WEBVTT

NOTE
This file was generated by Descript 

00:00:00.029 --> 00:00:03.740
Judith: Welcome to Berry's In the
Interim podcast, where we explore the

00:00:03.740 --> 00:00:08.140
cutting edge of innovative clinical
trial design for the pharmaceutical and

00:00:08.140 --> 00:00:10.449
medical industries, and so much more.

00:00:10.899 --> 00:00:11.640
Let's dive in.

00:00:13.289 --> 00:00:14.579
Scott Berry: Welcome everybody.

00:00:14.579 --> 00:00:19.469
Back to in the interim today
we're actually gonna take you.

00:00:20.054 --> 00:00:25.274
In the interim, we're gonna take you
into many interims, uh, a unique look

00:00:25.274 --> 00:00:29.354
back, uh, at a trial, a drug, a disease.

00:00:29.774 --> 00:00:36.704
And I have, uh, a wonderful guest,
Don Berry is here, my father, to join

00:00:36.704 --> 00:00:41.114
me, and we're gonna talk a little
bit about the story of Band 2 4 0 1.

00:00:42.599 --> 00:00:47.639
And the, the story of a phase two
trial and the before and the after.

00:00:47.639 --> 00:00:49.799
So welcome back to In the Interim.

00:00:51.120 --> 00:01:01.230
so let's go back to approximately
2013 and the, sorry, state of

00:01:01.230 --> 00:01:04.410
Alzheimer's disease drug development.

00:01:05.250 --> 00:01:13.230
Something like 14 straight failed phase
three trials, uh, promise of drugs.

00:01:13.230 --> 00:01:18.540
But all of these trials failing
and take us back to that time and

00:01:18.540 --> 00:01:19.350
set the stage a little bit, Don.

00:01:22.710 --> 00:01:23.190
Don Berry: Okay.

00:01:23.340 --> 00:01:30.750
Um, it was, as you indicate, uh,
a sorry time, but Alzheimer's,

00:01:31.650 --> 00:01:33.360
uh, researchers to their credit.

00:01:33.360 --> 00:01:34.140
Understood.

00:01:35.460 --> 00:01:36.870
That they were doing something wrong.

00:01:37.290 --> 00:01:38.160
They didn't know what.

00:01:39.540 --> 00:01:47.190
Um, and the Alzheimer's Association
put together a, a round table, uh,

00:01:47.190 --> 00:01:58.620
a meeting in Washington, which had
the goal to study and learn from

00:01:58.980 --> 00:02:03.000
the bit of a workshop, uh, the 14.

00:02:04.170 --> 00:02:09.900
Latest phase three trials in
Alzheimer's, all of which failed.

00:02:10.860 --> 00:02:17.460
And so there was a, each of the
trialists were asked to present

00:02:18.510 --> 00:02:25.500
what the trial was and to the extent
that they could, why it failed.

00:02:26.250 --> 00:02:28.740
Was it the drug, was it the trial?

00:02:30.300 --> 00:02:34.230
They had invited me to
give a presentation.

00:02:34.230 --> 00:02:42.660
I mean, they were aware of,
uh, IFI two, which, um, which

00:02:42.660 --> 00:02:44.940
was a trial in, in cancer.

00:02:45.540 --> 00:02:49.595
But, uh, did, uh, you know, smart
things along the way, and so they.

00:02:50.535 --> 00:02:56.715
Like, uh, many, uh, uh, researchers
at the time sort of salivating over

00:02:56.715 --> 00:02:58.335
doing something like ice flight tool.

00:02:58.425 --> 00:03:06.465
So they, I presented, uh, my
analyses of what the problem was.

00:03:07.395 --> 00:03:13.365
Um, one of the, uh, I won't go
through all 14 of course, but

00:03:13.365 --> 00:03:16.005
one I want to mention is a large.

00:03:17.310 --> 00:03:23.130
A pharma company, uh, that had a
drug that they were looking at for

00:03:23.130 --> 00:03:31.380
Alzheimer's, and they had a phase two
trial, and it led to a large phase

00:03:31.380 --> 00:03:33.360
three trial that bombed completely.

00:03:34.410 --> 00:03:42.420
The person presenting was one of the
instrumental people in the company who was

00:03:42.420 --> 00:03:45.030
part of the design of the phase two trial.

00:03:45.090 --> 00:03:52.290
It was part of the analysis to go
forward into phase three, and he

00:03:53.220 --> 00:03:57.210
admitted, said explicitly, you know,

00:03:58.140 --> 00:04:02.400
we didn't know that we
had a good drug here.

00:04:03.180 --> 00:04:07.500
We recognized that,
that, uh, we might fail.

00:04:07.500 --> 00:04:09.900
The data would not all that.

00:04:11.130 --> 00:04:13.800
Uh, uh, consuming of course, what the.

00:04:14.220 --> 00:04:26.010
The, the, the, the down part was in the
cognition and, uh, function and, you

00:04:26.010 --> 00:04:29.970
know, the drug was getting in, it was
doing what it was supposed to be doing and

00:04:30.630 --> 00:04:32.610
the biomarkers that they could measure.

00:04:33.570 --> 00:04:35.545
Um, but

00:04:35.885 --> 00:04:43.350
it was not, you know, big enough so to
speak, uh, to show cognition and function.

00:04:44.595 --> 00:04:51.435
And, uh, so it was an, an admission
and the, and the person wa was asked,

00:04:51.435 --> 00:04:53.445
okay, so why did you do the trial?

00:04:53.465 --> 00:04:53.705
Scott Berry: trial?

00:04:54.675 --> 00:04:58.005
Don Berry: Why did you do the phase three
trial if you didn't have enough evidence

00:04:58.005 --> 00:05:00.705
that the drug was going to have an effect?

00:05:01.575 --> 00:05:08.235
And he essentially said, well,
yes, we were playing the lottery.

00:05:10.065 --> 00:05:13.215
Um, that is, we're taking a chance.

00:05:14.100 --> 00:05:17.610
And, you know, the chance
might be best based on a, a

00:05:17.610 --> 00:05:22.590
phase, uh, uh, three trial that

00:05:22.620 --> 00:05:22.890
Scott Berry: that.

00:05:23.820 --> 00:05:27.570
Don Berry: just randomly, you know, 5%
of the time, even though you don't have

00:05:27.570 --> 00:05:30.900
anything, you have a, a positive trial.

00:05:31.500 --> 00:05:38.700
So, um, uh, he was, uh, I, I,
I thought surprising, and this

00:05:38.700 --> 00:05:41.160
happens a lot, uh, that you.

00:05:41.595 --> 00:05:44.415
You don't know that you, you
sort of hope for the best.

00:05:44.985 --> 00:05:47.025
So we're in World Series mode.

00:05:47.025 --> 00:05:54.075
So, uh, we have a, a, a pitcher
that's a great pitcher, uh, and you're

00:05:54.075 --> 00:06:00.375
swinging for the fences and you're
hoping to both hit the ball and hit

00:06:00.375 --> 00:06:02.655
it right and hit it into the fences.

00:06:02.655 --> 00:06:04.215
So there's a lot of randomness.

00:06:04.875 --> 00:06:06.225
Uh, let's hope for the best.

00:06:07.305 --> 00:06:13.335
Um, the result of that was at, at
the meeting was Veronica Vinky,

00:06:13.395 --> 00:06:15.765
who was at the time at Eisai.

00:06:16.695 --> 00:06:23.145
Um, and she asked if she could
discuss with me, uh, their

00:06:23.145 --> 00:06:25.365
circumstance with, uh, the.

00:06:26.670 --> 00:06:31.950
The, uh, ban, uh, 24 0 1.

00:06:32.640 --> 00:06:40.650
Um, and uh, so I agreed and I visited
them and, uh, they were enthusiastic

00:06:40.650 --> 00:06:44.430
and that's really what it takes is
enthusiastic part of the company.

00:06:44.940 --> 00:06:51.600
They were enthusiastic about doing
something different, um, and um,

00:06:51.840 --> 00:06:53.850
they ended up doing something.

00:06:54.975 --> 00:06:55.755
Different.

