The $10 billion unicorn in INmune Bio: On The Possible, part 1

Introduction

INmune Bio (INMB) is developing a novel therapy targeting neurodegenerative diseases, particularly Alzheimer’s disease (AD). Placebo-controlled results are expected for June 2025.

As a reminder, most INmune Bio shorts are institutional. They short stocks low on cash and target AD as a basket case, without a concrete bear thesis.

Very recently, we saw active short positioning. Biotenic came with a thesis, see the correction here. Excidium came with a thesis, here is the correction. Paurana Capital came with one, which was rebutted here. Excidium and Paurana are no Wall Street analysts, for the record; both are university-related.

A large part of the short theory appears to hinge on XPro not being brain-penetrant and being identical to non-selective TNF inhibitors. That thesis defies science, as I’ll explain below. Another thing stands out across the short theses: a scientific battle that took place years ago between the proponents of the amyloid thesis and those that believe Alzheimer’s may be an immunologic disease, now finds its way to the investor world, though the academic field has meanwhile moved on. Finally, many investors are not up to date on biomarkers the field is looking at. In this regard, the neurological space is in full evolution.

This is not to ridicule any institutions or persons who are short. I appreciate fair and honest debate. Some of the shorts have put in a good amount of work. The intricacies of the bear thesis on Cassava existed since 2021, and that thesis was strong. To a large extent, Martin’s Shkreli’s essay ‘On The Impossible’ was a follow-up of the same.

The bear theses on INmune Bio are recent. The fact that the stock has not budged since they came out – contrary to SAVA’s crash in 2021 – should already indicate a lot. I think they are wrong, and also think academics hold INmune Bio in much higher regard than Cassava Sciences.

Finally, there appears to be consensus that the stock valuation can shoot up drastically on a successful readout. If stable cognition would be announced, a $10 billion move on short notice, which would be a ~55x from current valuations, may be within the realm of the possible. Alzheimer’s disease, as Martin Shkreli stated, is one of the final frontiers for pharmaceutical companies, with six to seven million  patients in the US alone, and may become the largest pharmaceutical market of all time. “A great AD drug could easily command a $100,000 price”, which would mean “that a truly transformative AD drug could be the first medicine ever to break the $100 billion in annual revenue barrier.” From that angle, the game the shorts are playing is more risky than that of the longs.

Several successes in Alzheimer’s and related diseases have also been reported in the past 4 years, either from big pharma or from biotech companies, casting a different image on the historical failure rate. INmune Bio’s trial is done with rigor. Biogen’s and Eli Lilly’s trials already showed success at the six-month mark.

Anybody in biotech would be silly to think that the trial is a certain success. We have to be humble in the face of science. But I believe the chances are very high. Nothing is impossible. Tiziana Life Sciences just announced stabilization or improvement of na-SPMS in 10 patients, with an CNS-mediated immune-based mechanism of action (preprint publication here).

This coverage is split in two parts: a first part concerns a look at INmune Bio’s XPro from an analysis shared by Martin Shkreli on simufilam’s chances of success. The second part covers the content of his stream on INmune Bio. Including both in one post would be too much, so I have split them in two posts.

On The Possible

Introduction

Some months ago, Martin Shkreli presented a bear thesis against Cassava’s simufilam, arguing that its mechanism lacks biological plausibility, that its clinical data may be unreliable, and that there are concerns about data integrity.

I appreciated this work. The short thesis for Cassava Sciences, which is pinned to his X account, is accessible here: https://ontheimpossible.vercel.app/OnTheImpossible.pdf

 INmune Bio’s XPro (dominant-negative TNF inhibitor), in contrast, has a well-established mechanism rooted in immunology and neurodegeneration research.  I have been blogging about recent developments in Alzheimer’s disease and more specifically on INmune Bio since 2021, allowing some referrals to those blog posts.

This analysis examines whether the criticisms against simufilam apply to XPro, breaking down the key arguments. This will be its conclusion:

Let’s dive in.

 

Broader comparison

a) Mechanistic plausibility & scientific foundations 

Simufilam was claimed to restore the normal function of filamin A (FLNA), a scaffolding protein that Cassava alleges is disrupted in Alzheimer’s. 

