Donanemab EU Label: Why Subgroup Data Matter for Alzheimer's Drugs

WEBVTT
00:00:00.320 --> 00:00:07.120
Imagine like waiting years for VIP access to this incredibly exclusive, highly anticipated event.
00:00:07.360 --> 00:00:08.480
You've been waiting forever, right?
00:00:08.720 --> 00:00:09.119
Oh, absolutely.
00:00:09.279 --> 00:00:09.359
Right.
00:00:09.599 --> 00:00:10.800
The anticipation is huge.
00:00:11.119 --> 00:00:11.359
Right.
00:00:11.599 --> 00:00:16.559
And thousands of people were actually let in during this sort of um soft launch.
00:00:16.719 --> 00:00:16.879
Yeah.
00:00:17.199 --> 00:00:19.920
But then the grand opening happens.
00:00:20.079 --> 00:00:23.120
You walk up to the velvet rope and there's this new bouncer.
00:00:23.280 --> 00:00:26.320
And suddenly the VIP list is incredibly strict.
00:00:26.640 --> 00:00:27.920
It's a completely different standard.
00:00:28.320 --> 00:00:28.719
Exactly.
00:00:28.879 --> 00:00:32.560
It relies on this hyper-specific medical and genetic profile.
00:00:32.640 --> 00:00:35.600
And if you don't match it perfectly, you're just turned away at the door.
00:00:35.920 --> 00:00:37.679
Which is incredibly frustrating for patients.
00:00:38.000 --> 00:00:38.399
It is.
00:00:38.479 --> 00:00:41.679
And that is exactly what is happening in neurology right now.
00:00:41.840 --> 00:00:46.000
So today we're pulling from a 2026 editorial by Timo Grimmer.
00:00:46.159 --> 00:00:49.200
This was in uh the Journal of Prevention of Alzheimer's disease.
00:00:49.280 --> 00:00:50.799
Aaron Powell A really fascinating piece, yeah.
00:00:51.039 --> 00:00:51.520
It really is.
00:00:51.600 --> 00:00:55.840
And we're using it to ask this one mission critical question for you today.
00:00:56.079 --> 00:01:09.280
When regulators change the VIP list for a groundbreaking Alzheimer's drug, you know, drastically limiting who can actually take it, does the drug still work for the people who actually make it past that velvet rope?
00:01:09.519 --> 00:01:10.879
Okay, let's unpack this.
00:01:10.959 --> 00:01:16.159
Aaron Powell Well, it is arguably the most pressing question in neurodegenerative medicine today, frankly.
00:01:16.319 --> 00:01:22.400
Because to set the stage here, the European Union recently approved two anti-amyloid immunotherapies for Alzheimer's disease.
00:01:22.640 --> 00:01:24.400
Right, denanomab and licanimab.
00:01:24.640 --> 00:01:24.959
Exactly.
00:01:25.200 --> 00:01:27.840
And these drugs, I mean, they represent a monumental shift.
00:01:28.000 --> 00:01:30.480
They actively clear amyloid plaques from the brain.
00:01:30.719 --> 00:01:33.439
But the EU didn't just approve them and call it a day.
00:01:33.599 --> 00:01:36.319
They attached a really severe label restriction.
00:01:36.400 --> 00:01:38.400
Aaron Ross Powell They shrank the pool of who gets it.
00:01:38.480 --> 00:01:39.200
Aaron Powell Massively.
00:01:39.439 --> 00:01:39.599
Yeah.
00:01:39.760 --> 00:01:46.480
Aaron Ross Powell They narrowed the eligible population significantly compared to those pivotal clinical trials that you know originally proved the drugs even worked.
00:01:46.560 --> 00:01:57.200
Aaron Powell And since you listening likely know that the APOE epsilon four gene is like the primary genetic driver of Alzheimer's risk, the irony here is just palpable.
00:01:57.280 --> 00:01:58.400
Aaron Powell It's a huge paradox.
00:01:58.480 --> 00:02:03.519
Aaron Powell Right, because the very people at the highest genetic risk are the ones being barred from the treatment.
00:02:03.760 --> 00:02:11.680
Under this new EU label, if you are an APOE epsilon four homozygote, meaning you carry two copies of that variant, you are entirely excluded.
00:02:11.759 --> 00:02:12.800
Aaron Powell Completely shut out.
00:02:12.879 --> 00:02:13.039
Yeah.
00:02:13.120 --> 00:02:18.960
Trevor Burrus, you can only get the drug if you are a non-carrier or uh a heterozygote, so you just have one copy.
