The $10 billion unicorn in INmune Bio: reported biomarkers allow stabilization or improvement of cognition

Introduction

Novel biomarkers are changing the field drastically. They allow a view into the brain, whether through PET, MRI or blood-based measurements. Biogen gave a presentation in this regard in 2021.

We know that the ambitious goal of INmune Bio is to stabilize cognition in Alzheimer’s patients.

If biomarkers would be able to predict cognitive efficacy, where would XPro’s efficacy end up? I am doing an effort to make that exercise, on the CDR-SB rating scale, lacking comparative data for EMACC. The biomarker and cognition data for lecanemab (Leqembi) and donanemab (Kisunla), two recently approved therapies, have been taken into account. Chosen biomarkers, insofar as available, are NfL, ptau-217, and neurogranin.

This is an exercise, and per definition it is inexact and fallible. On the other hand, it would be unexpected not to see biomarkers align with cognitive efficacy. From that perspective, the exercise is useful in my eyes.

Introduction to biomarkers for Alzheimer’s disease

Reported biomarkers from the Phase 1b trial to be taken into account here are NfL, ptau-217 and neurogranin. These three are leading biomarkers recognized by the field.

o   NfL identifies neurodegenerative decay. The higher, the more neurodegeneration.

NfL predicts cognitive decline (see also here: comparison of NfL and total tau and associations with cognition and neuroimaging outcomes). Plasma NfL in MCI and AD as reported in one study is about 15-18 pg/ml.

o   Ptau-217 is considered the biomarker of excellence for Alzheimer’s. The FDA has granted Breakthrough Therapy Designation to a ptau-217 based blood test, and has just approved another blood-based biomarker tests for AD based on this biomarker. It identifies the amount of ptau in the blood, a biomarker of Alzheimer’s, and the neurodegenerative decay. Tau is a structural element in the cell. Phosphorylated tau is a hallmark of Alzheimer’s disease.

Ptau-217 predicts and identifies cognitive decline in AD. Plasma p-tau217 is about three-fold higher in amyloid-positive patients (12.4 pg/mL; 7.3—19.2 pg/mL) compared to amyloid-negative patients (3.7 pg/mL; 2.8—4.1 pg/mL) (data from the TIMC-Brain cohort, an Irish longitudinal research initiative with as primary goal to advance the understanding, diagnosis, and treatment of neurodegenerative diseases such as Alzheimer's disease, Lewy body disease, and frontotemporal dementia, recruiting approximately 1,000 participants over five years. For reference, plasma ptau-181 in MCI and AD as reported in one study is respectively 3 and 4 pg/ml, but in the TIMC-Brain cohort it was lower.

One should differentiate between plasma and CSF biomarkers. For example, in the TIMC-Brain cohort, for amyloid-positive patients, median p-tau217 is 12.4 pg/mL, but in the CSF it is 659.2 pg/ml. NfL in plasma in that cohort is 132 pg/ml, but in the CSF it is 2.9 ng/ml.

o   Neurogranin is a postsynaptic protein essential for memory, learning, and synaptic plasticity. It is considered to reflect synaptic degeneration.

Neurogranin is a predictor of memory and functional decline in MCI, and can predict cognitive decline in AD (also here and here).

Expected placebo decline in MINDful trial

The patients enrolled in INmune Bio’s MINDful trial are similar to those in Biogen/Eisai’s trial for Leqembi. Hence, we can assume a similar decline. Although patients with inflammation see a faster decline, we will disregard that for the purposes of the present exercise.

The placebo decline that was seen in the trial for Leqembi is shown below. At 18 months, the -0.451 difference translated to 27% slowing of cognitive decline. Absolute changes were 1.66 points for placebo and 1.22 for Leqembi at 18 months, both on CDR-SB.

Placebo decline for Kisunla was higher, coming close to 2 points in the low/medium tau population and being 2.3 in the combined population. I assume that the placebo decline was 0.5 at three months and 0.8 at six months, looking at the combined population.

Exercise relating cognition to biomarkers for lecanemab (Leqembi)

Placebo decline for Leqembi on CDR-SB was about 0.4 points at 3 months, about 0.6 points at 6 months and 1.66 points at 18 months. At that moment, Leqembi patient scores on CDR-SB had increased by 1.22 points, a difference of 0.451 points, showing a 27% slowing of decline vs. placebo.

I set out in this blog post what the estimated trajectory for NfL, ptau-217 and neurogranin was for patients on Leqembi.