The criticisms against that were: 

  - there is limited independent validation that FLNA dysfunction is a primary driver of AD; 

  - the hypothesis that fixing FLNA reverses neurodegeneration is not widely accepted in the Alzheimer’s research community;

  - some scientists argue that the proposed mechanism is too indirect to explain the reported cognitive improvements. 

XPro’s mechanism of action, however, is well-established:

- XPro is a dominant-negative TNF inhibitor that selectively neutralizes soluble TNF (sTNF), a key driver of neuroinflammation, while sparing transmembrane TNF (important for immune function); 

  - TNF’s role in neuroinflammation is well-documented in AD, multiple sclerosis, and other neurodegenerative diseases; 

- Precedent drugs: Anti-TNF therapies (e.g., etanercept, infliximab) are already approved for autoimmune diseases, validating the pathway;

- Preclinical data shows that sTNF inhibition reduces biomarkers associated with AD, such as microglial activation, synaptic toxicity, and amyloid-related inflammation; 

- Unlike simufilam, XPro’s mechanism is not an outlier—it aligns with the growing consensus that neuroinflammation is central to AD progression. 

Conclusion: while simufilam’s mechanism is novel but unproven, XPro’s approach is backed by decades of TNF biology research. 

b) Preclinical & clinical data consistency 

Simufilam’s data issues  were

- Lack of independent replication: Most positive data came from Cassava-sponsored studies; there was no third-party validation. 

- Unusual biomarker claims: rapid, dramatic reductions in phospho-tau and neuroinflammation markers without a clear mechanistic explanation. 

- Open-label trial biases: early cognitive improvements were reported in small, open-label studies, which are prone to placebo effects and subjective bias. 

 XPro’s data Strengths 

- Preclinical reproducibility: INmune Bio’s TNF inhibition approach has been studied by independent researchers in AD models; 

- Biomarker coherence: Reductions in neuroinflammatory markers (e.g., YKL-40, GFAP) align with the expected pharmacology of TNF blockade; 

- Phase 1 data: showed target engagement (reduced sTNF) and favorable safety, supporting further development. 

Conclusion: XPro’s data follows a more rigorous and reproducible path, whereas simufilam’s results have raised skepticism due to unverified claims and methodological concerns. 

c) Target engagement 

In the case of Cassava Sciences’ simufilam, there was an alleged lack of clear target engagement: it was unclear how binding FLNA would lead to the claimed downstream effects (e.g., reduced p-tau, Aβ improvements). 

As for XPro:

- There is a clear TNF pathway modulation: sTNF inhibition is measurable, and downstream effects (e.g., reduced neuroinflammation markers) are predictable. 

- There is an alignment with AD pathophysiology: neuroinflammation is a known contributor to neurodegeneration, making biomarker changes (e.g., reduced YKL-40) biologically plausible. 

Conclusion: XPro’s target engagement is mechanistically grounded. 

d) Data Transparency 

 Simufilam’s concerns were:

- lack of peer-reviewed publications for key claims. 

- unusual statistical choices in data presentation (e.g., selective endpoints). 

 XPro’s more robust approach shows:

- Abundant peer-reviewed preclinical work in a number of preclinical indications, supporting the mechanism;

- transparent biomarker strategies (e.g., CSF and imaging data in Phase 1). 

Conclusion: INmune Bio follows conventional, rigorous clinical development. 

e) Management & credibility concerns 

Cassava’s controversies were:

- allegations of data manipulation (e.g., Western blot concerns in preclinical studies). 

- SEC investigation into misleading statements. 

 INmune Bio’s track record shows:

- No controversies regarding data integrity;

- Academically rooted third-party validation of data;

- collaborations with academia and reputable institutions. 

f) Trial design

Cassava allegedly intervened too late in the disease, enrolling patients with mild to moderate Alzheimer’s disease.

Additionally, ADAS-Cog was not a good rating scale.

On the contrary, INmune Bio’s Phase 2 trials enrolls patients with MCI and mild AD.

EMACC is a much more precise and objective rating scale, designed to assess cognition in earlier stages of disease in line with FDA guidance. CDR-SB is also a better rating scale than ADAS-Cog, having yielded good results in similar patient populations in successful trials done by Eisai/Biogen and Eli Lilly.