00:02:19.039 --> 00:02:21.840
Aaron Powell And you know the restrictions don't even stop at genetics.
00:02:22.000 --> 00:02:25.199
The label strictly excludes anyone taking anticoagulants, too.
00:02:25.280 --> 00:02:26.240
Aaron Powell Blood thinners.
00:02:26.319 --> 00:02:26.960
Aaron Powell Right.
00:02:27.280 --> 00:02:29.599
And Denanomab actually goes even further.
00:02:29.759 --> 00:02:35.599
It excludes patients with superficial ciderosis, which is a really specific type of prior brain bleeding.
00:02:35.759 --> 00:02:38.560
And it excludes patients with uncontrolled hypertension.
00:02:38.639 --> 00:02:43.120
Aaron Powell Okay, but regulators don't just shrink a target market for fun.
00:02:43.199 --> 00:02:44.400
Aaron Ross Powell No, no, they don't.
00:02:44.479 --> 00:02:46.960
Trevor Burrus If this is a targeted response to aria, right?
00:02:47.039 --> 00:02:50.319
The amyloid-related imaging abnormalities we've seen in these trials.
00:02:50.560 --> 00:02:57.199
But I really want to dig into the mechanics of that for a second, because we know aria involves brain swelling and these microhemorrhages.
00:02:57.280 --> 00:02:58.719
Trevor Burrus, Jr.: Yeah, edema and bleeding.
00:02:58.800 --> 00:03:04.400
Aaron Ross Powell But why does having two copies of that Apoe gene make clearing amyloids so uniquely dangerous?
00:03:04.560 --> 00:03:06.560
Like why did regulators have to step in and stop it?
00:03:06.639 --> 00:03:15.360
Aaron Ross Powell Well, it really comes down to where the amyloid is physically located in the brain and the biological mechanics of how these antibodies remove it.
00:03:15.680 --> 00:03:19.840
Because amyloid doesn't just form these floating clumps in the brain tissue.
00:03:20.560 --> 00:03:25.039
In many patients, it actually aggressively deposits in the walls of the brain's blood vessels.
00:03:25.120 --> 00:03:25.759
Aaron Powell Oh, wow.
00:03:25.919 --> 00:03:26.319
Aaron Ross Powell Yeah.
00:03:26.479 --> 00:03:29.680
It's a condition called cerebral amyloid angiopathy.
00:03:30.000 --> 00:03:36.719
And ApoE epsilon-4 homozygotes, they typically have this massive systemic burden of this vascular amyloid.
00:03:36.800 --> 00:03:38.879
Aaron Ross Powell So the vessels are just coated in it.
00:03:38.960 --> 00:03:39.680
Aaron Ross Powell Exactly.
00:03:40.000 --> 00:03:51.199
So when you introduce a really powerful monoclonal antibody that rapidly strips that amyloid out of the vessel walls, you are temporarily compromising the structural integrity of those actual blood vessels.
00:03:51.360 --> 00:03:56.560
Aaron Ross Powell I mean, you're essentially pulling the mortar out of a brick wall before it even has a chance to heal.
00:03:56.639 --> 00:03:58.560
Aaron Powell Precisely the right mental image.
00:03:59.280 --> 00:04:02.479
That rapid clearance leads to increased vascular permeability.
00:04:03.199 --> 00:04:04.000
Things leak.
00:04:04.639 --> 00:04:04.800
Right.
00:04:05.039 --> 00:04:11.599
Fluid leaks out, which causes the edema, or red blood cells leak out and that causes the microhemorrhages.
00:04:12.479 --> 00:04:25.439
So the E-regulators looked at this data and concluded that for patients with heavy vascular amyloid burdens, the homozygotes, or patients whose blood can't clot effectively because they're on anticoagulants.
00:04:25.759 --> 00:04:25.839
Right.
00:04:26.000 --> 00:04:27.839
The risk of catastrophic bleeding is just too high.
00:04:27.920 --> 00:04:28.959
Aaron Ross Powell Exactly.
00:04:29.439 --> 00:04:31.920
It simply outweighs the clinical benefit of the drug.
00:04:32.000 --> 00:04:39.600
Aaron Ross Powell Which, I mean, that makes perfect biological sense, but statistically, it introduces a massive problem.
00:04:39.680 --> 00:04:40.639
Aaron Powell It really does.
00:04:40.720 --> 00:04:49.279
Aaron Powell Because if you take the clinical trial data and you just strip away all these high-risk patients, do you accidentally strip away the evidence that the drug actually works?