Leqembi did not have a significant effect on NfL values.

For Neurogranin, only reported by Biogen/Eisai, Leqembi’s reduction at three months is approximately 12.5%, and at six months approximately 25%.

Biogen/Eisai did not report numbers for Ptau-217.

Exercise relating cognition to biomarkers for donanemab (Kisunla)

Placebo decline on CDR-SB for donanemab in the total patient population was about 0.5 points at 3 months, about 0.8 points at 6 months, and 2.42 points at 18 months. At that time, Donanemab patients’ CDR-SB scores had increased by 1.72 points, a difference of 0.70 points to placebo.

I set out in this blog post what the estimated trajectory for NfL, ptau-217 and Neurogranin was for patients on Kisunla.

For plasma NfL, no treatment difference was observed at Week 76, and an increase was observed in the donanemab arm at weeks 12 and 24.

P-tau217 was significantly reduced from baseline with donanemab treatment compared with placebo in the low/medium tau and combined population. The difference vs placebo was −0.25 in the low/medium tau population and −0.22 in the combined population at 76 weeks. At 3 months, the decrease is approximately ~13%. At 6 months, the decrease was approximately ~30%. 

I did not find data for neurogranin (seemingly not in the publication, and neither in the briefing document to the FDA advisory committee).

For the record: a note on Anavex 2-73

For Anavex 2-73, according to the publication, the secondary endpoint CDR-SB was significantly improved for blarcamesine group vs. placebo (difference of − 0.483), representing a 27.6% reduction in clinical decline at 48 weeks.

A mean change from baseline to week 48 of plasma NfL of 1.65 pg/ml versus 4.92 pg/ml was reported. The mean difference versus placebo was 3.28 pg/ml. Taking into account that plasma NfL values are about 132 pg/ml (data from the TIMC-Brain cohort) for amyloid positive patients, I believe that change is not meaningful.

Exercise relating biomarkers to estimated cognition for XPro

Reported biomarkers from the Phase 1b trial to be taken into account here are NfL, ptau-217 and Neurogranin. These three are leading biomarkers recognized by the field, which correlate best to cognition.

The percentage reductions seen at three months for these three biomarkers, for the 1 mg/kg dose of XPro, were respectively: -84% (NfL), -46% (ptau-217) and -56% (Neurogranin).

On that basis, we can start making an estimation of what cognition would look like.

As a reminder, for estimated placebo decline, we use the data from the Lecanemab trial.

	NfL	Ptau-217	Neurogranin
Xpro (at 3 months)	-84%	-46%	-56%
Leqembi (at 3 months)	No change.	/ (did report on a less precise ptau-measure)	-12.5%
Leqembi (at 6 months)	No change.	/ (did report on a less precise ptau-measure)	-25%
Kisunla (at 3 months)	No meaningful change.	-13%	/
Kisunla (at 6 months)	No meaningful change.	-30%	/
Anavex 2-73 (at 11 months)	No meaningful change.	/ (did report on less precise ptau-measures)	/

 - For NfL, both Leqembi and Kisunla reported no change at 18 months, whereas XPro treatment reported a 84% reduction at three months. Plasma NfL levels in AD patients are 132 pg/ml. Plasma NfL levels in AD patients are about 35% higher than in normal patients. XPro’s NfL Reduction by −84% would drastically lower NfL levels, suggesting a strong neuroprotective effect and a high likelihood of cognitive improvement.

Leqembi and Kisunla achieve a 0.576-point average CDR-SB difference at 18 months with NfL unchanged. To compare the relative effect, I cannot divide XPro’s NfL reduction by that of Leqembi and Kisunla because there is none. But XPro’s NfL measurements are 6.25x lower than those of Leqembi and Kisunla. As Leqembi and Kisunla have achieved a 0.576 difference to placebo, and NfL measurements are 6.25 times stronger for XPro, then XPro should see a 3.6-point difference to placebo on the CDR-SB scale at 18 months (which would be a cognitive improvement of about 2 points taking into account the Leqembi placebo decline, and 1.3 points taking into account the Kisunla placebo decline). At 6 months, that difference to placebo would be about 1.2 points for treatment on Xpro versus a 0.6 or 0.8 decline of placebo seen in the Leqembi and Kisunla trials, or a cognitive benefit of 0.6 points taking into account the Leqembi placebo decline and 0.4 points taking into account the Kisunla placebo decline.

