Advances in Dementia Imaging: Insights from Experts

The Journal of Nuclear Medicine presents a new podcast on advances in dementia imaging. Panelists discuss the importance of amyloid and tau PET imaging in dementia diagnosis and treatment, highlighting their roles in clinical trials and future therapies. They emphasize the significance of early detection and the potential for personalized medicine approaches.
Panel: 
• Ken Herrmann, MD, MBA (Universitätsmedizin, Essen, Germany)
• Henryk Barthel, MD, PhD (Leipzig University)
• Gil Rabinovici, MD (University of California - San Francisco)
• Victor Villemagne, MD (University of Pittsburgh)

Transcript

Ken Herrmann (00:01.782)
Hello, my name is Ken Herrmann. I'm a nuclear medicine physician from Essen in Germany. And I'm very happy to welcome you to the second edition of the J&M podcast. First of all, thank you very much for looking at the first edition of the J&M podcast. We got a lot of feedback, valuable feedback. One of the key feedback was to not spend again 45 minutes on the podcast, so we will do better today, I promise. We have a very exciting topic. We're going to talk about dementia imaging.

And as all of you know, I'm pretty clueless about dementia imaging. The nice thing about being in J&M Associate Editor is that I can actually access the most talented and knowledgeable people. And that's why it's my great pleasure to introduce the moderator of today's podcast, Professor Henrik Bartel, who is a nuclear medicine physician from Leipzig in Germany, and clearly one of the two dementia experts in the editorial board of J&M. Henrik, welcome to the podcast.

Henryk Barthel (00:57.634)
Thanks so much, Ken, and thanks to everybody for setting this up and organizing that. And it's a great honor and pleasure for us to welcome Gil Rabinovich and Victor Wilmanier. Gil, you are a neurologist and the Edward Fine and Pearl Landwest Distinctive Professor and Director of the Alzheimer's Disease Research Center at UCSF.

You are known as a worldwide leading expert in biomarker based Alzheimer's diagnosis and have many important tasks and you are for instance, PI of the extremely important IDEA study, which we hopefully will have time to discuss. Victor Wilmania, as to my knowledge, you are a nuclear physician by training. Is that correct?

Victor Villemagne (01:54.076)
Unclear Medicine, yes.

Henryk Barthel (01:55.614)
Ah, yeah, yeah. And you are likewise a top expert, especially in the testing and implementation of amyloid and top-head imaging to improve the diagnosis of Alzheimer's and other neurodegenerative disorders. You are the Levidov-Pittsburgh Foundation Chair in Alzheimer's Disease and Dementia Disorders and a Professor of Psychiatry at the moment at Pittsburgh University.

So thanks so much again to both of you for giving us the chance to talk to you today.

Ken Herrmann (02:32.978)
And Henrik, before I let you...

Gil Rabinovici (02:33.149)
It's a real pleasure to be here. Sorry. I just said it was a, it's a real pleasure to be here. Thank you for the invitation.

Henryk Barthel (02:35.566)
Excuse me.

Victor Villemagne (02:40.416)
And thank you for the invitation.

Ken Herrmann (02:43.198)
And Henrik, before I let you go into the details, I would like to kick off and repeat the experiment from the first podcast about the Amazon Sinking Backwards experiment. I would ask all three of you very quickly, no more than 30 seconds to tell me if you would have to have to issue a press release in five years about where we are with dementia imaging, where we are. Please, maybe Henrik, you as a moderator, you start.

Henryk Barthel (03:06.338)
Oh, what an honor. My hope is really that in five years, we managed to get amyloid imaging into clinical care. That's my big hope. We all know that amyloid imaging is suitable to improve the diagnosis, which is extremely important to patients and caregivers and treating physicians. And we also.

I also expect amyloid and tau imaging to really qualify patients in the future in terms of prescription of anti-amyloid and potentially also anti-tau drugs. And that is really my big hope that we will see that coming. Also, in respect to this exciting developments, of course, in terms of fluid biomarkers, but I'm very optimistic that we will see PET imaging really entering.

the clinical arena in five years.

