The $10 billion unicorn in INmune Bio: the road from amyloid-inhibition to GLP-1 agonists, part I

Aloïs Alzheimer, 1915

Summary

Amyloid aggregates and phosphorylated tau were the traditional hallmarks of Alzheimer’s disease. Efforts focused on them failed to yield results for decades, until recently, though some contest it is actually the removing of amyloid that leads to minor beneficial results.

I sketch the history of drugs targeting amyloid below. The recent successes do not translate into strong sales numbers, and rightfully so in my opinion.

Then I address one point that I consider crucial: at least in preclinical models, even with amyloid, there is much less neurodegeneration if one removes the microglia, which could imply glial cells are much more essential for neurodegeneration than amyloid.

This is the first part of two blog posts. The second one will address the potential of GLP-1 agonists for the treatment of Alzheimer’s disease.

The history of amyloid in Alzheimer’s disease

Introduction

One can write a book on the history of amyloid in Alzheimer’s disease. I am keeping it short, and am adding a little word on tau as well.

As discussed before, XPro’s biomarkers were far more impressive than those of anti-amyloid therapies and other drug candidates, and could allow stabilization or improvement in cognition in my eyes. Amyloid plaques and phosphorylated tau tangles are the traditional hallmarks of Alzheimer’s.

P-tau

As for phosphorylated tau, the Phase 1 yielded impressive results indicating lowering of ptau following treatment of XPro. I find that important because p-tau in Alzheimer’s is not just considered toxic, it is also the result of structural decay inside neurons. In fact, tau as such is a structural element inside the cell: a protein that stabilizes microtubules in neurons. When tau phosphorylates, it often misfolds and detaches from microtubules. This is where loss of function occurs. Hyperphosphorylated tau starts to self-aggregate into soluble oligomers, which are considered toxic. These oligomers eventually form insoluble fibrils and neurofibrillary tangles, the so-called NFTs.

Amyloid plaques

For sake of simplicity, when I refer to amyloid here, I do not refer to the functional ubiquitous amyloid protein, but rather to amyloid aggregates which are considered toxic in Alzheimer’s disease. Writing the history of amyloid is not the goal of this blog, but the subject should be addressed, already because it is the primary hallmark of Alzheimer’s, because INmune did not report amyloid reduction in the Phase 1, but expects its levels to go down in the phase 2 study.

My personal opinion on it is that whether or not amyloid goes down does not matter much.

For that, digging a bit deeper into the amyloid hypothesis on Alzheimer’s and its failures helps. Ironically, not the failures but the modest successes are finally leading to that hypothesis being abandoned, at least insofar as some have considered that reducing amyloid could solve the disease entirely.

The amyloid hypothesis

Timeline

Reduced to its essence, the amyloid hypothesis of Alzheimer’s alleges that about 15-20 years before the onset of symptoms, amyloid peptide begins to aggregate, and when it aggregates, it causes damage  to the other cells of the brain, ultimately leading to the loss of neurons.  That’s supposedly where the dementia comes from. As such, that timeline does not make much sense. The onset of amyloid is just too distant from the cognitive impairment seen in Alzheimer’s.

Neither does the fact that about 30% of cognitively unimpaired individuals have high levels of amyloid in their brains (and that’s a high number). Nor does the fact that many cognitively impaired (dementing) people do not have high levels of amyloid in their brains.

How amyloid would be toxic

Amyloid is, among other things, considered to disrupt synaptic plasticity, interfere with glutamate receptors, lead to mitochondrial dysfunction and damage the BBB, but the mechanisms by which that takes place are not entirely clear and debated.

What is clear, is that Aβ aggregates activate microglia and astrocytes, leading to chronic inflammatory responses. In Alzheimer’s, microglia fail to effectively clear amyloid, remaining chronically activated and pro-inflammatory. This is how microglia stimulated by amyloid look like.

Amyloid precursors and subspecies

APP, BACE1, γ-secretase

Functional amyloid is generated by cleavage of the amyloid precursor protein or APP. It can be cleaved via two pathways, the ‘good’ and so-called non amyloidogenic pathway where it is cleaved by α-secretase, and the ‘bad’ pathway where it is cleaved by β-secretase (BACE1), thereby releasing a soluble APP-β fragment. Subsequently, γ-secretase leads to the aggregation thereof into Aβ peptides, which are mainly Aβ40 and Aβ42.