- Leqembi’s -25% neurogranin reduction at 6 months corresponds to a 0.451-point CDR-SB difference at 18 months (placebo: 1.66, Leqembi: 1.22). XPro had a −56% neurogranin reduction at 3 months. To compare the relative effect, I divided XPro’s neurogranin reduction by Leqembi’s (56/25=2.24) and doubled that given the 3 month time frame: 4.48. This means XPro’s Neurogranin reduction at 3 months is 4.48 times greater than Leqembi’s at the same time point.

On that basis, an estimated 18-month effect of Xpro would be 0.451 × 4.48 = approximately a ~2.02-point difference to placebo on CDR-SB. As a reminder, placebo decline in the Leqembi trial was 1.66 points. At three months, that would lead to about a 0.337-point difference to placebo (2.02 ÷ 6). At 6 months, that would lead to about a 0.673-point difference (2.02 ÷ 3), with placebo at a 0.6 point decline and XPro at a ~0.073 point improvement.

- For donanemab, a 30% p-tau-217 reduction at 6 months corresponds to a 0.70-point CDR-SB difference at 18 months (placebo: 2.42, donanemab: 1.72). XPro showed a −46% p-tau-217 reduction at 3 months. Again calculating the relative effect, XPro’s reduction of ptau-217 at three months is about 3.06 times greater than Kisunla [(1.53x donanemab’s 6-month reduction, 46 ÷ 30) x 2 given the halving of the treatment period].

On that basis, an estimated 18-month effect would be 0.70 × 3.06 = approximately a ~2.142-point difference to placebo on CDR-SB. Placebo decline in the Kisunla trial was 2.3 points. At 3 months, that would lead to about a 0.357-point difference to placebo (2.142 ÷ 6). At six months, that would lead to about a 0.714-point difference to placebo (2.142 ÷ 3), with placebo at a 0.8 point decline and XPro at a ~0.086 decline (or almost stabilization).

Combining the data

Combining the projected effects of these biomarkers may give better insight in potential efficacy. If we were to combine all three biomarkers, and try to calculate cognitive efficacy at 6 months, we would end up with a 1.187 point difference to placebo decline of about at six months [(1.2+0.073-0.086)/3=1.187 points)]. Placebo in the Leqembi trial declined by about 0.6 points after six months. Therefore, combined data for NfL, Neurogranin and ptau-217 may lead to improvement in cognition.

Conclusion: improvement or stabilization of cognition may be possible

The conclusion is that stabilization of cognition may be possible. That would align with anecdotal evidence that saw drastic improvements in cognition (33% of all patients on 1 mg/kg dose after four years, so certainly not irrelevant). It would also align with the ambitious goal INmune has set.

As I mentioned above, this is a theoretical exercise based on three reported biomarkers. From that perspective, the exercise is logically fallible. But it offers perspective, and is based on the three leading biomarkers which have been shown to correlate with cognition. Two out of three factors indicate stabilization of cognition. One of them indicates strong improvement, which is also the biomarker that led to accelerated approval of Qalsody (without the related trial showing any efficacy), and which remains unaffected by currently approved anti-amyloid antibodies.

This outcome is realistic. There is no indication that patients in the Phase 1b trial saw better results than what can be expected in the MINDful trial. These biomarkers are just some of many that have been reported. The difference is that the others that have been reported are not (so clearly) linked to cognitive results, or are too novel to have been reported by other companies making comparison impossible.

The fact that the Phase 1b trial included patients at the further and more severe stages of disease, a patient population which is generally considered too far gone to recover, lends credence to the above. It is likely to see better efficacy in a patient population which has less cognitive deterioration.

A potential synergistic effect of the combination of these biomarkers is also plausible, but has been disregarded for the exercise.

A caveat is that the trajectory may not be linear. But the cognitive trajectory of both placebo and treatment arms for the anti-amyloid therapies indicates otherwise.

Finally, anti-amyloid antibodies do not change NfL. Some say they may not alter neurodegeneration. The absence of NfL change could be indicative in that regard, and as such it is worrying. The change in ptau-217 may be purely related to removing amyloid from the brain. Many scientists are worried about the brain volume loss, apparently even accelerated brain volume loss, that coincides with anti-amyloid antibodies, and the lack of data that have been published in this regard. Treatment of XPro led to an increase in brain volume.

