2025
From NP Summer 2023 Issue
STROKE
The tide may finally be turning for drug development in stroke, a relatively recent, and dramatic departure from what had been a nearly complete abdication of responsibility by the industry regarding this substantial, undertreated area. More than in any other area of neuroscience, this revival of interest tends, with a few exceptions, to involve a combination of drugs and devices. Mechanical thrombectomy has revolutionized ischemic stroke therapy, and now pharmaceutical treatments are seen as potentially building upon that success. To be clear, Big Pharma remains largely on the sidelines, but its more opportunistic members are now looking more closely at stroke.
The Societal Scope
Stroke is third only to Alzheimer’s and Parkinson’s in its scale as a neurodegenerative disease, and is a leading cause of disability in adults. The population continues to gray, and estimates regarding the annual incidence of stroke in the US are approaching 800,000 cases, about 610,000 of these cases constitute ‘new’ stroke patients. Worldwide, an estimated 13.7 million individuals have an ischemic stroke annually (2016). In the US, at least two percent of individuals 80 years of age or older suffer a stroke, and being black, male, and/or hypertensive adds to the risk. 140,000 deaths in the US are linked to stroke each year. While approximately 80% of those who survive their stroke return home by the time a month has passed, they often need continued nursing care, their level of impairment presenting an enormous burden for them and their families. The economic burden of stroke in the US alone was estimated (by CDC) in 2022 to be $56 billion, a toll growing as the size of the aged population continues to expand. Contrary to common belief, stroke is not solely the province of the aged; a Dutch study found that 10-15% of first-occurrence stroke patients were in the age 18-50 category. For such younger patients, any resulting disability has a duration measured in decades.
Impact on the Individual
The impact of a stroke can range from transient to disabling to fatal, depending on the area, extent, and duration of the infarct. A stroke with widespread impact can leave an individual in a vegetative state, while focal strokes leave damage that varies by location. The impact of a stroke also depends upon the extent to which the stroke infarct extends into the white matter of the brain. When language centers are destroyed, the result is an aphasic deficit, compromising or eliminating a patient’s ability to express and/or comprehend verbal communication. Damage to the ‘motor strip’ leads to disability in contralateral motor functions. Destruction of some areas of the parietal area can lead to prosopagnosia, the inability to recognize the faces of family and friends, or the inability to recognize that one’s limbs are one’s own. At the seemingly most innocuous end of the stroke spectrum, transient ischemic attacks (T.I.A.s) appear ephemeral in terms of their duration, but they provide a warning sign flagging heightened vulnerability: One-third of individuals who experience a T.I.A. will have an ischemic stroke within the following twelve months. With 240,000 Americans estimated to experience a T.I.A. each year, that means it is a precursor to 80,000 ischemic strokes. Furthermore, recent work at University of Alabama has linked T.I.A.s to as steep a decline in cognitive capabilities over the ensuing five years as is seen with ischemic strokes themselves. As a University of Calgary researcher noted in the same issue of JAMA Neurology, this means that a T.I.A. is not really transient, but sparks a steepening trajectory of impairment that is not yet understood.
The Neuropathology
There are two main types of stroke; hemorrhagic and ischemic. Hemorrhagic stroke involves arterial rupture, causing a loss of blood supply to those areas downstream of the hemorrhage, while also initiating destructive compression effects and the ‘corrosive,’ inflammatory effect of blood within the brain parenchyma, no longer shielded by endothelial cells. We are only going to consider this category in a relatively limited fashion; it applies to about 13% of strokes, and diverges significantly from the neuropharmaceutical emphases of prospective treatments for ischemic stroke. Hemorrhage is pertinent to the treatment of ischemic stroke in terms of differential diagnosis: The presence or absence of hemorrhage dictates the parameters of what can be without risking the exacerbation of cerebral bleeding. It is partly the necessity of using CT scans to rule out hemorrhage that restricts the use of tPA to a small minority of the stroke population, even with attempts to accelerate stroke detection and treatment. Hemorrhage is also a limiting factor in another way: Using t-PA beyond its 4.5 hour post-stroke window risks hemorrhagic transformation, due to inflammation at the thrombolytic locus, producing iatrogenic hemorrhage. This had historically limited the use of tPA to 3-5% of the ischemic stroke population, although improvements in acute stroke care have increased that percentage to the 15-20% range. While mechanical thrombectomy has become an option for some cases, that is an expensive, skill-dependent procedure, generally available only in urban areas.
“Time Is Brain” (Goyal et al 2016)
During an ischemic stroke, a silent event rapidly escalates into an acute crisis. The degree to which perfusion is disrupted varies from stroke to stroke, but if more than 80% of the blood supply to a region is interrupted for more than a few minutes, that area dies. Such primary stroke damage is currently unavoidable, since most stroke victims are not triaged and treated until several hours have elapsed after the stroke onset. But inevitability is not necessarily the case for secondary stroke damage, the progressive expansion of cell death that occurs in the penumbra around the core of dead tissue: It is in the penumbra that the battle for survival is waged against pathological forces, where clot-busters and neuroprotective drugs seek to have effect. In animal models, neuron loss can be greatly reduced via neuroprotective agents, with significant benefit for retained brain function. This is of little consolation to the rodents subjected to the procedure, and has thus far been of equally little relevance to the human condition. In trial after human trial, compounds have failed to show efficacy after having seemingly shown promise in mouse studies.
The legacy of stroke pharmacotherapy has remained one of consistent failure and disappointment, though very few have ventured into clinical testing in recent years. In contrast, we have seen a wave of improved mechanical clot-retrievers establish themselves as the standard of care for those who qualify in terms of stroke location, size, and speed of access to the necessary expertise. Unfortunately, these factors reduce the patient candidate population considerably: One estimate is that, even with broadened inclusion criteria, 50,000 of the 800,000 Americans who suffer an ischemic stroke each year might be endovascular therapy candidates.
While some early trials showed negative or equivocal results, as treatment protocols and rapidity of intervention improved, so did the results. A number of large-scale clinical trials have shown endovascular therapy produces impressive results in the patients who qualify: In a 500pt Dutch study, 25% of patients receiving tPA alone regained independent functioning, compared with 33% of those receiving both tPA and endovascular therapy. An international study was terminated when an assessment of the first 200 patients found that, while 35% of tPA patients regained independent functioning, 60% of the tPA/endovascular combo group did. A 206pt study showed 44% of the endovascular group achieved functional independence 90 days later, compared with 28% of the tPA arm. The issue now is not whether endovascular treatment works, but how to broaden the range of stroke patients eligible to receive it, largely by accelerating triage and treatment as much as possible–the standard for endovascular care is thrombectomy within 6-8 hours of the stroke. A 2023 paper in Lancet by Hankey concluded that use of tPA prior to thrombectomy does not add therapeutic benefit compared with thrombectomy alone.
The process by which damage expands from its initial locus has been the key target for most therapeutic development programs: In laboratory studies, compounds can reduce the size of the final infarct by 70-90%, although animal models of ischemic stroke have been dreadfully poor predictors of clinical outcomes. Ironically, the best results were in reperfusion models that do not emulate durable ischemia, but now patients having a mechanical thrombectomy are indeed reperfused. This offers a fresh opportunity to document clinical neuroprotection that mimics the animal models, and has raised interest in combining neuroprotection with thrombectomy. The downside is that this limits the prospective patient population to the 40-50,000 fortunate cases who qualify for thrombectomy, about 5% of the ischemic stroke population.
Issues for Treatment
The mechanical removal of obstructions created a new gold standard for a small subset of the stroke population. Eventually, optimized stroke therapy will be polymodal, and clinical trials are changing accordingly.
Some of the principles governing effective pharmacotherapy are consistent across-the-board, the treatment of ischemic stroke often (when neuroprotection is the goal) adds a critical timing component.
1) Target area access: The fact that the blockage prevents oxygenated blood from reaching infarcted tissue also means that a drug carried within the bloodstream will also not reach that area easily, if at all.
2) Access in time to provide cellular rescue: This refers to the elusive window wherein a drug must be delivered in order to have effect. The original three hour window for tPA, within which a patient must be identified as possibly having a stroke, transported to an ER, and scanned for hemorrhage, for many years made tPA a therapy in theory more than one in practice. Published recommendations expanded the window to 4.5 hours, and real-world practice sometimes stretches that window slightly further, but with the risk of triggering hemorrhage. A small subset of patients can be identified by advanced imaging who can benefit even nine hours post-stroke, but that does not apply to the vast majority of stroke patients. The window for endovascular therapy is framed more loosely, with usual practice emphasizing a six-eight hour window, but a subset of patients may benefit from endovascular treatment 12 or even 24 hours post-stroke.
3) Selecting patients who have a penumbra that is not yet so damaged as to be doomed: Salvageability is the key to neuroprotection. Efficacy data can be compromised in many ways, but two factors of immediate relevance are the inclusion of mild stroke patients likely to recover no matter what was , or the inclusion of those without a rescuable penumbra, for whom there is nothing that can be . The current state-of-the-art strategy for identifying patients who might benefit from neuroprotective rescue uses imaging to compare the volume of tissue with impaired perfusion vs. the volume of tissue with a major loss of metabolic activity: The difference between the area of diminished perfusion and the area of metabolic failure constitutes the zone whose fate hangs in the balance.
These principles hold true for neuroprotective efforts, but there has been some interest in regeneration, where the paradigm is completely different. Rather than emphasizing speed and rescue, neural repair strategies are predicated upon waiting until the initial destructive onslaught has ebbed, so that growth-inducing measures then have a shot at repairing and rebuilding in the aftermath.
Biomarkers
The ability to definitively establish the presence of stroke, and even more importantly, to discriminate ischemic from hemorrhagic stroke without requiring imaging, would be an important forward step in stroke treatment, To this point, no diagnostic panels have established themselves in clinical use. At present, the biomarkers most commonly considered potentially relevant to the assessment of diagnosis and prognosis are, as was summarized by Dias et al (2022): Tau (for neuronal damage), neurofilament light chain (axonal damage), and GFAP and S100-beta (astroglial damage).
In an early attempt, Proteome Sciences licensed stroke biomarkers from University of Geneva, a panel of five plasma protein markers they claimed had 90% sensitivity and specificity (albeit from a tiny sample of ten patients) in identifying those who have suffered stroke: GSTP, CRP, NDKA constituted a first panel of proteins with which patients with symptoms not due to a stroke could be triaged out. A second panel, combining GSTP with VCAM and IL-1, was claimed to confirm the presence of stroke with 96% specificity, 86% sensitivity–but did not address the issue of hemorrhage risk, which minimizes the panel’s potential value. Eventually, Proteome signed a licensing agreement with Galaxy CCRO covering the Proteome IP for the stroke biomarkers, focused on the development of a bedside blood test that uses GSTP as a single marker differentiating ischemic from hemorrhagic stroke, zeroing in on clotting/platelet dysfunction. Six years later, they have not made tangible progress towards a clinically-usable test.
Other biomarker models have produced a more complex array of biomarker changes. A Canadian group (Penn et al) compared levels of 147 already-published plasma protein markers in a sample of 20 acute non-hemorrhagic stroke patients compared with 20 control subjects. Thirty proteins were significantly divergent between the two groups, 47% of them were neuroinflammation markers, another 40% were associated with coagulation.
A Korean group (Lee et al, 2020) identified eleven proteins linked to ischemic stroke, including two coagulation factors, FN1 and VTN. Complement c1 is also one of several other potential markers, with elevations associated with ischemia.
A UC Davis group published a report in 2023 that cited changes in gene expression patterns (peripheral blood samples) at various time points (e.g. 0-24 hours, 24-8 hours) post-stroke. The patterns pointed to transitions in immunomodulatory functions over time, varying with the type of stroke, with both activator and suppressor genes playing a role. Upregulated IL-4 and IL-8 expression was noted across all timepoints.
Changing the Context
The legacy of stroke pharmacotherapy has centered on the repeated failure of neuroprotectant candidates to provide therapeutic benefit, but as Fisher and Savitz noted in their 2022 overview (Nature Reviews Neurology), these failures may have been due to the challenge of protecting cells that remain occluded by a clot, without benefit of reperfusion. Reperfusion had been available only to the 3-5% of AIS patients who were evaluated soon enough to be administered the clot-buster/thrombolytic, t-PA. Then, for the past fifteen years, those suitable for mechanical thrombectomy via intra-arterial devices could be thereby reperfused, but the thrombectomy eligible population constituted only about 10% of the hundreds of thousands of American patients with AIS (Acute Ischemic Stroke), only half of those actually receive thrombectomy. But this has begun to approach a level of access that could permit the testing of protectant drugs requiring reperfusion to be successful.