00:06:56.535 --> 00:06:58.035
And Scott, back to you.

00:06:59.265 --> 00:06:59.625
Scott Berry: Okay.

00:07:00.015 --> 00:07:01.575
Okay, so sorry.

00:07:01.575 --> 00:07:06.375
State of drug development question
of whether, whether it's the drugs,

00:07:06.585 --> 00:07:10.425
whether the trial designs, whether
it's the development plan, whether

00:07:10.425 --> 00:07:13.185
it's actually patients and endpoints.

00:07:13.530 --> 00:07:16.110
Do these patients actually
have Alzheimer's disease?

00:07:16.230 --> 00:07:18.000
Are we measuring the right endpoints?

00:07:18.180 --> 00:07:20.610
Lots of blame going around.

00:07:20.970 --> 00:07:27.930
So as you mentioned, uh, ASCI and
Veronica Lavinsky, Andy Satlin, Chad

00:07:27.930 --> 00:07:33.330
Swanson, SBA Dadada Jinping, uh uh, also.

00:07:33.605 --> 00:07:38.765
Uh, was involved in this Jinping Wong, so
a large number of people, large effort.

00:07:38.765 --> 00:07:43.805
We built a phase two trial
and the notion was this is, we

00:07:43.805 --> 00:07:46.175
want a better phase two trial.

00:07:46.175 --> 00:07:48.875
We don't want to just jump
into a phase three trial.

00:07:49.445 --> 00:07:57.575
So I'll, I'll describe the trial somewhat
briefly, but the trial design had five

00:07:57.575 --> 00:08:02.975
different arms of band 2, 4 0 1, the drug.

00:08:03.615 --> 00:08:06.165
It had two different frequencies.

00:08:06.615 --> 00:08:11.745
The bimonthly frequency had
three doses, and the once monthly

00:08:11.745 --> 00:08:16.945
frequency had two doses, so five
active arms and one placebo arm.

00:08:17.880 --> 00:08:24.900
In order to incorporate the five arms and
not have a much larger trial response.

00:08:24.900 --> 00:08:29.430
Adaptive randomization was used in the
trial, so interim analysis would take

00:08:29.430 --> 00:08:34.830
place and allocation would be focused
on the arms that were doing better.

00:08:36.165 --> 00:08:40.485
The endpoints in the trial sort
of plays an interesting role here.

00:08:40.515 --> 00:08:46.605
12 months and 18 months change from
baseline in a cognitive measure in

00:08:46.605 --> 00:08:48.890
this patient population called Adcoms.

00:08:49.335 --> 00:08:57.435
Adcoms is a collection of, uh, questions
from MMSE, Addis Cog and CDR sum of

00:08:57.435 --> 00:09:02.595
boxes that was thought to be, uh, highly
sensitive in this part of the disease.

00:09:02.595 --> 00:09:04.530
And so this was the, the
endpoint in the trial.

00:09:05.715 --> 00:09:12.225
In order to drive the adaptations, the
RAR, we built a model, a longitudinal

00:09:12.225 --> 00:09:18.615
model that would look at three months, six
months, nine months, 12 months, 15 and 18,

00:09:18.615 --> 00:09:24.855
and forecasting forward where the patient
was to identify within patient that

00:09:24.855 --> 00:09:30.075
forecast to identify how well the arms
are working and to drive the adaptation.

00:09:30.075 --> 00:09:34.215
So a key part of this is what
we'll call the longitudinal model.

00:09:35.025 --> 00:09:41.475
Within that at, uh, adaptive analysis,
the trial could stop for futility.

00:09:41.715 --> 00:09:46.515
They wanted to be realistic about
this, given the, the, the, sorry

00:09:46.515 --> 00:09:48.795
state of, of Alzheimer's development.

00:09:49.125 --> 00:09:53.835
They also wanted triggers for
accelerating drug development.

00:09:54.585 --> 00:09:59.655
So if a particular trigger was hit,
they would go quickly to phase three.

00:10:00.825 --> 00:10:06.705
If, if it wasn't futile and it wasn't
this trigger of phase three, it

00:10:06.705 --> 00:10:10.515
would go to the next interim analysis
and it would get more data and more

00:10:10.515 --> 00:10:12.735
data during the course of this.

00:10:13.275 --> 00:10:17.475
So this is largely the
trial design within it.

00:10:17.835 --> 00:10:24.075
Now, the sample size also was,
was adaptive 300 was the minimum.

00:10:24.795 --> 00:10:27.225
And then 800 was the maximum.

00:10:27.495 --> 00:10:31.605
If we didn't know the answer yet and
we kept going, we'd enroll to 800.

00:10:31.875 --> 00:10:36.945
We would then continue interim analyses as
we're getting exposure through 18 months.

00:10:37.545 --> 00:10:38.475
The trial design.

00:10:39.015 --> 00:10:45.285
Now, this is a unique trial
in terms of disclosure.

00:10:46.245 --> 00:10:49.125
Uh, ACI has published.

00:10:50.340 --> 00:10:55.800
The result of every one of the interim
analyses during the course of this trial.

00:10:55.950 --> 00:11:00.360
And there were 16 of them
during the course of this trial.

00:11:00.360 --> 00:11:04.710
So many interim analyses during the
course of this trial, and I think we'll

00:11:04.710 --> 00:11:09.120
take you through the story of the trial,
what happened in the going forward,

00:11:09.480 --> 00:11:14.340
but any other comments about the design
before we sort of walk through the,

00:11:14.760 --> 00:11:17.100
the, the trial and what happened, Don?

00:11:19.590 --> 00:11:21.690
Don Berry: So the, um,

00:11:24.270 --> 00:11:28.110
the interesting thing is, is,
is what happened in the trial?

00:11:31.350 --> 00:11:34.245
uh, no, I think you just get into that.

00:11:34.365 --> 00:11:34.665
Scott Berry: Yeah.

00:11:34.665 --> 00:11:34.995
Yep.

00:11:35.055 --> 00:11:37.335
Yeah, let's, let's, I, I know
where you're going and let's,

00:11:37.515 --> 00:11:39.015
let's not quite go there yet.

00:11:39.165 --> 00:11:40.035
Um, within it.

00:11:40.125 --> 00:11:40.725
Okay.

00:11:41.235 --> 00:11:44.205
So all of this is, has been disclosed.

00:11:44.415 --> 00:11:47.775
Um, interestingly in the
course of this is that.

00:11:48.615 --> 00:11:52.845
Uh, in the trial design, I was
unblinded and part of a group that

00:11:52.845 --> 00:11:54.645
was reviewing the interim analysis.

00:11:54.645 --> 00:11:55.785
Don was blinded.

00:11:55.875 --> 00:11:57.705
Don did not see the data.

00:11:57.885 --> 00:11:59.295
He was involved in the design.

00:11:59.295 --> 00:12:02.415
We were both involved in
the design, but he was.

00:12:02.715 --> 00:12:06.105
Blinded to any results going on
during the course of the trial.

00:12:06.105 --> 00:12:12.225
The trial was conducted by a company
called Ella that they ran the interim

00:12:12.225 --> 00:12:16.665
analysis, they, they curated the
data, ran the interim analysis, and

00:12:16.665 --> 00:12:20.385
then I was involved in reviewing
them with Christine Broo, who's

00:12:20.385 --> 00:12:21.975
also here at Berry Consultants.

00:12:22.935 --> 00:12:29.235
So the first interim analysis in the
trial happens 196 patients enrolled.

00:12:29.955 --> 00:12:34.215
It happens in the date of this,
I don't want to get this wrong.

00:12:34.215 --> 00:12:37.425
The date of this is March 2nd, 2014.

00:12:37.785 --> 00:12:43.725
So the trial design, as Don described,
happened in the time of this, and at

00:12:43.725 --> 00:12:49.125
that time, randomization is updated
for each of the arms in the trial.

00:12:50.265 --> 00:12:54.795
Even at this first interim
analysis, the randomization.

00:12:55.110 --> 00:12:59.370
And this is published, um, with
this, and we'll, we'll add a link

00:12:59.370 --> 00:13:03.870
to this is randomization to the two.

