Transcript
MIC CAVAZZINI: Welcome to Pomegranate Health, a podcast about the culture of medicine. I’m Mic Cavazzini for the Royal Australasian College of Physicians. Australia has just joined America and the European states in approving a second amyloid-targeting therapy for patients with incipient Alzheimer’s dementia. But the limited clinical benefit of these drugs for quite substantial dollar costs concerns some in the field, while others question outright the importance of amyloidosis in the disease cascade.
Remember that amyloid-beta fibrils are secreted by neurons where they eventually mat together into insoluble lesions. These are the “senile plaques” that were associated with ageing and dementia by the great neurohistologists at the turn of the 20th century. The modern quiver of amyloid-targeting monoclonal antibodies are designed to recruit the brain’s microglia to phagocytose amyloid polymers and plaques. Aducanumab, developed by Eisai and Biogen was shown to cause significant reduction of amyloid over an 18 month trial in selected patients with mild cognitive impairment, MCI, or early dementia caused by Alzheimer’s. While this was heralded as the first real disease-modifying therapy, its listing with the US Food and Drugs Administration in 2021 was considered controversial because only patients receiving the highest doses showed any protection from cognitive decline.
In January 2023, the same firms received accelerated approval from the FDA for lecanemab after findings from a large phase 3 trial with the same design. In that study, the cognitive scores of patients receiving the amyloid-targeting therapy declined more slowly than those on placebo, by a few months on average. Lecanemab was, on the first pass, rejected by Australia’s Therapeutic Goods Administration because of safety concerns and lack of effect in a sub-set of AD patients with the infamous APOE4 mutation. But just as I was making the final edits on this story, the TGA announced that they and the sponsors had agreed to list the drug, with APOE4 homozygosity as a contraindication. That was only after Eli Lilly’s donanemab had been given the green light in May, which as we’ll hear in a minute yielded somewhat more impressive outcomes than lecanemab in trials. But even so, the protective effects from donanemab are modest, too modest for the Committee which determines whether to publicly subsidise drugs on the Pharmaceutical Benefits Schedule. The PBS continues to offer the decades-old symptom-alleviating drugs memantine and cholinesterase inhibitors, though these have a marginal effect in just a third of patients and another third cease taking them because of adverse events.
In today’s podcast we’ll weigh up the benefits and burdens of amyloid-targeting therapies as best we can at this stage. Not only is their cost very high but there’s a huge rigamarole around screening, delivery and monitoring of effect. Even before that is considered, Australia’s memory clinics are capable of diagnosing only about 5 per cent of all incident cases of MCI and early dementia every year. This capacity bottleneck was described in a recent piece for the Internal Medicine Journal co-authored by geriatrician Professor Michael Woodward in Melbourne. I also chased up the author of a helpful editorial published alongside the study, Dr Chrysanth Pulle in Brisbane.
Before bringing them in I should just do a bit of definitional housekeeping. In the 2015 DSM-5, the American Psychiatric Association rebadged MCI and dementia with the terms mild or major neurocognitive disorder to shift some of the focus from Alzheimer’s Disease, and even age as necessary preconditions for decline. But in the literature we’re discussing today, patients and trial participants must have tested positive for cerebral amyloidosis. The clinical criteria for mild cognitive impairment include deficits on neuropsych testing that are more than one standard deviation below age and education-matched comparators. And these can be on cognitive domains other than memory, including executive function, attention, language, and visuospatial skills. Finally, while these cognitive deficits may make instrumental tasks of daily living more difficult, they don’t remove a person’s independence. It’s only when they start to interfere with functional independence that a diagnosis of AD dementia is considered, confirmed by performance on neuropsychological tests that’s more than two standard deviations below appropriate norms. With all that out of the way, let’s hear from our experts.
CHRYS PULLE: My name's Chrys Pulle. I'm a geriatrician. I'm the clinical director of the Internal Medicine Dementia Research Unit here at the Prince Charles Hospital. I also am chair of Dementia Trials Australia, which is a network of clinical trial sites across Australia and New Zealand now, which has been involved in clinical trials for the past 25 years.
MICHAEL WOODWARD: And I'm Professor Michael Woodward. I'm the Director of Dementia Research at Austin Health. I've been working in dementia research now for close on 32 years and we've done about 200 trials on various therapies. I've also directed our memory clinic for best part of a quarter of a century, and I've got a strong still ongoing clinical practice in dementia, assessing a dozen or so patients per week in my rooms in private practice.
MIC CAVAZZINI: Fabulous, thank you. And thank you both for making the time. It does just occur to me maybe I should ask you to declare any interests in the industry that you might want to put on the table before we start.
CHRYS PULLE: So as a clinical trialist, I have participated in clinical trials sponsored by a lot of the pharmaceutical companies involved in Alzheimer's dementia research. So, I have been involved in clinical trials with Lilly, Novo Nordisk, Eisai, Biogen. So, most of the clinical trial specialists and I have been on advisory boards for them.
MICHAEL WOODWARD: Very much the same as Chrys. I've had some speaker’s honoraria also for various talks I've given on these issues from the companies that are making the monoclonal antibodies either in trial stage or currently marketed.
MIC CAVAZZINI: Excellent. So, to understand the promise and the limitations of amyloid-targeting therapies let’s start with the primary outcomes of the TRAILBLAZER and the CLARITY trials. TRAILBLAZER-ALZ 2, published late 2023 was an RCT on donanemab, in over 1800 patients living with MCI or AD with mild dementia. Over the 18-month trial period over three quarters of participants reached amyloid clearance on PET. Patients were assessed primarily using a well-validated tool known as the Clinical Dementia Rating Sum of Boxes and all patients had deteriorated, but those receiving donanemab had declined only 36 per cent as much as those on placebo. This made for dramatic headlines, but the absolute difference on the 18 point CDR-SB scale was only 0.67. Similarly, in the CLARITY-AD trial of lecanemab, published almost a year earlier, a 0.45 point difference between treatment groups was observed and reported as a protective effect of 27 percent. So after all that, can you as geriatricians tell us are these differences in scores functionally meaningful?