There are three pharmacotherapy developments en route that could further transform the therapeutic calculus for AIS: One would dramatically expand the proportion of the AIS population for whom reperfusion can be offered, the second is a neuroprotectant that may further improve the likelihood that areas of brain whose survival hangs in the balance may be rescuable. The third is the NIH sponsored SPAN (Stroke Preclinical Assessment Network) program that is revisiting some failed neuroprotectant candidates that may have had their value hidden by the lack of reperfusion. There were five drugs evaluated by SPAN in ‘SPAN 1.0’, going through rigorous, systematized preclinical protocols to see if they warranted a return to the clinical stage (fingolimod, fasudil, tocilizumab, uric acid, velparib). Uric acid has been reported by SPAN to have shown evidence of efficacy at multiple steps of the animal model vetting, none of the other four candidates made it that far.
Span 2.0 is now underway, following the same preclinical assessment protocol: A GSK FAK-inhibitor that is linked to deleterious IL-6 signalling (GSK2256098); a GSK epoxide-hydrolase inhibitor (GSK2256294) that slows the breakdown of protective arachidonic acid metabolites, yielding anti-inflammatory and antioxidant effects; BPN-27332, which targets the enzyme ALOX15; Neurexis‘ tatCN19o, a CaMKII inhibitor; and NanO2, which increases blood oxygenation and showed a positive signal in a 24pt pilot study.
Founded by Australia’s George Institute and the University of Calgary, the ACT-Global Platform Study should be mentioned in this context. It is reminiscent of the MGH Healey Center Platform study program in ALS, in that it establishes systematized, simultaneous evaluations of treatment interventions, though ACT-Global takes a Bayesian adaptive approach to assessing various domains of post-stroke treatment, including thrombectomy, thrombolysis, and neuroprotection. It is also open-label, in contrast to the Healey Center’s shared control group model.
The Therapeutic Targets
The damage from a stroke unfolds over time, sometimes euphemistically referred to as the ischemic ‘cascade’: The loss of oxygenation triggers cellular destabilization, beginning with the overrelease of glutamate, which accelerates the entry of calcium into postsynaptic neurons. This in turn disrupts anaerobic energy systems, preventing neurons from pumping out calcium in order to restore homeostasis. Sodium and potassium ion channels also open, adding to the toxic overrelease of excitatory neurotransmitters. All of this constitutes a domino effect of depolarization within the cell and its neighbors, leading to an avalanche of radical oxygen species. Free radicals cause damage to DNA, undermine mitochondrial energy systems, and stimulate both necrotic and apoptotic mechanisms. Inflammation ensues, and the process culminates in a wave of neuron dysfunction and death.
These steps offer, at least in theory, a plethora of targets, but there has been no clarity as to how early in this sequence intervention must occur to make a difference. Alternatively, one can dispense with damage prevention, in which case the question turns to–what is the optimal time post-injury for reparative efforts to take place? There is a period of time where residual destructive forces may still be salient, and could short-circuit repair efforts. Yet it also makes intuitive sense that there would be a post-stroke period of plasticity during which repair might be more likely, before changes in structure are ‘set in stone.’ In the following discussion of drug therapy efforts, these targets are ordered roughly from ‘upstream’ to ‘downstream’, from early to relatively later in the sequence of events.
NXY-059, the free radical scavenger developed by Centaur (acquired by Renovis, later acquired by Evotec), partnered with AstraZeneca, that caused the plug to be, for all practical purposes, pulled on stroke therapeutics for the next two decades. AstraZeneca spent more than $200 million on a program wherein the first Phase III trial enrolled 1700 patients, and showed efficacy on a revised version of the Rankin. However, even these results showed that the drug helped less than 10% of patients receiving it. In the forlorn hope that scale would amplify a faint signal, AstraZeneca then expanded the second Phase III trial to 3200 patients. When completed in 2006, the topline results showed a complete washout. The failure led an AstraZeneca researcher to declare that the concept of neuroprotection for stroke was dead, and the area of stroke remained moribund for Big Pharma, although that may now be changing.
Thrombolytic
Basic ‘clot-busting’ and reperfusion is the raison d’être for recombinant tPA, a class of drug with three members, two relevant to stroke in the US: Alteplase has been approved in the US for ischemic stroke, TNK (tenecteplase) was initially approved for myocardial infarction, but had increased use in AIS off-label, before finally being approved for AIS in early 2025.
In Phase III, tPA was given within 90 minutes of the stroke, and patients receiving tPA were 30% more likely to show full recovery at three month followup. Originally, tPA was labeled for use within three hours of the stroke, and the precise time of stroke is often not known. In hemorrhagic stroke, tPA worsens bleeding and could be fatal, thus patients must have a CT scan first, further shrinking the available treatment window. For most stroke patients, tPA is moot: Only one-third of stroke patients reach the hospital fast enough to even be assessed for tPA usability, and the percentage of patients that actually receive the drug had been variously reported as being between 3-5%. Many stroke units have switched to TNK (tenecteplase) because of the greater ease-of-use (a single bolus) and more rapid thrombolysis. Thanks to vastly improved coordination of stroke treatment, estimates of thrombolytic use now range as high as 15-20%.
There was a long debate regarding the possibility of extending this narrow time window, offset by the increased risk of intracerebral bleeding. An Australian group completed a small trial looking at the use of tPA with a nine hour window, with imaging to identify those with a rescuable penumbra, and found positive trends, but the downside risk of hemorrhagic transformation has kept real-world expansion to a 4.5 hour window in Primary Stroke Centers (PSC), whose specialization affords more comfort with this risk-benefit calculus.
The history of clot-busters that sought to improve on tPA has been dismal. Failures included Knoll’s spider venom derivative, dubbed Ancrod, which failed in a large Phase III. Neurobiological Technologies took on Ancrod, renamed it Viprinex, but their trial also came up empty. DSPA (desmoteplase) was a recombinant version of the anticlotting agent found in vampire bat saliva. DSPA went through a torturous marathon of clinical development, first by Bayer, then PAION (with and without Forest), and eventually by Lundbeck. A Phase III failure finally led to the discontinuation of DSPA’s development, more than a decade after it began.
TMS-007 was developed by Japan’s TMS. TMS-007 is a tissue plasminogen activator, but it has an additional anti-inflammatory component, which appears to prevent hemorrhagic transformation that otherwise risks intracerebral bleeding. The 90 patient Japanese PhIIa involved three dose-levels (1, 3 and 6 mg/kg); 52 patients who received TMS-007 and 38 patients who received placebo. The study included patients with acute ischemic stroke within 12 hours after onset and ineligible for tissue Plasminogen Activator (tPA) or thrombectomy. The primary endpoint was evaluation of safety with secondary endpoints included reperfusion and clinical outcomes at 90 days. At 90 days, significant improvement was reported in functional independence; 40 percent of TMS-007 patients achieving scores of 0 or 1 on the modified Rankin Scale, indicating either no residual symptoms or no significant disability, compared to 18 percent of patients who received placebo. Imaging indicated reperfusion in 58.3 percent of patients who received TMS-007 compared to 26.7 percent for patients who received placebo.
Biogen took a pre-PhIIa option in 2018, paying $4 million, and then exercised it in 2020, when the TMS PhIIa data was released, paying another $18 million. The potential milestones cited in that deal totaled $355 million. It was not until early 2023 that a 760pt PhII had been designed. However, the trial was first delayed, then cancelled entirely by Biogen. While this could be attributed to cost-cutting and a shift of corporate focus, it would also be admittedly challenging to run a successful trial for a novel thrombolytic versus the current available options. At this point, unless a new thrombolytic were shown to achieve more complete clot clearance, it would be unlikely to achieve much uptake given the two that already approved.
Lumosa Therapeutics: LT3001/oldatretide is a fusion-product that combines a novel thrombolytic with reduced hemorrhagic effects, and believed to have better clot-busting impact, along with a free radical scavenger to attenuate reperfusion injury. A pilot study, using a 24hr treatment window, showed evidence of improved functional outcomes. A PhII assessed LT3001 in conjunction with thrombectomy, but PhIII will be in patients without access to thrombectomy. Their partner in China (Lumosa is Taiwan-based), Shanghai Pharmaceuticals, is running a 300pt PhII as well, to finish late this year.
Acticor: Glenzocimab is a monoclonal antibody targeting glycoprotein VI platelets, which was tried as an adjunct to tPA and/or thrombectomy, within the tPA 4.5 hour window in Phase II. The Phase II trial was a placebo-controlled trial that enrolled 106 patients. The primary outcome was the proportion of patients with intracranial hemorrhages during the first 24 hours, and the glenzocimab groups altogether had a 0% incidence of symptomatic hemorrhage, less than the 10% in the placebo arm. Imaging showed asymptomatic hemorrhage was also lower (31%) than 48% in the placebo group. There were positive efficacy trends on the NIHSS and Rankin, numerically fewer glenzocimab patients died or suffered severe disability, though statistical significance was not achieved.
The Company then ran a PhII/III study that enrolled 436 patients, and was unsuccessful in reducing poor outcomes. A second PhII/III in thrombectomy patients was halted after a futility analysis of the first 78 enrollees. The program has demised.
There are a few other legacy programs that somehow survived. Pro-urokinase is a thrombolytic originally developed by Abbott that, in spite of the use of intra-arterial administration and concerns about bleeding risks, was not dropped until it reached Phase III. Active interest has continued in China, where IV-administered Pro-urokinase was compared by Tasly Pharmaceuticals to alteplase in a 663pt study, Pro-urokinase achieved the same benefit with less bleeding risk. But while it has received acceptance in China, it does not appear that it will go through further development in the US.
The thrombin-inhibitor argatroban, approved in Japan for stroke, was associated with increased intracranial bleeding in a monotherapy trial. But a 90pt trial run by a University of Texas group tested it in conjunction with tPA, and reported the addition of argatroban did not increase the frequency of hemorrhage, and the trends were in favor of argatroban improving outcome. The NINDS funded a 514pt study run by University of Cincinnati, argatroban was one of the two treatment arms. The other arm utilized eptifibatide, a platelet aggregation inhibitor. The results showed that neither drug improved clinical outcome, and argatroban was associated with a tripled risk of death, leading to an early termination of the trial. That should be the end of the road for argatroban for stroke in the US.
Quelling Excitotoxicity
During the 1990’s, stemming the excitotoxic tide was considered the wave of the future for stroke treatment, based on the hope that the impact of glutamate overrelease could be ameliorated. This search veered into a molecular minefield: Merck’s MK-801 was a competitive NMDA blocker, which unfortunately, in addition to its ability to reduce infarct size, caused brain vacuoles. The parade of failures killed off more than a dozen programs. Magnesium infusion (which has effects at the NMDA receptor) had been associated with decreased mortality in a pilot study. NINDS eventually ran a 1700 pt PhIII trial, ironically titled FAST-MAG. The study was exceptionally slow to enroll, and when it finished eight years later, it came up completely empty.
NoNO: One of the neuroprotective, clinical-stage drug NCE candidates left standing for ischemic stroke was NoNO’s nerinetide, an IV-administered peptide that blocks PSD-95, co-localized with NMDA receptors, from activating downstream neurotoxic pathways (e.g. nitric oxide) otherwise triggered by NMDA receptor hyperactivation. In 2020, NoNO initiated a Phase III trial aimed at enrolling 1120 AIS patients, with nerinetide administered via infusion. That trial showed nerinetide missed its primary endpoint, but patients who did not first receive tPA showed a 40% reduction in mortality. As it turns out, tPA and nerinetide have a negative chemical interaction, tPA deactivating nerinetide. NoNO then initiated two PhIII studies, one utilizing standard ER triage protocols, the other a Phase III where nerinetide would be given first, by EMTs. Nerinetide failed in the PhIII where standard ER triage was , which had delayed treatment onset. But when administered by paramedics in the ambulance, the average delay from first symptom to intervention was 1hr. In this subgroup, the bottom line was clinically and statistically meaningful: a 30-40% better chance of zero or minimal longterm impairment.
However, the need to either be administered very soon after the infarct develops, or withheld from patients who are receiving a thrombolytic (testing for compatibility with tenecteplase would be time-consuming and a drain on limited resources) sharply diminishes the potential value of the drug. NoNO has thus begun clinical testing of NoNO-42, the d-enantiomer of nerinetide, as a next-generation PSD-95 antagonist that does not create a problematic interaction with tPA/alteplase, opening up the full gamut of stroke patients to treatment and improving access to the penumbra. NoNO-42 successfully went through PhI testing and has entered PhIIb, under the auspices of the ACT-Global trial discussed earlier.