00:13:03.870 --> 00:13:10.530
Higher doses is increased and each of
the other three doses is decreased.

00:13:10.590 --> 00:13:16.650
The placebos held constant, the
trial does not hit futility, uh, it

00:13:16.650 --> 00:13:22.770
can't hit success at this time or, or
flagging to ACI to go to phase three.

00:13:22.950 --> 00:13:24.780
So the trial continues on.

00:13:26.415 --> 00:13:32.835
One comment about, uh, flagging to ASCI
is they were interested if there was a

00:13:32.835 --> 00:13:40.605
high posterior probability that their
effect was at least a 25% slowing.

00:13:40.755 --> 00:13:43.545
That was kind of this
important trigger for them.

00:13:44.145 --> 00:13:49.245
If they were highly likely of a
25% slowing, they wanted to know

00:13:49.245 --> 00:13:52.965
that, be told that, and they
would accelerate to phase three.

00:13:54.600 --> 00:14:00.090
If there was a low probability of a 25%
slowing, they would stop for futility.

00:14:03.540 --> 00:14:03.960
Okay?

00:14:03.960 --> 00:14:07.200
That continues on every 50 patients.

00:14:07.470 --> 00:14:18.240
So at two 50 and 300 similar analyses go
on now at, at, um, 300 is where, uh, an

00:14:18.240 --> 00:14:21.420
an unplanned event happens in the truck.

00:14:23.115 --> 00:14:26.925
Don Berry: So can I interrupt you and,
and, and say a little bit more about, uh,

00:14:28.020 --> 00:14:28.440
Scott Berry: Yep.

00:14:28.575 --> 00:14:33.345
Don Berry: the 300, um, when we talk about
these things, a lot of people say, oh,

00:14:33.345 --> 00:14:35.325
well you need of course, complete data.

00:14:36.075 --> 00:14:38.325
Um, this is not complete data.

00:14:38.325 --> 00:14:40.935
This is all the patients in the trial.

00:14:41.535 --> 00:14:45.195
Nobody is at the, uh, 12
month point at this time.

00:14:45.195 --> 00:14:46.245
The first analysis.

00:14:47.055 --> 00:14:50.265
Um, but we're using the longitudinal
model to predict what's gonna

00:14:50.265 --> 00:14:52.755
happen in the, in, in the future.

00:14:53.415 --> 00:14:58.485
So, um, and the, the most
interesting thing that happened at

00:14:58.485 --> 00:15:00.495
this first analysis was that the

00:15:00.915 --> 00:15:05.715
algorithm was already learning
that the two lowest doses,

00:15:05.725 --> 00:15:06.075
Scott Berry: doses,

00:15:07.040 --> 00:15:07.260
Don Berry: um,

00:15:07.455 --> 00:15:07.675
Scott Berry: um,

00:15:08.235 --> 00:15:09.495
Don Berry: were not effective.

00:15:10.065 --> 00:15:12.795
And so it essentially zeroed out.

00:15:12.795 --> 00:15:14.325
It didn't, it wasn't quite zero.

00:15:15.030 --> 00:15:20.400
Uh, the first time the adaptive
randomization, um, and it was built

00:15:20.400 --> 00:15:25.920
so that they could come back if the
longer, uh, term data pointed out that,

00:15:25.920 --> 00:15:30.030
you know, you're missing something
and, and now it's the low doses are

00:15:30.030 --> 00:15:32.550
actually doing well that never happened.

00:15:33.360 --> 00:15:42.695
Um, the something to address
Scott and something that, um.

00:15:43.605 --> 00:15:48.225
The journalists addressed and
quoted people in the research

00:15:48.225 --> 00:15:51.405
community was the many interims.

00:15:51.405 --> 00:15:55.515
I mean, it was just taken as a negative,
you know, you're, you're doing all

00:15:55.515 --> 00:15:58.995
these things and you're gonna be misled.

00:15:58.995 --> 00:16:06.735
And, uh, there was lots of criticism about
simply the multitude of it, of analysis.

00:16:07.515 --> 00:16:09.255
Uh, would you comment on that?

00:16:10.605 --> 00:16:13.575
Scott Berry: Yeah, and, and, and we'll
come back to this, uh, in, in part

00:16:13.575 --> 00:16:19.545
of the story, um, and it's general
and, and now there's this pervasive

00:16:19.575 --> 00:16:25.155
viewpoint that it's too complicated,
these number of interim analysis, but

00:16:25.155 --> 00:16:30.315
every 50 patients and analysis was
done, and a key part of the design is.

00:16:31.285 --> 00:16:34.315
We're, we're only stopping at that point.

00:16:34.315 --> 00:16:39.805
If we have strong evidence, a 95%
chance that we have at least a 25%

00:16:39.805 --> 00:16:45.335
slowing or less than a seven and a
half percent chance that we have a.

00:16:45.885 --> 00:16:47.055
25% slowing.

00:16:47.865 --> 00:16:52.005
So, and, but if, if we're not there,
we're gonna continue on in the trial.

00:16:52.485 --> 00:16:57.225
So, and then we're gonna update the
randomization as Don described in that.

00:16:57.225 --> 00:17:05.145
So this trial and its ability to update
was critical and the analysis every 50

00:17:05.145 --> 00:17:07.755
patients enrolled was critical as we'll.

00:17:07.755 --> 00:17:10.755
Describe what happened
in, uh, as the trial goes.

00:17:11.900 --> 00:17:16.640
Don Berry: So, as Scott
mentioned, um, the,

00:17:17.090 --> 00:17:17.660
the,

00:17:17.690 --> 00:17:21.650
fact that these are available, the
interim analysis are available, one

00:17:21.650 --> 00:17:26.930
of the places is, um, a publication
that, uh, I was first author and

00:17:26.930 --> 00:17:30.020
Scott was last author in JAMA Network.

00:17:30.590 --> 00:17:31.460
Uh, open.

00:17:31.890 --> 00:17:38.100
And you, what you should do, uh,
dear viewers, is go to that, go

00:17:38.100 --> 00:17:42.330
to the supplemental material, and
you can actually watch a movie.

00:17:42.870 --> 00:17:48.780
You can, uh, show, uh, a movie of what
happened in all of these analyses and

00:17:48.780 --> 00:17:56.070
how did things change Naturally, there's
correlation from one to the next, but you,

00:17:56.070 --> 00:17:59.010
you're looking at the tendency of, uh, of.

00:17:59.580 --> 00:18:04.470
Of where it's going, what it's doing,
it's assigning to the higher doses.

00:18:05.040 --> 00:18:09.300
Uh, and control, along with it
control had the same probability

00:18:09.300 --> 00:18:15.810
as the highest, uh, uh, probability
experimental, uh, uh, drug.

00:18:16.530 --> 00:18:21.270
Um, and it's, it, it's very instructive.

00:18:21.450 --> 00:18:25.260
It's, it's not, these are
obviously not independent.

00:18:25.320 --> 00:18:26.520
We're taking, uh.

00:18:28.035 --> 00:18:35.895
Uh, uh, these serial looks, uh, and it's
one of the few studies that, uh, all of

00:18:35.895 --> 00:18:38.535
the interim analyses are are published in.

00:18:39.705 --> 00:18:40.450
Back to You, Scott.

00:18:40.815 --> 00:18:41.085
Scott Berry: Yep.

00:18:41.355 --> 00:18:46.185
So the first interim analysis, and, and
Don described that it, it's triggered by

00:18:46.185 --> 00:18:48.555
patients enrolled and I, I'll back up.

00:18:48.555 --> 00:18:53.385
The, the, the, there's two Bayesian
models that are critical to this.

00:18:53.655 --> 00:19:01.065
One is a dose response, so the 12 month
change from baseline and Addis Cog is

00:19:01.065 --> 00:19:06.315
modeled in a two dimensional, normal,
dynamic linear model over the five.

00:19:06.660 --> 00:19:08.310
Arms and placebo.

00:19:09.270 --> 00:19:14.730
The other one that turns out to be
even more critical is we built a

00:19:14.730 --> 00:19:21.540
model between each visit and 12 months
for the correlation between them.