Gil Rabinovici (04:08.613)
Yeah, I share your optimism, Henrik. So I think this year we're celebrating 20 years for the first paper on the amyloid pet in humans. And it's been over 10 years since the first TauPet paper was published. And yet still amyloid and TauPet in the US and Europe and many other countries is still primarily used in research centers and academic centers. It's not widely used in clinical practice. And...

I share your optimism. I actually think that five years from now, Amyloid and Taupet will be an integral part of our clinical decision-making around diagnosis and implementation of new therapies using more precision medicine approaches than we have currently to identify the right patients for the right treatments to determine whether drugs are working and changing the course of the disease. I will also make a bold prediction that we will have a pet tracer.

either for alpha-synuclein or for TDP43, within five years that will be performing very well in humans and will give us even more insight into the protein aggregates that are driving late life cognitive decline. So we'll see how well that prediction ages, but that's my prediction today.

Henryk Barthel (05:06.211)
Wow.

Victor Villemagne (05:24.955)
Okay, I will echo you both. My prediction is for repeat tau tracer, more than TDP43. TDP43, I think it's going to be much more difficult. There are five different subtypes. It doesn't form an amyloid. It aggregates without forming an amyloid. So if we follow the same tracers that we are

using today might not work because they're beta-sheet treasures. It might be better if they look at the RNA binding side of the TDP-43. But I agree totally with you both about the future of imaging. But I will also add that we need to incorporate some inflammatory markers because that so we can see the whole picture.

of the regional distribution of amyloid, tau, and inflammation in Alzheimer's disease and other neurodegenerative conditions.

Henryk Barthel (06:33.494)
So fantastic, thanks a lot. So Victor, if we now get these PET technology being much more sensitive and much better in terms of speed, so we can inject very low doses of tracers nowadays and do multiple scans maybe on the same day in the same patients, would you predict that we will?

Do multiple scans in the same patients like to identify different proteins or is that possible? Do you think that's it?

Victor Villemagne (07:11.442)
Especially if you use short-lived isotopes like carbon 11. But the other thing that we have to make sure when we use these digital new cameras is that the quantification is going to be different. And when we did a head-to-head comparison of PET cameras, PET-CDs, and against the digital camera, there was a 10% difference in the centiloy quantification in the same patient.

especially at the high levels of centroid. So if these cameras are used for patient selection or to evaluate treatment, we need to be careful with that too.

Henryk Barthel (07:55.222)
mean, Gil, the big advantage of PET imaging is that we get information on the topology, so the distribution of the pathology within the brain. And groups are thinking about potentially mixing tracers, like to get a mix of amyloid and tau, for instance, and to get both information one of our scan. What do you think about these ideas?

of like mixed pathology tracers.

Gil Rabinovici (08:29.681)
Yeah, absolutely. I mean, I think that the contributions of PET imaging to dementia care can't be overstated. You know, MRI was a huge revolution in neurology, but MRI is not able to tell us about some of the molecular changes in a very sensitive way, including these proteins that are driving late life cognitive decline. So amyloid plaques and tau tangles, which define Alzheimer's disease.

I mentioned alpha-synuclein, which of course defines Parkinson's and Lewy body disease. TDP-43, which Victor, which we mentioned seems to be a very important driver of late life amnestic disorders in particular, but also frontotemporal dementia, ALS. And then Victor mentioned the four repeat tauopathies, progressive supranuclear palsy, corticobasal degeneration.

provided us for the first time the ability to visualize and quantify these proteins in living people. And it completely changed how we study Alzheimer's in clinical studies, how we do clinical trials. It helped validate cerebrospinal fluid markers of amyloid beta and phosphorylated tau. And now has, I think without PET, the blood biomarkers that are so, I think we're all excited about even as imagers.

were excited about blood biomarkers because they make biomarker testing so much more accessible. These would not have been developed so rapidly without cross-validation with amyloid PET or TauPET as the standard of truth. And so the contributions really can't be overstated. Now the spatial distribution is interesting. Amyloid plaques early on aggregate in large areas of the neocortex. And...