Avilex Pharma is a University of Copenhagen spinout also targeting PSD-95, in the service of reducing excitotoxicity and oxidative stress. AVLX-144 is a peptide that is reportedly 1000X more potent than nerinetide, more effective at reducing infarct sizes in mouse models, and with better BBB penetrance. In 2021, Simcere partnered AVLX-144 for China, making an undisclosed upfront payment which, along with undisclosed milestones, could total up to US$175 million. The PhI was not completed until mid-2023, and there has been no announcement around PhII timing.
Selective NR2B subtype inhibitors developed at Emory were licensed to the startup NeurOp, animal data showing that infusing the prototype compound prior to arterial occlusion reduced infarct volume. NeurOp developed NR2B antagonists designed to be active only in the acidified environment found post-stroke, and NP10679 entered PhI, to be aimed at subarachnoid hemorrhage, and apparently completed two PhI studies. But there has been no news since 2021, their viability is in question.
Argenica Therapeutics is the commercial spin-off of work at the University of Western Australia that created ARG-007, which uses glutamatergic inhibition to reduce excitotoxic calcium influx. Argenica claims downstream antioxidant effects and mitochondrial protection as well. No serious adverse effects were reported from PhI. ARG-007 has fully enrolled (n=92) PhII, treating patients with large vessel occlusion strokes and received thrombectomy. Topline results should be available during 3Q:25.
Aarhus University is running a trial of Novo Nordisk‘s semaglutide in stroke, as part of the effort to see if semaglutide may be applicable to the full range of human diseases, and based on the reduction of stroke incidence (by 39%) in diabetes studies. The GLP-1 agonist is believed neuroprotective, this 380pt study in acute stroke should finish in late 2025.
A group from University of Queensland identified a peptide inhibitor of ASIC1a (acid-sensing ion channel 1a) in the venom of the Hadronyche infensa spider. In a mouse model, icv injection of the Hi1a peptide delivered within eight hours of onset reduced infarct volume by 65%. However, the group shifted to using this mechanism in heart disease.
Anti-Inflammatory
There is a ‘cytokine cascade’ of inflammatory factors that is precipitated in stroke: Interleukin-1(IL-1), interleukin-6 (IL-6), and Tumor Necrosis Factor-alpha (TNF-alpha) fuel the inflammatory process by recruiting leukocytes to the area. The several companies that had targeted IL-1 and IL-6 in multiple sclerosis did not turn their attention to stroke.
CCR5 (CC chemokine receptor 5) was proposed by a UCLA group as a target for addressing neuroinflammation in stroke (and TBI), and to improve stroke rehabilitation outcomes. In mice, blocking CCR5 expression led to a better functional outcome in stroke models, and showed improved dendritic spine density. The UCLA group initiated a Phase II/III trial using the CCR5 blocker Selzentry/maraviroc, which is marketed for the treatment of HIV infection. The trial was to enroll 60 patients who are within 4 weeks of their stroke (ischemic or subcortical hemorrhagic), however Covid-19 so devastated enrollment that the trial was terminated. UCLA is now collaborating with Canadian groups on a 120pt PhII trial, expected to complete around year-end 2025.
Revalesio’s RNS60 (a ‘charge-stabilized nanobubbles’ asset, produced from saline) has been claimed to reduce inflammation and function as a mitochondrial protectant, though it has not fared well in ALS testing. However, in stroke there is a face-valid rationale for the mechanism: These nano-bubbles deliver oxygen to a hypoxic parenchyma. They reported an 82pt study in stroke patients receiving thrombectomy: 48 hours post-thrombectomy, the RNS60 high-dose group showed 50% less infarct growth than the thrombectomy-only group. Numerical advantages were reported on a number of other secondary endpoints. Revalesio will go into PhIII before year-end.
AptaTargets has developed an aptamer, ApTOLL, that blocks TLR4, thereby modulating the inflammatory response to the stroke. In animal models, ApTOLL reduces ischemic damage by 65%, using a 12 hour treatment window. A JAMA-published PhIb/II study, which used a 6 hr window, reported that the administration of ApTOLL just prior to thrombectomy yielded a mortality rate of 5% (compared with 18%), and produced improved functional outcome at 90 days. AptaTargets had raised over $9 million in 2024 with which to initiate a PhIIb study, but thus far has not announced the start of a trial. There is a credible rumor that Merck is partnering with AptaTargets on this program, if so, that might explain the delay, since it can be assumed that Merck would want to oversee Phase IIb. However, there is no confirmation from either company, so this remains speculation.
Aruna Biomedical developed exosomes (vesicles) obtained from neural stem cells as a means of treating stroke, Its lead exosome product, AB126, was tested in pig models of stroke, following mouse model testing which showed decreases in infarct size and improved functional outcomes. Aruna’s working hypothesis is anti-inflammatory, steering macrophages to an anti-inflammatory role and reducing pro-inflammatory T-cell generation. Aruna plans to first enroll patients who have had a thrombectomy but were not treatment responders, based on the fact that reperfusion allows AB126 to access the infarct, but in theory allows for better signal-detection in a nonresponder population. ArunA received IND clearance in January 2024, eighteen months later, PhIb/IIa has yet to begin, though they now claim a 3Q:25 start. Financing is the likely problem, they have not disclosed a round since 2021, when they raised $10.8 million.
Cai et al reported (2022) that a key transcription factor in Tregs, FOXP3, mediates macrophage clearance of cellular debris, with knockdown of FOXP3 associated with negative outcomes. Immunomodulation could thus be aimed at enhancing FOXP3 production.
Complement
A Columbia group cited complement component-1q (C1q) as a novel target in stroke, as C1q knockout mice show less damage in a stroke model. Based on work since that time, it appears that C1q operates as a key initiating factor for the classical complement pathway, which triggers microglial activation, the release of inflammatory cytokines, and the tagging of synapses for pruning. C1q is a major target for Annexon, but they do not list stroke as an indication of interest.
A University of Colorado/Medical University of South Carolina collaboration on complement system pathology was spun out into AdMIRx, with stroke as the lead CNS indication. They developed a technology for inhibiting complement activation locally, in areas contiguous to a stroke infarct. However, the company was rebranded at Q32Bio, CNS is no longer cited as a focus.
Myrobalan cites stroke as a potential venue for its preclinical neuroprotective repertoire, which includes small molecules targeting CSF1R, TYK2, and GPR17. The CSF1R program is in IND-enabling phase, but Myrobalan appears to be prioritizing MS.
Some of the molecular culprits pursued in other neurodegenerative disorders have also been proposed as playing a destructive role in stroke. A University of Wisconsin researcher reported alpha-synuclein knockout reduces infarct volume and improves survival/functioning in the MCA stroke model. None of the major groups in human trials with alpha-synuclein mAbs have evidenced any interest in expanding into stroke. A U. South Wales team cited tau as a target for stroke therapy–knocking out tau reduces neurological deficits and time to recovery in mouse models. This has not translated into any broad interest in tau therapeutics for stroke.
An attempt to repurpose an MS drug turned out to be a road to nowhere for Biogen, exploring natalizumab for stroke. They ran this integrin alpha(4) inhibitor through Phase IIb, hoping for an anti-inflammatory impact, but the trial failed to hit any of its endpoints, and the program was terminated.
Apoptosis-inhibition
The inhibition of programmed cell death has received attention as a way of addressing a variety of degenerative diseases, based on the production of apoptotic factors like caspases and bax resulting from the disruption of the mitochondrial energy system. However, apoptosis is at its essence an adaptive culling mechanism, and to the degree to which it clears doomed cells from the scene, may have a useful role in stroke as well. On the other hand, it is possible that in stroke, apoptotic mechanisms are amplified, to the point of eliminating cells that otherwise might have survived and been functional. Until it is clear which of these is more characteristic of apoptosis in stroke, the value of its inhibition will remain ambiguous.
Edaravone: Given its primary mechanism of free radical scavenging, this is the logical place to discuss Mitsubishi’s edaravone/MCI-186, which has been approved in Japan for the treatment of stroke since 2001, based on a Phase III which utilized the modified Rankin at 3 months, showing 22% of edaravone patients to be symptom-free, compared with 10% of the control group. Other stroke efficacy results have been mixed, with Lapchak’s 2010 review concluding that a 24 hr treatment window is optimal, and that the ability to increase dosing is circumscribed by the risk of renal toxicity. In China, sublingual edaravone, combined with dexborneol, a herbal anti-inflammatory, was successfully run through a sizable stroke efficacy trials by Simcere, using a 48hr window, and a 14 day BID IV schedule. In a study published in JAMA, 64.4% had a good functional outcome, compared with 54.4% of the placebo group: ‘Bicun’ has been approved for stroke treatment in China. Simcere is running another 880pt PhIII, transitioning from an initial injected bolus to 14 days of sublingual edaravone. The treatment window is 48 hours as well. Simcere has not updated the clinical record since 2023, nor have they published the site roster, but they are adding US sites for the sake of satisfying the FDA if and when they file an NDA. Simcere is also running a 200pt PhII in China utilizing sublingual edaravone/dexborneol for six months, expected to finish at the end of 2026. To avoid the confound of a study including patients with and without thrombectomy, these studies exclude patients who have had a thrombectomy.
In Japan, the edaravone treatment protocol uses 14 days of infusions, which Mitsubishi considered incompatible with treatment protocols in Europe, so an alternate dosing regimen providing 72 hours of continuous infusion was tested, with satisfactory results. However, with edaravone having been subsequently approved in the US and EU for ALS, Mitsubishi Tanabe has not been inclined to take on the risk and cost attendant to the further development of edaravone in stroke for those regions.
Silver Creek Pharmaceuticals is running a 120pt trial for Scp776, an IGF-1 fusion protein, that they postulate has extended bioavailability and anti-apoptotic effects. Three dose levels are being sequentially evaluated during the first stage, the optimal dose will then be used with forty patients in the second stage. All patients have a thrombectomy. It should wrap up late this year.
DiaMedica has been developing DM199, a recombinant form of the protease kallikrein-1, believed to play a role in mediating a variety of downstream mechanisms, in particular increasing collateral blood flow in acute ischemic brain. DM199 was on an FDA hold, but that hold was lifted, and a 350pt PhII/III is underway. The drug is being given to patients unable to receive thrombolytic treatment or thrombectomy, administered with a IV bolus followed by daily subQ injections. DiaMedica plans to run an interim analysis during 1H:26, on the first 200 completers.
Minocycline has been an antioxidant/mitochondrial protectant candidate for seemingly all varieties of neurological pathologies, and stroke is no exception. In a 2018 review of the literature by Malhotra et al, they concluded (from a total of 7 trials, 426 patients) that minocycline showed positive trends indicative of efficacy vis-a-vis the preservation of functioning. Sheng et al published a competing review using a different methodology and patient sample, and concluded that minocycline reached statistical significance in its impact on the Rankin and NIHSS, but not the Barthels. Several years ago, in a search for proprietary turf, NeuroVive had developed an IV formulation for TBI, but felt they needed a formulation with better BBB access for stroke, and partnered with Isomerase Technologies to work with novel chemical structures. Nothing appears to have come of that, and stroke is not on the renamed Abliva‘s list of target areas. Despite the tentative meta-analytic support for minocycline as a neuroprotectant in acute stroke, it seems unlikely that this is going to go anywhere other than generating publications for academia.
Uric acid is a free radical scavenger that passed the first SPAN 1.0 preclinical vetting for possible clinical testing in stroke. Whether anyone is going to invest in clinically testing a treatment without any substantive IP protection is an entirely separate question, one which is almost certainly answered in the negative.
A number of potential protective targets have been identified, but not yet well validated: A novel protectant pathway candidate was identified by a German academic group: the Von Willebrand Factor (VWF)/glycoprotein-IB pathway. Higher levels of VWF have been linked to higher stroke risk, and appear to recruit monocytes to areas of infarct, creating neuroinflammation. Mice with VWF knocked out showed smaller infarct sizes in the MCA model. In 2021, a KU Leuven group reported that nanobodies against the A1 domain of VWF was neuroprotective in a mouse stroke model, via anti-inflammatory effects.
Salt-inducible kinase 2 deficient mice show less damage in an ischemia model than do wildtype controls, but SIK2 inhibitors have yet to be devised and tested; and NADPH oxidase 4 inhibitors have produced infarct-reduction in animal models, albeit given intrathecally.
Recombinant Activated Protein C (APC) was championed as a stroke therapeutic for several years by a University of Rochester researcher (Zlokovic), now at USC; the work was spun-out as Socratech, later renamed ZZ Biotech. Believed to reduce tPA-induced bleeding, and initially hypothesized to be neuroprotective via blockage of MMP-9 and inflammatory processes. APC was used as a monotherapy in a NINDS sponsored, 72pt trial, but the trial was terminated due to poor enrollment.