00:19:21.540 --> 00:19:26.430
Built a linear model between each one
of these, that a patient's change from

00:19:26.430 --> 00:19:33.900
baseline at 3 6 9, wherever their last
exposure was, is used to estimate what

00:19:33.900 --> 00:19:36.210
is their value going to be at 12 months.

00:19:37.380 --> 00:19:42.360
Multiply, imputed through the
model to estimate the dose

00:19:42.360 --> 00:19:44.460
response and drive conclusions.

00:19:45.480 --> 00:19:51.030
So, as you're flipping through this, as
Don describes at 300 patients, it's a bit

00:19:51.030 --> 00:19:56.010
of a a fun movie to watch at 300 patients.

00:19:56.070 --> 00:20:00.660
The data's at its worst
relative, the arms to placebo.

00:20:01.080 --> 00:20:04.620
The placebo's actually at
that point estimated with high

00:20:04.620 --> 00:20:05.715
uncertainty to be the best arm.

00:20:07.380 --> 00:20:13.830
And the probability that the 10 bimonthly
dose is better than placebo, I'm sorry,

00:20:14.010 --> 00:20:23.220
is as a 25% improvement over placebo
is 18% had that have dropped below 5%.

00:20:23.220 --> 00:20:27.120
The trial would've stopped for
futility, and that's the closest

00:20:27.120 --> 00:20:28.650
it's ever got to futility.

00:20:30.420 --> 00:20:32.400
It moves on to three 50.

00:20:33.405 --> 00:20:42.225
So now by the way, we are in
the, uh, September 26th, 2014.

00:20:42.675 --> 00:20:45.165
So we've moved on about seven months.

00:20:45.165 --> 00:20:49.605
They've enrolled 250 patients,
they're on their fourth interim

00:20:49.605 --> 00:20:54.310
analysis, and in enrollment is
temporarily paused in the trial.

00:20:55.740 --> 00:21:01.770
There's a safety committee that's, that's
looking at the data and they notice

00:21:01.860 --> 00:21:07.620
a, not a, a, a radiological finding.

00:21:08.580 --> 00:21:15.450
It's, we now understand this in the
diseases aria, and it is a radiological

00:21:15.450 --> 00:21:22.710
finding that looks like microbleeds in the
brain, and specifically for the highest

00:21:22.710 --> 00:21:26.190
magnitude dose, which is 10 bimonthly.

00:21:27.150 --> 00:21:31.050
They are noticing this for
patients that are a POE four

00:21:31.050 --> 00:21:34.890
positive, they are asymptomatic.

00:21:35.850 --> 00:21:38.790
So these are not symptomatic
bleeds in the brain.

00:21:38.790 --> 00:21:42.630
They're asymptomatic, but they
were concerned about what is this?

00:21:43.200 --> 00:21:52.890
To the extent that they said ASCI could
not enroll that high dose, one of the

00:21:52.890 --> 00:21:59.040
five arms could not enroll patients
A POE four positive on that arm.

00:22:01.290 --> 00:22:04.380
So we've built this
adaptive randomization.

00:22:05.205 --> 00:22:10.995
For, um, all of the different
arms, one of the stratification

00:22:10.995 --> 00:22:13.125
factors is a POE four positive.

00:22:14.445 --> 00:22:19.365
At that time, the decision is made by
people blinded in the trial that we're

00:22:19.365 --> 00:22:24.585
just gonna separate that randomization
for a POE four positives and a OE four

00:22:24.585 --> 00:22:30.795
negatives, and there'll be different
randomization, adaptive randomizations,

00:22:30.795 --> 00:22:32.955
which will zero out the 10 bimonthly.

00:22:35.655 --> 00:22:37.605
And the trial continues

00:22:38.240 --> 00:22:42.065
Don Berry: Uh, more over the patients
who had been assigned to that.

00:22:42.930 --> 00:22:47.220
Within the last period
of time would be stopped.

00:22:48.000 --> 00:22:52.830
So it was, you know, it was like
a, an a bomb dropping on the trial.

00:22:54.090 --> 00:22:56.490
And, uh, a I was very concerned.

00:22:56.490 --> 00:23:01.320
They knew, of course, about this,
they got the message from the

00:23:01.860 --> 00:23:03.420
regulators and they had to do it.

00:23:04.140 --> 00:23:09.060
Uh, and they talked to me about,
you know, canceling the trial.

00:23:09.060 --> 00:23:13.710
It was such a huge, uh, negative Scott.

00:23:15.015 --> 00:23:18.825
Scott Berry: So the, the, the, this is
one of those cases where an unplanned

00:23:18.825 --> 00:23:23.745
adaptation was able to, very quickly,
by the way, Ella did a fabulous job

00:23:24.075 --> 00:23:29.835
that now it's randomizing patients with
this, this, uh, separated randomization

00:23:29.835 --> 00:23:32.715
for them and the trial continues on.

00:23:33.450 --> 00:23:39.390
At this point, by the way, at 350
patients in it wants to randomize.

00:23:39.420 --> 00:23:41.580
It is absolutely zeroed out.

00:23:41.580 --> 00:23:44.550
The three low doses, the 2.5,

00:23:44.550 --> 00:23:48.990
the five monthly, and the five
bimonthly, it wants the two high doses.

00:23:50.175 --> 00:23:54.765
So a POE four negatives are getting
the 10 bimonthly at a high rate.

00:23:54.765 --> 00:24:00.405
The positives are getting, uh, the
only, the 10 monthly at that point,

00:24:00.615 --> 00:24:05.655
and it is already focused at this
point in the trial on the high doses.

00:24:06.555 --> 00:24:10.365
And now the estimated efficacy,
the probability of at least

00:24:10.365 --> 00:24:14.205
a 25% slowing is about 55%.

00:24:15.450 --> 00:24:21.120
Again, if that's 95, they would've
signaled to ACI back in 2014 to run

00:24:21.120 --> 00:24:23.040
phase three, but it did not signal.

00:24:23.730 --> 00:24:26.700
It says, you know, the
data's good, but keep going.

00:24:27.390 --> 00:24:29.130
Its estimated efficacy.

00:24:29.130 --> 00:24:32.640
That point is approximately a 25% slowing.

00:24:34.170 --> 00:24:40.440
This continues on, and interestingly,
and so now at this point, only about 15

00:24:40.440 --> 00:24:43.080
patients per dose have passed 12 months.

00:24:43.980 --> 00:24:48.960
And it's estimating that that
high dose has about a 25% slowing

00:24:49.200 --> 00:24:51.390
through this longitudinal modeling.

00:24:52.320 --> 00:24:57.180
The early patients enrolled are
starting to move Past 12 months, nobody.

00:24:57.285 --> 00:25:01.845
Don Berry: Data and the Scott
Berry's model for missing data,

00:25:02.280 --> 00:25:02.880
Scott Berry: Yeah, so

00:25:02.880 --> 00:25:08.820
on the 10 bimonthly dose, a POE,
four positives that had less than six

00:25:08.820 --> 00:25:11.520
months exposure were removed entirely.

00:25:12.285 --> 00:25:14.985
So there were different
kinds of missing data.

00:25:14.985 --> 00:25:17.715
There were patients in
the trial who went away.

00:25:18.675 --> 00:25:23.625
There were patients that were forced to
go off drug, and then there were patients

00:25:23.625 --> 00:25:28.515
that were enrolled so frequently that
we just didn't have their 12 month data.

00:25:29.235 --> 00:25:34.635
All three of these types were being
imputed based on their last cognitive

00:25:34.635 --> 00:25:37.660
status in the multiple imputation model.

00:25:40.275 --> 00:25:42.735
It 500 patients enrolled.

00:25:42.735 --> 00:25:46.935
So this is, um, June 19th, by the way.

00:25:46.935 --> 00:25:51.105
These interims took two
or three days to run.

00:25:51.705 --> 00:25:57.165
The data was there, the, the algorithms
were run that largely took a day.

00:25:57.675 --> 00:26:02.175
They were reviewed in 12 hours a day.