MICHAEL WOODWARD: Maybe I could start. Yes, look, a difference between, say, 0.5 and 1 on the memory subscale of the CDR is a difference between very mild forgetfulness and quite obvious forgetfulness, so that's a 0.5 difference in just the memory domain. Now this was Sum of Boxes rather than Averages, which means that it might be that there wasn't much change in memory—it could have been in other areas.
MIC CAVAZZINI: Chrys, you sent me that very helpful chart of the different domains. Do you want to skim over that to give the listeners something?
CHRYS PULLE: Yeah, sure. So, the CDR Sum of Boxes is used in clinical practice, but mainly in clinical trial settings. And it tries to ascertain a meaningful difference for these patients with early Alzheimer's disease or with mild cognitive impairment and essentially looks at the cognitive domains of memory, orientation, judgment, problem-solving; and then functional outcomes of community affairs, home and hobbies and personal cares. And they are graded between no major abnormality to severe, where none is zero and severe is three. Now, the Sum of Boxes actually totals up to 18 for severe dementia and zero for normal cognition. Now the issue has always been in trying to analyse the differences for these patients at these early stages and as Michael's pointed out earlier it has been the in these early stages.
MIC CAVAZZINI: Yeah, in a paper for the MJA last year Dr Balala and colleagues indicated that the better scores observed by the end of this trial would mean that a person with mild symptoms might have had their independence preserved, representing “about 50-75 hours of carer time over a period of 18 months.” So does that sound sort of plausible to you, the benefits of the...
MICHAEL WOODWARD: Absolutely, and it's not just carer time. I mean, the difference between forgetting constantly what we were told yesterday or where you put your glasses and only having very occasional lapses—and that's a point five change on the CDR memory scale—can be really significant in terms of your quality of life and your impact on the people around you. You know, to not have to tell dad all the time that there's a major event on tomorrow and you might have to tell him five times in a day, that's different to having to tell him just once or twice.
It's also, I think, important to note that it's not a linear decline. You don't go from 0 to 18 on the CDR Sum of Boxes down a straight line. And in the early stages, the decline is actually quite small. So, 0.67 in the early stages on this 18-point scale can be quite significant, whereas in the more severe stages, you might expect a decline of two or three points per year. The other thing I think we need to temper our understanding of this with is the fact that it’s very unlikely that we're going to see an improvement in a disease where the damage has already been done. So, I would say that on the basis of what we should be expecting that these are important clinical results.
MIC CAVAZZINI: By contrast to these encouraging responses to the data, there’s a 2024 metanalysis in the Annals of Family Medicine that looked not just at these trials but 24 studies of eight different amyloid-targeting agents where the authors concluded that; “In no case did the results of any single study, of all combined studies for an individual drug, or of all combined studies overall find a change in cognition or function that exceeded the minimal clinically important difference.” Now MCID has been defined as “the smallest difference in score in the domain of interest which patients perceive as beneficial and which would mandate…a change in patient's management.” And there’s a 2019 study suggesting that on the Clinical Dementia Rating Sum of Boxes you need to achieve a difference of 1 to 2 points to register as significant to a patient. Why are these commentators so much more conservative?
MICHAEL WOODWARD: Well, my take of this is that there's an assumption behind this that you actually want a patient to improve because that's what they will notice. People don't notice a slower rate of decline. They can't. They don't know what they would have been like otherwise, but they can certainly notice an improvement. And this idea that you've got to actually improve to be clinically noticeable or clinically important is incorrect. So, I again come back to the fact that the slower rate of decline, by a magnitude of 27 per cent and we'll turn that into time saved in a moment, that is significant but that's something we can measure in trials. It's not something that will necessarily be clinically noticeable by other people.
CHRYS PULLE: Yeah, and I would tend to agree with Michael here. And I think the other important thing in mild cognitive impairment or early Alzheimer's disease is that it’s an individual progression, not everyone deteriorates at the same rate. And so, there will be patients that do better on therapy than those and then it becomes an important individual clinician decision as to whether to continue with therapies or not.
MIC CAVAZZINI: So, you're saying that this concept of MCID—and there is other writing about how it's potentially controversial or subjective—you're saying it's not suited to measuring the what-if of how much they wouldn't have declined.
MICHAEL WOODWARD: My other comment on this, Mic, is that there's almost an assumption here that if a drug doesn't have dramatic effects in an older person, we forget about it. They're old, they've got a disease that's not very pleasant, let them stew in their juices. I think that negative ageism, almost, attitude, that we've got to expect almost miraculous cures or miraculous improvements in a chronic neurodegenerative disease is unacceptable. We can achieve that with heart failure, can achieve that with some cancers, we can achieve that with many other diseases but we're not likely to be able to turn this ship around. What we really want to do is slow down decline. We can't just ignore the fact that a quarter of a million new people are being diagnosed or developing MCI/ dementia each year. To say that, well, we'll just provide a few more care packages and grin at and bear it, we've got to make some headway.
MIC CAVAZZINI: As we’ve heard, it’s not easy to evaluate comparative protection from cognitive decline. Proponents of these therapies talk about ‘time saved’ as a more intuitive framing of their impact. That’s to say how much has disease progression been slowed by the intervention. For example, an analysis of the TRAILBLAZER data showed that cognitive decline at trial end for patients in the donanemab arm was 5.2 months behind those in the placebo arm. For patients starting with a low or medium load of tau protein, the time saved was seven and a half months. If you don’t recall your path lectures from med school, tangles of hyperphosphorylated tau fibrils are another hallmark of advanced Alzheimer’s disease that precede neuronal death.
