.3 gives a p-value of 1e-7
z score of 5.7 gives a p-value of 1e-8

Using an approximation formula, Pr(z<=5.525)= 3.4 e-8

So the updated p-value for the ADAS-cog 14 assuming that it also hit ~47% reduction in decline (equal to the phase 2 result) would be p< 0.000 000 1 or less than one ten-millionth. (using the approximation, = ~ 3 hundred-millionth).
Last edited by J11 on Sun Oct 09, 2022 12:38 pm, edited 1 time in total.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

Dona 2.PNG
Dona 1.png
You do not have the required permissions to view the files attached to this post.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

Dona 4.PNG
Dona 3.PNG
You do not have the required permissions to view the files attached to this post.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

New Donanemab article has been published; there are some new perspectives on the amyloid/tau dataset.

Top figure shows how change in amyloid relates to baseline amyloid. I added in the red line which shows "0" amyloid; not sure how some of the points are below the line. Is there negative amyloid?

Up till now we have had a largely flat world view of the CDR-sb vs SUVR amyloid plane. This article and others add in other dimensions. Thinking less in terms of absolute amyloid removal and more in terms of percentage amyloid removal from baseline is a good first step towards a more complex view of how amyloid relates to cognition.

Second from the top is another good one; it shows how the percentage of patients achieving an amyloid clear status relates to baseline amyloid level. It might be helpful if they considered ARIA rates as a function of initial amyloid level or the ARIA rate versus change in amyloid.

First in the post directly above is another great one. On the right it shows how cognition change versus % amyloid change relates to e4 and non e4 genotype. ZERO change for the non e4s!! I had not thought about how the Clarity result will continue to be refined. In an earlier post, I estimated that Clarity should read out a ~0.60 benefit on CDR-sb for the APOE e4s. The figure above shows how the benefit would distribute according to the amyloid percentage cleared for the e4s. What is of interest is that in the above figure at 100% amyloid clearance the benefit is 45% for the e4s. Yet, the ~0.60 inferred for Clarity (~40% benefit versus placebo) is only the average for the e4s. This does suggest the question what would be the 100% amyloid clearance CDR-sb benefit for the Clarity e4s? Possibly could think this through, though for a complete guess 1.0 or greater CDR-sb does not seem unlikely. The Aducan trials had substantial numbers who achieved 1.0 or greater benefit. The subgroups in Clarity could have biggish numbers.

Figure directly above shows how tau changes in various brain regions relates to partial or complete amyloid clearance.
Notice how the frontal cortex has the largest response to complete amyloid clearance (as it relates to tau).

This line of research could help to deepen our understanding of the response to mabs. I had thought before on thread about rotating to specific brain regions. I wonder if they are thinking about perhaps moving to a bespoke brain benefit analysis for mabs. Instead of thinking merely in terms of a crude metric such as CDR-sb, one could imagine that PET amyloid/tau scans could reveal where the amyloid/tau was most removed and then provide clinically accurate assessments of what this would imply for cognitive and functional effects.

That could be extremely powerful. This might also help to reduce the confusion that now occurs with the various tests that give different results. With a specific cognitive test for the frontal lobe, the linkage between neuropathology and cognitive/function could be better determined.

Really looking forward to some open science in the upcoming readouts. All the data points in this article along with supplemental data should be provided in a csv file. Having multivariate datasets for patients would allow a much richer view. As it is now, all of this deeper insight is lost in 2D space. Providing this additional information would seem advantageous to the companies because observers could then understand how until now hidden variables were largely adding noise into the result. At some point revealing this noise by providing richer data could allow the entire result to pop into focus. There continues to be a fair amount of noise in the results; as of now, clear focus is only achieved when the numbers are averaged at the group level. Patients are ultimately more interested directly in their own specific outcomes. Providing more data could help to resolve down to the level of the individual response.
Last edited by J11 on Sun Oct 09, 2022 12:47 pm, edited 1 time in total.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

There is so much to keep up with! One small corner of the universe and there is so much happening!

Up till this point I have tried to stay mostly focused on the science and not much else. However, it has occurred to me that there is also the political perspective to consider. In this respect I have thought that perhaps rebundling the Alzheimer's file within government might be helpful. Typically restructuring organizations does not always seem to be that useful, yet with AD there could be synergies available to harvest.