00:04:49.360 --> 00:04:51.360
Aaron Ross Powell That is the exact dilemma.
00:04:51.439 --> 00:04:52.000
Aaron Powell Right.
00:04:52.079 --> 00:04:56.800
Because the original headlines, you know, the big success rates, those were based on the broad crowd.
00:04:56.959 --> 00:05:00.959
If you drastically alter the test subjects, the test results have to change.
00:05:01.040 --> 00:05:05.120
Aaron Powell And that is exactly what clinicians across Europe are grappling with right now.
00:05:05.360 --> 00:05:10.639
Does the efficacy actually hold up in a real-world clinic population governed by these new rules?
00:05:10.959 --> 00:05:11.120
Right.
00:05:11.839 --> 00:05:16.800
So to answer this, Grimmer's editorial highlights a brand new study by Yessin and colleagues.
00:05:17.040 --> 00:05:19.279
They performed a retrospective reanalysis.
00:05:19.519 --> 00:05:24.160
They basically took the 76-week placebo control trial for Donanimab.
00:05:24.480 --> 00:05:27.439
That's the Trailblazer ALZ2 trial.
00:05:27.759 --> 00:05:31.680
And they retroactively applied the new EU bouncers rules to it.
00:05:31.759 --> 00:05:43.199
Aaron Powell So they effectively filtered the original trial data, like they deleted the homozygotes, deleted the patients on blood thinners, deleted the ones with siderosis or the uncontrolled hypertension, and just looked at who was left.
00:05:43.360 --> 00:05:44.639
Aaron Powell Yes, exactly.
00:05:44.959 --> 00:05:52.000
And this resulting quote unquote EU eligible population, it made up roughly 87% of the non-homozygous trial cohort.
00:05:52.079 --> 00:05:53.439
Aaron Powell Okay, so still a big chunk.
00:05:53.519 --> 00:05:54.639
Aaron Powell A very big chunk.
00:05:54.959 --> 00:05:57.920
But they couldn't just average out the remaining scores.
00:05:58.160 --> 00:06:04.480
The European Medicines Agency, the EMA, they actually required them to use a highly specific statistical framework.
00:06:04.639 --> 00:06:07.600
It's called a conservative hybrid imputation method.
00:06:07.680 --> 00:06:08.160
Aaron Powell Okay.
00:06:08.639 --> 00:06:20.079
When I see a phrase like conservative hybrid imputation in a methodology section, my immediate translation is that the statisticians are deliberately trying to make the drug look worse just to stress test it.
00:06:20.160 --> 00:06:21.360
Aaron Powell You're not wrong.
00:06:21.759 --> 00:06:24.720
I mean, imputation is all about how you handle missing data, right?
00:06:24.879 --> 00:06:27.439
Because in any 76-week trial, people drop out.
00:06:27.839 --> 00:06:28.240
Constantly.
00:06:28.560 --> 00:06:31.360
They move, they get sick with something else, they just stop showing up.
00:06:31.839 --> 00:06:32.959
That's the core of it, yeah.
00:06:33.199 --> 00:06:41.120
In standard analyses, if a patient drops out, you might assume, well, their cognitive decline probably would have continued at the same average rate as everyone else in their group.
00:06:41.439 --> 00:06:42.240
Sure, that makes sense.
00:06:42.720 --> 00:06:47.759
But the EMA's conservative imputation basically says, no, let's assume the absolute worst.
00:06:48.079 --> 00:06:48.480
Oh wow.
00:06:48.720 --> 00:06:48.879
Yeah.
00:06:49.199 --> 00:06:55.519
If a patient on the active drug drops out, this mathematical model heavily penalizes the drug's efficacy scores.
00:06:55.680 --> 00:06:58.319
It essentially assumes that patient had a terrible outcome.
00:06:58.399 --> 00:06:59.519
Aaron Powell So it's rigorous.
00:06:59.920 --> 00:07:05.360
It is a rigorous, brutal way of handling missing data, so you don't accidentally overstate the benefits.
00:07:05.839 --> 00:07:11.040
Okay, so they run this filtered data through this brutal worst-case scenario math.
00:07:11.199 --> 00:07:12.079
And the result?
00:07:12.560 --> 00:07:13.439
The drug still worked.
00:07:13.759 --> 00:07:14.560
It really did.
00:07:14.959 --> 00:07:24.319
Under these incredibly strict parameters, dynanomabs still produced an 18 to 28% slowing of decline across five independent clinical scales.