Conclusion: INmune Bio does not face the same credibility risks as Cassava Sciences. 

In summary, while the bear case against simufilam hinged on unproven mechanisms, questionable data, and credibility issues, none of these apply to XPro. INmune Bio’s approach is scientifically sound, transparent, clinically rigorous, independently validated, the trial intervenes earlier in the disease, and trial design fits the enrolled patient population.

I come to this conclusion:

In-depth comparison of XPro with Mr. Shkreli’s bear thesis on Cassava Sciences

I include excerpts from Mr. Shkreli’s X posting and report on Cassava Sciences below. These do not cover his introductory remarks on Alzheimer’s disease, nor professor Wang’s research for Cassava Sciences, as this would not add any value here. I have also rearranged some of the arguments on the basis of the subjects they belong to.

1. On the etiology of Alzheimer’s disease (1)

“The central actor in Alzheimer’s is amyloid-β (Hampel, 2021).”

Wrong. Amyloid-β is more a biomarker of disease, and if a trigger, it is probably one of many triggers for neurodegeneration. The disease is multifactorial. TNF is a driver of amyloid and tau.

 “Why is the amyloid hypothesis accepted as canon? While there have been some early papers on amyloid and Alzheimer’s retracted for errors (and even outright fraud), the amyloid hypothesis has been proven. How? There are three inherited forms of Alzheimer’s disease which have mutations in amyloid-β pathways (PSEN1, PSEN2, APP) (Scheltens, 2020) that cause early amyloid deposition. These patients quickly get Alzheimer’s at young ages. Another proof of amyloid’s involvement in Alzheimer’s is Down Syndrome. In Down Syndrome, patients have a third copy of chromosome 21 (trisomy 21). APP is located on chromosome 21! Hence, Down patients have three doses of BAPP when the rest of us have two. Down Syndrome patients accumulate amyloid and have an Alzheimer’s like phenotype at an early age.”

Wrong. The amyloid hypothesis has not been accepted as canon. Read recent book coverages here and here. Moreover, the above is not ‘proof’ that the amyloid hypothesis is proven. There is a high number of patients with high amyloid burden that are not cognitively impaired, and vice versa, there are a number of patients with dementia without high amyloid burden.

2. On the etiology of Alzheimer’s disease (2)

“But amyloid deposition does not coincide with cognitive deficit—in fact, it precedes it by decades.”

Correct, which already indicates that amyloid cannot be the sole culprit in Alzheimer’s disease. At the latest conferences, the prevailing idea currently seems to be that amyloid is a biomarker of disease much more than it is a driver of disease. The consensus here is rather new, but the academic thoughts are not, which is why there are already mainstream books that have propagated this. Already in 2018, authors wrote (Questions concerning the role of amyloid-β in the definition, aetiology and diagnosis of Alzheimer’s disease):

“Aβ pathology is a risk factor for AD, but does not guarantee it

Although unexplained, the evidence does suggest Aβ pathology in cognitively normal and MCI is associated with a higher likelihood of progression to MCI or AD dementia [51, 221]. Moreover, a correlation has been reported between amyloid PET positivity and subjective cognitive decline in cognitively normal elderly in some [3, 202] but not all studies [44, 45]. Others have reported amyloid positivity in cognitively normal individuals is associated with low memory scores, but not Mini Mental State Examination (MMSE) scores [130] (see [8, 182]). Furthermore, there have been recent suggestions that the relative amount of amyloid plaques (i.e. a dose-response) [22, 79, 94], or the rate of accumulation [153], rather than just amyloid ‘positive’ or ‘negative’ status, is linked to cognitive decline.

Hence, the presence of amyloid in cognitively normal individuals may be useful for predicting a risk of conversion from non-symptomatic to symptomatic stages. However, these studies merely suggest Aβ pathology to be a risk factor for dementia, not necessarily a cause, let alone the sole cause. Some studies, for instance, have shown as high as 80% a non-conversion rate of amyloid positive cognitively normal individuals to MCI or dementia 2–3 years later [280], with some individuals remaining cognitively stable for up to 6 years after follow-up [44]. Furthermore, other evidence suggests injury markers, rather than amyloid markers, are better predictors of progression from MCI to AD [276]. Collectively, these results suggest the utility of Aβ pathology alone to predict cognitive decline may be limited, questioning its applicability as a disease defining biomarker.”