ZZ Biotech turned to developing the APC recombinant variant, 3K3A-APC (invented by a Scripps Research scientist), and ran a 110pt trial in stroke patients receiving tPA, thrombectomy, or both. The incidence of hemorrhage was lower in the APC group, 67.4% compared to 86.5%. Zlokovic’s USC group published a report that 3K3A-APC spurs neurogenesis, postulating a possible role as initiating neuronal repair, with functional recovery demonstrated in mice implanted with neural stem cells when accompanied by 3K3A-APC.
However, Zlokovic’s work has been tarred by persistent reports of data manipulation and research fraud in his work on APC. Several papers have been retracted, a $30 million grant from NINDS to USC to help fund a Phase III trial in stroke was canceled, and he has been placed on ‘indefinite leave’. This renders all of this moot.
A UCLA group has reported on a disconnection phenomenon in stroke, impacting areas distant from the actual infarct damage. This had previously been noted in our Alzheimer’s review, where they described a similar pathology in AD. Based on mouse models and stroke patients, they believe that this reflects a loss of brain ‘rhythm’ mediated by gamma oscillation. The group recently reported that DDL-920, a small molecule which originated at UCLA, and is a NAM for GABA-A subunits regulating tonic inhibition of parvalbumin-interneurons, improved gamma oscillation power and enhanced cognition in a mouse stroke model. DDL-920 has not yet been vetted for potential use in humans.
Reperfusion Injury
One of the ironies of treating ischemic stroke is that ‘undoing’ the vascular blockage and reoxygenating affected tissue itself is problematic, there are negative consequences to reperfusion as well. Before thrombectomy became more widely available, reperfusion injury was a preclinical phenomenon, a risk yet to be made clinically tangible other than in those patients fortunate enough to receive t-PA. But the downside of reperfusion, which involves inflammatory factors (via leukocyte entry and cytokine release) and oxidative damage, becomes a potential target for neuroprotective intervention, hence the inclusion of a free radical scavenger in the Lumosa program described earlier. There is another potential negative aspect to reperfusion–hemorrhagic transformation, where an ischemic process morphs into intracerebral hemorrhage. The later in the stroke process that reperfusion occurs, the greater the risk of hemorrhagic transformation, which is why t-PA has had such a limited time window for deployment.
Downstream: Edema
Remedy Pharmaceuticals’ glibenclamide is aimed at reducing cerebral edema, a contributor to stroke mortality and poor functional outcomes in large hemispheric stroke (a class that comprises about 15% of all ischemic strokes). Work by Marc Simard at the University of Maryland had identified this drug as a mediator of cerebral edema and hemorrhage. In stroke, it is administered IV, the treatment window is 10 hours post-infarct. Remedy ran an open-label trial of glibenclamide in severe stroke, and reported that 12.5% of patients experienced edema, versus 88% of historical controls; 12.5% died or suffered severe disability at 30 days post-infarct, compared with 67% of their historical control group. An 83pt Phase II showed that while the primary endpoint–the modified Rankin–was a washout (42% vs. 39%), mortality at six months was 37% in placebo patients, compared with 14% in the glibenclamide group; significant effects on death due to edema (2% vs. 22%) and due to ‘neurological causes’ (7% vs. 25%) were also cited.
Biogen shocked most observers when it acquired the glibenclamide program from Remedy in 2017, paying $120 million upfront. Designated BIIB093, it went into Phase III, a trial that originally aimed to enroll 765 patients globally, using the modified Rankin score at 90 days post-treatment as the primary endpoint. In early 2023, Biogen pulled the plug on the partnership as part of the Viehbacher revamp. The results showed no benefit, though only a tiny proportion (86 patients) of the target sample had been enrolled, still fewer had outcome data.
Remedy went ahead and completed what turned into a 306pt PhIII for glibenclamide in large hemispheric stroke, which showed a clear failure. Remedy spent the rest of 2024 mining the data for anything that might suggest a post hoc signal. They made a case for impact in smaller-size infarcts, particularly when thrombectomy was also utilized. Remedy claimed positive post hoc findings from the 147pt subgroup with lesion volume under 126ml, while in the 34 patients who also had a thrombectomy, Remedy claimed that the glibenclamide subgroup doubled the number of patients able to walk independently at 90 days post-stroke, and the mortality rate was 5.6%, compared to 31% in the placebo arm. Combining their PhIIb and PhIII populations post hoc, Remedy reported a treatment effect of 2.42 at 90 days. But with the data strongly supporting thrombectomy, and the percentage of patients receiving it increasing, funding may be hard to find to run a prospective trial for glibenclamide given the post hoc aerobics the data needed to extract this ostensible signal.
Gene Therapy
A University of New Mexico researcher reported that blocking miRNA-155 48 hours after occlusion in a mouse model decreased infarct volume and improved functional recovery. A Xiangnan University researcher published similar data regarding the contributory role played by miR-155 in ischemic stroke, knocking it out attenuates damage in a mouse model. A China Southern University group identified micro-RNA-143 as a target in stroke–inhibiting its expression reduced neurological injury markers and infarct size in a mouse model. On the other hand, augmenting miRNA-24 levels appears to reduce neuroinflammation, oxidative stress, and apoptosis.
There have been reports of potential value for the AAV-delivered NeuroD1-gene. A Penn State group conducted mouse studies where the NeuroD1 payload caused the conversion of proliferating astrocytes into neurons (whose native proliferative capability is nil). In these studies, the regenerated cells replaced one-third of neurons destroyed by the initial ischemic assault, while providing protection to one-third of the neurons in the penumbra. The new neurons appeared to be functionally interwired, with improvements in both motor and cognitive capabilities. A Kunming Institute-led group demonstrated similar results in nonhuman-primates, 10-30 days post-stroke, reporting that 90% of transfected astrocytes converted to neurons, and that the astrocyte ‘supply’ was maintained. A neuroprotective effect was also documented, interneuron levels were preserved, while microglia and macrophage levels were reduced.
However, ischemic stroke is not a target for any of the high profile gene therapy companies, preoccupied as they are with the treatment of rare diseases.
Microbiota
Manipulating the gut-brain microbiome is a tactic under consideration for Parkinson’s and Autism, but a Weill Cornell group ran a stroke mouse model using fecal transplants modified via antibiotic pretreatment to increase the gut population of Proteobacteria, and reduce two bacterial populations (Clostridiaceae and Bacteriodetes); they reported that it reduced infarct volume. Microbiota species have been reported to mediate levels of homocysteine and creatinine, both linked to stroke risk.
Neuronal Restoration/Regeneration
With protection of neurons and glia so elusive, some turned to the task of undoing the damage , aiming for regeneration. This has not been an easier path. Timing may be a central variable here as well, but in the reverse direction: Given the neurotoxic environment of the post-stroke brain, initiating recovery/regeneration might be doomed if started too early, thus programs have tried a variety of time delays, ranging from hours to weeks.
AbbVie ran a PhII trial of their RGM-A mAb, elezanumab. Using a 24hr post-stroke window, the 121pt study administered elezanumab monthly, for a total of thirteen IV administrations. The stroke trial finished last December, but other than positive trends for biomarkers like NfL and GFAP, no results have been disclosed, which even for close-to-the-vest AbbVie, is not encouraging.
Post-stroke rehabilitation was pursued in Phase II (in Japan only) by Fujifilm Toyama with T-817MA/Edonerpic. Initially described as a sigma receptor activator, T-817MA has been better known as an Alzheimer’s candidate. Toyama reported preclinical stroke model results (mice and nonhuman primates) where T-817MA bound to the CRMP2 (collapsin-response-mediator protein2) and promoted recovery of motor functions, partly by recruiting AMPA receptors; it is now described as a ‘plasticity enhancer’. The Japan-only Phase II was fully enrolled back in 2022, 65 patients with upper limb functional impairment, change from baseline in motor control was the primary endpoint. Three years later, the results have not been disclosed, which almost surely means burial after failure.
Shionogi’s redasemtide is a small molecule which is aimed at promoting the release of endogenous mesenchymal stem cells into the circulation, in theory available for regenerative effects in the locale of injury. A preliminary PhII was reported to have met its primary endpoint, now a dose-ranging PhIIb is underway, aimed at enrolling 627 patients who are not candidates for thrombolysis or thrombectomy. It was still enrolling patients in January, and with a 90 day treatment period, it is going well past its projected completion date in 1Q:25.
Pharmazz has an endothelin-beta agonist, sovatelide/PMZ-1260, that they believe increases the availability of neuronal progenitor cells (and protects mitochondria). It is approved for the treatment of stroke in India, and is starting an international PhIII aimed at enrolling 514 patients. With just six sites, and a 90 day readout, this trial is guaranteed to not come anywhere near its projected completion date in 2026.
Chronic stroke recovery is the focus of Constant Therapeutics‘ TXA127,an angiotensin 1-7 peptide, in a 50pt PhII that enrolls patients 6-24 months post-stroke.
A Karolinska group reported that notch-1 is reduced in a mouse model of stroke, and that when this reduction was blocked, post-stroke neurogenesis was also impaired. In a subsequent 2020 study, they reported that blocking Notch activation in stroke models increased post-stroke neurogenesis 3.5 fold. However, the adverse event legacy of erstwhile gamma-secretase inhibitors in Alzheimer’s, which also blocked Notch, suggests that this target is problematic.
Animal studies suggest that the ‘inhibiting the inhibitor’ strategy long pursued without success in the treatment of spinal cord injury might be applicable to the upregulation of recovery in stroke. Nogo-A inhibition followed by an exercise regimen improved motor function recovery in rats. The researchers reported that the sequential treatment led to organized neuronal sprouting, whereas simultaneous treatment led to disorganized growth, which raised the question of whether one must trigger and institute growth before trying to utilize and train those new ‘circuits.’ A Loyola/University of Chicago group, who reported that stroke-model mice given anti-Nogo therapy nine-weeks post-occlusion regained 78% of their forelimb function, compared with 47% of mice who received no post-occlusion treatment. But the same group reported that anti-Nogo treatment did not improve hippocampal neurogenesis post-stroke, while a University of Colorado group reported hippocampal neurogenesis was increased in a pediatric stroke mouse model. If this tactic is only applicable to juvenile neural systems, that makes it irrelevant to the vast majority of ischemic stroke patients.
A UCLA group (Carmichael et al) reported (Nature Communications) believed that a circuit disconnection occurs in stroke, leading to a loss of functional synchrony, and attribute this to failure of parvalbumin neurons to generate gamma oscillations. They reported that a drug developed inhouse, DDL-920, produced significant motoric function in a mouse model. DDL-920, a NAM selective for GABA-A alpha1/beta2/delta receptor subtypes, has yet to be tested for safety in humans.
Rho is an growth-mediating factor also studied in SCI without success, the Rho inhibitor fasudil was assessed by the NIH Span 1.0 program, without success.
Cell Therapy
There are two avenues by which cell implants could in theory provide benefit in ischemic stroke. First, one might hope that implanted cells could migrate to areas that have been damaged, guided by signaling factors, beacons of cellular distress, where they would interwire, replacing cells that were lost. Rodent studies indicated that human ES cell derived stem cells could travel to an infarct, differentiate into neurons and endothelial cells, and appeared associated with improved function. The question remained whether they can achieve this across the greater physical expanse of the human brain, and whether they could actually ‘wire in’ and replace damaged cell circuits. One related safety question is whether cells sufficiently undifferentiated as to be able to respond to such signals in the heterogeneous brain areas affected by stroke, might also be more at risk of tumor generation than their less flexible, more differentiated peers. While no cases of brain tumors as a consequence of implanted pluripotent cells have been reported, there is experimental evidence suggestive of teratoma risk: Two decades ago, Layton Biosciences moved aggressively with its tumor-derived stem cells into a stroke PhII, but the initial benefit appeared to diminish over time, the teratogenicity concerns never abated, and the Company ran out of money after the FDA demanded more primate trials. Immunoreactivity has also been an ongoing concern for programs that utilize allogenic cell sources (in contrast to autologous, induced pluripotent stem cells derived from a patient’s own cells).
The indirect route to therapeutic impact is contextual, likely involving neurotrophin upregulation and/or anti-inflammatory effects, providing a pro-growth alteration of the context wherein the stroke damage has occurred and regeneration sought.
There are numerous poorly controlled trials of cells of various origins, usually mesenchymal stem cells, primarily being conducted in China and Korea. This is more a function of a lax regulatory environment than any genuine therapeutic or scientific merit, we are not going to devote any space to them here.
The early returns on cell therapy per se for ischemic stroke were modestly encouraging. In a meta-analysis of cell therapy trials published in 2017 by Detente and Moisan, which reviewed controlled trials with six-month efficacy results, they concluded that the effect size was between .45 and .81, without any clear link to a particular type of cell or delivery. Which makes the area of cell therapy for stroke potentially fruitful, but leaves open all the questions regarding the optimal technologies to be deployed.