00:07:24.480 --> 00:07:32.079
I mean, we're talking the IADRS, the CDRSB, 8S COG-13, ADCSIADL, and the MMSE.
00:07:32.160 --> 00:07:45.920
Aaron Powell And seeing that consistent benefit across five independent scales, scales ranging from, you know, cognitive recall tests to functional daily activity assessments, that is a profound signal of true biological efficacy.
00:07:46.399 --> 00:07:47.439
It wasn't just a fluke.
00:07:47.759 --> 00:07:48.160
Exactly.
00:07:48.319 --> 00:07:50.480
It wasn't a fluke on one specific test.
00:07:50.720 --> 00:07:54.319
But more importantly, the safety profile didn't just hold up.
00:07:54.480 --> 00:07:56.160
It actually improved dramatically.
00:07:56.480 --> 00:07:58.319
Okay, here's where it gets really interesting.
00:07:58.480 --> 00:08:07.759
Because in this restricted EU eligible group, the incidence of ARIA dropped from 36.8% in the overall trial down to 32%.
00:08:08.160 --> 00:08:08.399
Right.
00:08:08.639 --> 00:08:11.759
And we aren't just talking about a drop in total numbers either.
00:08:12.000 --> 00:08:13.519
We are talking about severity.
00:08:13.839 --> 00:08:14.399
That's the key.
00:08:14.639 --> 00:08:24.319
The vast majority of those cases were mild to moderate and completely asymptomatic, meaning the patients felt fine and the brain swelling was only caught because of routine MRI monitoring.
00:08:24.399 --> 00:08:24.480
Yeah.
00:08:24.800 --> 00:08:30.480
And the critical data point, there were zero ARIA-related deaths in the EU eligible group.
00:08:30.560 --> 00:08:35.440
Aaron Powell Which serves as this incredibly powerful validation of the regulator's mandate.
00:08:35.759 --> 00:08:40.399
By narrowing the label, they didn't destroy the drug's efficacy, right?
00:08:40.559 --> 00:08:45.840
They successfully curated a subgroup where the risk-benefit ratio is unequivocally favorable.
00:08:45.919 --> 00:08:46.879
Aaron Powell I hear that.
00:08:47.039 --> 00:08:52.559
And I totally respect the rigorous math, but I do want to push back on the framing of the efficacy just a bit.
00:08:53.039 --> 00:08:53.200
Okay.
00:08:53.279 --> 00:08:53.600
Go ahead.
00:08:54.080 --> 00:08:58.879
We just threw out 18 to 28% slowing of decline on continuous scales.
00:08:59.120 --> 00:09:01.759
And as a statistical proof of concept, that's fantastic.
00:09:02.320 --> 00:09:12.720
But for you listening right now, if your parent is dealing with an Alzheimer's diagnosis, what doesn't 18% slowing on the CDR SB scale actually look like on a Tuesday morning?
00:09:13.039 --> 00:09:13.840
And it's hard to visualize.
00:09:14.159 --> 00:09:14.320
Right.
00:09:14.480 --> 00:09:16.720
Continuous scales are notoriously abstract.
00:09:16.879 --> 00:09:18.879
They don't really help a family plan for the future.
00:09:18.960 --> 00:09:24.799
Aaron Powell Well, this raises a really important question, and the editorial actually pivots to address this exact frustration.
00:09:24.960 --> 00:09:30.879
Because in clinical practice, we have to translate those continuous variables into discrete threshold events.
00:09:31.200 --> 00:09:31.840
Milestones.
00:09:32.320 --> 00:09:33.039
Exactly.
00:09:33.279 --> 00:09:38.159
Families don't measure the disease in point drops on some 144-point scale.
00:09:38.320 --> 00:09:40.720
They measure it in milestones of lost autonomy.
00:09:41.120 --> 00:09:52.559
And the Justin analysis really delivered on that front, too, because when they looked at disease stage progression, they found a 40.3% reduction in the risk of progressing to the next CDR global stage.
00:09:52.960 --> 00:09:54.000
A massive finding.
00:09:54.399 --> 00:10:00.879
And they also found a 47.2% reduction in the risk of progressing to moderate dementia.
00:10:00.960 --> 00:10:01.440
Aaron Powell Yeah.
00:10:01.519 --> 00:10:05.120
And the editorial draws this brilliant parallel here to oncology.
00:10:05.279 --> 00:10:10.480
Because in cancer medicine, the field relies heavily on a metric called progression-free survival.