In inflammation, TNF is the first cytokine going up. See here.

3. On the etiology of Alzheimer’s disease (3)

“Inflammation and brain atrophy are common in Alzheimer’s. Remember that the brain’s immune system differs substantially from the rest of the body. The brain uses microglia to act as the primary immune system. Lewy bodies and TDP43 are also found in AD. Many, including myself, believe that the main damage caused by Alzheimer’s is collateral damage via the immune system. Still, we do not completely know how involved microglia are in causing or even suppressing AD.”

The immune system of the CNS and microglia are indeed essential. Read my blog: It’s the (micro)glia, stupid! 

Noted: “Many, including myself, believe that the main damage caused by Alzheimer’s is collateral damage via the immune system.” This statement would contradict the thesis that amyloid is the central actor in AD,  unless when considering that amyloid and damage by immune cells correlate. Read here and here.

4. On the etiology of Alzheimer’s disease (4)

“The strong success of APOE protective alleles raises critical questions. The dramatic protective (or negative) effect suggests that new Alzheimer’s pharmaceutical research should focus here for the best results. TREM2 mutants which cause Alzheimer’s decrease binding of TREM2 to ApoE, which clears amyloid plaques. Many Alzheimer’s risk genes relate to microglial response. To me, it is obvious that the brain struggles to clear amyloid-β, often causing damage in the process, akin to peripheral autoimmune diseases. Note that Cassava’s putative mechanism of action, which we will review later, does not have any interaction with ApoE, TREM2 or any of the microglial response.”

Read here for some further information on APOE and sTREM2.

5. On the etiology of Alzheimer’s disease (5)

“Those not close to this space may have heard that the amyloid hypothesis is wrong or debunked. This could not be further from the truth. In fact, in recent years, the amyloid hypothesis has been completely confirmed, and I will convincingly prove this.”

Outdated, read recent book coverages here and here. Amyloid is one of many triggers. Read here, here and here. Alzheimer’s is a multifactorial disease. After numerous failures, the most an anti-amyloid antibody can do is about 30% slowing of cognitive decline, as proven by Eisai/Biogen and Eli Lilly.

6. On background information (1)

“When I get really interested in a drug, I start from the beginning. Who invented it? Why? What else were they working on? What was their approach?”

Malù Tansey, a renowned professor and speaker, developed XPro. Not Rudolph Tanzi, to be clear.

7. On the Mechanism of Action (MoA) (1)

XPro’s MoA is selective inhibition of TNF. TNF inhibitors have been the biggest drug franchise in the world. Humira peaked in 2021 with $21 billion in revenue, having been the top earning in pharma for six years straight, and having about 40% of the TNF inhibitors market. Total revenue for Humira alone was estimated around $208 billion in 2023. Meta analysis show significantly reduced risk of developing AD when treated with TNF inhibitors. XPro’s selectivity for soluble TNF allows restoration of original microglial function. The mechanism of action has been confirmed in multiple publications, many of which are independent and peer-reviewed.

8. On the MoA (2)

“But a drug is a drug. They’re all the same—the molecule can be characterized, it can be manufactured, and it can be determined to work or not work. The truth is all we care about.”

XPro’s mechanism of action and efficacy have repeatedly been seen to work.

9. On the MoA (3)

“The actual drug discovery and clinical development components of drug development, in my view, is often comparatively easier. After all, we’ve been making accurate, structure-based medicines for more than 30 years. We have tools like small molecules, antibodies, nucleic acids and other methods to interrogate the usefulness of interrupting or otherwise regulating a protein’s function. The real question is: what protein do we want to make a drug for, and which disease do we hypothesize that will benefit?”

XPro’s Phase 2 trial is based on all preclinical and clinical evidence to data.

10. On the MoA (4)

“As you read the following pages, I want you to understand that the likelihood a medicine can work without a precisely defined mechanism of action is very low.”