I can recommend a review in the journal Alzheimer’s Dementia that elaborates on the concept of time saved, titled ‘Expectations and clinical meaningfulness of randomized controlled trials’. As the US authors put it, “A statistically significant difference seen after 6 months would be considered very important, but a statistically significant result of the same magnitude at the end of a hypothetical 60-month study may not be considered clinically meaningful.” They go on to describe a progression model for repeated measures which suggests that a 25 per cent slowing of disease would be an appropriate benchmark for clinical meaningfulness. The time saved accumulates with every visit, such that after a year and a half, the treatment group is 6 months behind the placebo group, while at 3 years, the saving has extended to 9 months.
This was certainly the view of esteemed US neurologist Professor Jeffrey Cummings in a recent review for the journal Neurotherapeutics. He wrote that the 18-month exposure in the trials of lecanemab “artificially truncates the potential long-term effects of therapy possible in real world settings.” Professor Cummings also pointed to modelling suggesting that in patients with early Alzheimer’s, treatment with lecanemab would delay progression to mild and moderate AD by two to three years. “The model also predicted a lower lifetime probability of admission to institutional care for those on therapy (25 % versus 31 %).” Professor Michael Woodward picks up from here.
MICHAEL WOODWARD: Absolutely. I mean, Jeff Cummings, who I know extremely well was quite right. The longer you use these treatments and probably the earlier you use them, the greater time you save, the greater time it takes for a person to reach the same degree of severity of symptoms than those who are on placebo or no treatment. So, time saved is a nice way to turn our understanding of the benefits of these treatments into a concept that the general population can understand. Just like with cancer therapies, we talk about months saved, we can talk about months saved with these monoclonal anti-amyloid therapies.
CHRYS PULLE: Yeah so, these trials were still only 18 months. So, we do need to wait in terms real-world data or post-marketing surveillance to see whether or not that, as Jeff Cummings mentioned, extends over a longer period of time.
MIC CAVAZZINI: When you consider years of independence and dignity retained, you can see why some patients would be prepared to fork out serious money for treatment. As of writing, donanemab, marketed at Kisunla, goes for about $56,000 for a year’s worth of supply, while lecanemab, also known as Leqembi, is pitched at closer to $40,000. You can also understand why the Pharmaceutical Benefits Advisory Committee in July rejected a submission from Eli Lilly to publicly fund donanemab.
It’s not just the cost of the drug that Australia’s Medicare would have to consider. There is also the organisational machinery needed to support delivery of monoclonal therapies. As already hinted at, the first eligibility criterion for treatment of patients with a clinical diagnosis is the presence of amyloid-beta in the brain. This is determined by use of positron emission tomography after infusion of amyloid tracers. Tracer uptake is then normalized and quantified on what’s known as the Centiloid scale. Centiloid values above 30 consistently identify the presence of neuritic amyloid plaques and above 40 centiloids you start to see deposition of neocortical tau protein. In clinical trials of amyloid-targeting therapies, clearance is deemed to occur when the centiloid score is brought down to between 20 and 30.
So amyloid-PET scans are required in the screening phase and then again after 12 or 18 months to determine the effectiveness of treatment. As Dr Pulle told me, each scan cost around 2000 dollars, give or take depending on setting. ($3600-5000 combined) You might also test for amyloid subtypes and tau from CSF samples though this is not essential. The eligibility criteria require genotyping for apolipoprotein E to exclude the e4 homozygotes who are less responsive to treatment and at greater risk of adverse events such as bleeding and brain swelling. That gene test costs only $150, but it’s no small detail. APOE4 homozygosity increases risk of developing Alzheimer’s Disease twelvefold over the wild type variant and this exclusion criterion reduces the field of eligible patients by a quarter.
Then, as we discussed in detail in the earlier episode, MRI scans are required to identify cerebral microbleeds. Any more than two of these at baseline and the patient is determined to be at too high a risk of further bleeds or other complications. For patients accepted onto the treatment regime, brain MRIs are repeated on four further occasions, costing $800 per scan. ($4000) There’s also the rigmarole around drug administration. Donanemab is infused intravenously once a month, and for lecanemab it’s every fortnight. So, there are all these day beds and staff required which cost around $500 per infusion ($6000py $9000 total). Throw in a couple of specialist consults and all told, you’re looking at up around 20 thousand dollars in addition to the drug costs mentioned earlier. [see also US data on the delivery of lecanemab over recent years].
It’s one thing to consider these costs on a case-by case basis but it takes on a different complexion when you look at overall burden of disease. There are an estimated 245 000 incident cases of MCI and dementia each year in Australia. Now, the narrow indications for these drugs select for no more than 8 per cent of these, which is 20,000 people. If we were able to treat all of these patients, and as we’ll hear in a minute that’s a big if, we’d be looking at a total annual outlay of 1.4 billion dollars at the undiscounted rate. This is two or three times what the PBS spends on any of the top 5 most costly drugs today. That includes COVID antivirals, cystic fibrosis combo therapy, aflibercept for macular degeneration or the cancer immunotherapies pembrolizumab and nivolumab.
And that cost is on a steeply rising trajectory. According to modelling by the Australian Institute of Health and Welfare, there are today, an estimated 440 000 Australians living with dementia and that prevalence will double by the year 2050. The only reason the increase isn’t steeper is that dementia has recently become the country’s leading cause of death. But on the other side of the ledger we have to consider what the cost to society is of dementia left untreated. The AIHW has calculated that in the 2021 financial year almost $3.7 billion was spent on health and residential care for people with dementia. I’m surprised that figure isn’t bigger, given a report from the UK’s Alzheimer’s Society which estimates that the cost to that country’s economy is 7 billion pounds in healthcare and 17 billion pounds in social and residential care. After adjusting for population, that’s twice what we spend per capita, and even more remarkable was their assessment that off the books was $21 billion pounds worth of unpaid care from families. These staggering numbers suggest that cost effectiveness could eventually be reached even for monoclonal therapies if costs come down, and more patients become eligible and responsive. I asked Professor Woodward and Dr Pulle for their read on this complex equation of dollars per quality adjusted life years.