  

Efficacy starts with similarity to anti-amyloid antibodies

Stabilization of cognition is not necessary to conquer this multi-billion market. XPro is safe and easy to administer, and therefore if it performs similar to anti-amyloid antibodies, it should be able to at least compete with them or allow for (synergistic) combination therapy. On the basis of the above, however, efficacy will likely be higher than what has been reported by the two successful anti-amyloid antibodies.

Anti-amyloid antibodies are unsafe and require intravenous infusion. A safe and easy to administer therapy would be a game-changer. And that would translate to a valuation revision.

Biogen’s stock added 40% when Leqembi’s Phase 3 results came out in 2022, adding $11 billion in market value. Lilly’s stock rose by about $40 billion in a week’s time after the company had announced Kisunla’s Phase 2 results.

 

To those in disbelief

Now, some may say that other (non-selective) anti-inflammatories have failed, and therefore XPro will fail too. That’s pretty much the short thesis right now from what I understand. They disregard the selective profile of XPro, modulating glial activity in a different manner, which is crucial. I would refer to the previous blog post for that, but it basically disregards established science and everything INmune is built on insofar as it relates to XPro.

Furthermore, following that same reasoning, Leqembi and Kisunla should have also failed, after the numerous earlier failures of anti-amyloid therapy candidates. But they did not.

Confirming metrics

Furthermore, without reiterating the entire biomarker reporting INmune Bio has done across several press releases, webinars, poster and oral presentations, I want to point out the following.

- INmune Bio has performed testing with five different tools from five different companies on patients in the phase 1b trial. All of them showed improvements of metrics.

- INmune Bio used biomarkers from five different companies which all showed that XPro treatment led to changes in the brain: Imeka biomarkers showed white matter changes, Oxford brain Diagnostics showed grey matter changes, Proteome Sciences showed proteomic data, the Olink cytokine panel assay showed cytokines in the CSF with every detectable datapoint showing a reduction with the exception of interferon gamma (flat), and further biomarkers came from the Roche Neurotoolkit.

- In the cerebral spinal fluid, the 15% reduction of inflammatory markers as measured by the Olink panel showed reduction of all but one inflammatory biomarkers, insofar as detectable (IFNG was flat).

- Measurements from Imeka, such as  white matter, apparent fiber density and tissue radial diffusivity kept improving over the course of twelve months of treatment. All are relevant to AD. Free-water differentiates NC, MCI, and AD patients and is associated with global cognition, apparent fiber density rent differentiates normal controls, MCI, and AD patients, is associated with episodic memory and with plasma biomarkers of axonal degeneration, and tissue radial diffusivity in the fornix is predictive of conversion to AD.

- INmune Bio has correlated white matter and grey matter measurements from treatment with changes in cognitive function;

- GFAP was reduced in the Phase 1b trial, expects to see it reduced in the MINDful trial, and hopes that this decrease correlates with clinical response in the Phase II trial.

Conclusion

Biomarkers should correlate with cognitive efficacy.

Next to several other biomarkers, INmune Bio has reported on three biomarkers which the field considers as leading and correlated to cognitive efficacy. Some of these biomarkers have also been reported by Biogen/Eisai and Eli Lilly for approved anti-amyloid antibody therapies Leqembi and Kisunla.

I have tried to use data from the trials for Leqembi and Kisunla, looking at their efficacy versus placebo and the reported biomarkers, to make an estimation of what results for XPro at six months could look like. The combined analyses lead me to conclude that XPro may be able to report improvement of cognition in MCI and mild AD patients. Stabilization of cognition should logically be possible as well. This exercise is theoretical and disregards, among others, potential synergistic effects and the fact that the Phase 1b trial treated patients at the further end of the disease spectrum. It is however useful for indicative purposes; biomarkers should correlate with clinical efficacy. A pooled analysis of biomarkers should better reflect the potential effect on cognition.

These biomarkers are not standalone; they are corroborated by many other biomarkers that INmune Bio has reported over the past years, from tests from several different companies.

Whereas cognitive improvement or stabilization may be possible, success in the MINDful trial actually requires much less efficacy. XPro is safe and easy to administer; if efficacy would be in line with what has been reported by Biogen/Eisai and Eli Lilly, namely about 30% slowing of cognitive decline, there should be a place on the market for XPro next to or in replacement of the anti-amyloid antibodies. That should translate to valuation that is more in line with XPro’s potential. Right now the stock is fundamentally mispriced.

But I’ll leave that to another blog post…if time is on my side, with results expected in the second half of June 2025.

---

INmune Investors Server Discord: https://discord.gg/JEA8r7wCGY