A variant of the neurotrophin theme was developed by SanBio. Based on Kyoto University work, this program provides neurotrophic support via allogenic mesenchymal cells transfected with Notch. Increases in multiple factors were described, including IL-5, IL-6, IL-8, PDGF-AA, VEGF). In Phase IIa, the implanted cells were implanted via stereotactic surgery 3-12 months post-infarct; SanBio eventually published a report that claimed that the limited durability of these cells is believed to be a positive factor in their therapeutic impact; while demising, they release FGF2 in amounts thought to be pro-proliferative. SanBio postulated that these cells would help with the reconstitution of the extracellular matrix, promoting a possible regenerative effect. There was also a putative anti-inflammatory effect, these cells triggering an increase in IL-5, and a reduction in activated microglia. Vandefitemcel/SB623 completed a Phase I/II trial that enrolled eighteen patients, 7-36 months post-stroke, followed for twelve months. In this open-label trial, SanBio reported ‘highly significant’ benefit on the NIHSS and two other stroke scales, at six, nine, and twelve month followups.
SanBio then initiated a 60 site, 163pt Phase IIb trial, in conjunction with Dainippon. Patients had to be 6-60 months post-stroke and continue to show motor deficits. There were two dose levels (2.5 million cells and 5 million cells), and a sham surgery control group. The trial failed to meet its primary endpoint, which was the proportion of patients significant improving from baseline. SanBio conducted posthoc analyses and claimed that in the 47% of patients with a relatively smaller infarct size, and defining treatment response with a complex matrix of motor score improvements (all smaller scale than the original endpoint required), 49% of the SB623 group met those criteria, compared with 19% of the placebo group. SanBio announced that based on this, they would continue the clinical development of SB623 for both ischemic and hemorrhagic stroke. In June 2024, SB623 was given ‘conditional and time-limited’ approval in Japan for the treatment of TBI.
So far as stroke is concerned, threading the needle with new criteria derived from posthoc datamining has been a consistent route to clinical disappointment, we see little reason to expect otherwise here. More than six years after their PhIII failure, the stroke program is still described as “planning for a PhII or III study (Japan).”
Athersys ran a Phase II trial in stroke using their mesenchymal allogenic progenitor cells, a trial that enrolled 126 patients, administering their ‘Multistem’ cells 24-48 hours post-stroke–the goal being neurotrophic factor production. The trial failed. Athersys ran a Phase II/III that enrolled 300 patients, involving a single IV infusion of 1.2 billion cells, 18-36 hours post-stroke. In late 2023, Athersys announced that the DSMB had deemed the sample size too small to demonstrate therapeutic benefit, which means that any separation between groups was less than planned for. Athersys did not have the resources to add patients, and filed for bankruptcy.
ReNeuron had prioritized stroke for years, using their fetal-source, immortalized cells. They eventually ran a dose-ranging trial of their stereotactically-injected cells (having undergone a proprietary proliferation and conditional immortalization program, they can be ‘turned off’). Safety findings were positive, and an open-label Phase II trial was run in patients 2-12 months post-stroke, followed for a year. The cells were implanted near the area of the brain where the ischemic event had been centered, this yielded results that were fragmented and inconsistent: ReNeuron claimed that fifteen of the twenty-one patients showed improvement on some efficacy measure (but not the same for each patient). A Phase IIb aimed at enrolling 110 patients was initiated, but then was terminated, partly due to the impact of Covid-19. ReNeuron recently came through bankruptcy and has reorganized, with the plan to outlicense their stem cell lines. We doubt that any of them will be returned to stroke development.
NeuralStem’s founding management set up a Chinese subsidiary in order to use NeuralStem’s fetal-source cells in an ischemic stroke clinical trial, a trial intentionally not placed under the purview of the FDA. Their intention was to use their spinal cord stem cells (NSI-566) in areas related to motor function, since rat studies showed functional motor recovery four weeks post-implantation. Aimed at enrolling 18 patients, that trial stopped enrollment with 9 patients on board, and reported that there were no serious adverse events. NeuralStem subsequently ceased operations.
None of the new wave of cell therapy companies, Aspen Neuroscience, Bayer/Blue Rock, or Treefrog Therapeutics, mention stroke as a potential venue for development.
The Rise of the Machines
NeuroPerspective covers devices only in passing, given our focus upon pharmacotherapy, gene, and cell therapeutics, and what we have seen as a limited role for devices in the treatment of epilepsy, depression, pain, and Parkinson’s. But in the case of stroke, devices have taken a dominant role in transforming stroke care. Endovascular therapy now plays a central role for a significant minority of stroke patients.
1) Thrombectomy: It is perhaps slightly ironic that the relatively more primitive (in concept, not engineering) technologies aimed at removing clots via mechanical means moved to the forefront of stroke treatment. Stent-retrieval devices benefited from the lower regulatory bar that applied to them, where they had to simply show that they do what they claimed in terms of direct effect, ‘recanalizing’ the vasculature, permitting the resumption of blood flow, without having to prove any impact on clinical outcome. But a new generation of endovascular devices became dominant, exceeding the results produced by their predecessors, yielding an overall success rate in the 35-40% range. One KOL has told us that he considers these devices (which he did not play a role in developing) one of the great advances in clinical medicine of recent years.
To quickly recap the endovascular story: The first generation of endovascular devices included the ‘Merci’ device, marketed since 2004 by Concentric Medical and then Stryker, and the eponymous device from Penumbra. There was some degree of trial and error involved with this first generation, since imaging was not utilized, leaving some diagnostic uncertainty regarding the nature of the blockage, and no clear guidelines regarding the time window for intervention. Covidien then introduced the first stent-retrieval device for the endovascular treatment of stroke, Solitaire, followed by Stryker’s Trevo. Covidien ran a head-to-head trial of Solitaire versus ‘Merci’, wherein Solitaire produced recanalization in 80% of patients receiving it, compared to 57% of those for whom the Merci device was utilized. In other comparison studies, the stent retrieval devices produced recanalization in the 68% range, compared with 30-40% for Merci (Fisher and Saver, 2015). For comparison, tPA provides substantial recanalization in 30-40% of patients receiving it. The Solitaire device also was shown to produce significantly better likelihood of a benign outcome: 58% of the Solitaire patients had a positive (Rankin score of 0-2) outcome, compared with 33% of the Merci patients. Trials have shown Solitaire and Trevo have consistently bettered the results provided by tPA alone. Given that this device has an eight hour window, this has offered a new standard of care for ischemic stroke, but with an important caveat: As is the case with tPA, patient selection is critical. Patients with particularly large infarct areas and very impaired perfusion not only did not respond well to the recanalization, they had worse outcomes. Neuroimaging is now routinely utilized to document the occlusion and size of the penumbra at risk, and the time between stroke and treatment has become recognized as a critical determinative feature. As a patient moves towards the far end of the eight-hour window, more sophisticated imaging assessment of the location of the occlusion (preferably in large to medium size arteries) and the ischemic core becomes essential to identifying potential treatment responders.
A number of competing stent retrieval devices have been developed: Anaconda Biomed‘s novel device (ANA System) combines both clot aspiration and retrieval, and achieved recanalization in 83.3% of patients, 45.2% achieved ‘excellent’ outcomes at 90 days. Another device, Skynor‘s ‘Skyflow’, tested head-to-head in China with Solitaire for efficacy, achieved a recanalization rate of 88.4%, compared with Solitaire’s 82.5%.
Time is of the essence in endovascular and tPA therapies, and one potential avenue to expanding the portion of the stroke population for which endovascular treatment is feasible would be to slow the pace of ischemic core expansion. If a neuroprotectant drug could be validated as slowing progression within the first few hours, it could expand the population of treatment candidates. One estimate is that, with the current six-eight hour window and limited population of specialized stroke centers able to do endovascular interventions, out of the 800,000 strokes annually in the US, about 50,000 patients may qualify. If a neuroprotectant drug could even modestly slow infarct growth during the critical hours immediately post-stroke, that might double (or more) the proportion of stroke patients for whom this technically challenging but effective therapy would be viable. In addition, neuroprotection could provide a modest but meaningful boost to the overall success rate provided by thrombectomy.
2) Hyper-oxygenation had been hoped to slow down lesion growth, aimed at reducing the proportion of the penumbra at risk due to oxygen deprivation, and expanding the time window for the thrombolysis/thrombectomy. However, supplemental oxygen for patients with adequate endogenous oxygenation had no impact on survival or disability three months later. The Revalesio ‘nano-bubble’ pharmacotherapy is a sophisticated, targeted variant on this theme, and has some potential.
3) Hypothermia has long been considered as an approach to reducing damage post-stroke, ostensibly via anti-inflammatory and anti-oxidative effects. It is a technically challenging intervention to implement, requiring significant hospital infrastructure, and there are adverse events, including the risk of pneumonia. A review of the literature in 2016 by Tahir et al concluded: “Although hypothermia has been used for various purposes over several decades, its efficacy in the treatment of ischemic stroke is debatable”. A consortium received funding from the EU’s Framework Program 7 to conduct a 1500 pt Phase III trial of hypothermia in ischemic stroke. However, the protocol had to be altered to account for the increased reliance upon thrombectomy, and the task of convincing stroke centers to add a burdensome intervention that required twenty hours turned out to be untenable. Enrollment was finally ended in 2018, after a total of 98 patients had been enrolled in fifty-two months. No difference was found between the two 49pt cohorts on treatment outcome or safety, though given the small size, this was not definitive.
Programs to Watch
NoNO Therapeutics: The first PhIII was un by unexpected interference between tPA and nerinetide, patients who did not have tPA first showed a 40% reduction in mortality. A subsequent PhIII showed a 30-40% improved chance of minimal disability when administered by EMTs in the ambulance, cutting the mean delay to one hour. The need to either be administered very soon after the infarct develops, or withheld from patients who are receiving a thrombolytic sharply diminishes the potential value of the drug. NoNO has moved on to NoNO-42, the d-enantiomer of nerinetide, as a next-generation PSD-95 antagonist, opening up the full gamut of stroke patients to treatment and improving access to the penumbra. NoNO-42 has entered PhIIb.
Revalesio’s RNS60: Has not fared well in ALS testing, but in stroke these nano-bubbles deliver oxygen to a hypoxic parenchyma. They reported an 82pt study in stroke patients receiving thrombectomy: 48 hours post-thrombectomy, the RNS60 high-dose group showed 50% less infarct growth than the thrombectomy-only group. Revalesio will go into PhIII before year-end.
AptaTargets: ApTOLL blocks TLR4, and in animal models, reduces ischemic damage by 65%, using a 12 hour treatment window. A JAMA-published PhIb/II study, which used a 6 hr window, reported that the administration of ApTOLL just prior to thrombectomy yielded a mortality rate of 5%. AptaTargets had raised over $9 million in 2024 with which to initiate a PhIIb study, but thus far has not announced the start of a trial. There is a credible rumor that Merck is partnering with AptaTargets on this program, but this remains speculative.
Simcere: In China, sublingual edaravone, combined with dexborneol, a herbal anti-inflammatory, was successfully run through a sizable stroke efficacy trials by Simcere, using a 48hr window, and a 14 day BID IV schedule. In a study published in JAMA, 64.4% had a good functional outcome, compared with 54.4% of the placebo group: ‘Bicun’ has been approved for stroke treatment in China. Simcere is running another 880pt PhIII, transitioning from an initial injected bolus to 14 days of sublingual edaravone. They are adding US sites for the sake of satisfying the FDA if and when they file an NDA.
Conclusions
The seemingly unending parade of clinical failures for stroke therapeutics may be approaching an end. Endovascular devices have changed the conversation for a minority of the AIS population, while the emphasis on the development of dedicated stroke centers has offered patients a better shot at squeezing through a narrow treatment time window. There are small company assets that are showing promise, albeit largely in conjunction with thrombectomy. Enhanced adjunctive treatment is a start, and may begin to rewrite the legacy of failure that had taken ischemic stroke off the developmental menus for most biopharm companies: The treatment of ischemic stroke is on the cusp of radical transformation over the next five years.
2023
From NP Fall Issue 2023
Huntington’s
Huntington’s provides a blood-chilling signature which affords a level of definitiveness seldom seen in brain-based disorders. The huntingtin protein, which is essential for early development and longterm neural maintenance, is turned by mutation into a bomb with a very long, inexorable fuse, though the pace of progression, and the time of eventual detonation, is not so easily predicted. But the certainty of diagnosis has not yet translated into unambiguously effective treatment; clinical results have demonstrated that, even when the target has been identified and attacked, the outcome is nowhere near certain. This partly reflects the decades-long timeline for disease progression: Save for a small proportion of cases on the cusp, the CAG sequence repeat characteristic of HD leaves little doubt about what will happen, only ample ambiguity regarding when. The number of repeats predicts the speed with which the disease will manifest itself–albeit with variability attributed to the impact of other modifier genes on the emergence of the HD phenotype. Individuals with a critical number of repetitions in the gene encoding the huntingtin protein eventually begin a slow slide into depression, dementia, and a movement disorder which includes impaired voluntary motor control and involuntary, choreiform writhing.
