The $10 billion dollar unicorn in INmune Bio: on fast-progressors in AD, APOE4, TREM2 and EMACC
APOE4, involved in different mechanisms in the brain including lipid homeostasis, is the strongest risk factor for Alzheimer’s disease.
It is expressed in more than half of Alzheimer’s patients and sits at the interface of neuroinflammation and glial activity.
Alzheimer’s patients with biomarkers of neuroinflammation have faster cognitive decline, with less variance.
To measure cognition as exactly as possible, existing measuring scales were in appropriate given their considerable ceiling effects.
EMACC was developed for MCI, and has shown to be considerably more precise than existing measuring scales of cognition.
In trying to understand INmune Bio in Alzheimer’s, the deeper you look, the more things start making sense. This blogpost efforts to take a deeper dive in that science.
In the briefest possible manner, I wanted to express some of my thoughts on why exactly I believe INmune Bio’s short trials are not at risk of failure, and why EMACC is probably the most appropriate good scoring scale. That reasoning includes thoughts on fast progressors in Alzheimer’s, APOE4 as the most high-risk gene in Alzheimer’s disease and its link with neuroinflammation, and TREM2. If you haven’t read some of my earlier pieces, I would suggest reading/skimming the one on CJ Barnum’s teachings, the one on patient subgroup selection, and the one on microglia first. And this is a blog with my own understandings.
Though I read a lot, I am not a scientist, so I can’t assure everything I write is exactly right or fully up-to-date. Nor am I a financial advisor. That’s the disclaimer. Without further ado.
Apolipoprotein E or APOE is primarily known as a key regulator of lipid homeostasis. Within the brain, it plays a critical role in transport of cholesterol and lipids. As the brain is the most lipid-rich organ, that’s not a minor thing. APOE comes in three forms: APOE2, APOE3 and APOE4.
The APOE4 allele is by far the strongest genetic risk factor for sporadic Alzheimer’s disease, which accounts for 95% of all Alzheimer’s cases. APOE4 is expressed in more than half of all Alzheimer’s disease patients.
Astrocytes – glial cells like microglia - are the main source of APOE in the brain.
The APOE4 allele primes microglial cells toward a pro-inflammatory state, both in normal aging as well in AD.
APOE4 and the hallmarks of Alzheimer’s disease
Alzheimer’s is a disease of aging, where multiple things seem to go wrong at the same time, which could present differently from one person to another. APOE4 seems to be of indirect influence to many of them.
APOE4 influences disease onset and progression by interacting with the major pathological hallmarks of AD. APOE4 carriers in Alzheimer’s disease show earlier Aβ deposition, heavier Aβ plaque burden and increased brain atrophy compared to non-APOE4 carriers. Higher amyloid-β burden is even seen in APOE4 carriers with mild cognitive impairment and cognitively healthy elderly. BaselineAβ and tau burden is higher in APOE4 carriers, and Aβ and tau accumulation rate is higher in various brain regions as well. This has been well-known for a while.
APOE4 disrupts lipid homeostasis in the brain, which had already been remarked by Alois Alzheimer when he considered that ‘many glial cells show adipose saccules’, but has since been further confirmed. In animal models, APOE4 has been seen to disrupt glucose metabolism (the brain has high needs of glucose), and has been suggested to decrease insulin signaling by impairing recycling of the insulin receptor which in turn could be leading to the decreased glucose metabolism seen in Alzheimer’s patients. Energy storage in the brain is lower compared to the rest of the body. Energy metabolism in the brain is also regulated in part by astrocytes. Glycogen from astrocytes is important for maintaining healthy neurons and overall brain function, as an energetic buffer during periods of low glucose availability.
APOE4, glial cells and neuroinflammation
More recently, the incidence of neuroinflammation and glial activity has been uncovered. Neuroinflammation and glial activity are mentioned at once, not just by me but also in literature, because they are so interrelated.
While microglia and astrocytes are the main modulators of inflammation in the brain, they are also the major sources of ApoE in the brain.
As I have mentioned before, glial cells have both defense and nurturing functions in the brain. Microglia and astrocytes have different housekeeping functions that promote neuronal well-being. Oligodendrocytes are the main source of myelination in the CNS, or insulation allowing perfect signaling within neurons. Microglia and astrocytes are also the two major immune cells in the brain, at a permanent immune-vigilant state providing immunological defense if so needed, and as such are also master regulators of cytokines in the CNS. In case they are triggered, they move to a pro-inflammatory state, but in normal conditions, they revert to an anti-inflammatory state (lowering inflammation) once there is no further need for them to be activated, which is when they will revert to a resting state.