Aβ42 and Aβ40

Amyloid proponents dug deeper into amyloid subspecies, and found that Aβ42 is more toxic and more prone to aggregation than Aβ40. On that basis, the Aβ42/Aβ40 has seen the light, which is claimed to be a predictor of efficacy: if it goes up, it would be indicative of potential efficacy. How does this work?

In Alzheimer’s disease, Aβ42 accumulates in plaques, thereby reducing its levels in the cerebral spinal fluid, while Aβ40 remains more stable. Therefore, in Alzheimer’s, contrary to normal individuals, the CSF Aβ42/Aβ40 ratio declines and therefore a low ratio becomes a biomarker of amyloid pathology, even before cognitive symptoms appear. Hence, seeing the CSF Aβ42/Aβ40 ratio go up could be indicative of less toxic amyloid accumulation in the brain, and hence less cognitive decline. The only thing is: you have to be a strong believer of the amyloid hypothesis, and expecting any drug to necessarily having to impact amyloid to be successful, to consider this relevant. Which is why in trials with a mechanism of action other than amyloid-reduction, I believe this ratio is close to irrelevant.

For the record, not every agrees. Professor of neurology Alberto Espay, for one – cfr. infra – does not. And like many others, I give his voice some weight as we know the amyloid hypothesis is prone to groupthink.

Amyloid subspecies based on aggregation

Furthermore, science has also differentiated between subspecies of amyloid that are considered more toxic than others. This is more or less where we’re at.

Evolutions in the amyloid approach

BACE and γ-secretase-inhibitors

On the basis of the understanding of APP and BACE1, if one could inhibit BACE1 or γ-secretase, Alzheimer’s would probably be solved.

In the early 2000’s, pharmaceutical companies developed BACE-inhibitors, such as verubecestat and lanabecestat.

People ended up getting sicker in many of these trials, which showed no cognitive benefit, leading science to abandon the BACE inhibitors.

But the theory remained unchanged, and nobody thought to wonder how people could be worse off on such a therapy (answer, in my eyes: it’s the microglia).

Amyloid vaccines

Another approach that was taken was to develop a vaccine that would work against toxic amyloid. Amyloid vaccines challenged the immune system to make an amyloid antibody, so that it mounts an immune response and tries to get rid of the amyloid. Those generally came with a lot of inflammation (immune response), something you may want to avoid in neurodegenerative diseases because you want the glial cells to be functional.

One of these vaccines was called AN1792. It did not lead to cognitive benefits. In fact, it induced meningoencephalitis in 6% of patients due to T-cell activation. That’s inflammation of the brain ánd of the tissue surrounding the brain (meninges).

The second-generation vaccines were designed to stimulate other immune cells than T-cells. They were safer, but not efficacious.

Anti-amyloid antibodies

Considering that some amyloid species may be more toxic than others, big pharma started developing antibodies which should reach the brain in sufficient amounts to allow amyloid to be removed.

Anti-amyloid antibodies are basically pre-engineered antibodies, which should subsequently bind to toxic amyloid (subspecies) and allow them to be removed by microglia.

For clarity, and as explained here, these antibodies do not remove amyloid themselves, they call upon microglia to do so.

These antibodies are huge, and even if the blood-brain-barrier is impaired in Alzheimer’s disease, need high quantities to be present in the brain in sufficient amounts.

Most of them failed. Aduhelm (aducanumab) – now taken off the market – got approved and showed how much the FDA was able to bend to Biogen’s lobbying. It did not show efficacy. It was approved through the accelerated approval pathway because it removed amyloid.

Then Leqembi (lecanemab) and Kisunla (donanemab) finally made it, causing about a 30% slowing of cognitive decline. Remternetug is still in Eli Lilly’s pipeline. To put those wins into context, about 30-40 trials in Alzheimer’s disease have succeeded in lowering amyloid in the brain, and about two of them actually saw an effect on cognition. Those two, remarkably, only allowed patients that have high amounts of amyloid in their brain – a novelty which I believe is integral to their success and which is, mutatis mutandis, also implemented by INmune Bio. Two (little) wins in 30-40 trials is a surprisingly low number for the predominant theory on the etiology of Alzheimer’s disease.

Flaws and criticism

ARIA is the elephant in the room for these therapies. There have not been any anecdotes of people who experience stable cognition on these drugs, but there have been anecdotes of deaths, however.