00:26:02.745 --> 00:26:06.345
Uh, they were signed, that
they were good to go, and they

00:26:06.345 --> 00:26:07.935
were implemented immediately.

00:26:09.240 --> 00:26:12.570
The information was passed to a
data safety monitoring board, but

00:26:12.570 --> 00:26:17.760
these were done essentially fully
automated with a review to make

00:26:17.760 --> 00:26:20.665
sure nothing weird was going on and.

00:26:21.525 --> 00:26:22.845
The trial continued.

00:26:22.845 --> 00:26:28.845
At this point, it's estimating about
a 25% slowing, which happens to be in

00:26:28.845 --> 00:26:31.455
this really interesting part for aci.

00:26:31.755 --> 00:26:35.985
If there's high probability, it's better
than 25, they wanna run phase three.

00:26:36.135 --> 00:26:39.855
If there's low probability, they
would wanna stop for futility, and

00:26:39.855 --> 00:26:42.465
it's hovering around that uh, point.

00:26:42.815 --> 00:26:47.285
Don Berry: And that's on the basis of
the imputation of the missing data.

00:26:47.675 --> 00:26:54.305
Had you looked at just the raw data
and ignored this, you would've had

00:26:54.305 --> 00:26:59.435
something like, I don't know, 10%
or something, a lot less than 25%

00:27:01.145 --> 00:27:04.210
and the the DSMB um.

00:27:05.595 --> 00:27:10.095
they happy with the the model, or
did they react to the model because

00:27:10.095 --> 00:27:11.325
everything's driven by this model.

00:27:12.360 --> 00:27:12.900
Scott Berry: Yes.

00:27:12.900 --> 00:27:16.110
They, they, they had no
concerns with the model.

00:27:16.380 --> 00:27:21.600
Um, in it, we had, we had, we
actually, uh, created and we only

00:27:21.600 --> 00:27:24.420
interacted with them infrequently.

00:27:24.720 --> 00:27:26.310
The interims took place.

00:27:26.430 --> 00:27:29.400
We said, statistically
they're, they're, they're good.

00:27:29.405 --> 00:27:31.740
And they signed off on it and continued.

00:27:31.950 --> 00:27:36.180
It was this one aspect of the
a OE four positives, and that

00:27:36.180 --> 00:27:37.590
continued throughout the trial.

00:27:40.065 --> 00:27:44.655
This continues on at every one of
these interim analyses, and so the

00:27:44.655 --> 00:27:46.845
trial gets to its full enrollment.

00:27:46.845 --> 00:27:53.895
It actually over-enrolled 856 patients,
uh, ended up over enrolling in the trial.

00:27:54.495 --> 00:27:58.215
And at this point in the
trial, this, this takes place.

00:27:58.215 --> 00:28:02.805
Full enrollment takes
place in September of 2016.

00:28:03.015 --> 00:28:05.295
All patients are enrolled at this point.

00:28:05.595 --> 00:28:08.055
The only thing it can stop for futility.

00:28:08.415 --> 00:28:12.225
And it never got close to futility
again after that third interim,

00:28:12.585 --> 00:28:14.745
or it could signal to aci.

00:28:15.645 --> 00:28:20.835
Now, there was the ACI built in a really
cool part of this trial that this all

00:28:20.835 --> 00:28:26.085
is being built, uh, is being triggered
on 12 month change, but all patients

00:28:26.085 --> 00:28:28.065
are being followed through 18 months.

00:28:29.130 --> 00:28:33.480
And the trial's gonna run and get
full exposure through 18 months.

00:28:33.480 --> 00:28:39.600
And at 12 months, this was billed
publicly as, uh, you know, we'll find

00:28:39.600 --> 00:28:41.610
out if this is gonna go to phase three.

00:28:41.670 --> 00:28:44.040
And at this point they have that.

00:28:44.040 --> 00:28:46.590
If there's an 80%, they
lowered the threshold.

00:28:46.710 --> 00:28:52.500
If there's an 80% probability
of a 25% slowing, they wanted

00:28:52.500 --> 00:28:54.690
to go to phase three in that.

00:28:54.690 --> 00:28:56.700
And again, they could stop for futility.

00:28:58.230 --> 00:29:04.740
They found out at this point it's not 80%
and the actual value, let me find out what

00:29:04.740 --> 00:29:11.250
the actual value of this was at the time,
um, that this interim analysis happened.

00:29:12.540 --> 00:29:13.080
Oh, shoot.

00:29:13.260 --> 00:29:16.710
Uh, at the time this interim
analysis happens when all patients

00:29:16.710 --> 00:29:26.130
are enrolled, the probability of at
least a, um, uh, 25% slowing was 70.

00:29:27.015 --> 00:29:33.285
Was 65%, so that needed
to be 80 and it was 65%.

00:29:33.915 --> 00:29:36.195
So it didn't jump that 80.

00:29:36.345 --> 00:29:43.875
So ACI publicly disclosed this that
we're continuing exposure in the trial.

00:29:43.875 --> 00:29:46.155
We don't know the answer yet.

00:29:47.955 --> 00:29:51.735
And interestingly, I'll,
I'll go to a snapshot of.

00:29:52.290 --> 00:29:56.160
A, an article that was written
by the Alzheimer's in the

00:29:56.160 --> 00:29:58.680
Alzheimer's forum about this result.

00:29:59.100 --> 00:30:03.660
And they interviewed you,
Don, and asked you about this.

00:30:03.660 --> 00:30:07.650
And they also interviewed
a neurologist, a well-known

00:30:07.650 --> 00:30:10.710
Alzheimer's trialist, uh, Paul Aen.

00:30:11.160 --> 00:30:15.690
So they asked you about this result
that at 12 months, we don't know

00:30:15.690 --> 00:30:18.030
the answer yet, and Donna's blinded.

00:30:18.725 --> 00:30:24.755
And he says the ban 2 4 0 1 trial
results so far to the extent that

00:30:24.755 --> 00:30:28.235
we know them, and all he knew
was the trial was continuing.

00:30:28.235 --> 00:30:35.105
It wasn't 80%, and it wasn't less than
seven and a half indicate the therapy

00:30:35.105 --> 00:30:42.125
is neither extremely effective nor
clearly ineffective on balance and in

00:30:42.125 --> 00:30:46.385
context of the, sorry, state of drug
development in Alzheimer's disease.

00:30:46.385 --> 00:30:47.555
That's good news.

00:30:49.065 --> 00:30:55.005
Was Don's quote at this, and we are now
six months from full readout of the trial.

00:30:56.025 --> 00:31:00.765
Uh, and all they know is it's not 80
and it's not less than seven and a half.

00:31:02.595 --> 00:31:07.035
Paul Azen takes an opposite view,
and, and I'll read you his quote and

00:31:07.035 --> 00:31:08.535
I think it's kind of interesting.

00:31:09.165 --> 00:31:12.195
Um, so what do these
results mean for the field?

00:31:12.195 --> 00:31:16.645
Not everyone is convinced that adaptive
trials are the wave of the future.

00:31:17.310 --> 00:31:22.590
Paul Aisen of the University of South
Carolina, uh, California in San Diego

00:31:22.590 --> 00:31:29.760
commented that while adaptive designs
have proven useful in the oncology field

00:31:30.300 --> 00:31:32.580
where endpoints are more objective.

00:31:33.885 --> 00:31:38.715
They may be problematic in ad
trials, Alzheimer's disease trials.

00:31:39.165 --> 00:31:44.025
The great and now quote, the
greatest concerned about a concern

00:31:44.025 --> 00:31:48.645
about Bayesian adaptive approaches
in ad trials relates to our

00:31:48.645 --> 00:31:51.735
cognitive clinical outcome measures.

00:31:52.545 --> 00:31:58.425
These measures are highly
variable and noisy, adapting

00:31:58.425 --> 00:32:00.315
on interim incomplete data.

00:32:02.100 --> 00:32:10.260
Risks chasing noise and making erroneous
decisions regarding efficacy or futility.

00:32:11.855 --> 00:32:16.505
Don Berry: So I'm, I'm just compelled
to say I know that it has noise.