MICHAEL WOODWARD: I think Mic you've made some very valid points. These are going to be expensive drugs however you measure them in but, nevertheless we've got here the major cause of death, we've got the most feared disease of older people, how much are we prepared to spend to keep quality of life from deteriorating as rapidly, to keep people out of residential care for a bit longer. I don't know that we can actually answer that question easily.
But I think one other concept though is that with donanemab the proposal is that in fact we check amyloid status at 6, 12 and 18 months and we know from the trials that one third of people by six months had achieved an amyloid centiloid value below 22 which is regarded as normal and therefore were switched over to placebo. Likewise at 12 months there was another third that's two-thirds in total and by 18 months about 84 per cent had achieved essentially amyloid negativity. So not everyone's going to be treated for 12 months and some will only be required requiring treatment for six months.
And we also need to be aware that on the horizon we've got drugs that are even more effective in removing amyloid. One drug trontinemab, still under phase 1 and soon moving into phase 2 and 3 studies, removes almost all amyloid in almost all patients after about 12 weeks. So, we are likely to have drugs costing less either because of less time of use or eventually, as we know with most other pharmaceutical areas, the cost of the drug will come down mainly as the pharmaceutical companies recoup their initial upfront costs in doing all these trials.
CHRYS PULLE: Yeah, I agree, Mic. These are good questions and one for, I think, for society and not just health economists to decide. And the other issue is, the dollars are high because the numbers are high with Alzheimer's Disease. I suppose the numbers of Alzheimer's disease is higher than prostate and breast cancer combined. And I don’t think there's a there's ever a question of whether we should treat patients with prostate or breast cancer. So, then it boils down to a societal issue of, you know, is this a subcurrent of ageism? There is an onset of younger onset Alzheimer's disease patients but they're still only about five per cent. So, I do see 50 year olds with established Alzheimer's pathology who would like to start treatment but for a variety of reasons are not yet eligible.
MIC CAVAZZINI: Yeah, when the benefits of these drugs are framed as time saved, which is reminiscent of the way we look at modern cancer therapies—you know, I found an analysis published last year showing that for PBS-listed publicly funded drugs against solid tumors, the median overall survival gain was just 4.1 months and some of these cost as much as the donenemab we’ve described. But in order to calculate the cost per QALY there are all sorts of assumptions you have to make; what screening modalities are going to be used, how sensitive are they, how many patients does that filter in or out, how early do you start treating. And I couldn't find a good review of this. One for the earlier-listed drug, lecanemab, showed that at the market price it translates to $250,000 US dollars per QALY gained. How much do you think Eli Lilly would drop the price before PBAC gives it the tick? Remembering that our public funder has an informal bar of about 70,000 Australian dollars per QALY.
MICHAEL WOODWARD: It's hard to say because it has international consequences and also the restrictions on use. I mean they're not entirely happy that E4 homozygotes have been excluded. So, there are a number of factors there. I would have thought though that if at some stage donanemab can be got down to $10,000 a year, there'd be a lot more take, but I'm sure that's not going to be very palatable to Lily. Or indeed to Trump but that's another story. But we need to perhaps drill down and look at where these drugs might be much more cost effective. We know from the lecanemab data that if it was used early enough in the disease process that the actual rate of decline of the patient is markedly reduced.
MIC CAVAZZINI: I found a review that suggested that brand name originator drugs in Australia do cost a fifth of the US, which surprised me. I thought maybe they'll drop it by 25 per cent or something. I guess when the PBAC holds off and plays a hard bargain, even though there might be lobby groups champing at the bit, eventually that does produce results for the taxpayer-funded system.
MICHAEL WOODWARD: Mic, I think what's happening actually is PBAC are waiting for lecanemab to get TGA approval, then they've got two drugs that they can consider and to some degree play off against each other. And they will eventually have other drugs, as we've mentioned trontinemab, the Roche drug, and there's others in the category that are being developed. So, I think there's a little bit of strategic play going on with PBAC, as there has been with other drugs in the past. They've waited for more than one drug in the same class to be considered. But maybe that's my cynical point of view. What do you think, Chrys?
CHRYS PULLE: I have not the experience that you have with the PBAC. We talked about other drugs, so you know, the EVOKE trials will report top line results in December. And that's looking at semaglutide in Alzheimer's disease. So, watch this space.
MIC CAVAZZINI: As mentioned at the start, that TGA approval for lecanemab did come through right after recording this interview and we’ll have to wait and see how this influences pricing. But one my reviewers shared this shrewd reading of the intentions of the Pharmaceutical Benefits Advisory Committee when it turned down funding for donanemab. He’s an endocrinologist, so it was with interest that he observed that at the same PBAC meeting in July, they also knocked back the GLP-1 agonist tirzepatide for the management of type 2 diabetes. Not a good day at the office for Eli Lilly, but notably, the decision on tirzepatide was “deferred” until a better price could be negotiated. There’s no real doubt now that tirzepatide is superior for glycaemic control and weight loss than the blockbuster drug semaglutide.
By contrast donanemab was marked as “not recommended,” and given this blunt evaluation; “The PBAC concluded that the high burden of treatment on both patients and the health system, combined with the risks and modest clinical impact, makes the drug unsuitable for PBS subsidy”. Reading between the lines, one might infer that even at a lower price, donenemab won’t be given a second chance unless more compelling patient outcomes can be demonstrated.
With all this talk of dollar costs, there’s one important detail missing. Can we even service all the patients that would be eligible for amyloid-targeting therapy in Australia? I’ve not mentioned Aotearoa-New Zealand today, because the sponsors have not yet sought registration in that country. The Australian capacity question that was addressed by Professor Woodward and colleagues in the July edition of the Internal Medicine Journal. The article’s first author was Johannes Michaelian and the last author was Professor Sharon Naismith, both with the Healthy Brain Ageing Program at Sydney Uni’s Brain and Mind Centre. The backdrop comes from an earlier review through the Australian Dementia Network which found that the average memory clinic could assess just over 5 new people each week, totalling about 12,000 a year across all services in the country. Most don’t actually end up with a diagnosis of MCI or early dementia, but even if they did, that would be just 5 percent of all incident cases in a year.