ARIA – amyloid-relating imaging abnormalities – is the most notable euphemism one can find in the science of Alzheimer’s to identify either brain swelling (ARIA-H) or brain leakage (ARIA-E). ARIA is the reason why the EU has been so reluctant to approve these therapies, even if there is nothing out there – except for Anavex, perhaps (…?) - which may readily be approved and could allow patients to see some efficacy from therapy. ARIA is a nice word to say that parts of the brain such as the blood-brain-barrier are being destroyed further – the BBB is already dysfunctional in Alzheimer’s - by the forced removal of large aggregated amyloid structures which are considered toxic.

Strangely, the mechanism of action of both drugs is quite different, which may indicate that the theory behind certain toxic species of amyloid may not be entirely correct. Leqembi targets soluble amyloid protofibrils. Kisunla targets amyloid plaques and fibrils. Remternetug, still in Eli Lilly’s pipeline, targets amyloid plaques.

Some say, moreover, that it is not the removal of the allegedly toxic species of amyloid that are causing the slowing of cognitive decline, but the increase of Aβ42 levels. In fact, amyloid aggregation is considered by some as a mechanism that is supposed to be protective. One of the proponents of this idea is Professor of neurology Alberto Espay, who stated:

“Here is an alternative explanation for how lecanemab and other anti-Aβ monoclonal antibodies may slow cognitive decline in Alzheimer’s —by increasing Aβ42 levels. By analyzing 24 monoclonal antibody trials involving 25,000 patients, we showed that a slower decline in cognitive and clinical function (as measured by ADAS-Cog or CDR-SB) can be equally predicted by reductions in brain amyloid and increases in cerebrospinal fluid Aβ42 levels. https://lnkd.in/gtDdR5tz We conclude that increases in Aβ42 may represent a mechanism of potential benefit of anti-Aβ monoclonal antibodies in AD.All stories have two sides —including the one we have told ourselves about how anti-amyloid treatments work: by lowering amyloid. Even if the increase in Aβ42 is unintended, it is why there may be any benefit from anti-amyloid treatments.But do we give patients an anti-amyloid infusion to increase their levels of the normal protein Aβ42? No—the end doesn’t justify the means. The removal of amyloid is toxic. Donepezil is better and safer.”

The basis for that idea was that the supplementary material to Leqembi’s data showed that CSF Aβ42 levels were markedly increased in Leqembi’s study.

His full series of tweets on the matter is worth reading and can be found here.

The future of amyloid therapies

So after all these failures, is big pharma now finally targeting the right type of amyloid, and should better results be forthcoming? I believe that is unlikely.

In the end – in my view – amyloid is only one of several inflammatory/neurodegenerative triggers. Remove some, and you may end up seeing neurodegeneration reduced, remove many and you may even see cognition returning. I see no other way to explain all the other links or theories for Alzheimer’s, such as the entire viral hypothesis of dementia (cfr. my blog on that topic). There are so many clear links between viral load of certain brain-persistent viruses and dementia, and one sees reduced incidence of dementia after vaccination for only just one of these viruses.

Big revenue estimates, yet people don’t buy it

In light of all of this, anti-amyloid antibodies are a sales flop. Aduhelm, originally priced at $56,000 per year but denied Medicare coverage, was expected to hit $9 billion in peak sales. Sales were terrible, many hospitals refused administration, and Biogen eventually took Aduhelm off the market.

Leqembi comes with a yearly cost of $26,500. Initial projections for Leqembi were also optimistic, with Eisai estimating peak sales of $7.3 billion by 2030 and GlobalData forecasting $12.9 billion in revenue from 2023 to 2028. Actual sales have significantly underperformed these expectations.

Leqembi has repeatedly missed sales targets, and 2024 sales for Leqembi were only $280 million, with the U.S. market contributing $170 million. Kisunla commercialization started later, so it’s hard to assess that at this time.

But to put those Leqembi sales into context: at a $26,500 price tag, with $170 million in sales in the US and $280 globally, Biogen treated 10,566 Alzheimer’s patients in 2024.

There are 7.2 million Americans living with Alzheimer’s in the US. If you would give all of them Leqembi at a $26,500 price tag, you would have 190.8 billion in sales.

Also, though exact numbers are missing, about 55% of all patients with Alzheimer’s are using non-disease-modifying drugs such as Donepezil. These are the most frequent ones:

In conclusion: the Alzheimer’s market remains largely untapped. People don’t buy the anti-amyloid story, and rightfully so. No do many of the scientific leaders in the field. I quote some.