00:10:10.720 --> 00:10:11.120
Right.
00:10:11.360 --> 00:10:18.639
If you are treating a tumor, the absolute millimeter difference in tumor diameter from week to week, that isn't the primary focus for the patient.
00:10:18.720 --> 00:10:20.559
Aaron Powell No, nobody cares about three millimeters.
00:10:20.960 --> 00:10:21.519
Exactly.
00:10:21.679 --> 00:10:29.360
The focus is on how long that patient can remain in the current manageable stage of life before the disease aggressively advances.
00:10:29.440 --> 00:10:36.159
Aaron Powell Right, because delaying a stage transition in Alzheimer's is basically the neurological equivalent of progression-free survival.
00:10:36.559 --> 00:10:36.720
Exactly.
00:10:37.120 --> 00:10:42.879
We are talking about concrete thresholds here, of the difference between mild cognitive impairment and mild dementia.
00:10:43.039 --> 00:10:46.480
That is the difference between driving your car safely and having your keys taken away.
00:10:46.960 --> 00:10:52.399
And the transition to moderate dementia, that means losing the ability to live independently without daily supervised care.
00:10:52.559 --> 00:10:52.720
Trevor Burrus, Jr.
00:10:52.879 --> 00:10:59.120
It is the difference between a statistical abstraction and a profoundly tangible lived experience.
00:10:59.519 --> 00:11:10.159
And the editorial argues that this 40.3% lower risk of stage progression should be viewed with the exact same clinical reverence as progression-free survival in oncology.
00:11:10.240 --> 00:11:10.399
Trevor Burrus, Jr.
00:11:10.559 --> 00:11:11.200
It's a huge deal.
00:11:11.360 --> 00:11:12.080
Trevor Burrus, Jr.: It is.
00:11:12.480 --> 00:11:21.200
The fact that this delay in the loss of autonomy remains so robust under the strict EU rules, that is the most vital piece of information for prescribers.
00:11:21.279 --> 00:11:21.759
Aaron Powell Okay.
00:11:22.000 --> 00:11:29.360
Let me play devil's advocate for a second, though, because the source material translates that 40.3% risk reduction into a unit of time.
00:11:29.600 --> 00:11:29.679
Right.
00:11:30.000 --> 00:11:34.159
It translates to roughly five additional months before reaching a given level of decline.
00:11:34.240 --> 00:11:35.440
Aaron Powell Five months, yes.
00:11:35.519 --> 00:11:45.600
Aaron Powell So if I'm looking at this objectively, you've got the bi-weekly intravenous infusions, the constant anxiety of MRI monitoring for brain bleeds, the sheer logistical burden of all of it.
00:11:45.840 --> 00:11:48.879
Is all of that actually worth just five extra months?
00:11:48.960 --> 00:11:50.639
Aaron Ross Powell It's a very sobering question.
00:11:50.879 --> 00:11:54.879
And it represents the very real conversation that has to happen in the clinic.
00:11:55.120 --> 00:11:58.799
But we really must contextualize neurodegenerative timelines here.
00:11:59.039 --> 00:11:59.679
How do you mean?
00:12:00.080 --> 00:12:04.960
Five months of preserved independence at the mild stages of Alzheimer's is not trivial.
00:12:05.279 --> 00:12:05.440
Right.
00:12:05.759 --> 00:12:19.440
That is five more months of managing your own finances, of recognizing your environment, of actively participating in your own life before crossing a threshold of cognitive decline that you can never uncross.
00:12:19.840 --> 00:12:20.639
You can't go back.
00:12:21.039 --> 00:12:21.519
Never.
00:12:21.919 --> 00:12:29.360
For a disease that is historically offered zero disease-modifying delay, five months of preserved high function is a paradigm shift.
00:12:29.679 --> 00:12:36.879
That distinction right there between just living longer and living longer independently completely changes the math on those five months.
00:12:37.200 --> 00:12:37.919
It really does.
00:12:38.240 --> 00:12:42.639
Now, we focused heavily on denanomab so far because of the Justin reanalysis.
00:12:42.799 --> 00:12:54.080
But if this restricted label success is real, we really have to ask if this is just a unique quirk of Denanomab or if it's a universal truth for this entire new class of amyloid clearers.
00:12:54.320 --> 00:12:56.240
So let's look at Lacanimab.
00:12:56.559 --> 00:13:01.200
What's fascinating here is that a parallel analysis points in the exact same direction.