00:32:16.985 --> 00:32:18.485
I know this variability.

00:32:18.755 --> 00:32:19.775
I recognize it.

00:32:19.775 --> 00:32:20.645
I modeled it.

00:32:20.645 --> 00:32:21.190
I live with it.

00:32:22.575 --> 00:32:28.305
And I can tell and build into the
algorithms the uh, and the design,

00:32:28.995 --> 00:32:31.665
what's real and what's noise.

00:32:34.395 --> 00:32:39.825
Scott Berry: And what, what was
so misunderstood by Paul is that,

00:32:39.825 --> 00:32:42.405
what did the adaptive design say?

00:32:43.035 --> 00:32:45.945
It said, we don't know the answer yet.

00:32:46.305 --> 00:32:47.505
Let's keep going.

00:32:47.505 --> 00:32:48.375
We need more data.

00:32:49.845 --> 00:32:54.375
Uh, we, we, we don't know the answer,
and his criticism was, they're, they're

00:32:54.465 --> 00:33:00.375
too risky of making erroneous decision,
claiming success or futility, when in fact

00:33:00.645 --> 00:33:06.645
it knew that the variability and it knew
keep going, go to 18 months, uh, in it.

00:33:06.705 --> 00:33:11.145
So think there was this sort of doubt
and it was built in and part of it.

00:33:12.555 --> 00:33:13.215
Okay.

00:33:13.335 --> 00:33:15.555
Flash forward now six months.

00:33:16.500 --> 00:33:20.520
And the trial finally reads out 18 months.

00:33:21.210 --> 00:33:25.380
And the fi by the way, the, the
randomization, so the adaptive

00:33:25.410 --> 00:33:31.680
randomization, what's the role of
this adaptive randomization there?

00:33:31.680 --> 00:33:38.010
Out of the 856 patients enrolled
in the trial, 238 go to placebo.

00:33:39.540 --> 00:33:42.630
The lowest dose gets 52 patients.

00:33:43.560 --> 00:33:45.390
The next gets 48.

00:33:47.160 --> 00:33:49.770
The other five milligram gets 89.

00:33:49.770 --> 00:33:56.040
Those are the three doses that very
quickly are moved away from the two

00:33:56.040 --> 00:34:04.020
high doses, get 2 46 and 152, and the
only reason it was 152 is because it

00:34:04.020 --> 00:34:05.965
couldn't enroll a POE four positive.

00:34:07.680 --> 00:34:13.830
But at that time, out of the, uh,
out of the band 2 4 0 1 doses,

00:34:13.830 --> 00:34:17.310
587 get put on band 2 4 0 1.

00:34:18.060 --> 00:34:24.030
Two thirds of them, 67%, 398
are on the two high doses.

00:34:25.155 --> 00:34:29.985
So the response adaptive randomization,
when Don goes back to the Alzheimer's

00:34:29.985 --> 00:34:34.845
and says, you need better dose
finding trials with better exposure.

00:34:34.845 --> 00:34:39.525
To understand efficacy, you
need to explore dose it.

00:34:39.555 --> 00:34:41.595
It did that and through response.

00:34:41.595 --> 00:34:45.405
Adaptive randomization very
quickly focuses on the high dose.

00:34:47.460 --> 00:34:49.785
Don Berry: And so you get
greater precision associated

00:34:49.785 --> 00:34:51.945
with the high doses, exactly.

00:34:51.945 --> 00:34:54.495
The doses you're thinking
about moving forward.

00:34:55.035 --> 00:34:56.400
So it's, it's a huge benefit.

00:34:57.090 --> 00:35:01.200
But I wanna point out that this
is an attraction for patients.

00:35:02.070 --> 00:35:08.670
They can be told in the design that
if there is a dose that is doing well,

00:35:09.360 --> 00:35:10.830
we're gonna keep looking at that.

00:35:10.860 --> 00:35:16.170
And so you are, uh, if that has happened
in the trial, uh, you're more likely

00:35:16.170 --> 00:35:17.910
to get a dose that's doing well.

00:35:18.540 --> 00:35:23.790
And not only in this trial, but in other
trials that we use adaptive randomization.

00:35:24.375 --> 00:35:30.225
We find out that that's exactly true, that
the, the end result is that the patients

00:35:30.225 --> 00:35:35.775
in the trial, uh, you know, this has been
something that has been vetted around

00:35:35.775 --> 00:35:39.645
for, uh, decades, uh, in jama, et cetera.

00:35:39.645 --> 00:35:45.555
Is are, are patients in clinical
trials treated better than patients?

00:35:45.555 --> 00:35:48.855
Not in clinical trials, and
some say yes, some say no, and

00:35:48.855 --> 00:35:50.385
there's data on both sides here.

00:35:50.385 --> 00:35:51.450
It's clearly the case.

00:35:52.380 --> 00:35:57.570
That if you're in an adaptively randomized
trial and there is something that's

00:35:57.570 --> 00:36:00.960
positive, you're more likely to get it.

00:36:01.140 --> 00:36:06.540
So it should be an attraction
to not only physicians, um,

00:36:06.630 --> 00:36:08.310
uh, but also their patients.

00:36:11.025 --> 00:36:13.965
Scott Berry: So the trial
reads out at 18 months.

00:36:14.850 --> 00:36:20.820
And the two high doses that I described
with the larger sample size, the 18

00:36:20.820 --> 00:36:26.910
month, uh, estimate for the 10 monthly.

00:36:26.910 --> 00:36:30.930
So this is the, the high dose that's
given monthly, and then there's the

00:36:30.930 --> 00:36:33.060
high dose that's given bimonthly.

00:36:33.630 --> 00:36:37.380
The 10 monthly is estimated
to be an 18% slowing.

00:36:39.840 --> 00:36:46.740
And its probability of being
better than placebos, 93%, but

00:36:46.740 --> 00:36:48.660
it isn't estimated to be 25.

00:36:48.990 --> 00:36:54.870
The high dose given twice monthly
is estimated to be a 27% slowing.

00:36:55.770 --> 00:37:02.430
And again, think about where this is
relative to their targeted 25 and a 76%

00:37:02.430 --> 00:37:07.740
probability that it's at least a 25%
slowing is the estimate of that dose.

00:37:09.645 --> 00:37:13.125
So Asci at this point
does a couple things.

00:37:13.485 --> 00:37:19.095
Uh, it starts a phase three trial
and it selects the 10 bimonthly.

00:37:19.095 --> 00:37:24.375
At this point, aria is much better
understood, and there's actually

00:37:24.375 --> 00:37:29.895
some sense that this is a level of
efficacy of the clearing of amyloid.

00:37:29.895 --> 00:37:34.785
So in the background, the other thing
is the mechanism of this treatment is

00:37:34.785 --> 00:37:37.425
the removal of the pathological amyloid.

00:37:37.425 --> 00:37:45.225
It and they're doing pet scanning and they
it, the high dose is absolutely removing

00:37:45.225 --> 00:37:50.505
amyloid within this and this ar Aria's
thought perhaps to be in effect of this

00:37:50.505 --> 00:37:55.710
in some sense, almost that it's showing
benefit by this demonstration of Aria.

00:37:57.315 --> 00:37:58.725
Uh, it's still a concern.

00:37:58.725 --> 00:38:01.785
And, and of course, uh, uh, if
symptomatic bleeding or a higher

00:38:01.785 --> 00:38:04.965
level of bleeding would of course
be a bad thing within this.

00:38:05.415 --> 00:38:09.795
So they run this phase three trial,
but they also submit the data to the

00:38:09.795 --> 00:38:15.255
agency, the phase two data to the
agency, um, uh, and present to them.

00:38:15.285 --> 00:38:21.105
So they plan a trial of the 10,
bimonthly, enrolling both a POE

00:38:21.105 --> 00:38:22.695
four positives and negatives.

00:38:23.370 --> 00:38:28.470
And, uh, 1,795 patient trial is enrolled.