Professor Karl Herrup:

- [On Aduhelm] “Biogen lobbied hard and the FDA, despite its mandate, caved into the pressure. When a regulatory agency goes against the near unanimous advice of a group of experts called in to advise them (and is so egregiously wrong in its actions that three of the 11 experts resign) it’s pretty clear that ‘something is rotten in the state of Denmark’. That was the breach that caused the entire regulatory dam to fail. After that approval, based on amyloid reduction as the salient outcome measure, there was no way to stop the approval of the other monoclonals. To be blunt, money talked,”

- [On the amyloid hypothesis] “I remain open to new information, but I basically spent an entire book explaining why claiming amyloid as a “core” biomarker is not supported by the data. […] Biomarkers only make sense in one of two conditions. The first is where we are sure that the marker reflects a process that is in the direct line of causality of a disease. In this situation, by measuring the marker we are measuring a necessary step in the disease process. I feel 100% confident in saying that this is not true for amyloid in any stage of aggregation. […] The second situation is when the biomarker is tightly linked with the disease process, even if it is not in the direct pathogenic pathway. Amyloid fails in this realm as well. As I argue in the book, you can have amyloid deposits without dementia, and you can have clinical dementia of the Alzheimer’s type without amyloid. A third requirement that is valuable but not necessary is that the biomarker has to work in both directions. It goes up in the presence of disease and goes down if and only if a treatment is successful. Amyloid fails this test as well since the lecanemab/donanemab trials show clearly that people continue to get worse for months after their amyloid is cleared.”

- George Perry, founding and current editor-in-chief for the Journal of Alzheimer’s Disease: “Amyloid is a downstream response, not irrelevant but not the primary driver. All efforts to remove amyloid have shown no significant benefit. It is likely metabolic/inflammatory abnormalities are upstream.”

- Prof. Rudolph J. Castellani: “This has resulted in the promotion of marginal results, media campaigns to sell a toxic drug, rationalization of iatrogenic deaths, rationalizing or minimizing toxicity. […] It is clear enough to me that if a company makes a drug such as an anti-amyloid antibody, such that will be given to millions of people at enormous expense (not just the drug but other expensive accompaniments like PET scan), a third party should run the clinical trial, and a third party should monitor the toxicity.”

- I have quoted professor Rob Howard on the matter before. Here he is once more, outspoken as always: "Framing the effects as a percentage also subtly suggests that patients who took the drug could still be doing 35 percent better than their peers years later, says Howard. But there’s little reason to think that should be the case. Within 18 months, both lecanemab and donanemab wiped most patients’ brains clear of excess amyloid beta, so taking the drugs for longer may not provide any additional benefit. And though the data is limited, there’s some preliminary evidence that patients don’t see any additional slowing of their cognitive decline after finishing their course of lecanemab, and from this point decline at the same rate as untreated patients—making any relative benefits of the drug smaller over time."

 _____________________________

I wanted to take a little detour here focusing on research from Phd., Dr. Jack Auty, a researcher who’s been focused on the immune system in Alzheimer’s disease. This is work in animal models, but it brings together but the glial and amyloid story. What has Dr. Auty reported on?

Amyloid is toxic in cultures where astrocytes and microglia are present.

However, amyloid appears much less toxic without microglia.

“What we see is that this medium dose of amyloid is now not toxic at all. So if you remove the microglia, the amyloid is now not very toxic. We don't get that neuronal loss in these cell culture dishes. However, they did show that the very high dose of 10  micromolar dose of amyloid was still neuronal toxic, so it seems like a very very high dose of amyloid is neurotoxic, but a medium to low dose is not directly neurotoxic.”

They went further into this, and tested whether a protein knocking out microglial phagocytosis could have an influence. Apparently it did.

“[…] 30 to 40 percent of neurons are dying, remember this is in the presence of microglia and here they're put on annexins, so they put annexin, some protein to block the microglial phagocytosis of the neurons that's phosphatidylserine dependent. And what we see is, annexin completely blocks the number of neurons dying, it's returning back to 100. So now we know that annexin and the phosphatidylserine flip is involved in the neurons dying due to microglia.”

To be continued…

INmune Bio discord investors server: https://discord.gg/JEA8r7wCGY