Ergo: the likelihood a medicine can work with a precisely defined mechanism of action is logically much higher. The MoA for XPro is clear. It also builds on the information that non-selective TNF inhibitors, although they are too big to pass the blood-brain-barrier and come with several side effects such as demyelination, show some efficacy in preventing Alzheimer’s, as has been confirmed repeatedly in several meta analyses.  To quote some of these studies:

- A study of 8.5 million insured adults in the United States (US) reported increased risk for Alzheimer’s disease in patients with rheumatoid arthritis. It also reported that etanercept, a TNF inhibitor, was associated with reduction of risk for AD.

- A  large, retrospective case-control study from 56 million unique adult patients showed again that patients with rheumatoid arthritis, psoriasis, ankylosing spondylitis, inflammatory bowel disease, ulcerative colitis and Crohn’s disease had an increased risk for Alzheimer’s disease. The risk for Alzheimer’s disease in patients with rheumatoid arthritis and psoriasis was lower among patients treated with different TNF inhibitors.

-  A study of 2510 patients with rheumatoid arthritis reported an association of TNF inhibitor use and reduced dementia risk, consistent as the study period increased from 5 to 20 years after diagnosis. TNF inhibitor use showed a long-term effect in reducing the risk of Alzheimer’s disease during the 20 years of follow-up.

- Here, here and here are some further meta analyses.

One study reported that patients with rheumatoid arthritis have treated with etanercept, adalimumab, or infliximab had a 66%, 72%, and 48% lower risk of developing Alzheimer's disease, respectively, compared to those not treated with these TNF inhibitors. Another study mentioned a 30% risk reduction.

Various TNF-blocking agents including etanercept, infliximab and adalimumab have furthermore demonstrated the ability to reduce microgliosis, neuronal loss, tau tangles, and Aβ accumulation. Both pre-clinical and clinical studies have indicated their potential in improving cognitive function.

The evidence that TNF-inhibition, even non-selectively, impacts risk and therefore probably also progression of Alzheimer’s, is therefore too big to ignore. It is, in my opinion, so much clearer than the all-in-all muddy waters in which the amyloid hypothesis has been developed.

Recently, a new element to that research has been added. On April 11, 2025, a non-placebo-controlled study was published showing a statistically significant association between cognitive improvement, particularly executive function and memory, and TNF inhibition. Test duration was the same as INmune Bio’s Phase 2 trial. The patients in that trial had mild cognitive impairment and should hence decline by 1-2 points on average on the MoCa scale (Montreal Cognitive Assessment) over a year’s time. The study reported, on the contrary, an improvement in cognition by a point on the same scale (p = 0.001).  Significant improvements were also observed in the digit span forward test (p = 0.003), digit span backward test (p = 0.021), Stroop-W test (p = 0.040), Stroop-C test (p = 0.014), and Stroop-CW test (p = 0.035). Improvements in the MoCA were associated with average CRP, the standard biomarker for levels of inflammation in the blood, and one of four selection biomarkers of INmune’s MINDful trial (hsCRP).

As the short thesis hinges to a large extent on the idea that traditional TNF inhibitors have failed to show benefit in Alzheimer’s disease, I believe this study in MCI patients shows differently. One could argue that it’s not placebo-controlled. Furthermore, traditional TNF inhibitors are not brain-penetrant, and broad TNF inhibition leads to immune dysfunction, which is undesirable particularly if one wants to target the immune cells of the brain. INmune Bio’s study will be placebo-controlled, XPro is brain-penetrant and selective for soluble TNF only, allowing the benefits of tmTNF signaling.

11. On the MoA (5)

“Whenever I look at a drug, I want to know exactly what the “binding event” looks like.”

XPro selectively binds to soluble TNF, as shown time and again in preclinical studies.

12. On the MoA (6)

“To work, a putative drug must make some molecular interaction with one or more targets in the patient. The targets should be related to the patient’s illness. I like to study this closely.”

Soluble TNF is related to Alzheimer’s disease and other neurodegenerative diseases. See the preclinical work and read here.

13. On the MoA (7)

“When designing drugs, it is critical to understand the protein target thoroughly.”