This bottleneck occurs even before considering whether existing clinics are capable of screening for eligibility for amyloid-targeting therapies. In the recent paper, medical specialists from 30 different memory clinics around the country, public and private, were interviewed to identify how prepared they were to implement the treatment guidelines. For example, only 40 per cent of clinics were testing for AD markers at all. Lumbar puncture for CSF sampling was rarely done. Likewise, APOE genotyping. Even MRI was performed in only 60 per cent of patients. I asked Professor Woodward whether that was because until now they haven’t been critical to deciding on management.
MICHAEL WOODWARD: I think it's a number of factors that are contributing to this low rate of biomarker testing that you're referring to. Certainly, for those centres outside of capital cities, it's more difficult to access some of these. Secondly, there is a cost of these to the health system, if not to the patient directly. Thirdly, some clinicians rightly or wrongly feel that a clinical diagnosis is accurate. In fact, it's not. It's only accurate in about 60 to 65 per cent of cases, even in the best memory clinic so, we absolutely need to use more biomarkers.
I can't tell you how many times, Mic, I've been sure that a person has Alzheimer's disease on their phenotypic, their clinical presentation, and I do an amyloid-PET on most of my patients, and a very substantial number of these people do not have amyloid. They might have other mimics of Alzheimer's and that's another whole question. Will these monoclonal antibodies work where a patient has a mixture of Alzheimer's pathology and another disease as the primary driver? That's a whole further discussion. But we can't rely on our clinical acumen, even in memory clinics, to make an accurate diagnosis. And these monoclonal antibodies are focusing our attention on the need for an accurate diagnosis of the underlying pathology.
CHRYS PULLE: I would agree, Mic, that in the past that there hasn't been therapies available and so the clinicians, especially away from the clinical trial sites or the major centres in metropolitan cities, said there's no need for these modalities. But what I'm noticing is a lot of patients and the carers are now demanding this and I think that's what's been apparent. In Australia, CSF in public settings and also even in clinical trial sites outside of the biggest centres in Melbourne and Sydney, patients have been reluctant for CSF and would rather have blood biomarkers or amyloid PET scans.
But short of funding appropriate clinicians, you're not able to see the number of patients at an earlier stage where the benefits of this treatments are available. Memory clinics themselves are not able to diagnose and manage all the patients coming through and there needs to be an overall health strategy, you know a brain health review when someone turns 50 as part of a screening process. And if anything, these newer therapies being available has shown the public that there are potential treatments if patients are found earlier.
MIC CAVAZZINI: And you might have both touched on this already, but Chrys, in your editorial with Dr Maher, you alluded to some imminent developments in biomarker screening and drug administration that may lower the barriers and costs. Do you want to add?
CHRYS PULLE: There are significant cost issues. As the biomarkers become more available, more reasonable priced—there’s an interesting blood test, for example P-tau 217 is very promising—this could be a screener almost. And then if it's positive proceeding to an amyloid PET or CSF. I think as potentially other newer treatments become more available, subcutaneous versions—although at the moment the subcutaneous version of lecanemab is almost as expensive, I think, as the intravenous infusion—the costs may change with time.
MICHAEL WOODWARD: The blood biomarker, Mic, is looking really promising. If we look at phospho-tau 217, as Chrys has just been talking about, the AUC, the area under the curve for diagnostic accuracy, is about 0.9, in fact a little bit higher. And it obviously depends on your pre-test probability of the disease. This is not rolling out to the general population, this is in a population of people who have significant concerns about cognition or where the doctor has detected cognitive decline. But that's a very accurate test, accurate for determining the presence of amyloid in the brain which has to be confirmed at this stage by an amyloid-PET or CSF. So, it's likely that we'll have much better biomarkers and much more accessible and cheaper biomarkers and that's why we're testing these in general practice right now. There are three general practices in South-Eastern and South-Central Australia where we're testing these blood-based biomarkers in general practice.
MIC CAVAZZINI: In your paper, Michael, it was estimated that public memory clinics would currently only be able to treat 5 per cent of the 20,000 new patients every year who have the right indications for amyloid-targeting. And that raises the question about equity of access. As described in a piece for JAMA, patients with less socioeconomic advantage typically get diagnosed at later stages of disease and have to travel further to get to memory clinics or imaging centres —that’s, of course, the case in Australia as well. They’re less able to advocate for themselves to get onto drug trials and more likely to express comorbidities that exclude them from these therapies. So, these social and moral questions are nothing new in healthcare but they do become more stark when we're talking about very expensive treatments. And then even aside from the dollar cost of these drugs, there's also the opportunity cost of giving up those beds to deliver therapies that currently have marginal benefit. Do we have some sort of algorithm to say whether those resources should be invested here or perhaps towards patients with multiple sclerosis who say would be younger and more able to contribute “productively” once treated.
MICHAEL WOODWARD: Again, that sort of borders onto ageism and what's the value of an 80-year-old deteriorating more slowly? I think an 80-year-old who can get to their granddaughter's wedding or their wife’s 90th birthday party is, I think, as important as being able to contribute through the workplace if you're much younger. But that's a social judgment. I don't think we should undervalue quality of life at 80 and overvalue quality of life at 20. That's my personal opinion as a geriatrician.
MIC CAVAZZINI: I mean, when spending public funds, we know that the payback is evaluating cost and contribution when they evaluate QALYs. Is that a necessary evil?