00:13:01.440 --> 00:13:03.759
So another group of researchers, Perry et al.
00:13:08.240 --> 00:13:08.480
Okay.
00:13:08.720 --> 00:13:18.720
And they isolated the 1,521 participants who fit the Apo Epsilon-4 genetic restrictions, which was about 85% of their total trial population.
00:13:19.200 --> 00:13:21.039
And the results were strikingly consistent.
00:13:21.279 --> 00:13:28.320
The clinical benefit on their cognitive scales held firm, and the incidence of ARIA was roughly halved compared to the overall population.
00:13:28.399 --> 00:13:29.759
Aaron Powell Halved, yes.
00:13:30.240 --> 00:13:40.240
So we have two independently developed antibodies run by totally different sponsors in different trials, proving that genotype restriction makes the drug safer without erasing the benefit.
00:13:40.320 --> 00:13:40.879
Aaron Powell Yes.
00:13:40.960 --> 00:13:44.639
However, the editorial forces us to look closely at the methodology here.
00:13:44.720 --> 00:13:45.200
Aaron Powell Right.
00:13:45.279 --> 00:13:45.840
The fine print.
00:13:46.240 --> 00:13:47.440
Always read the fine print.
00:13:47.679 --> 00:13:54.320
The peri-analysis of Lacanimab is currently just a quote genotype-based sensitivity analysis.
00:13:54.559 --> 00:13:58.480
They only stratified the data by the APOE Epsilon-4 restrictions.
00:13:58.799 --> 00:13:59.200
Oh, I see.
00:13:59.519 --> 00:14:04.639
Whereas the Jessin analysis of Donanimab applied the full suite of EU contraindications.
00:14:04.799 --> 00:14:09.200
So genotype, siderosis, anticoagulants, and hypertension.
00:14:09.279 --> 00:14:14.639
Aaron Ross Powell So we're still waiting for a fully matched, strict EU label reanalysis for Lacanimab.
00:14:14.960 --> 00:14:15.279
Exactly.
00:14:15.600 --> 00:14:23.279
And that methodology gap brings us right to the heavy caveats, the scientific fine print that Grimmer is very careful to outline in this paper.
00:14:23.519 --> 00:14:27.039
Because both of these analyses are post hoc and exploratory.
00:14:27.600 --> 00:14:27.759
Right.
00:14:27.919 --> 00:14:32.799
Meaning the researchers went back and sliced the data after the trial was already completed.
00:14:33.039 --> 00:14:37.200
Basically, to answer questions that the trial wasn't statistically powered to ask from the beginning.
00:14:37.600 --> 00:14:37.759
Right.
00:14:37.919 --> 00:14:45.759
It's like uh watching a marathon, waiting for the race to totally end, and then deciding you're only going to count the finish times of the runners wearing red shoes.
00:14:45.840 --> 00:14:47.360
Aaron Powell Yeah, that's a great way to put it.
00:14:47.440 --> 00:14:57.759
Aaron Powell Retrospective data is inherently dangerous, you know, because if you slice a massive data set enough ways, you can almost always find a positive pattern by pure mathematical chance.
00:14:58.320 --> 00:15:06.639
And the editorial explicitly notes these analyses are unadjusted for multiplicity, which is just the formal statistical term for exactly that problem.
00:15:06.720 --> 00:15:09.919
Aaron Powell It's a critical vulnerability in the evidence space, for sure.
00:15:10.559 --> 00:15:15.200
Furthermore, Grimer points out a major flaw in Denatomab's long-term extension data.
00:15:15.279 --> 00:15:16.320
Aaron Powell What happened there?
00:15:16.399 --> 00:15:25.600
Aaron Ross Powell Well, once a clinical trial formally ends, it is generally considered unethical to keep the control group on a placebo if you have already proven the drug works.
00:15:25.679 --> 00:15:26.159
Aaron Powell Right.
00:15:26.240 --> 00:15:27.200
So everyone gets the drug.
00:15:27.279 --> 00:15:30.000
Aaron Ross Powell Therefore everyone is offered the active drug.
00:15:30.159 --> 00:15:35.440
But that means you no longer have an internal randomized control group to measure long-term efficacy against.
00:15:35.519 --> 00:15:42.639
Aaron Powell So to figure out if the drug keeps working in like year two or year three, they had to create an external propensity weighted cohort.
00:15:42.960 --> 00:15:43.039
Yes.
00:15:43.360 --> 00:15:53.200
They pulled data from an external database called ADNI, the Alzheimer's Disease Neuroimaging Initiative, and they tried to match historical patients to the actual trial patients.