Under disease conditions, glial cells become permanently activated and progressively dysfunctional in regulating metabolic and immunoregulatory pathways, thereby promoting chronic inflammation-induced neurodegeneration. Microglia can respond to virtually all foreign factors in the brain typically described as danger-associated molecular patterns (DAMPs) or pathogen-associated molecular patterns (PAMPs). As mentioned before, microglia change phenotype, whereby M1 is the pro-inflammatory phenotype. So do astrocytes, moving to a pro-inflammatory A1 phenotype. Other scientific articles see more subtypes and a gliding scale here. An upcoming blog will focus on recent learnings, mostly from the lab of Prof. Shane Liddelow (the Liddelow Lab), on the interaction between microglia and astrocytes.
Until recently, inflammatory processes driven by microglia and astrocytes were not considered essential in Alzheimer’s disease. Those days are gone. Neuroinflammatory changes are among the earliest brain manifestations in Alzheimer’s disease, even before the onset of Ab accumulation, and are increasingly considered as critical in pathogenesis initiation and progression. Glia respond to, and clear, misfolded proteins in the brain. Those misfolded proteins may include Aβ and tau. Reports have suggested microglia surround the plaque forming a barrier with the ultimate goal of preventing further spread while also attempting to clear the Aβ. However, if there is a buildup of Aβ in the microglia, this can subsequently lead to microglial cell death and an increase in inflammation and recruitment of more microglia, thus continuing this inflammatory cascade. Additionally, the activated microglia can respond with a pro-inflammatory response, releasing cytokines such as tumor necrosis factor-α (TNFα) and interleukin 1β (IL-1β) as well as other factors.
APOE4 sits at the interface of inflammation and neurodegeneration via glial-mediated mechanisms, by altering specific pathways in astrocytes and microglia in the context of Alzheimer’s disease and the aging brain.
Genome-wide association studies and meta-analyses have indicated that the majority of genetic variants conferring risk for late onset sporadic AD are immune-related and enriched in microglia, implicating disease associated microglia or DAM.
APOE is a key regulator of the microglial transcriptional signature. The presence of APOE4 appears to confer an increase in pro-inflammatory cytokine production.
TREM2, APOE4 and inflammation
Recently, the importance of microglia and TREM2 has been brought to the forefront, as mutations in TREM2 increase the risk for Alzheimer’s disease.
TREM2 is a triggering receptor expressed on microglia. It has been reported to have an anti-inflammatory role, being implicated in a wide range of stimuli including Aβ and lipoproteins. Without Alzheimer’s pathology, TREM2 is expressed to clear damaged or apoptotic neurons and clear cellular debris while downregulating the pro-inflammatory response.
TREM2 is also a genetic risk factor for Alzheimer’s disease, and a 2013 study has reported that such risk exists ‘through impaired containment of inflammatory processes’. It was later found that several rare variants, R47H, R62H and H157Y, are linked to Alzheimer’s disease (so, targeting TREM2 as a whole may come with a risk – cfr. Alector).
In Alzheimer’s disease, TREM2 has been shown to be highly expressed on microglia surrounding neuritic plaques and would be important in clearance of Aβ.
APOE can bind TREM2 and, either directly or indirectly, APOE can alter TREM2 signaling or function. Both APOE and TREM2 are implicated in key steps in the homeostatic to a disease-associated microglia or DAM phenotype.
Recent studies suggest an interaction between microglia and astrocytes finding that once microglia become activated, they can lead to activation of astrocytes, resulting in a feed-forward loop which is detrimental to the surrounding environment. The mechanism showed that when activated, microglia release IL-1α, TNFα and C1q and astrocytes become activated.
Rapid progressors in Alzheimer’s disease
APOE4 carriers in Alzheimer’s disease show earlier Aβ deposition and clinical disease onset. They also show faster disease progression, heavier Aβ plaque burden and increased brain atrophy.
Increased tau levels, decreased amyloid-β42 and APOE allele predict progression from mild cognitive impairment to Alzheimer's disease.
A 2017, twoarticles on Alzforum on the basis of an article in Nature, stated that APOE4 interacts with tau to exacerbate the pathogenic cascade underlying neurodegenerative disease, independently of Aβ, and suggested ApoE may fuel the entire neurodegenerative process across multiple diseases. Obviously microglia and inflammation are very much implicated in that story. That article not only mentioned faster disease progression, but also – as commonly understood – a much earlier disease onset.