MICHAEL WOODWARD: Well, remember Mic that these people in their 80s or so have also spent between the ages of 20 and 60 a life of working. Do we undervalue that? It's only what they're currently able to do and not what they have done that matters. This is a very moral question. To come back to your comment about access, you're right we can't accept that only 5 per cent of the 20,000 new patients every year will get to be properly assessed in a memory clinic. I think we really need to look at other ways of assessing people and that's what a group of us called the Brain Health Collective have been trying to do and lobbying the Australian government and the Health Department to say there are other ways that we can assess people with cognitive decline. We need to upskill GPs. We need to make specialist services much more accessible for people in rural and regional areas. So, we do need to rejig the whole system anyway even without consideration of monoclonal antibodies. But we particularly do, where we need an accurate diagnosis for people to begin these medications. Maybe Chrys has got some views on this.
CHRYS PULLE: Yeah, I think it is a vexed question. The editorial that we that I did with Sean Maher also looked at these focuses as the concern would be that if the finances are focused towards these therapies, does that take away from the 75 per cent of patients with Alzheimer's disease that don't meet the criteria. You know, it's a vexed question, but I still think that patients should have that choice.
MIC CAVAZZINI: We talked about that Lancet Commission report from 2020 suggesting that 40 per cent of all cases of dementia might be preventable. They highlighted 12 risk factors, some of which, are sadly predictable, obesity, hypertension, smoking and excess alcohol. Traumatic brain injury contributes 3 per cent, depression another 4 per cent. But at the top, you've got things like poor education, 7 per cent, and unaddressed hearing loss at 8 per cent These would be eminently easy and not very costly to deal with and yet we don't. Maybe we think of those as too distal from the problem of dementia. But even when you can draw those links there isn't enough impetus to trigger that funding and investment.
MICHAEL WOODWARD: Mic, there's now a 2024 version of that Lancet Commission. They've increased it up to 14 risk factors. They've added visual and cholesterol risk factors. And you're right, many of them are modifiable. In fact, a recent study showed that for those who do have impaired hearing, using hearing aids regularly as opposed to not using hearing aids, did actually reduce the risk of dementia. It's only one study, but it does show that not only are we identifying risk factors, but we can actually attack these risk factors and reduce the risk of dementia at a population level—there's no guarantee it's going to do so at a personal level. So, this is the background. This is what we should be pouring our money into. And we have online and other programs that have been developed in Australia, but sadly, some people will have their brain accumulating amyloid and we know that amyloid causes, eventually, Alzheimer's disease phenotypically in most patients, given enough time.
[MIC CAVAZZINI: To go back to drugs and what we can pay for, in a review for Australian Prescriber magazine, Geriatrics Professor, Louise Waite noted that we've been paying for drugs like memantine and the cholinesterase inhibitors (donezepil, rivastigmine, galantamine) for decades. In combination, they may delay transition into residential care by a little bit, but they’re not particularly effective. They’re just very cheap, ranging from $20 to $65 for a box of tablets.
Prescription of these drugs is restricted to clinicians with special authority and must only be continued if there’s a measurable response within six months. But a 2012 post-market review of the program showed that 60-80 per cent of scripts for cholinesterases and memantine are continued, which is problematic as that would represent a response rate more than twice what had been reported in trials. And the authors suggest that prescribers are “likely to be interpreting improvement in symptoms more subjectively or are applying… [cognition] scales with some degree of flexibility…it is likely that over half of these prescriptions are not cost effective as originally assessed by PBAC.” What would you say about the responsibility of clinicians in this space and can they be trusted with the more costly anti-amyloid therapies?
MICHAEL WOODWARD: So, Mic, the problem there is that the endpoints, or the requirement, I should say of the PBAC was an improvement. You had to improve two points on the MMSE or four points on the ADAS cog. And again, these drugs are unlikely to cause improvement. So, many clinicians are making an assessment about whether their patient would have been worse off without the drug, which does not mimic what the PBAC initially required or indeed what were the endpoints in these clinical trials. So, I think this highlights the difficulty of making an assessment of response without biomarkers in a chronic neurological progressive disease. If we could more accurately measure response biologically, such as a removal of amyloid to below 22 centiloids, which is what's proposed for these drugs, and then stop therapy or go down to some lower maintenance therapy we'd be probably more likely to align with our decisions about ongoing treatment with what's actually happening with the patient.]
MIC CAVAZZINI: We won’t go into the adverse events associated with amyloid-targeting therapies in too much detail because we already talked about those in the recent [IMJ On-Air] episode about cerebral microbleeds. Amyloid-related imaging abnormalities, known as ARIAS, include oedema and haemosiderosis and in the TRAILBLAZER trial they were detected in more than a third of patients receiving donanemab. A quarter of those with oedema had noticeable symptoms that included headache, confusion, visual impairments and gait-disturbance. These usually resolved within 4 months but could sometimes could be quite serious, such as stroke-like episodes and seizures. The treatment discontinuation rate was 13 per cent, almost twice what it was in the CLARITY trial for lecanemab. And there were three treatment-related deaths associated with donanemab only.
One odd side-effect that accompanies successful clearance of amyloid is a reduction in brain volume. This was established in a meta-analysis of MRI data published in 2023 in the journal Neurology. An accompanying editorial speculates whether this is caused by what’s been labelled ‘pseudoatrophy’ or perhaps neurodegeneration. This has fuelled one of the many sceptical takes I have come across about the outcomes of these trials. In a 2024 review for the journal Ageing Research Reviews, thirteen authors spread around the globe argued that decreases in amyloid-PET signal could be related not just to removal of the protein but also therapy-related cell death. They wrote that the specificity of the various tracers is poorly understood and that some are taken up also during inflammatory processes other than Alzheimer’s disease.
I ran this concern past a friendly nuke med registrar and he told me he did put faith in the very recently updated Appropriate Use Criteria for amyloid PET published in the Journal of Nuclear Medicine. The consensus seems to be that while some details remain to be ironed out, the technique is relatively well standardized. The paper links to a few different validation studies, using different amyloid tracers, that demonstrate good congruence between centiloid scores or other semi-quantitative measures on PET, and the beta-amyloid load as determined by post-mortem histopathology.