00:15:53.279 --> 00:16:04.399
Aaron Powell And while propensity weighting attempts to match patients on things like age, genetics, and baseline cognitive scores, it is materially weaker than having a randomized internal control.
00:16:04.480 --> 00:16:05.519
Aaron Ross Powell Of course it is.
00:16:05.759 --> 00:16:08.639
Because there are always hidden covariates you just aren't measuring.
00:16:08.960 --> 00:16:09.200
Always.
00:16:09.519 --> 00:16:18.799
I mean, you can match their age and genetics, but you cannot match the sheer psychological and physical impact of being enrolled in a massive modern clinical trial.
00:16:19.120 --> 00:16:19.519
Absolutely not.
00:16:19.840 --> 00:16:25.039
Trial patients get intense medical attention, regular check-ins, rigorous schedule management.
00:16:25.279 --> 00:16:29.039
A historical database patient simply didn't have that environment.
00:16:29.200 --> 00:16:30.799
It inherently skews the comparison.
00:16:31.120 --> 00:16:31.519
Exactly.
00:16:31.679 --> 00:16:35.519
And that methodology is exactly where the skepticism comes in from a lot of researchers.
00:16:35.919 --> 00:16:36.399
Makes sense.
00:16:36.799 --> 00:16:40.240
We also have to factor in real-world clinical delivery.
00:16:40.480 --> 00:16:50.879
Because the editorial highlights that in practice, these drugs will likely be given using gradual titration regimens, meaning slowly upping the dose over time to further mitigate that ARIA risk.
00:16:50.960 --> 00:16:52.559
Aaron Powell Which they didn't do in the trials.
00:16:52.879 --> 00:16:53.440
Right.
00:16:53.840 --> 00:17:00.639
We don't yet know how that slower dosing will impact the actual efficacy numbers compared to the fixed doses they used originally.
00:17:00.720 --> 00:17:06.160
Aaron Powell So we have this 40.3% risk reduction in stage progression, which is amazing.
00:17:06.400 --> 00:17:13.119
But we also have serious post-hoc caveats, propensity weighting flaws, and these real-world titration unknowns.
00:17:13.519 --> 00:17:14.720
It's a lot of mixed signals.
00:17:14.880 --> 00:17:20.799
Aaron Ross Powell With all this conflicting weight, how do we actually determine if this drug is a success?
00:17:21.359 --> 00:17:28.400
The editorial argues we basically have to step outside the Alzheimer's bubble entirely and look at how medicine judges success in other fields.
00:17:28.480 --> 00:17:42.400
Aaron Powell If we connect this to the bigger picture, Grimmer compares this Alzheimer's data to a massive success story in cardiovascular medicine, specifically the Fourier trial, which tested a monoclonal antibody called Evaluomob.
00:17:42.559 --> 00:17:43.039
Aaron Ross Powell Right.
00:17:43.200 --> 00:17:45.759
Evalucomab is a major cholesterol-lowering drug.
00:17:45.839 --> 00:17:48.240
It's used for uh secondary cardiovascular prevention.
00:17:48.319 --> 00:17:48.880
Aaron Powell Yes.
00:17:49.039 --> 00:17:56.720
Aaron Ross Powell And in that landmark trial, evaluomob reduced the risk of the primary composite cardiovascular endpoint by 15 percent.
00:17:56.799 --> 00:17:57.200
Aaron Powell Okay.
00:17:58.160 --> 00:18:08.160
Aaron Ross Powell And for the key secondary endpoint, which combined cardiovascular death, myocardial infarction, and stroke, the risk reduction was 20 percent compared to a placebo.
00:18:08.319 --> 00:18:11.359
Aaron Powell A 15 to 20 percent reduction.
00:18:11.920 --> 00:18:16.160
I remember when that trial published, cardiology just went wild over those numbers.
00:18:16.240 --> 00:18:17.119
Aaron Powell They absolutely did.
00:18:17.440 --> 00:18:22.319
That drug immediately influenced international guidelines and is universally considered practice changing.
00:18:22.480 --> 00:18:32.720
The cardiology community, which is practically the gold standard for evidence-based medicine, views a 20% relative risk reduction not as a borderline result, but as a total clinical triumph.
00:18:33.200 --> 00:18:36.720
So a 40.3% reduction in dementia progression.
00:18:37.119 --> 00:18:37.599
Exactly.
00:18:37.920 --> 00:18:38.079
Yeah.