To oversimplify it, the longer the genetic stutter, the earlier the disease will reach critical mass and then symptomatically explode. Even when individuals have as many as 35 CAG repeats, the huntingtin protein functions normally. It is in the range of 36-40 repeats wherein there is a borderline zone: HD symptoms may never appear; it may be that in those cases, the accumulation of mutant Htt protein does not reach a pathological threshold during a normal lifetime. Smaller numbers of CAG repeats also increases the likelihood that the choreiform movement disorder that is a key diagnostic sign may not manifest early, and their diagnosis may in turn be considerably delayed. The presence of 40 repeats or more means that Huntington’s symptoms will eventually develop, with higher repeat counts associated with earlier onset and a higher salience of motor symptoms. At a thankfully rare 60 repeats and above, the onset is so accelerated that these individuals (5-10%) can develop juvenile-onset HD.
But there is a caveat: An MGH group found that the number of CAG repeats in Huntington’s is modified over a patient’s lifetime by the activity of genes associated with DNA maintenance and repair. Increases in CAG repeats drive phenotypic disease onset, different haplotypes can have onset-accelerating or onset-delaying effects. The number of CAG repeats is thus changeable, the absolute repeat number at a given timepoint does not reliably predict whether and when the phenotype will manifest itself. Individuals with 27-35 CAG repeats have some risk of that number expanding into the active disease range. Additionally, CAGs may be modified ‘en route’, impacting disease-emergence and the pace of progression.
The Epidemiology
There are approximately 42,000 patients with Huntington’s in the US. Estimates of the number who have the mutation but have not yet become symptomatic with manifest illness range up to 200,000. The intensity and duration of custodial care eventually needed, and the fact that the modal patient experiencing symptom-onset is an adult in the prime of life (between 30-55) magnifies the socioeconomic burden; HD is an orphan disorder with an outsized impact. For most Huntington’s patients (early onset cases tend to progress more rapidly) the expected lifespan after diagnosis is in the ten to twenty year range, an eternity for the families who have to watch their loved ones writhe and wither. Estimates of the overall costs attributable to HD in the US have been in the $2.5 billion range annually; CHDI and IQVIA have been collaborating on a project to refine the estimates of the cost-burden for HD in the US, part of a global observational study funded by CHDI (a vital nonprofit that spends $80-120 million per year on HD research).
HD has crystallized the mixed blessing proffered by genetic testing untethered to treatment choices. While it can be identified, given that there is nothing that can be done at present to alter the course of the disease, such knowledge is accompanied by a sense of helplessness. Depending on their own temperament and familial circumstances, some patients at risk for HD prefer to know, in order to prepare themselves and their families. On the other hand, some who are aware that they may carry this mutation prefer to tolerate the desperate anxiety of not knowing their diagnosis, in the hope that fate will be kind to them.
Brain scans can detect striatal anomalies in HD patients up to fifteen years before overt behavioral symptoms display themselves, but the slow pace of deterioration means that the assessment of changes in response to intervention is difficult. There has also been a lack of standardized models for tracking HD deterioration, though the ‘Huntington’s Disease Harmonization Consortium‘ has assembled a database of 8000 MRIs from 2000 HD patients, in the service of establishing structural biomarkers for disease-status and progression. Prodromal cognitive deficits have been identified via neuropsychological testing seven years before the disease declares itself, changes in mutant huntingtin assays have been explored as potential biomarkers, but have yet to be validated.
One interesting ancillary element is work that has been done on the relationship between the age of onset (of manifest symptoms) and hypertension. There had been a hypothesis that hypertension delays HD onset, but over time the hypothesis has changed, with some European data suggesting that the use of calcium channel blockers to treat hypertension might instead accelerate onset.
Knowing the molecular culprit has translated into the major theme for disease-modification. Finding a way to delete or neutralize mutant huntingtin, while sparing enough wild-type huntingtin for necessary neural functioning, has been at the heart of the divergent approaches to gene therapy in HD that have begun to yield clinical results, mixed though they are.
The Clinical Picture
Misdiagnosed psychiatric disturbances are often the first ambiguous sign of emerging HD, with fits of anger or depressive episodes the first warning flag, ushering in dramatic changes in personality. But for other patients, the motor signs precede the psychiatric symptoms. It had been thought that early motoric manifestations are associated with higher CAG counts, but a European observational study of over 6300 patients showed that this is not the case, and that those symptoms are more likely to come first in patients with later onset. In that study, 42.4% had a psychiatric or cognitive symptom before any motor signs emerged; the majority had motor symptoms first, or concurrent with nonmotor signs. As was noted earlier, choreiform-silent HD tends to be associated with shorter CAG repeats.
Mood symptoms are of course highly nonspecific, but when choreiform movements are observed, they herald something much more dire than a depressive disorder, and tend to be the impetus (along with a family legacy of HD) for the genetic testing that finally confirms the grim prognosis. Writhing and impaired voluntary motor control eventually leave patients wheelchair-bound, often helmeted in order to prevent inadvertent self-injury. Personality changes and depression are preceded or followed by motor dyscontrol and cognitive deterioration, which eventually devolve into dementia. Between depression and the devastating prognosis, suicide is a frequent concern, with over 25% of HD patients attempting suicide at some point; some 5-7% of HD patient deaths are attributable to suicide. The dramatic movements thought to be emblematic of the disease are not correlated with the rate of cognitive decline. Those are instead associated with a growing impairment in voluntary movement control. The tragic endstage is epitomized by Parkinsonian rigidity and eventual immobility, the patient finally demising due to secondary medical complications.
The Neurobiology
The motif of critical proteins misfolded and then aggregated into toxic conglomerates, is generally, albeit not universally, thought to provide the pathological understructure for most neurodegenerative disorders. The role of such a toxic aggregate appears more clearcut in HD than the still debated contributions of amyloid and/or tau in Alzheimer’s, alpha-synuclein in Parkinson’s, or TDP-43 in ALS. There are those (e.g. Humbert et al, 2020)who consider the lifelong impact of mutant huntingtin to constitute a neurodevelopmental disorder, not just a degenerative one.
The signature CAG stutter leads to the expression of abnormally lengthened huntingtin; such mutant huntingtin (mHtt) misfolds and aggregates together and cannot be sufficiently disposed of via the usual maintenance/clearance operations. Instead, the mHtt coheres into toxic oligomers. Post-mortem examination of HD patient brain tissue shows huntingtin oligomers within cells in the striatum, where they produce their effects upon motor function via the destruction of GABAergic tracts, and in the neocortex, which accounts for the changes in cognitive capabilities and affective control that become so salient as the disease progresses. Mutant huntingtin aggregates also travel between cells, thereby propagating degeneration.
But it is more complex than that. There are multiple isoforms of mutant Htt, associated with variants of SNPs. Wave Life Sciences at one time had projected that their first two SNP targets in HD would cover 80% of the HD population, then they suggested that ‘over two-thirds’ would be accounted for. When their programs for both those SNPs failed, they refocused on a third SNP, found in about 40% of the HD population. But beyond the three or more mHtt isoforms that, if targeted selectively, could be required in order to cover the bulk of the HD population, there are other processes relevant to this target array:
1) CAG repeats can change over time, and with them, the profile of an individual’s phenotypic status and progression. The past few years has featured considerable work, particularly by the GeM-HD Consortium, on explaining the variance in age-of-onset not determined by the number of CAG repeats, the latter accounting for about 60% of the variance. Variations in genes associated with gene repair has drawn particular emphasis: As was noted earlier, in 2018 Gusella et al (MGH) reported that the number of CAG repeats in Huntington’s can be modified over a patient’s lifetime by the activity of two haplotypes. The first alters the FAN1 gene, resulting in the acceleration or delay of onset, depending on modifiers either reducing FAN1 gene function or increasing its expression respectively. The other haplotype involves the acceleration of onset via either the RRM2B or UBR5 genes. But in their work, these two sources of modification appear to account for just 1.6% of the variation in onset, thus there are numerous other factors yet to be delineated.
2) A group at U.Florida (Ranum et al) reported that, beyond mutant huntingtin, there are four more proteins produced by ‘repeat-associated non-ATG-translation’ (RAN). All of these rogue proteins (the various isoforms of mHtt and the four RAN cousins) aggregate and are neurotoxic. Longer CAG repeats appear to be associated with higher RAN levels; healthy control brains do not show any of these proteins. In theory, though the toxic contribution of each subtype has yet to be delineated, these RAN-proteins could mean that other toxic targets could be, or must be, addressed. However, the originators have focused on the salience of RAN proteins in c9orf72 mutations in ALS and FTD, Huntington’s has not been their priority.
3) A 2016 paper from a UCLA group noted that Huntington’s is associated with increased methylation in brain tissue, and proposed that this reflects an epigenetically induced acceleration in neural ‘age’ in some brain areas. The authors found that patients with HD showed an overall 3.2 year increase in neurobiological age. There was a negative correlation between the number of CAG repeats and epigenetic aging, the authors suggested that the effects of epigenetic aging may be primarily observable in early-stage disease, and that high CAG repeat patients tended to show more advanced cellular dysfunction. It remains to be seen whether epigenetic aging predicts earlier disease onset, accelerated progression, or both–and whether targeting the process can alter either one.
4) The MutSB complex is comprised of two subunits, MSH2 and MSH3, enzymes involved in DNA mismatch repair (MMR), mediating the pathological expansion of Htt into mHtt. There is discovery stage work being done by CHDI and LoQus23 Therapeutics on modulating that activity in order to delay onset.
Current Therapeutics
One key clinical question is–by what point in the disease must intervention start in order to have an appreciable impact on its subsequent course? There is no proven disease-modifying therapy for Huntington’s, but there have been tantalizing signs that targeting huntingtin is the right track for doing so. Up to now, treatment has been limited to compensatory measures of circumscribed impact and value. Antipsychotic drugs with dopamine-blocking components have often been used for the treatment of motor symptoms. The first HD-specific, approved symptomatic treatment was Xenazine/tetrabenazine, a VMAT2 inhibitor originally developed by Roche, which addresses only choreiform movement symptoms. Tetrabenazine provides modest symptomatic improvement in more than 80% of patients receiving it. Drowsiness and Parkinsonian symptoms are the main adverse effects, along with the risk of depression and suicidal ideation (the drug is similar to reserpine), concurrent treatment with antidepressants is often recommended. Xenazine exceeded expectations in terms of market-penetration and payer-acceptance, but the drug’s patent exclusivity ended.
Auspex developed a deuterated form of tetrabenazine that was acquired, along with the company, by Teva. Phase III results showed a significant impact on chorea, with a 4.4 point reduction compared to 1.7 points with placebo, 51% were rated as very improved, compared to 20% of the placebo arm. Tolerability appeared considerably better than that seen with tetrabenazine, very few patients needed to have dose reductions due to AEs, and depression/anxiety were not elevated. Austedo was finally approved for HD in 2017, as well as for Tardive Dyskinesia. In HD, it produced a 7.4pt reduction on the UHDRS, compared with 3.4 points for placebo, with 11% of patients experiencing somnolence. It has had a rapidly accelerating ramp-up in sales aided by the introduction of an XR formulation: By mid-2023 it had reached a $1.2 billion annualized pace, sold for both HD and TD, with sales predicted to more than double from there by 2027.
Neurocrine Biosciences’ Ingrezza/NBI-98854 is a once-daily VMAT2 inhibitor that was approved for Tardive Dyskinesia in 2017. Eventually, Neurocrine successfully ran a 120pt PhIII for Ingrezza in Huntington’s, yielding a UHDRS reduction of 4.6 points compared with 1.2 points on placebo. 42.9% of Ingrezza patients were judged ‘much improved’ or better, compared with 13.2% of the placebo group. The FDA approved that sNDA in August. Even before that approval, Ingrezza had maintained its lead over Austedo, reaching an annualized sales rate of over $1.6 billion in 2Q:23.
Future Therapeutics: Symptomatic
The Odyssey: Pridopidine
Pridopidine aka Huntexil aka ACR16 aka TV7820 has been on as bizarrely elongated a developmental odyssey as we can recall, helmed by a succession of owner/sponsors: Carlsson Research, NeuroSearch, Teva Pharma, and finally, Prilenia Therapeutics.