Correct. Additionally, enrollment should align with that target. INmune Bio’s Phase 2 trial enrolls patients with inflammation, in line with the drug target. See here, here and here.

14. On the MoA (8)

“The Company’s theory is that Alzheimer’s patients have an ‘altered’ form of filamin A.”

This theory was never proven by Cassava Sciences. The involvement of soluble TNF and transmembrane TNF has been proven in a number of neurodegenerative indications. Read here, for example.

15. On the MoA (9)

“The company’s theory is filamin A interacts with three different receptors: nicotinic alpha7, TLR4 and the insulin receptor. The company suggests that by simufilam binding filamin A, the binding of filamin A to the three receptors is terminated. While this is unlikely to be true, we must explore it further. By terminating the “association” between filamin A and nicotinic alpha7, the company claims that the aberrant signaling caused by amyloid-beta at the nicotinic receptor is also abrogated. This syllogism (simufilam binds filamin A effectively enough to have physiological consequences à simufilam binding results in filamin A dissociation from a binding partner à the dissociation induces a distal conformational change sufficient to alter orthosteric binding) is shaky, at best. There is little support for any of the claims, which are outlandish. One of the reasons to disregard this theory is the extensive work the pharmaceutical industry conducted on the nicotinic space. Unfortunately, pharmacological interventions against 17 the nicotinic receptor family in Alzheimer’s have come up empty (Gault, 2016) (Lenz, 2015) (Frolich, 2011).”

“I colored the alleged binding region of simufilam in green. To those untrained in this field, this binding region is almost entirely “solvent-exposed”. The solvent in most of these cases is simply water. This is not an “orthosteric” site to my knowledge. Nothing is known to bind specifically at this site, which also makes sense because it would be difficult to form a high quality bond at this flat surface. Medicinal chemists hate protein surfaces like this because there is very little ability to create “shape complementarity.”

“This is another bombshell. Cassava claims in various places that simufilam has “femtomolar” potency. This is not something any serious drug developer would ever claim, as even picomolar potency is out of the realm of possibility for most small molecules. Even in Cassava’s own patents, simufilam is disclosed as a micromolar binder. I still do not believe this data, as it is suspiciously presented.”

There is abundant support for the XPro’s MoA. Read the preclinical work and here, for example.

16. On the MoA (10)

“Phase I SAD - NCT03784300 The results of this SAD were never published, but the short half-life of simufilam should have given Cassava pause and put priority on a “back-up” compound with better pharmacokinetics. A T1/2 of 4 hours is untenable for a serious pharmaceutical product.”

“In a non-public investigator’s brochure, Cassava disclosed critical details of simufilam problems that it did not disclose to Wall Street. In Figure 4, you can see that Cassava noted that in male rats, simufilam widely distributed to the typical organs: kidney, stomach, liver, etc. But it also disclosed the organs with the lowest concentration: bone and brain. One does not have to be an expert pharmacologist to know that it is not a good thing if your Alzheimer’s drug partitions in very low amounts to the brain.”

In several earnings calls and presentations, INmune Bio has mentioned that XPro can cross the blood-brain barrier and that the drug achieves measurable levels in the cerebrospinal fluid (CSF) after subcutaneous dosing.

This slide shows that more than 99.9% of the soluble TNF is blocked by the 1 mg/kg dose.

This is the reduction of cytokines in the cerebral spinal fluid, i.e. in the brain, seen when patients were given the 1 mg/kg dose.

Additionally, many of the brain biomarkers INmune has used look in the brain specifically. It is impossible that these results were obtained without the drug passing the blood-brain-barrier.

Additionally, there may be beneficial interaction with the periphery. Crosstalk between immune cells of the CNS and those of the periphery has been established. That may explain the meta-analyses showing reducing dementia risk in patients on TNF inhibitors, which are non-brain-penetrant. That may also explain the recently published  study (April 2025) showing statistically significant cognitive improvement from TNF inhibitors in patients with mild cognitive impairment.

Finally, the blood-brain-barrier is dysfunctional / broken down in Alzheimer’s disease, which could also explain the quick efficacy.