There are some areas that they appear very early, including prefrontal cortex, posterior cingulate, and percuneus. But by and large, long before people develop symptoms, you can see widespread neocortical uptake of the tracer. And there doesn't seem to be a very good correlation between the location of plaques and the clinical symptoms. But it's a very different story with Taupet. With Taupet, there's actually a remarkably strong correlation

Gil Rabinovici (10:49.637)
between the symptoms and the severity of the symptoms and the distribution and intensity of PET uptake. So just to give you a couple of examples, when you look within patients with Alzheimer's disease at their performance on memory tasks, their performance on verbal memory tasks correlates with top head uptake in the left medial temporal lobe and with visual memory tasks in the right medial temporal lobe. So...

The location of the tau is very strongly associated with the symptoms. When people have strong language problems, that correlates with the amount of tau pet uptake in the left hemisphere in language areas. When people have trouble with visual spatial or visual perceptual tasks, that correlates with tau pet in occipital areas. And so there's a remarkable correlation between the symptoms.

and the location of Taupet and that spatial distribution can really help patients understand their symptoms and can actually, as we're seeing, predict symptoms that are yet to come. So, you know, brain areas where tau has spread are the areas where we're likely to see in the future neuronal dysfunction or neurodegeneration, and you might even be able to predict specific symptoms that are going to arise in individual patients.

based on where the tau is spread. And tau pet is a very strong prognostic marker. So it's very clear that Alzheimer's patients who have a lot of tau pet uptake are going to progress more rapidly than those who have less tau pet uptake. And so it really is quite a remarkable biomarker. And there's so much information there beyond just elevated or non-elevated in terms of the spatial pattern and in terms of the overall intensity of signal.

Victor Villemagne (12:37.542)
Thank you.

Henryk Barthel (12:41.696)
Yeah.

Victor Villemagne (12:42.31)
Yeah, I totally agree. That's why I think that pet imaging of tau will survive all the fluid biomarkers, because the regional distribution is so important. And then they predict the short-term cognitive decline. Might not be the same for amyloid imaging, because we didn't see much difference pattern.

Victor Villemagne (13:11.678)
autosomal Alzheimer's, but the patterns are always the same when we see with different traces and in Alzheimer's disease. So that what would be interesting now having PET and plasma biomarkers is to study those that have abnormal amyloid in plasma with a PET negative scan.

Henryk Barthel (13:37.383)
Mm-hmm. Right.

Victor Villemagne (13:40.846)
and follow them up because those are the fluids being a different biochemical pool of amyloid will change earlier. And plasma amyloid is not a very good target because less than 30% come from the brain and there's a huge overlap between controls and ideas. It might be a very good early predictor of amyloid.

Victor Villemagne (14:10.306)
when we look at the new therapies, the earlier they administer, probably the highly likelihood of success. Yeah. Of preventing, actually, preventing the disease.

Henryk Barthel (14:21.29)
Yeah, yeah, yeah.

Ken Herrmann (14:24.145)
Henrik, I started my nuclear medicine career 20 years ago. And since 20 years, I hear next big thing is dementia imaging, amyloid imaging. So first of all, my question is what

Henryk Barthel (14:33.418)
Yeah, it's the new, it's the new theranostics actually, I would say, combining amyloid pet with anti-amyloid, anti-plat, bodies can be considered theranostics. And, and that's what we all want to do, yeah, to do theranostics.

Ken Herrmann (14:47.97)
But my question is, why is it now happening after 20 years? And then when I listen to your experts, I hear so much tau. Should I even care about amyloid imaging or should I go straight to tau or?

Henryk Barthel (14:54.018)
Right.

Henryk Barthel (14:57.35)
Of course, of course, of course, yeah, yeah.