Interestingly, there seems to be an implication for anti-amyloid antibodies as well. Recently, Seeking Alpha author Lane Simonian mentioned that non-APOE4 carriers barely benefit from treatment with Lecanemab, and that Lecanemab reduces the rate of progression of APOE4 carriers closer to that of non-carriers in mild Alzheimer's disease. This is from a 2018 on Alzforum, on Lecanemab: “On the ADCOMS… where the highest dose group declined 30 percent less than placebo, APOE4 carriers declined 63 percent less than placebo and non-carriers only 7 percent less.”
INmune Bio’s view on cognitive decline in inflamed patients
INmune Bio basically places APOE4 at the same level as neuroinflammation, and as shown above, this is logical. The brief summary of INmune Bio’s MCI trial mentions: “The goal of this Phase 2 MCI study is to determine whether 1.0 mg/kg XPro1595 is superior to placebo at improving measures of cognition, functioning and brain quality in individuals with MCI and biomarkers associated with neuroinflammation (APOE4) and to evaluate safety, tolerability, and efficacy of XPro1595.”
In quarterly calls, questions from analysts on the subject have arisen more than once. Tom Shrader from BTIG is one of these analysts, whose focus seems to be there.
During the latest quarterly call, he asked whether the ADi (Alzheimer’s disease with inflammation) group would be progressing faster or slower, and what the placebo group would look like. INmune Bio answered that, on the basis of statistical analysis of the large Alzheimer’s database ADNI, focusing on patients with elevated CRP or APOE4, patients with these biomarkers indeed progress more rapidly and with less variance.
“So patients with neuroinflammation have more rapidly progressive cognitive decline in patients without neuroinflammation. Secondly, is that rapid progression occurs in a way that has less variance. And what I mean by variance is mean there’s less statistical noise around that decline. And that is what gives us the third advantage. It provides us with an ability to model the response compared to the treatment group in a way that allowed us to do both the shorter trials and the smaller trials.”
In the Q1 2022 call, the subject had been addressed as well, where INmune Bio had stated that the stage of disease does not matter. “If they have the biomarker of inflammation and you look over the course of the next six months, 12 months or however you want to look at it, their disease progresses faster and the most important part is that the variance in disease progression between patients is low, and that gives you enormous power advantages to do smaller and shorter trials.”
And in the Q4 2021 call: “So we believe in fact that APOE4 is a biomarker that predicts inflammation in the patients. Now whether that inflammation is independent of peripheral inflammation, we can't tell at this point. But we are very comfortable that in fact, this is one of the driving, this is the genetic marker that we can easily identify that is identifying a group of patients that have neuroinflammation. And we like it so much that the only enrichment criteria or enrollment criteria for the MCI trial is you have to be APOE4 positive.”
As an aside, XPro also decreases insulin resistance.
A word on measuring scales
Tom Shrader from BTIG has also repeatedly asked INmune Bio to explain a bit about the measuring scale EMACC, namely during the latest quarterly call and the Q4 2021 call. Those questions seem to be related to an abundance of measuring scales, making it harder to compare one to another. And that’s correct.
The answer to that question is also intertwined with the knowledge that a measuring scale to detect changes in early Alzheimer’s patients needs to be sufficiently precise.
Over the last decade, FDA has released two non-binding guidances for Alzheimer’s disease (AD) trials. A draft guidance in 2013 recommended the use of composite cognitive and functional scores like the Clinical Dementia Rating–Sum of Boxes (CDR-SB) to measure mild cognitive improvement in early-stage Alzheimer’s disease trials. A 2018 draft guidance that shifted their recommendations to “a broader array of endpoints including biomarkers, coprimary endpoints, and integrated scales” dependent on Alzheimer’s disease type.
That same year, an article appeared suggesting that using the ADAS-Cog scale in pre-dementia patients may be compromised, leading to the risk of trials incorrectly concluding that a novel treatment approach is not beneficial, and suggesting to improve measurement performance by altering scoring methodology or adding tests of memory, executive function, and/or daily functioning.
The diversity in measuring scales at this point is enormous, and comes forth from trial failures and the understanding that ADAS-Cog11, the first measuring scale, had ceiling effects and needed to be adjusted. Adas-Cog11 is an 11-task measuring scale to assess the severity of cognitive and noncognitive dysfunction from mild to severe AD. The full ADAS takes about 45 minutes to administer, and is scored from 0 to 150 by summing the number of errors made on each task so that higher scores indicate worse performance, and reduction in points means cognitive improvement. Seven of the eleven tasks have severe ceiling effects, meaning The term ceiling effect they are limited in that the highest possible score or close to the highest score can easily be reached, thereby decreasing the likelihood of precise testing.