Carlsson Research conducted a 58pt PhII with ACR16, at the time believed to be a dopamine stabilizer, which they claimed showed improvement in both cognition and motor function. NeuroSearch bought Carlsson in 2006 to acquire Huntexil/pridopidine, seeing this as a route to supplanting tetrabenazine. The 2010 results of a 437pt Phase III trial, using 45mg and 90mg doses, initially were described as showing statistically significant improvement on both involuntary and voluntary movement measures, and the separation of drug from placebo increased over time, suggesting the disease process itself had been modified. However, this dataset utilized a post hoc adjustment that was not part of the pre-specified primary endpoint analysis, and a second 227pt Phase II/III (using 20mg, 45mg, and 90mg doses) ended up with the primary endpoint missing statistical significance (p=.078) at three months. NeuroSearch was eventually told that another Phase III would be required, and their hope of going higher in dosing in order to maximize therapeutic effect was thwarted by dose-related QTc effects. With its fiscal resources strained, NeuroSearch finally sold the program to Teva Pharmaceutical in 2012.
With HD-specialist Michael Hayden as CSO,Teva ran a 408pt PhII trial (PRIDE-HD) using doses ranging from 45 mg to 112.5mg, twice daily. That trial finished in 2016. It failed to show significance at 26 weeks or 52 weeks on the UHDRS motor score, or on a physical performance test. However, Teva claimed a positive impact for the 45mg dose (an effect not seen with other doses) at 52 weeks on a total functional capacity measure. Teva said that this was evidence that they were impacting disease progression–several HD KOLs went public with their disagreement with this claim, which was based on a cherry-picked secondary endpoint. Additionally, 18% of the pridopidine group had discontinued treatment due to side effects. Because of a high placebo response rate, the results were not seen as definitively ruling out a motoric benefit, but there was no credible support for the claim of disease-modification.
In the course of Teva’s work with pridopidine, the mechanism had been further defined as a sigma-1 agonist (along with binding at D3, adrenergic-A2C, and 5HT-1a receptors). A 2019 paper in Cell reported that sigma-1 activation induces autophagy, which could in theory reduce toxic levels of mutant huntingtin aggregates.
In the meantime, Teva acquired Auspex for $3.5 billion, and thus owned deutetrabenazine, renamed Austedo/SD-809. As was previously discussed, Austedo received FDA approval in 2017, which made pridopidine a lame duck program, and Teva terminated its development.
When Teva eliminated its neuroscience R&D, including Michael Hayden’s role as CSO, he took pridopidine with him, and founded Prilenia Therapeutics. Prilenia’s only asset is pridopidine, which they claim has therapeutic potential in a broad spectrum of disorders. They initiated a PhII trial in PD-LID that was eventually terminated due to Covid-19. However, they were able to bring in the first tranche of a $62.5 million funding round, and MGH’s Healey Center also chose pridopidine as one of the compounds to be tested in ALS patients (for which they reported a signal in rapidly-progressing patients).
A 480pt PhIII for pridodipine as a disease-slowing intervention in Huntington’s reported top-line results in early 2023; pridodipine failed to hit its primary endpoint, but Prilenia emphasized ‘nominally significant’ trends on disease progression and secondary endpoints (including cognition, based on just one cognitive test, the Stroop) for patients not receiving neuroleptic and chorea-specific drugs. Prilenia did not provide detail regarding the size of the treatment-responsive group or the size of the reported treatment effect, stating they would disclose those over time.
We have long wondered if the literary metaphor apropos to pridopidine may have shifted from The Odyssey to Moby Dick: Hayden’s work is not to be dismissed out-of-hand, but after all this time, pridopidine still has not proven its value in any of the neurodegenerative disorders for which benefit has been envisioned, neither modifying disease-progression or affording symptomatic benefit.
SAGE Therapeutics’ NMDA receptor PAM, SAGE-718, is believed to bind to a unique binding site on the NMDA receptor complex. The biomarker they are using (24S-HC), related to cholesterol processing and synaptic plasticity, has also been linked to cognition, and SAGE-718 has been expected to have procognitive effects. In a healthy volunteer PhI, Sage reported improved working memory and problem-solving, albeit on just two specific neuropsych tests. Improved executive function was reported in HD patients, in an open-label, six patient cohort.
Sage decided to run pilot studies in other procognitive contexts before proceeding in HD, but are now running three PhII trials in HD. A 178pt trial is administering SAGE-718 once-daily for 84 days, with changes in cognition the primary, functional status on the UHDRS the main secondary. The trial is expected to run through 2024. Results may be in a few months earlier from an 80pt study that is administering SAGE-718 over a 28 day dosing period, cognitive measures emphasized. A 300 pt open-label extended safety study is also ongoing.
PDE10 inhibition was once considered to be a route to improving cognition and motor function via downstream dopamine-system effects, but PDE10 has been a difficult target. Pfizer’s PF-02545920 went the farthest, but failed to produce improvement on any symptom domain in a 271pt Phase II in HD that reported results at the end of 2016. Lundbeck had a failure in a schizophrenia PhII with Lu AF11167, a PDE10A inhibitor. Omeros, Biocrea and Zenobia Therapeutics all had PDE10 inhibitors that demised early on.
The vasopressin system is interlinked with the better-known oxytocin system, and has been seen as a potential access point for mediating episodic anxiety/irritability and sociability. Azevan Pharma’s SRX246, a vasopressin-1a antagonist, went through a 106pt Phase II aimed at reducing irritability in Huntington’s patients. The trial was finished at the end of 2018, but when the results were finally published in late 2020, they were presented as if the trial was designed to only establish safety and tolerability, and was claimed to have shown both. There was no mention of impact on irritability or aggression, even though one or both had to be present for a patient to be enrolled, and standard assessments for both were done over the three month treatment duration (long enough for this mechanism to show an effect on neuropsychiatric symptoms, if it had any). The absence of any mention of activity in these domains was curious, begging the question of why Azevan claimed “exploratory efficacy endpoints demonstrated clinical benefit as well” when they were not even mentioned in the publication. NINDS apparently provided Azevan with funding for additional toxicology work and manufacturing in 2020, but that seems like a waste of resources. Neumora has a vasopressin-1a antagonist program, but they have not cited HD as a potential treatment area, focusing on AD-agitation.
Future Therapeutics: Disease Modification
Mutant huntingtin has been the target of choice, with relevant disease-modification options including: Reducing the production of mutant huntingtin; inhibiting its aggregation; increasing clearance; and altering mHtt in order to reduce its toxicity. But there is also a spectrum of tactics, both upstream and downstream of mutant huntingtin, for potentially impacting disease development and progression. This includes targeting other genes that mediate the number of CAG repeats and/or intervening in downstream processes (e.g. apoptosis, neurotrophin underproduction) triggered by mHtt.
Huntingtin’s Role
The huntingtin premise is made more complicated by ambiguity regarding the pathways linking mHtt and the HD phenotype. Work on identifying the modifier genes that impact the rate of symptom development in patients who begin with the same number of CAG repeats is ongoing. If genes that delay onset can be identified and mimicked pharmacologically, this could offer another array of potential targets.
In models, mutant Htt is toxic to neurons and glial cells via multiple routes of impact: It is known to alter gene transcription, dysregulate NMDA receptor functioning, impair potassium channel function and glutamate uptake; interfere with mitochondrial processes; trigger neuroinflammation; and to impair neurotrophin transport/availability.
The mechanistic hypotheses include:
1) that mHtt aggregates are harmful, a direct participant in the neurodegenerative process. Mutant Htt is cleaved into fragments that clump together, forming inclusion bodies. Additionally, mRNA is produced that leads to the production of toxic fragments within neurons.
Because the presence of mHtt aggregate fragments can occur without neuron loss, and conversely, neuron loss can occur without aggregates being present, some have postulated that it is mHtt in soluble, oligomeric form that is actually the toxic actor, reminiscent of the debate about the different forms of beta-amyloid in Alzheimer’s.
2) Other abnormal proteins are produced as a result of the presence of mHtt, including polycysteine and polyalanine, that themselves form toxic aggregates.
3) HD could involve a loss-of-function mechanism, wherein normal Htt’s prosurvival effects are lost, accelerating the slide down the ‘slippery slope’ of cell death. However, while mutant Htt outweighs Htt in HD patients, Htt still exists in considerable proportion. We had previously noted that the initial Roche/Ionis ASO data suggested that huntingtin could be significantly reduced without losing necessary functions provided by the wild-type protein. But the final data pointed towards the opposite, because the high-dose group in fact did worse than placebo, which could mean that there was a salient loss-of-function.
Beyond mutant huntingtin itself, the question is begged as to whether any of pathways downstream of its direct effects may offer a useful target for intervention. Among the several downstream processes that have been suggested as potential intervention points:
1) Work has linked mutant huntingtin to activation of the jnk inflammatory/apoptotic cascade.
2) Several mitochondrial protection points have been cited: A Cornell group published research in Nature Medicine that connected mutant huntingtin to cell death due to mitochondrial damage via DRP1, and a University of Barcelona team published work in 2020 also linking DRP1-driven mitochondrial fragmentation in mouse models of HD.
3)UCLA researchers suggested that mutant huntingtin reduces the flow of neurotrophic factors to oligodendrocytes, and that resultant demyelination is part of the process that leads to HD symptoms. Blocking that inhibition, or providing compensatory trophic support, could be a therapeutic option. This is a loss of function premise, where the absence of intact wildtype huntingtin diminishes the production of BDNF, which would otherwise play a neuroprotective role.
4) Huntingtin also plays a role in neurogenesis vis-a-vis the generation and differentiation of new neurons, thus mHtt may lead to a loss of neurogenic functions.
5) The caspases that cleave huntingtin take a role parallel to the secretases in Alzheimer’s, contributing to the aggregation of malformed proteins. Caspases 3 and 6 have been cited as putative culprits in this process, and their inhibition (usually facilitated by normal Htt) is cited as a potential treatment strategy. Zenobia Therapeutics had long ago identified caspase-6 as an excitotoxic bridge from mHtt to the HD phenotype, and reported that preventing caspase-6 induced cleavage reduced mHtt fragment accumulation and slowed the emergence of behavioral signs in animal models. However, this work did not come to a successful end, and Zenobia did not survive.
Chronic Administration: Gene-Editing
The road most frequently taken has been to knock-down the production of mutant Htt, reducing its pathological aggregation, while hopefully allowing for the continued expression of wild-type Htt (the proportions of Htt to mHtt vary between patients and between brain regions) for cell maintenance. The two main tactical paths are to either provide chronic administration of products (e.g. antisense oligonucleotides/ASOs) that temporarily reduce mHtt production, or to provide a gene that permanently alters the ratio of mutant to wild-type Htt production, in a more classic rendition of gene therapy.
Originally, it seemed impossible to parse out mHtt as a class, it exists in several isoforms, and knocking down all of them requires reducing huntingtin levels overall, including the wild-type form. In the past, NIR had espoused the concern that ‘chronic downregulation of all huntingtin protein, including wild-type, would probably be disastrous, due to the deletion of huntingtin’s role in development and neurotrophin production; huntingtin functions as a prosurvival factor.‘ Part of the thinking around the negative impact of reducing Htt overall was due to the role wild-type Htt plays in early neural development, but studies of adult neural function in animals has indicated that Htt is less critical at later stages of maturation. For the most part, companies have operated on the belief that a nonselective but partial ‘knockdown’ might attenuate the pathological effects of mutant Htt while permitting sufficient wild-type Htt activity to foster needed neural functions.
A pioneering program in this category is Roche/Ionis’ intrathecal antisense therapeutic, tominersen, which is not isoform-selective. For a while, it looked like a therapeutic miracle was in the making: In 2018, tominersen/RG6042 made headlines when the results from a 46pt Phase I/II trial were announced. The five dose levels produced reductions in overall huntingtin levels ranging from 20-42%, compared to a mean increase of 10% in placebo patients. The reduction in mHtt in the two highest dose cohorts averaged 40%, with the highest reductions around 60%. While no overall differences on functional and cognitive measures were seen between the tominersen and placebo groups, positive posthoc correlations were found between the amount of mHtt reduction and improvement on the composite UHDRS, motor functioning, and a measure of visuomotor processing. One anomalous finding was for neurofilament light chain (NfL), which had been expected to decrease over time with tominersen. Instead, there was a transient increase in neurofilament levels, which eventually returned to baseline. The high-dose groups also showed MRI signs of brain volume loss. But Roche set these biomarker anomalies aside, believing that the cognitive results and the seeming lack of safety issues rendered this ready for PhIII.