As it is now there is the usual government silos that manage AD such as CMS through Medicare/Medicaid, Department of Health, etc.. The ordinary logic then develops where each silo tries to minimize its costs relative to its own budget. My idea is to merge all of these separate silos into one mega silo. How would that help? CMS in its NCD for antiamyloid mabs went out of its way to limit coverage. This would seem to follow the "save" the budget model of thinking. They saw the potential of a $5 billion ... $10 billion .... $20 billion ... ... obligation for AD mabs and balked. Yet, it needs to be remembered that the hard dollar cost to the government is already ~$350 billion and rapidly increasing. "Saving" that $ 10 billion in silo 1 could then cost silo 2 $30 billion. It is only when you have these separate silos that any illusory "savings" materializes. From the macroview "saving" money is not saving money at all but costing money.

Reorganizing this so that there is an AD supersilo would solve the problem. Externalities could then be internalized and the optimal macrofiscal result could be achieved. There is roughly $350 + billion per year of Dementia Dividend to claim and the sooner it is claimed the bigger the payback. Interestingly in this framing one might expect that the strategy would not be so much delaying mab rollout but accelerating it. Once the AD conveyor belt is stopped then there would be an enormous fiscal windfall. The easiest way to achieve this would be to move towards pre-symptomatic dosing.

One suggestion that could help move us there quicker would be to create the concept of mab treatment eligibility.
I mentioned this idea before, though I think it was such a great idea that I'll mention it again. One of the problems that has been part of mab development was that biotech companies needed to show a cognitive benefit in a shortish clinical trial of 18 months and clinical trials are not cheap to run. This lead to driving the uptitration and having some ARIA side effects. These ARIA adverse effects are partially related to the comingled cerebral amyloid angiopathy that occurs in AD. Amyloid lines blood vessels and then the mabs interact with the amyloid and cause edema and hemorrhage. Once this CAA stage of the titration is completed and the more purely AD portion of amyloid removal starts the rate of ARIA decreases greatly (e.g., 26 of 29 serious ARIA events with Aducanuamb in the phase 3s occurred before the third 10 mg/kg dose). The ARIA problem can then be viewed more as a peripheral consequence of CAA than the actual AD pathology.

This has me wondering whether a pretitration phase could be introduced. A patient could be pretitrated perhaps 2 or more years ahead of actual therapeutic dosing. In this scenario there would no longer need to be the same level of urgency to fight against clear symptomatic decline. Pre-titration would occur before an MCI diagnosis. One would pre-titrate in order that when the established protocol began there would be much reduced ARIA (because the CAA amyloid would have already been burned through). In time one could imagine that this pre-titration protocol could be an eligibility requirement for therapeutic dosing. Such predosing would be especially helpful for 44s (and possibly 34s) who have higher ARIA risk along with higher amyloid levels.

The beauty of this predosing concept is that it avoids the question of clinical efficacy altogether. One would not need to establish a clinical benefit; the intention would be to establish that pre-dosing enhanced safety of the typical titration. This could be a fantastic way to roll back mab treatment, increase safety and possibly also enhance clinical efficacy. Such a plan might also be achievable over a reasonably short timeframe. One for example, might start by selecting patients that were 6 months from an MCI diagnosis, pretitrate them and then titrate per schedule once they reached official MCI onset. A year of pretitration, titration and then treatment could then determine whether this scheme reduced era risk.

As this was shown effective, pretitration could happen years in advance and would be a means of controlling CAA and to establish mab eligibility in the future. It will also presumably help to reduce the risk involved with the buildup of amyloid in the cerebral blood vessels.

It would be interesting to know what the FDA would think of this. Could the companies expand out the label in this way without it triggering some requirement to demonstrate efficacy? The endrun involved is that efficacy is not even the objective: the objective is to improve the safety of mab therapy by preemptively reducing the CAA that causes the ARIA. Would a safety enhancing innovation be considered worthy of an indication label?

The current game plan to roll back Lecanemab to an earlier stage is AHEAD 3-45. This trial is thinking in terms of proving clinical efficacy in those pre-MCI. That is heavy lifting. This requires much better cognitive tests such as the PACC and it will not readout for 5 years. Merely establishing safety with the pretitration suggestion might be doable over the more immediate term of ~2-3 years. The potential to miss the primary in a safety trial would seem less than for an efficacy trial.
J11
Contributor
Contributor
Posts: 3351
Joined: Sat May 17, 2014 4:04 pm

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

Post by J11 »

Last edited by J11 on Tue Oct 11, 2022 3:24 pm, edited 3 times in total.
NF52
Support Team
Support Team
Posts: 2772
Joined: Tue Oct 25, 2016 9:41 am
Location: Eastern U.S.

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

Post by NF52 »

J11 wrote: Sat Oct 08, 2022 6:18 pm There is so much to keep up with! One small corner of the universe and there is so much happening!
You've probably seen the Oct.7 AlzForum article on lecanemab, but here's my link to it:
Lecanemab: "Seismic" changes to clinical trials & MCI options?
4/4 and still an optimist!
Post Reply