Two Adas-Cog 11 tasks show milder ceiling effects. Accordingly, 71% of errors made by subjects with MCI occur on only two tasks, word recall and word recognition, which have also been ranked as the most difficult tasks.
31 adjustments to that scale have meanwhile been made, stemming, for example, from the consideration that ADAS-Cog-11 tasks do not have equal measurement precision, several subtask item response categories are disordered in terms of difficulty, and a difference of X points at the low end of the scoring range does not equal the same amount of difference in cognitive ability as a difference of X points at the higher end of the scoring range.
Those adjustments have resulted in several attempts to improve that measuring scale, some of which are Adas-Cog 13, Adas-Cog 12, iADRS, ADCOMS, and Adas-3, to mention some of the more known ones. For example, Adas-3 removed eight tasks from the ADAS-Cog-11 that demonstrate ceiling effects in MCI, resulting in a three-task ADAS-Cog: word recall, orientation, and word recognition.
Other measuring scales used are, for example, ADCS-ADL, or the Global Statistical Test combining Adas-Cog11 and ADCS-ADL23.
That leaves investors utterly confused. As these scales effort to be different from one another, and are hence not interchangeable, comparisons between them may logically not be made. Yet, that’s what appears to be done by some investors. Anavex proceeds down the same road and included ‘calculated Adas-Cog’ charts in its corporate presentation on the basis of other measurements taken, presenting them as if they were actual Adas-Cog measurements, and the question for the intelligent investor then arises: which Adas-Cog even??
Don’t get me wrong, I believe Anavex 2-73 may do something good, perhaps even outperform Lecanemab's 27% slowing of cognitive decline, but the above approach comes across as potentially misleading.
If there’s a lesson to be learned here, it’s that one should take open label studies in mild to moderate patients with Alzheimer’s disease with a serious grain of salt, and most certainly when Adas-Cog11 is applied (such as in the case of Cassava Sciences). I’m supportive of their work, but cutting-edge performance has been touted coming from the bluntest equipment in the toolbox. That comes at a risk, and I believe Cassava Sciences investors are not so much aware of this.
The need for a more precise measurement tool: EMACC
INmune Bio will use EMACC, ADCS-MCI ADL and CDR-SB as measuring scales for the MCIi trial, and EMACC, CDR and E-cog as measuring scales for mild Alzheimer’s patients in the ADi trial, with EMACC twice being the primary endpoint.
One understands from the above that, the sooner one intends to intervene in disease progression, the more important it becomes to measure precisely. One of the efforts by Clint Hagen at the Mayo Clinic is known as a compositive battery of tests known as PACC-R.
However, for MCI and early Alzheimer’s disease, that measurement also showed shortcomings, which is why EMACC has been developed.
EMACC, or the Early Mild Alzheimer’s Cognitive Composite, was developed by Judith Jaeger and presented on CTAD 2021, which I’ll summarize below. EMACC’s goal is not to have floor and ceiling effects, so that it can show sensitivity to change and accurate measures of change magnitude in the mild stages of the disease. EMACC consists of measures from six validated and widelyused neuropsychological test paradigms that in combination were shown to be maximally sensitive to disease progression. The EMACC score is the mean of standardized scores on a word list learning paradigm, immediate recall, digit span forward and backward, category fluency, letter fluency, trailmaking A&B and digit symbol coding.
Compared to Adas-Cog 13 and CDR-SB, using the EMACC scale led to an effect size difference which was substantially larger than for ADAS-Cog or CDR-SB, including in the subgroup with neuroinflammation. The differences in sample size required to determine a treatment affect were substantial when comparing EMACC, CDR-SB and ADAS-Cog 13 for ADi participants. To assess meaningful difference between groups of treated patients, sample size requirements for EMACC were therefore considerably smaller, and mostly for patients with neuroinflammation, which show a higher rate of cognitive decline and less variance than other Alzheimer’s patients. As a result, EMACC yields clear sample size advantages over ADAS-Cog13 and CDR, and is simply a more appropriate measure, that allows smaller sample sizes in faster progressors with biomarkers of neuroinflammation. Those patients with biomarkers of neuroinflammation are likely to be APOE4 positive, or in the case of the MCI trial, will certainly be APOE4 positive, as APOE4 sits at the interface of neuroinflammation and glial activity.
In short, quite some reasoning went into using this measurement scale on the side of INmune Bio. That reasoning was related to setting up the trials in patients with mild disease for success as much as possible, in light of earlier trials and failures. As such, I believe it is another paradigm-changing and de-risking effort, which I applaud as an investor.