Victor Villemagne (14:59.898)
is a lot that has to do with for the for the last 50 years, we have some therapies that might work on amyloid. So that is what's making the huge difference. That's why even is approved by to buy Medicare now to cover the amyloid imaging study. So but

Gil Rabinovici (15:20.645)
Yeah, Ken, so I think, I mean, to get to your question about why so long, I think we have to have a little bit of perspective on how we were doing Alzheimer's clinical trials 20 years ago, which is that we were doing clinical trials in people with mild to moderate dementia, which means they probably had the biology of the disease for 15 or 20 years in the brain. We were diagnosing them based on clinical symptoms.

which means that probably 25% of the people in the trial didn't even have, for example, if you were testing an amyloid therapy, didn't have amyloid in their brain. And so this is just to give you an idea of where we were in the field. The diagnosis was made on clinical basis. It was made in late stage disease. And what PET has enabled us to do is to detect disease much earlier, in fact, in the preclinical stage. And that's where we think therapies are probably gonna be most impactful.

is in the 10 or 15 years when people have protanopathy in the brain, but are not yet expressing even early clinical symptoms, don't have irreversible neurodegeneration. And then finally, we had no idea really about target engagement. So when you tested an anti-amyloid therapy, you didn't actually know if the therapy was hitting the target and having a biological effect. And so amyloid PET,

And I want to give credit to Bill Klunk and Chet Mathis, who developed the first amyloid beta pet tracer Pittsburgh compound B or PIB at Victor's University, revolutionized the field by enabling us to detect this pathology in living people. And that transformed how we do clinical trials. So we could do clinical trials in earlier stages. We could select subjects for clinical trials based on their biology, not just their symptoms.

or symptoms plus biology, and we could determine whether the drugs were hitting the target and how they were modifying the disease. And PET, of course, it's not only PET. So as Victor brought up, there are fluid biomarkers which are measuring monomers of amyloid beta or phosphorylated tau. And so therapies that are directed at soluble forms of these pathologies, you need the fluid biomarkers to determine the target engagement.

Gil Rabinovici (17:42.749)
drugs that are now approved in the United States and Japan and soon probably in Europe and other parts of the world. These are monoclonal antibodies that are moving fibrillar plaques from the brain and to determine whether that is occurring, amyloid PET is the standard of truth. And so why should you care about amyloid PET versus Taupet? They're both very important. Amyloid PET changes earlier.

So there is certainly a significant proportion of people who have plaques but not tangles detectable by PET. And amyloid PET is again, the theranostic target as Henrik brought up. So if you are treating someone with an amyloid antibody and you wanna know if you're removing plaque from their brain, you need amyloid PET. Taupet tells you more information about disease staging, how much has the disease spread.

Victor Villemagne (18:31.939)
Yeah.

Victor Villemagne (18:36.547)
Yeah.

Gil Rabinovici (18:38.073)
which brain areas are affected, the spatial information is more important. And in current trials, what we're really trying to measure is whether removing plaques from the brain can slow the progression of tangles. Because for your viewers who are less familiar with the biology, the amyloid cascade hypothesis basically states that the plaques are promoting the spread of tangles from the medial temporal lobes into the neocortex.

And then it's really the spread of tangles that drives synaptic dysfunction, neurodegeneration, and ultimately clinical decline and dementia. And so the idea is that by removing the plaques, you can slow the spread of tangles. And if you can slow the spread of tangles, you can slow cognitive decline.

Victor Villemagne (19:20.783)
Yeah.

And you can see that clearly in the phase two of the AniMAP trial that there was no much changes in areas that started with high tau. But if you look at the frontal lobe, which is an area that has laid the position of tau, it almost blunted the accumulation of tau in that region. And the advantage of using amyloid is that we also know that if we go above a certain level of amyloid, we'll start having tau.

Henryk Barthel (19:43.966)
Yeah, I.

Victor Villemagne (19:54.314)
And so if we need to probably to prevent the disease from happening, start treatment before we reach those levels of amyloid in the brain.

Henryk Barthel (20:05.278)
Yeah, I fully agree. And this important discussion brings us also to this other topic of how should we in the future analyze our PET data? Is it sufficient to do a read, a binary read? Or do we need quantification? What kind of quantification? Should the tau images be analyzed in a binary manner? Or should we do tau staging?