17. On brain penetration

At the dose used in the Phase 2 trial, XPro gets into the brain, blocking more than 99.9% of the soluble TNF. See picture above.

The log chart is a through levels, which means measured at day 6 after injection at day 1. That means a once-a-week therapy should neutralize all sTNF.

XPro is a modified version of soluble TNF itself, designed as a dominant-negative trimer. The dominant-negative technology means that XPro can still bind to TNF receptors, without activating them. Native human TNFα is a homotrimer, with each monomer approximately 17 kDa, making the trimer about 51 kDa. The protein is PEGylated, meaning it is conjugated with polyethylene glycol (PEG) to enhance its pharmacokinetic properties, such as half-life and stability. PEGylation will add about 20 kDa, making XPro’s total molecular weight probably about 71 kDa or half the size of anti-amyloid antibodies. To that exptent, I align with Paurana Capital’s view.

The practical perspective, both in preclinical and human models, shows that XPro achieves functionally meaningful brain exposure, enough to reduce neuroinflammation markers without huge doses. In studies where mice were injected with XPro (systemically, not into the brain), researchers found decreased microglial activation, decreased astrocyte activation (lower GFAP levels) and reduced amyloid plaque pathology. These effects were seen specifically in brain tissue. In a mouse model of multiple sclerosis, XPro was shown to reduce demyelination after peripheral (subcutaneous) dosing — not direct brain injection. In traumatic brain injury, XPro given systemically decreased neuroinflammation markers and improved cognitive outcomes. See also ‘Inhibition of soluble tumour necrosis factor is therapeutic in experimental autoimmune encephalomyelitis and promotes axon preservation and remyelination’, a study in which XPro was administered peripherally via subcutaneous injections, with this peripheral administration leading to significant improvements in clinical outcomes, including enhanced axonal preservation and remyelination, compared to both vehicle- and etanercept-treated groups.

18. On the validity of placebo-controlled trials (1)

“Clinical trials, done correctly, are deterministic. There is no chance. The outcome is preordained by the laws of physics: the chemistry of the drug, the biology of the patient, the medicine’s interference with the disease pathology. The carefully designed statistics allow us to blend away individual differences, the minute randomness that exists. If there is a potent effect: it will be seen.”

Noted.

19. On the validity of placebo-controlled trials (2)

“We will use what we know about medicine and logic to prove that Cassava’s simufilam cannot work. This is sobering news. I am no Cassandra, though. For 90% of medicines, I cannot 5 predict ahead of time whether they will work or not.”

Noted.

20. On external validation (1)

“I haven’t been able to find one such professional who thinks simufilam ‘has a chance’.”

There are numerous professionals who consider XPro to have a chance.

21.  On trial design (1)

“Keep in mind this very important issue: If it is too late to treat AD because the disease starts so early—is that a major barrier to success for any drug being developed in symptomatic AD? Could this reason be the primary driver of the 98% failure rate of Alzheimer’s drugs (Cummings, 2018) (Kim, 2022)”

Anti-amyloid antibodies have been shown to be efficacious in patients with MCI and mild AD, the same patient population of INmune Bio’s Phase 2 trial.

22. On biomarkers (1)

“The field must architect strong diagnostic and prognostic markers, to which regulators and physicians must be comfortable that they act as substantive surrogates for symptoms and disease progression.”

Nfl (Xpro: -84% in Phase 1 trial) and ptau-217 (XPro: -46% in Phase 1 trial) are such biomarkers.

NfL is an accepted biomarker in ALS for the purposes of accelerated approval.

Ptau-217 is the most predictable biomarker of disease, which is why the FDA granted Quanterix Breakthrough Device Designation to a pTau 217 blood test for Alzheimer disease four months ago (January 2025).

Also see here:

P-tau217 as a Reliable Blood-Based Marker of Alzheimer’s Disease (2024)

Plasma p-tau217 and tau-PET predict future cognitive decline among cognitively unimpaired individuals: implications for clinical trials (2025

And read here, here, here, here, here, here and here.

Many other biomarkers indicating disease-modification have been reported coming out of INmune Bio’s Phase 1 trial.