00:38:31.200 --> 00:38:41.820
The, um, the meanwhile, the FDA
reviews the data, they give accelerated

00:38:41.820 --> 00:38:47.955
approval to Ban 2 4 0 1, which
has now changed its name to lecan.

00:38:49.170 --> 00:38:54.510
They grant accelerated approval based
on the reduction in amyloid, and there's

00:38:54.510 --> 00:39:01.650
a stunningly cool data that if you
look at the reduction in amyloid for

00:39:01.650 --> 00:39:07.860
the lower doses relative to the higher
doses, and you look at the change

00:39:07.860 --> 00:39:12.600
in the cognitive values for those
doses, it's incredibly predictive.

00:39:13.875 --> 00:39:19.995
And I think the fact that the multiple
doses in the effect on amyloid re relative

00:39:19.995 --> 00:39:25.605
to the effect on the clinical outcome is
very highly predicted by the effect on it.

00:39:25.875 --> 00:39:28.905
Almost demonstrating, and I know
we shouldn't say the word, a

00:39:28.905 --> 00:39:33.855
surrogacy of this, I think enabled
them to trigger accelerated.

00:39:37.695 --> 00:39:44.805
Now the phase three trial reads out
1,795 patients reads out, and the

00:39:44.805 --> 00:39:52.095
press release, uh, for this comes out
in 2022 and it demonstrates highly

00:39:52.095 --> 00:39:58.785
statistically significant benefit on CDR
sum of boxes for that dose at 18 months.

00:39:59.145 --> 00:40:04.965
And the estimated slowing of
the rate of decline is 27%.

00:40:06.555 --> 00:40:08.175
It was nailed.

00:40:08.415 --> 00:40:13.635
Uh, normally we, you know, to variability
we wouldn't have expected exactly, but

00:40:13.635 --> 00:40:19.215
it was, it was exactly the estimate from
the phase two trial in that Bayesian

00:40:19.215 --> 00:40:23.475
modeling that's doing the longitudinal
modeling, that's doing the dose response,

00:40:23.625 --> 00:40:29.775
estimated it to be a 27% slowing at 18
months and it hit it right on the head.

00:40:30.150 --> 00:40:35.790
In that which ended up to be highly
statistically significant, and FDA

00:40:35.790 --> 00:40:41.280
gave, uh, kinumab full approval for
the treatment of Alzheimer's disease.

00:40:43.455 --> 00:40:48.135
Don Berry: I go back to that meeting we
had with the 14 failures in phase two.

00:40:49.215 --> 00:40:55.935
Uh, it, it's hard not to wonder,
uh, whether if they had done

00:40:55.935 --> 00:40:58.185
something in those 14 trials.

00:40:58.710 --> 00:41:04.320
Like this, you know, saw something
that was positive, uh, and increase

00:41:04.320 --> 00:41:05.940
the sample size accordingly.

00:41:06.450 --> 00:41:11.485
If we wouldn't have better treatments
today for Alzheimer's disease.

00:41:13.710 --> 00:41:16.710
Scott Berry: Yeah, it's, it's
hard to go back to this trial,

00:41:16.710 --> 00:41:21.330
but I, I describe a couple things
happened that I think was powerful.

00:41:21.690 --> 00:41:25.890
If you go back to the algorithms, and you
can go back to this and, and look at the

00:41:25.890 --> 00:41:33.210
posterior estimate of them when there are
500 patients enrolled in this trial, um,

00:41:33.240 --> 00:41:36.330
and this happens, 500 patients enrolled.

00:41:36.540 --> 00:41:38.595
This happens in 2014.

00:41:39.975 --> 00:41:46.305
The algorithm is estimating this
25% reduction, and it holds pretty

00:41:46.305 --> 00:41:51.165
constant all the way through the
end that it's giving this forecast.

00:41:51.375 --> 00:41:53.010
Now, uh, I, I think.

00:41:53.835 --> 00:41:58.455
A side did a great thing by getting this
exposure and this decision, but it, it

00:41:58.455 --> 00:42:00.525
pretty much forecasted at that point.

00:42:01.005 --> 00:42:06.015
But I also alluded to back at 300
patients when 300 were enrolled, this

00:42:06.015 --> 00:42:08.415
analysis took place for futility.

00:42:09.045 --> 00:42:10.995
The data were not promising.

00:42:11.565 --> 00:42:19.815
Had they run a 300 patient trial, a
small phase two, they might not have

00:42:19.815 --> 00:42:22.305
ever continued development of this drug.

00:42:22.380 --> 00:42:27.780
Okay, so the alternative designs that
could have been run may have led to

00:42:27.780 --> 00:42:30.690
this never being run in phase three.

00:42:30.690 --> 00:42:34.290
It's hard to know, and it's hard
to know the decisions that would've

00:42:34.290 --> 00:42:36.900
been made, uh, in this scenario.

00:42:38.340 --> 00:42:42.105
Don Berry: And also had they
used something other than.

00:42:42.525 --> 00:42:50.055
A reasonable and somewhat sophisticated
missing data imputation process.

00:42:50.535 --> 00:42:57.075
Instead of that 27%, uh, that, uh,
uh, they observed in the phase two

00:42:57.105 --> 00:43:03.855
study, uh, and they looked at just
completers, it would've been 12%.

00:43:06.495 --> 00:43:12.405
Part, and this is part possible to
explain on the basis of the, the fact

00:43:12.405 --> 00:43:19.575
that the missing data were precisely, if
you look at what the missing data were,

00:43:19.785 --> 00:43:24.945
uh, for patients who had been assigned
to the high dose, that distribution was

00:43:24.945 --> 00:43:31.455
very different from the missing data
distribution for the, uh, uh, other

00:43:31.455 --> 00:43:33.705
doses and for example, for control.

00:43:34.560 --> 00:43:44.370
Um, and so the model recognized that, um,
that it was, uh, that whether the data is

00:43:44.370 --> 00:43:53.790
missing, of course, the, the model didn't
know about this, uh, regulatory, uh, folks

00:43:53.790 --> 00:43:59.460
that, uh, mandated, uh, dropping that, uh,
dose from the future part of the trial.

00:44:00.120 --> 00:44:05.460
Um, it only knew what the data
were and was basing it on what

00:44:05.460 --> 00:44:08.550
the longitudinal modeling was.

00:44:08.610 --> 00:44:14.820
So it was, there, there were a number
of, uh, it's sort of a perfect storm,

00:44:14.820 --> 00:44:20.610
or not a storm at all, but perfect
weather, uh, that, uh, everything was,

00:44:21.270 --> 00:44:27.750
uh, moving in the right direction and,
um, it gave rise to a positive conclusion.

00:44:29.550 --> 00:44:32.340
What's the latest?

00:44:32.340 --> 00:44:34.830
I mean, the umab has been fully approved.

00:44:35.550 --> 00:44:42.900
Uh, now another, uh, phase, uh, phase
three trial has been, uh, observed

00:44:42.900 --> 00:44:45.720
with, uh, similar mechanism of action.

00:44:45.720 --> 00:44:48.125
And can you say something about
the current status, Scott?

00:44:49.170 --> 00:44:53.970
Scott Berry: Yeah, so Umab from
Eli Lilly has been approved full

00:44:53.970 --> 00:44:58.230
approval, uh, and it demonstrates
slightly better, uh, slowing of

00:44:58.230 --> 00:45:00.630
cognition, I think at, at 18 months.

00:45:00.630 --> 00:45:03.540
But now I think there are two
treatments that are disease

00:45:03.540 --> 00:45:05.190
modifying that have been approved.

00:45:05.610 --> 00:45:11.670
And interestingly now where Alzheimer's
was a bit of a a, a dead end.

00:45:12.090 --> 00:45:14.010
For pharmaceutical development.

00:45:14.010 --> 00:45:19.920
There's a huge investment now in part
that, that clearly that that, that

00:45:20.280 --> 00:45:25.170
affecting amyloid has an effect and
it has an effect on slowing disease.

00:45:25.500 --> 00:45:31.080
Now, the, going into tau going into
better effects of this there, there's much

00:45:31.080 --> 00:45:33.450
more in the, in the way of development.

00:45:34.380 --> 00:45:36.450
I, I, I wanna point out
something interesting.