Victor Villemagne (20:32.038)
No.

Henryk Barthel (20:34.136)
What is your take on that?

Victor Villemagne (20:37.058)
I would say that I would do both, because in an expert visual read, they tend to be more sensitive than quantification. Because for quantification, for example, if we use centilos, we use a huge area that might dilute the counts and you're below the pre-established or recorded threshold. And for tau, you need to for staging, because it's not the same to have

a lot of tau just in the mesial temporal lobe than having tau in the cortex. And how much tau do you have in the cortex? So and the new criteria proposed by Jack and others are using tau to stage the disease. Mesial temporal, moderate levels of cortical tau, and high levels of cortical.

Henryk Barthel (21:27.138)
This might even influence therapeutic decisions. Maybe it's a good idea to only apply anti-amyloid antibodies if the tower spread is not too much advanced. Maybe that's the best patient to treat.

Victor Villemagne (21:42.826)
I totally agree, but it's very difficult to convince people. I don't think that if you treat a patient with dementia, with an anti-amyloid, it will make any difference. If you look at the current trials, there's a difference in the slowing of disease progression. But they weren't looking at late MCIs and mild AIDS. If you start the therapies even before that,

And you see that in the Donanima phase two clearly, that the ones who respond are the ones with low doubt. So if it's true...

Henryk Barthel (22:19.882)
Yeah, yeah, I fully agree. I mean, yeah, I don't know what you think, but I would love to see trials preventing the occurrence of tau pathology in the brain. That should be the ultimate goal, I would say.

Gil Rabinovici (22:36.721)
Yeah, so there are clinical trials now that are looking at people who are asymptomatic, but have a positive amyloid biomarker. And then we'll have various levels of tau spread. Very few of them will have a lot of tau spread. Quite a few will have some tau in the medial temporal lobe and some won't show anything on the pets. It's important to recognize that the pet is, even though it's exquisitely sensitive, it's not picking up the earliest pathology. So we know that

The amyloid pet tracers are only picking up moderate to frequent density of amyloid neuritic plaques. And we know that the tau pet tracers are picking up the later stages of tau spread. So, BRK5, 6, potentially BRK4 on the sixth stage scale, but not the very earliest tau spread. But I think this is exactly where we need to go. So again, you know, to give Ken some perspective, right now we're...

stratifying patients in clinical trials based on things like their mini mental status exam score, which you know can fluctuate depending on what kind of good night's sleep they got, and whether they had coffee in the morning and all kinds of things like that have nothing to do with biology. And I think where we need to go, just like in any other disease, is to stage the disease on biomedical grounds. And that's where Taupet seems to be helping us understand.

OK, this person is in early clinical stage, but they're also in early biological stage of disease. And they may be more likely to benefit from therapy than someone who is in an advanced biological stage. I'm also a very strong proponent of quantification. I fully agree with Victor that the best interpretation of a PET scan, I think, relies on a combination of a visual read by a highly skilled clinician and quantification.

And the combination of the two will give you the most precise information. You know, maybe expert readers can do just as well or better on visual reads, but certainly as this starts to permeate nuclear medicine and as people with less experience reading these scans are trying to interpret them, I think they'll learn and be aided by quantification. And certainly when you're trying to determine whether your drug is working, like is it lowering clocks?

Victor Villemagne (24:59.674)
Yeah.

Gil Rabinovici (25:01.629)
You want that quantitative information and not just a qualitative assessment, you know, oh, you know, it looks like there's less amyloid in the frontal cortex. You actually want the numbers and the patients want the numbers too. So we're further ahead in amyloid PET. We have the scale that Victor has alluded to called the centyloid scale, which is, allows us to basically standardize amyloid PET measurements across different tracers.

Victor Villemagne (25:14.408)
Yeah.