The critical reviewers also took issue with the reported statistical interpretation of the outcomes from the TRAILBLAZER trial. Their own paper contains a Bayesian Factor analysis, don’t ask me to explain which suggests that there wasn’t enough of a difference in outcomes between the two arms to discriminate clinical efficacy of the treatment. The authors added that the effect on cognitive scores was smaller even than that seen with memantine and cholinesterase inhibitor therapy. I checked this with a couple of neurologists and trialists we’ve had on the podcast recently, and they batted away the suggestion that the guidelines from SPIRIT-CONSORT and the trial registries could have been flaunted. Or that you could compare in the same breath the effect of symptom-alleviating treatments from that of diseases-modifying ones.
But there’s the rub. For sceptics of the research, the fact that almost complete removal of amyloid plaques only has marginal effects on symptom progression, is evidence that the amyloid cascade hypothesis is inadequate to explain Alzheimer’s Disease dementia. But it’s a hypothesis that’s more grimly held onto than almost any other in biomedicine. A reviewer of mine described as “the original sin” the description in 1909 of neuritic plaques and tangles as preconditions for amnestic disease. It was, Leipzig psychiatrist Emil Kraepelin who named the disease after Alois Alzheimer, and had him publish his neurohistological findings, drawn from a single patient only.
Only with the explosion of omic techniques in the early 1980s was the amyloid composition of senile plaques determined, as was a link between some cases of familial AD and the precursor protein from which amyloid beta is cleaved. Located as it is on chromosome 21, this may also explain the early onset of dementia in some people with Down syndrome. But in the decades since, proof that deposition of amyloid-beta is a driver of the disease cascade has been elusive. Animal models of the pathology rarely play ball and a couple of dozen monoclonal antibodies had been piloted and abandoned before the meagre success of the aducanumab trials.
In recent years there have even been scandalous claims of doctored Western blots and immunochemistry images in a couple of foundational bodies of work from twenty years ago. This has bred feelings from some in the community that a generation of research, and hundreds of millions of dollars, had been wasted on some kind of Kool Aid; on a mere epiphenomenon of disease, or cellular “tombstone”. According to the most heated critics of the amyloid hypothesis, this recent flurry of approvals for monoclonal therapies is just the latest examples of confirmation bias and sunk cost fallacy, amplified by the weight of expectation from millions of patients and their families.
It’s not the intention of this podcast to adjudicate this tortuous debate but simply to take the recent approvals by the TGA at face value and reflect on what they mean for practice. I will say, that among all the people I spoke to, there was general agreement that one couldn’t expect miracles from first generation treatments that act on only one aspect of a multifactorial disease process. There are many details still to work out, chiefly how early in the molecular cascade can amyloid-beta be targeted and which of the accompanying protagonists need to be considered to get the best results.
The many mAbs that have been trialed against amyloid have targeted different epitopes. The donanemab antibody is most specific for amyloid plaques that form at the later stages of disease. Lecanemab, meanwhile, preferentially targets fibrils and protofibrils, which is why it needs to be administered continuously while donanemab should be stopped at plaque removal. Other monclonals are more specific for soluble monomers of amyloid-beta. With a better understanding of amyloid pathology it’s conceivable that prophylactic treatment could be started long before plaques even develop and clinical signs of cognitive impairment are detected. Here’s Dr Chrys Pulle again.
MICHAEL WOODWARD: In terms of staging Alzheimer's disease, absolutely. We've got the pre-clinical stage where amyloid is accumulating. Most people have some tau in their medial temporal lobes but it stays there. But then as amyloid accumulates, usually in other parts of the brain, it sets off a tau cascade. That's the crucial point, that's when we start to move to the next stage, the prodromal or mild cognitive impairment stage, where we start to get symptoms. And then that moves on to the dementia stage when tau and amyloid are wreaking havoc in the brain.
So, if we could target people in the earlier stages before the amyloid trigger sets off the tau bullet, with tau moving in whatever way from the medial temporal lobe to the rest of the brain, we may be able to get much more clinical effectiveness from amyloid targeting drugs and even using tau-targeting drugs at a very early stage might be clinically beneficial.
Theoretically the earlier the better. And there are studies now such as the AHEAD study and there's soon going to be a study with donanemab in people in the pre-clinical stage. In other words, amyloid accumulation, minimal tau, no symptoms. That could be where our biggest bang for the buck is. Of course, though, we're going to extend the treatment target population extensively so we need drugs that are affordable. And we also do need drugs that are relatively safe if we're going to be exposing a large proportion of the population to these therapies.
MIC CAVAZZINI: You suggested the idea that these would almost a prophylactic treatment. But there was a drug Eli Lilly developed called solanezumab, which failed in the trials, so to speak, because it went for the more soluble forms of misfolded amyloid beta rather than the fibrils or the plaques. And those misfolded amyloid betas precede pathology by 20 years. So yes, we could imagine treating people for 20 years before likely onset of symptoms. I mean, would that be for the whole population once they reach a certain age? Are there cost-effective screens that could be rolled out at scale to identify those most at risk and worthy of treating with lecanemab, say?
MICHAEL WOODWARD: So, at the moment, Mic, we don't recommend general screening simply because we don't have those treatments. But we do certainly need treatments that can be used at an earlier stage. We can find a population of people with a high risk of Alzheimer's disease, perhaps because of their own individual risk profile, perhaps because of a strong family history, perhaps because of subjective cognitive decline but no objective findings. These people are at a greater risk. They could be screened maybe in their 50s or 60s and if they're accumulating amyloid we need to do something to remove that amyloid before it sets off the tauopathy, the tau cascade or cataustrophe is word I like to use. If we can prevent the cataustrophe at an early stage, we may prevent them getting the symptoms of Alzheimer's ever. We know that once removed from the brain, the amyloid plaques only accumulate at about 2.6 centiloids per year. So, clearing amyloid at an early stage might put back the onset of symptoms by decades, theoretically.