00:45:36.860 --> 00:45:38.750
Don Berry: We always
wanted to change the world.

00:45:39.630 --> 00:45:39.750
Scott Berry: the, yeah.

00:45:39.875 --> 00:45:39.995
Don Berry: I.

00:45:40.500 --> 00:45:41.070
Scott Berry: Yeah.

00:45:41.070 --> 00:45:47.070
And so the interesting thing is that
trial design was run meanwhile, A,

00:45:47.460 --> 00:45:53.880
there were more than ACI that were
interested in Don's messaging of

00:45:53.880 --> 00:45:56.395
better phase two trials and Abbott.

00:45:57.165 --> 00:45:58.245
Was one of them.

00:45:58.245 --> 00:46:04.335
And Rob Lens was the clinical, um, in, in
charge of neurology at Abbott at the time.

00:46:04.575 --> 00:46:10.545
And they ran two trials, one of
them a BT 1 26, which is published.

00:46:10.905 --> 00:46:13.875
The results of that stopped for futility.

00:46:14.610 --> 00:46:19.050
And I think largely that the conclusion
was that the drug didn't work.

00:46:19.350 --> 00:46:27.090
So a very similar design run in that
scenario with a different drug sort of was

00:46:27.090 --> 00:46:29.310
pretty strong that the drug doesn't work.

00:46:29.310 --> 00:46:31.500
And I, I think that's
the right conclusion.

00:46:31.500 --> 00:46:34.290
So the design doesn't make the drug.

00:46:34.375 --> 00:46:39.715
But hopefully the design gets the right
answer for the drug, and I think it's

00:46:39.715 --> 00:46:43.885
two examples of putting in a different
profile to a very similar design.

00:46:44.305 --> 00:46:47.395
Gave very different answers
at the end of the day.

00:46:47.925 --> 00:46:52.425
Don Berry: There was an interesting
discussion with Rob and the, uh, clinical

00:46:52.425 --> 00:46:57.885
trial committee at, uh, uh, at at Abbott.

00:46:58.125 --> 00:47:00.585
They asked him at, when, at the beginning.

00:47:01.485 --> 00:47:04.455
So this is a complicated trial.

00:47:04.845 --> 00:47:08.625
I mean, it had many things,
many uh, aspects, very, uh,

00:47:08.625 --> 00:47:09.825
much in common with, uh.

00:47:10.965 --> 00:47:12.645
Uh, the 24 0 1 trial.

00:47:13.305 --> 00:47:18.765
Um, how much time did it take to do this?

00:47:19.455 --> 00:47:20.535
And Rob said,

00:47:21.225 --> 00:47:21.975
you're right.

00:47:21.975 --> 00:47:22.755
It took a while.

00:47:22.755 --> 00:47:27.105
It took us like six months working
with the Berry's to get what we wanted.

00:47:27.765 --> 00:47:34.875
Um, but on the other hand, the trial
ended two years sooner than it would

00:47:34.875 --> 00:47:39.675
have otherwise because we were doing
the right thing in investigating.

00:47:39.975 --> 00:47:45.885
And, and, uh, understanding that we
really did hit futility with that drug.

00:47:47.145 --> 00:47:50.055
So it's good in both directions.

00:47:50.835 --> 00:47:51.125
Scott Berry: Yeah.

00:47:51.130 --> 00:47:51.480
Yeah.

00:47:52.470 --> 00:47:56.040
I, I guess a last thing it would
be good to come back to is you made

00:47:56.040 --> 00:48:02.400
this comment, so in the trial there
were 17 analyses of this data.

00:48:03.270 --> 00:48:08.280
The data's accruing, it changed the
randomization, but this notion that.

00:48:08.790 --> 00:48:14.880
Uh, what's beautiful in the Bayesian
perspective and is that the analysis at

00:48:14.880 --> 00:48:19.350
the end of the trial is unaffected by the
fact that all these analyses take place

00:48:19.740 --> 00:48:22.800
within it, and it makes so much sense.

00:48:22.830 --> 00:48:29.435
And, and there's this view that
somehow these 17 analyses Infect the

00:48:29.435 --> 00:48:37.235
data that it no longer tells us with
the same level of strength and in in

00:48:37.235 --> 00:48:41.135
viewing this and seeing the results
go through and look at this and just.

00:48:42.255 --> 00:48:45.885
We should be doing this in
most of our phase II trials.

00:48:46.095 --> 00:48:49.005
It's good for patients, it's
good for drug development.

00:48:49.005 --> 00:48:52.995
It is not a bad thing that
these analyses take place when

00:48:52.995 --> 00:48:55.245
a snapshot of the data are done.

00:48:55.365 --> 00:48:59.205
It doesn't change the data, it
doesn't change the scientific

00:48:59.205 --> 00:49:04.005
conclusion, and I think this is a
beautiful example of exactly that.

00:49:05.085 --> 00:49:07.245
Don Berry: And in a way it's, uh, it's ai.

00:49:07.245 --> 00:49:09.945
You know, everybody's talking
about ai, but this is a

00:49:09.945 --> 00:49:11.655
prospectively designed trial.

00:49:11.655 --> 00:49:15.495
As an algorithms, it's run it, the
algorithm is like an automaton.

00:49:16.425 --> 00:49:24.495
Um, and so we can evaluate as we did, uh,
the type one error associated with it,

00:49:24.495 --> 00:49:37.425
the power of, uh, the study, um, and the,
uh, again to sell at Tom Park, um, working

00:49:37.425 --> 00:49:39.885
on the logistics associated with it.

00:49:40.725 --> 00:49:47.145
Uh, you know, we, uh, uh, hearkening back
Tom, first work, Tom and I were first

00:49:47.145 --> 00:49:53.895
work together in a Pfizer trial where
we had daily stroke trial, where we had

00:49:54.045 --> 00:50:02.055
daily updates and the algorithm would
re randomize, uh, would reassess the

00:50:02.055 --> 00:50:03.915
randomization probabilities every day.

00:50:04.605 --> 00:50:05.835
Uh, and this kind of thing.

00:50:05.895 --> 00:50:09.015
There is a logistical aspect of this.

00:50:09.840 --> 00:50:13.890
Um, but it's possible to do,
and it's possible to have

00:50:13.890 --> 00:50:15.120
something that's rigorous.

00:50:15.120 --> 00:50:23.550
And, uh, and the FDA has issued
guidance to, um, uh, uh, address the

00:50:23.550 --> 00:50:29.220
question of the, uh, complex, the
adaptive complex adaptive designs.

00:50:29.850 --> 00:50:33.540
Uh, so the FDA is wholly involved here.

00:50:33.570 --> 00:50:34.830
Not only the.

00:50:35.235 --> 00:50:39.795
Neurological part of the cancer
part, but other parts as well.

00:50:39.795 --> 00:50:39.825
I.

00:50:41.925 --> 00:50:42.135
Scott Berry: Yep.

00:50:43.485 --> 00:50:44.385
All right.

00:50:44.385 --> 00:50:47.805
So, uh, again, uh, jama.

00:50:48.495 --> 00:50:52.365
Network open, uh, Barry at all.

00:50:52.605 --> 00:50:57.075
Uh, Lecan map for patients
with early Alzheimer's disease.

00:50:57.075 --> 00:51:01.095
Bayesian analysis of a phase two
B dose finding randomized clinical

00:51:01.095 --> 00:51:08.655
trial has a really unique look at,
uh, uh, a trial that, uh, with a very

00:51:08.655 --> 00:51:10.845
cool story and a very cool impact.

00:51:10.845 --> 00:51:13.485
And so here on, in the interim.

00:51:13.860 --> 00:51:19.860
We've taken you through 17 interims and
a look back at a, at a pretty cool trial.

00:51:20.460 --> 00:51:22.680
So Don, thanks for joining me.

00:51:22.680 --> 00:51:26.850
In the interim, I know you could not
join me in the interim in that trial.

00:51:27.150 --> 00:51:29.370
Uh, but thanks for joining me here

00:51:33.360 --> 00:51:34.440
Till next time.