Gil Rabinovici (25:28.117)
and imaging processing pipelines. And it gives us a good grounding. So on that scale, 0 centiloids is the average in young, healthy people who don't have amyloid. 100 centiloids is the average uptake that we see in people with mild to moderate dementia. And somewhere between 15 to 25 centiloids is a threshold where the scan is turning.

positive from negative to positive when you're starting to see above the noise into something that clearly indicates pathology and so, you know Having that scale is very important. Victor. Do you want to talk about your Centaur? Project because you've really pioneered the idea of trying to standardize tau pet measurements so we can do the same thing with tau pet

Victor Villemagne (26:12.751)
Bye bye.

Victor Villemagne (26:19.65)
We are doing that. We just got the second paper accepted for the center. Then they have to use.

Henryk Barthel (26:26.038)
Can you quickly explain what you're doing, Victor, to get these centaur values?

Victor Villemagne (26:33.506)
Well, we did the same thing for centeloids to create masks that are the subtraction of AD patients from controls. And then we subdivided those masks in a missal temporal, a metatemporal, a temporal parietal, and a frontal. So we can not only get a universal measure, but also get a regional measure of that, because as the disease progresses, it changes the topology. So we tried it.

Henryk Barthel (27:01.323)
sick.

Victor Villemagne (27:02.33)
The first try was to use set scores because that will allow you to have universal thresholds that are the same for any tracer and incorporates the noise of that tracer in the standard deviation. So we are looking into that.

Henryk Barthel (27:16.311)
Yeah.

Gil Rabinovici (27:17.037)
And I think the name Centaur is particularly inspired and it's a great indicator of you, Victor, because you are truly a Renaissance man and you could probably do a podcast on Greek mythology just as you can on Amaloid and Taupet. And so combining the two is really brilliant.

Ken Herrmann (27:34.378)
This is actually, Gero, this is a fantastic comment because we are hitting now the 30 million mark and I think we see that there's a lot of topics. We haven't even talked about dedicated brain scanners. We haven't talked about

Henryk Barthel (27:34.724)
Right.

Victor Villemagne (27:34.963)
Okay, thank you.

Henryk Barthel (27:45.962)
Yeah, we haven't talked about idea study, which is a nightmare. So we need more time. Yes.

Victor Villemagne (27:49.214)
We need to talk about the idea study. No, no.

Ken Herrmann (27:51.51)
So what we learned is that this is just the first one. And I think if the audience like what we did, then please request and we come back together in the same composition and hopefully touch upon all the topics we haven't touched upon today.

Victor Villemagne (28:09.398)
Should I be wearing the same shirt to make it look like it's a container?

Gil Rabinovici (28:11.869)
I'm sorry.

Henryk Barthel (28:12.498)
Hahaha!

Ken Herrmann (28:13.672)
Actually, this is a good idea. We can keep the same background so no one sees that we have actually moved.

Victor Villemagne (28:18.366)
Okay. I've been wearing this for the last month anyway, so.

Ken Herrmann (28:23.431)
So Hendrik, I would ask you kindly to close up this beautiful vodcast.

Henryk Barthel (28:27.562)
Yeah, thanks Ken. So thank you. Thankful, Victor and Gil. This was most interesting. And yeah, time was not enough. I have so many other questions which are unanswered today. So we need to repeat that if you have time. And if the audience likes that. Yeah, we will. Yeah, so I will click it 20 times.

Ken Herrmann (28:50.119)
If you reach 2000 views, I think then this is an invitation to continue.

Gil Rabinovici (28:56.029)
Hahaha

Victor Villemagne (28:56.953)
That will only be with my creditors. Don't worry, yes.

Henryk Barthel (29:01.112)
So thanks so much again Gil and Victor for your time and it's been a pleasure and an honor and you are the stars, you are the heroes in the field and thank you very much.

Victor Villemagne (29:11.31)
The honor was ours. Thank you very much.

Ken Herrmann (29:13.494)
Thank you, Hendrik, for compiling these really high level, highly decorated and a very fun gentleman for this beautiful podcast. Thank you very much.

Gil Rabinovici (29:13.617)
Thank you so much.

Henryk Barthel (29:23.31)
Thank you. Goodbye. Bye-bye.

Gil Rabinovici (29:24.221)
Bye everyone.