MIC CAVAZZINI: Okay, but yeah, there are already strategic ways to identify high-risk patients, you're not providing every single person in the population with this drug. And I mean, Chrys, you pointed out in your editorial with Dr Maher that a 65-year-old man positive for amyloid beta has only 1.7 times the risk of someone with no amyloid beta deposits. So even with screening, there's some risk of overtreatment. I guess more efficient screening and targeting therapies will also require a better understanding of the other biomarkers and molecules in this cascade. We've mentioned apolipoprotein E in passing, and then of course there’s tau. And in the TRAILBLAZER trial it was found that after stratifying for tau load, patients with low or medium tau levels had a better response to donanemab. What more do we need to know to better use these drugs?
MICHAEL WOODWARD: I just wanted to pick up on two points if I can, Mic. One is, you know, as we start using drugs, we still better understand mechanisms. We don't want to wait until we have a perfect understanding of mechanisms before we start drugs. Look at breast cancer therapy fifty or sixty years ago where we had the taxol groups of drugs, which were a major breakthrough, but that was well before, I believe, we understood about the triple negative breast cancers that we understand now. So, we got a much better understanding of what was going on with breast cancer over time, but that didn't stop us using drugs in a sort of somewhat more blunderbuss way early on and they were still a major breakthrough. The same will happen with Alzheimer's but if we keep saying no until we've got the perfect drug combination with massive savings of dollars or improvements in quality of life, we might be still waiting another 10 or 20 or 30 years before we have any drugs of significance available for Alzheimer's disease.
You're right, we do need to better target those who are going to go on to develop Alzheimer's phenotypically and as Chrys has said in that editorial with Sean Maher, not everybody who's amyloid positive will go on to develop phenotypic Alzheimer's. That's a really interesting area. Some of these people seem to be making monoclonal antibodies against their amyloid and in fact it was from those people that we isolated aducanumab. It was from one such individual who had slowly-progressing Alzheimer's. We're now actually also understanding there are processes which stop the tau spread. There are certain genes that make proteins that seem to dampen down the spread of tau. So, we might want to only target those people who don't have natural protective factors that would slow down their progression from early amyloid positivity to phenotypical Alzheimer's disease. That might explain why there is only this 1.7 times greater risk because that's looking at everybody with amyloid but not everybody with amyloid has the same degree of protection from amyloid.
MIC CAVAZZINI: That's interesting. So you could imagine—future drugs could be helping the patient's own immune system better respond, or there's another company that's investigating an mRNA, interfering mRNA for the early amyloid precursor protein. So that could be cheaper and more tolerable than the hard core monoclonals we're talking about now.
MICHAEL WOODWARD: Absolutely, there's many other approaches, antisense oligomers, ASOs are also being looked at and we haven't given up yet on the enzymes that actually cause the cleavage of the toxic A beta from its precursor proteins. So, gamma-secretase modulators and even beta-secretase inhibitors have still not completely died as possible treatment options. There's a lot happening There are other processes such as tau and neuroinflammation and that's probably where the future is taking us. We need to target those as well, just like with cancer, we use multiple drugs in most patients, not just relying on a single underlying pathology to target. We can do the same with Alzheimer's disease.
MIC CAVAZZINI: I think you've said everything that needs to be said. mean, I was going to end with a long, complicated question about the three decades of the amyloid hypothesis. It's had a rocky time but, Chrys, is this the end of the beginning rather than the beginning of the end?
CHRYS PULLE: Yes. Hopefully it's the end of the beginning. I think we still have a long way to go in terms of fully understanding but these drugs are a start. You could argue that with the TRAILBLAZER-V, with the low tau patients do better than the high tau patients, because they're progressed in their disease. And we've seen that for a number of these trials that it's early onset, I think that's the take home message, early onset of symptoms, not putting that depression or that ageing as the cause of someone's cognitive decline, but being seen and diagnosed early and then hence with the use of biomarkers to appropriately target those patients.
And I think the other key point—you know, I'm a clinician. I can think nothing worse than, I see a lot of my patients come to me, they have done everything they can in terms of exercise, diet, know, risk factor reduction, and still they say, “I have Alzheimer's disease, what can you do for me, doctor?” And it's very hard to say to that patient, “There's nothing else more that we can offer you. The trials here do not show sufficient enough benefit as per the funders". So, I think the main key take home message that I would like to say is that patients deserve that choice. And I think as long as they're able to be aware of the risks and the potential benefits, then they need to be given that choice.
MIC CAVAZZINI: Many thanks to Dr Chrysanth Pulle and Professor Michael Woodward for contributing their time and expertise to this episode of Pomegranate Health. The views expressed are their own and may not represent those of the RACP. If you want to follow up on any of the facts and figures discussed there’s a transcript packed full of links to the academic literature. Go to racp.edu.au/podcast, click on the episode, then the dropdown tabs labelled ‘Transcript’ and ‘Further Resources.’
I have to thank all the people who gave me feedback during the development of this podcast. Neurology Profs Bruce Campbell, Mike Parsons and Amy Brodtmann and geriatrician Ronaldo Piovezan all gave me valuable perspectives on the clinical trial process. Nuclear medicine docs Joseph Lee, Jamie Bellinge and Karan Singh helped me understand some of the caveats around amyloid PET imaging. And as always, the physician on the podcast editorial group chimed in on early drafts of the podcast to help tighten it up. They were Aidan Tan, Hugh Murray, Rahul Barmanray, Simeon Wong and Sebastian Lambooy.
If you’ve been suitably infotained please tell your friends to subscribe via their favourite podcasting app or the mailing list at our website. And send any of your own thoughts about this or future stories to the address podcast@racp.edu.au. Pomegranate Health is produced on the lands of the Gadigal clans of the Yura nation. I pay respect to their elders past and present, and indeed all wise ones in our community who may not get the recognition they deserve. I’m Mic Cavazzini. Thank you for listening.