23. On alleged fraud (1)

“Burns presumably could not stand the idea of the drug demonstrating the exact opposite of what an Alzheimer’s drug could do. Not satisfied in manipulating the data only to show an equivocal outcome, Burns removed nearly half of the data to demonstrate that simufilam aided cognition. For those who do not follow scientific protocols or experiments, this is a cardinal sin. One practitioner once said it is akin to shooting an arrow and painting the bullseye after the arrow lands. It is extremely unethical, which is why Burns is no longer with Cassava. In my view, this is worse behavior than Wang’s!”

INmune Bio did not manipulate any data.

24. On the potential of a good Alzheimer’s drug (1)

“Even though it feels like reversal is out of the realm of possibility, I would counter that so long as neurons and synapse formation is stable, one can at least re-learn the lost information.”

Agreed: reversal is within the realm of possibility, as confirmed by INmune Bio. In its corporate presentation, INmune Bio mentions that stable cognition is the goal.

25. On valuation (1)

“At the same time, one could argue that a bona fide Alzheimer’s drug could be worth as much as $50 billion.”

Noted.

26. On valuation (2)

“Alzheimer’s is one of the final frontiers for pharmaceutical companies. With six to seven million patients (Alzheimer's Association, 2023) (Hampel, 2021) in the United States alone, Alzheimer’s Disease (“AD”) may become the largest pharmaceutical market of all time. With the most severe sequelae possible, a great AD drug could easily command a $100,000 price (around the price of immunooncology regimens). This means that a truly transformative AD drug could be the first medicine ever to break the $100 billion in annual revenue barrier, which no drug has come close to (Comirnaty, the COVID-19 vaccine reached $39 billion in its short-lived existence, Humira peaked at $20 billion).”

Noted: a truly transformative AD drug could be the first medicine ever to break the $100 billion in annual revenue barrier.

Probability analysis and conclusion for simufilam

“How do we synthesize our research to make a prediction? One reasonable avenue is the collect the necessary but not sufficient requirements for simufilam to meet the primary endpoint of the “RETHINK” trial. We must be careful to avoid dependent conditions. Necessary: • Simufilam binds to filamin with high affinity. • Simufilam partitions to the brain and has the stoichiometry to modulate filamin A proteins at scale. • Simufilam is a protein-protein interaction inhibitor, a “refolder” of filamin A, or both. • Filamin is relevant in Alzheimer’s. • The lack of efficacy in Phase IIa is irrelevant due to a small sample size. The lack of efficacy in Phase IIb can be explained: • The failure of the CWS can be overlooked. • Explanation for the lack of efficacy in Phase IIb is “follow-through” effect from open-label drug. • Explanation for the lack of efficacy in Phase IIb is mild patients do better than moderate patients. • Explanation for the lack of efficacy in Phase IIb is that open-label results suggest two-year stabilization of some Alzheimer’s patients. • The “RETHINK” trial is designed well and it is not too late in the disease to intervene.

I suggest that all nine conditions are necessary to overcome the burden of simufilam working in Alzheimer’s. I suggest that the probability of all of these being true is very low, which leads to a combined conditional probability of close to zero.”

Probability analysis and conclusion for XPro

• Xpro binds to soluble TNF with high affinity.

• XPro accesses the brain in sufficient amounts to inhibit 99.9% of its target.

• XPro is a pegylated selective TNF inhibitor.

• Inhibition of soluble TNF is relevant to Alzheimer’s disease, as is non-inhibition of transmembrane TNF.

• The efficacy seen in Phase 1 is relevant given the amount of impressive and consistent biomarkers, including those approved by the field.

• Efficacy seen in Phase 1 aligns with that seen in preclinical research, confirmed and peer-reviewed by external academia.

• MCI and mild AD is the right patient population to target.

• There was no lack of efficacy in any clinical trial to data with XPro, contrary to Cassava Sciences trials.

• The MINDful trial is designed well, and it is early enough in the disease to intervene.

Conclusion

As all nine conditions alleged to be necessary by Mr. Shkreli to overcome the burden of a drug potentially working in Alzheimer’s disease, XPro can be efficacious for the treatment of Alzheimer’s disease. A combined conditional probability of efficacy should therefore be high. Therefore, the thesis on INmune Bio should be: on the possible.