00:18:38.559 --> 00:18:40.319
So what does this all mean?
00:18:40.559 --> 00:18:52.799
If cardiology celebrates a 20% risk reduction in heart attacks, why is the neurology field endlessly debating a 40.3% lower risk of progressing to moderate dementia?
00:18:52.880 --> 00:18:54.720
Aaron Powell It's a staggering double standard.
00:18:55.119 --> 00:18:59.599
It means we hold brain health to a disparate, almost impossible standard.
00:18:59.759 --> 00:19:11.599
We are so historically conditioned to failure and disappointment in Alzheimer's research that when a drug comes along and performs at double the magnitude of a superstar heart medication, we almost don't know how to celebrate it.
00:19:11.759 --> 00:19:13.680
Aaron Powell It's a profound observation.
00:19:14.079 --> 00:19:26.400
I mean, preserving autonomy and delaying the transition to dependent stages of dementia is one of the most consequential goals in contemporary healthcare today, especially given the staggering societal burden of the disease.
00:19:26.720 --> 00:19:27.200
Absolutely.
00:19:27.599 --> 00:19:36.559
This data forcefully argues that for the specific subgroup of patients Europe has actually chosen to treat, that goal is now realistically within reach.
00:19:37.119 --> 00:19:44.960
Synthesizing all of this, for the clinicians and patients who are navigating this totally new landscape, this editorial really bridges the gap.
00:19:45.119 --> 00:19:51.440
It takes those massive, abstract trial statistics and translates them into credible, individualized data.
00:19:51.920 --> 00:19:52.880
It makes it tangible.
00:19:53.200 --> 00:19:57.200
It proves that the narrow EU label hasn't ruined the drug's potential.
00:19:57.359 --> 00:20:02.319
It has actively curated a population where the benefits clearly outweigh the risks.
00:20:02.720 --> 00:20:04.799
But you know, the work is far from over.
00:20:05.119 --> 00:20:05.680
Oh, for sure.
00:20:06.079 --> 00:20:10.640
What the fuel desperately needs next is prospective, registry-based follow-up.
00:20:11.519 --> 00:20:22.240
We need to track the patients who are actually being treated under this new label in real time across Europe, applying the exact same strict eligibility definitions for both Donanimab and Lacanimab.
00:20:22.640 --> 00:20:22.960
Right.
00:20:23.279 --> 00:20:34.720
We must continuously track those stage progression endpoints, those lived experiences we talked about, alongside the cognitive scales, just to see if the retrospective math actually holds up prospectively.
00:20:35.200 --> 00:20:35.359
Right.
00:20:35.440 --> 00:20:38.559
We need the live game to prove the replay was actually accurate.
00:20:38.799 --> 00:20:42.880
As we wrap up this deep dive, I want to leave you with one final thought to mull over.
00:20:43.039 --> 00:20:48.240
Building directly on Grimmer's point about that chasm between controlled trials and clinical reality.
00:20:48.559 --> 00:20:49.279
It's a big chasm.
00:20:49.599 --> 00:20:56.559
The source notes that we still don't know how real-world comorbidities and medication adherence will ultimately affect these outcomes.
00:20:56.720 --> 00:20:57.759
So think about this.
00:20:58.000 --> 00:21:01.279
In a clinical trial, patients are monitored relentlessly.
00:21:01.440 --> 00:21:02.880
Their adherence is basically perfect.
00:21:03.200 --> 00:21:04.079
Handheld the whole way.
00:21:04.400 --> 00:21:04.720
Right.
00:21:04.960 --> 00:21:18.000
But when these drugs enter the chaotic, messy reality of everyday human life, I mean, we're talking missed appointments due to winter colds, transportation issues, getting to the infusion center, or even just fluctuating blood pressure from daily stress.
00:21:18.160 --> 00:21:24.400
Will the strict bouncers at the door of the EU label be enough to protect that fragile 40% success rate?
00:21:24.640 --> 00:21:29.680
Or will the sheer friction of real life erode the benefit we saw in that pristine trial environment?
00:21:29.839 --> 00:21:32.960
It really is the ultimate test of science versus reality.
00:21:33.279 --> 00:21:37.200
It is the defining variable for the next decade of neurodegenerative care.
00:21:37.599 --> 00:21:41.680
And we'll be right here to analyze that real world data the moment it publishes.
00:21:41.920 --> 00:21:43.359
Thanks for joining us for this deep dive.
00:21:43.440 --> 00:21:44.160
We'll catch you next time.