Roche and Ionis quickly moved into what was to be a 909pt (up from an original goal of 660) Phase III trial. There were two dose-levels; intrathecal injections either every eight weeks, or every sixteen, with outcome measured at 101 weeks. After 791 patients had been enrolled, and 17 months of treatment, an enormous scale for a HD trial, Roche pulled the plug after a DSMB finding of futility. Not only was there no benefit for disease-arrest, but the higher-dose, every-eight-week group performed worse than placebo (the lower dose also trended negatively), and the high-dose was linked to more adverse effects, with increased, albeit reversible, ventricular enlargement, and increases in neurofilament light chain that were worst in the high-dose group. The reason for this disease-worsening is unclear, though excessive knock-down of wild-type Htt is one hypothesis. That had seemed to be the end for tominersen, but subsequent post hoc data mining yielded a signal of benefit in younger, prodromal HD patients with a lower disease burden. In January, Roche began enrolling a 360pt PhII, two lower dose-levels administered every four months, top-line data expected in mid-2025. But with the deleterious impact seen in the previous study, we expect interim analyses to be deployed, with fatal effect.
Then there is Wave Life Sciences, partnered with Takeda, whose raison d’être is specificity. Wave has focused on the development of stereopure ASOs, aimed at specific SNP variants. By targeting those specific mutations, Wave Life Sciences believed that wild-type Htt is spared, avoiding a loss-of-function scenario.
The SNP-1 (WVE-120101) and SNP-2 (WVE-120102) candidates entered the clinic in 2017, and originally were expected to produce topline results in early 2019. However, by the end of 2021, they had produced initial results from just the first 39 patients enrolled in the 120102 trial. That data was disappointing: The mean change in mHtt across the four dose groups was just 12.4%, and surprisingly, there was no change in total (Htt and mHtt) protein levels. There was also no impact on neurofilament/NfL. Wave stated that there was evidence that the response was dose-related, and added a top (32mg) group to both active studies, double the previous high dose. That was for naught, one week after the news about tominersen’s PhIII, Wave announced that both of their higher-dose subtype-selective trials had also failed to show any benefit. Wave attributed that failure to a dosing/potency issue, and devised a new chemical family that yielded an ASO against SNP-3, found in about 40% of HD patients. That ASO has gone into PhI/II, a single dose cohort showed mHtt reduction and preserved wild-type Htt levels. Additional multi-dose findings are expected soon. It does not appear that Wave is still pursuing SNP-1 or SNP-2 HD.
Ionis had developed allele-specific Htt ASOs as well, these do not appear to be prioritized, none have reached the clinic. However, U.Mass researchers reported last year that they have developed siRNA therapeutics that selectively silence specified SNP’s in mouse models, while maintaining wild-type Htt expression.
The simplicity of oral delivery accounts for much of the appeal of gene splicing modifiers. PTC Therapeutics released interim data for PTC518 (an oral splicing modifier) in June, reporting on 22 early-stage patients at twelve weeks of treatment, the end of the first phase of the trial, aimed at safety and pk measures (the second, nine month phase will focus on functional effects). With a long half-life yielding effects for 72 hours, there was a mean, dose-related reduction of 30% in mHtt levels, a trend towards NfL reduction. While no toxicity has been reported, PTC518 was placed on a partial clinical hold in October 2022, pausing US enrollment, as the FDA reviewed changes in the trial protocol (extending the placebo-controlled portion to a total of 12 months). EU and Australia enrollment continues, completion projected in 2024.
Novartis had its own oral splicing modulator, branaplam, but dropped development in HD due to the side effect profile. Branaplam had previously shown itself to not be problem-free, a PhI/II trial in Spinal Muscular Atrophy had been stopped due to peripheral nerve and spinal cord damage.
Vico Therapeutics has developed an ASO that blocks CAG repeat expansions in both HD and Spinocerebellar Ataxia (SCA1 and 3). In preclinical models,VO659 selectively knocked down levels of mHtt, Delivered intrathecally, VO659 entered PhI/II this past April, with an enrollment goal of 65 patients. The trial is dose-ranging, with the lowest two doses only administered to SCA1/3 patients, early HD patients will be added when the third dose cohort is reached. The drug is administered four times over thirteen weeks, patients then tracked for 23 weeks. Completion is expected in 3Q:25.
Alnylam has spent more than a decade developing a siRNA program (ALN-HTT) that preclinically, ameliorated pathology when infused into the striatum. This program was specifically brought into the joint development portfolio when Regeneron partnered with Alnylam is 2019, but it remains in preclinical stage. The work has evolved into the development of a C16-siRNA conjugate that has yielded (via a single intrathecal infusion)Htt lowering for three months, up to 90% reduction in the cortex, at a level that was tolerated in a mouse model. Six month animal results were similar, there were no adverse event findings in terms of NfL elevations or histopathological changes. The group has not yet assessed the impact on phenotypic features and whether there is sufficient activity in the putamen to produce benefit.
Biogen and Atalanta include HD in their partnership, a discovery-stage project aimed at utilizing divalent siRNA (two connected siRNAs) in order to decrease Htt production. In mice, Htt production was reduced for six months after a single intrathecal administration. Htt reduction was also reported in non-human primates, with broad CNS distribution. In a 2019 paper, the group claimed ‘minimal off-target effects’, though the broad and nonselective impact on Htt begs the question of whether longterm Htt suppression could prove problematic.
Sangamo’s zinc-finger DNA-binding protein strategy has been pursued in Huntington’s. CHDI provided $1.3 million to fund discovery work on ZFP transcription factors targeting mutant huntingtin expression, and Shire added this target to a 2012 deal covering monogenic disorders. In April 2019, Takeda/Sangamo reported that they had screened 1600 ZFP candidates and identified a lead. Their HD-patient iPS cell data showed that they could knock down mHtt levels by up to 99% while not excessively impacting (86% preserved) wild-type Htt. In mouse testing, this mHtt ZFP was delivered via an AAV vector, and produced similar reductions in mHtt, lasting for 33 weeks post-injection, and functional benefit was observed. An IND had originally been anticipated in 2015, but while the work was published in 2019 (Nature Medicine) the program is in limbo, still preclinical stage some eight years later.
In September, Roche partnered an undisclosed, early-stage Ionis RNA-targeting program relevant to HD.
Takeda and Anima Biotech have a discovery-stage program developing small molecule mediators of mRNA production of mHtt, reported to be highly-selective for the mutant protein and tissue-specific.
Alchemab, taking a page from Neurimmune’s work on disease-resistance in Alzheimer’s (which, judging by aducanumab, has not panned out that well) has identified endogenous antibodies in HD patients who are progressing more slowly than one would expect based on their genetic loading, on the premise that these antibodies (not found in controls or patients with normal progression rates) may provide resiliency. ATLX-1095, which binds to mHtt aggregates and facilitates microglial clearance, is in preclinical development.
Exicure had a discovery-stage program involving a spherical structure for Htt oligonucleotides they believe would enhance uptake in deep brain regions. Ipsen had signed a collaboration agreement in 2021, but as Exicure entered a period of fiscal peril and corporate reorganization, it ceased its internal work on the project just over a year later. The collaboration was dropped, though Ipsen retained the option to re-enter that agreement. Considering the dire straits where Exicure now finds itself, we suspect that is a moot point.
The emergence of RNA-targeting programs, and gene editing techniques such as CRISPR, offers the tantalizing prospect of editing-out gene defects–though it also raises the spectre of potential off-target effects. A Swiss group reported that they had been able to almost completely stop the production of aggregating mHtt in transgenic mice using CRISPR, reducing it by 90%. However, that work also suppressed wild-type Htt, which did create a loss-of-function issue. That U.Lausanne group hopes to edit the Htt gene at targeted SNPs, similar to what Wave Life Sciences sought to do, leaving the production of wild-type Htt intact. Based on the paucity of follow-up publications, that is still a work in progress.
In 2022, a UCSD group (Morelli et al) published preclinical work regarding a (CRISPR-cas13) system they used to reduce the production of mHtt. Administered to HD-model mice via AAV into the striatum, they showed reduced mHtt RNA and mHtt, with functional benefit and decreased brain atrophy observed as well over eight months. No off-target adverse effects were reported.
Another group (Yan et al, 2023) ran a CRISPR-Cas9 study in pigs to assess whether they could safely knock down mHtt levels. They claimed that delivering the AAV via stereotaxic injection, or via IV administration, reduced mHtt levels and produced functional benefit. The report that IV administration provided sufficient therapeutic levels without toxicity requires replication.
‘One-Shot’ Gene Therapy
UniQure: With semi-disaster having befallen both the Roche/Ionis and Wave Life Sciences Huntington’s programs, UniQure has achieved a higher profile in HD. One substantive vulnerability in the ASO story has been the reliance upon CSF delivery, a systemic approach whose lack of selectivity opens the door for both efficacy and safety issues. While the stereotactic gene therapy delivery used by UniQure has not been a panacea in Parkinson’s, UniQure’s protocol offers a degree of mechanistic precision lacking with its predecessors. AMT-130 utilizes a convection-enhanced AAV5 vector to deliver micro-RNA into the striatum with MRI guidance, three bilateral injections into the putamen and caudate nucleus, two dose-levels tested in PhI/II. Patients must have 40 or more CAG repeats to qualify. The goal is to block huntingtin expression where knockdown of mutant huntingtin might be most useful, while hopefully avoiding a potentially problematic knockdown of wild-type Htt throughout the parenchyma. If successful, this would constitute a one-shot gene therapy. On the other hand, it is not reversible, unlike the transient impact of ASOs on gene expression. The degree of technical skill required by the CED-facilitated neurosurgical implantation procedure is intimidating. A third cohort will explore shortening the time required for the implantation procedure.
Since this is a nonselective tactic, reducing both wildtype and mutant Htt, the possibility exists that this broad-spectrum suppression could have negative effects, as occurred with tominersen. UniQure believes that its stereotactic, convection-enhanced dosing skews knockdown to improve tolerability, reducing striatal Htt levels by 50% to 75%, but only by 25-50% in the cortex, hopefully avoiding a loss-of-function issue.
The first twenty patients have been treated, beginning with ten in the low-dose cohort, another ten in the high-dose. Ten of the twenty have received sham injections–there are another six to be enrolled in the high dose group. At twelve months post-infusion, both the high and low dose groups showed slowing of progression compared with the sham control group and historical controls. The good news in June 2023 was that the low dose of AMT-130 at two years slowed progression on the Total Motor Score scale, and the eighteen-month data for the high-dose patients indicated a dose-related improvement in slowing. Total Function and cUHDRS scores were also preserved for both dose-levels.
However, the two main CSF biomarkers, neurofilament light chain (NfL) and mHtt readouts were contradictory: After a transient post-surgical spike, NfL levels decreased from baseline, 24 month data for the low-dose group showed reductions compared to both the sham control and historical trends. However, mHtt levels increased by almost tenfold (40% vs. 4.7%) more in the high dose group compared with the control group at twelve months. There was considerable variability in the high dose group, which UniQure tried to explain as reflecting limitations of the mHtt assay process. However, the high mHTT readout in some high-dose patients caused some consternation among analysts, with UniQure’s valuation promptly cut in half.
There is a second, open-label PhIb/II trial that will enroll 15 additional patients, six in the low-dose cohort, nine receiving the high-dose.
Voyager Therapeutics’ VY-HTT01 used AAV9 to deliver an RNAi payload knocking down Htt production. In nonhuman primate studies, VY-HTT01 was reported to reduce Htt mRNA by 67-68% in striatal regions, 32% in cortical samples. It was one of the programs optioned by Sanofi, only to have those rights returned. Voyager terminated the program in light of a new inhouse generation of capsids that, in one animal model, provided a thousand-fold improvement in brain transgene expression compared with AAV9 vectors. The revised HD program is aimed at using an allele-specific, novel AAV capsid that spares wild-type Htt, can be delivered via IV, and has diminished liver targeting, enhancing its tolerability. The siRNA payload-in-development also targets MSH3, an enzyme involved in the pathological expansion of CAG repeats, and hence affecting the onset of manifest disease. The new HD capsid program is still in discovery stage.
AskBio (aka Asklepios Biopharmaceutical), acquired by Bayer, has an AAV-delivered gene therapy program in Huntington’s. AB-1001 is administered via bilateral injections in the striatum. A five patient PhI/II is fully enrolled, and includes some preliminary efficacy measures at 52 weeks, with longterm followup going out another four years. When two optimal dose levels have been identified, 3-6 patients will be enrolled in each cohort. When an optimal (in terms of dose-limiting toxicities) dose is finally selected, six additional patients will be enrolled at that dose.
Roche has an RNA editing partnership with Shape Therapeutics, which features Shape’s AI-driven AAV capsid platform, that they believe allows them enhanced tissue specificity. Shape emphasized Alzheimer’s and Parkinson’s, but probably also includes Shape’s program targeting Htt. This work is still in discovery stage.
Spark Therapeutics, spun out of the Children’s Hospital of Philadelphia, and acquired by Roche in 2019 for $4.3 billion, had a preclinical stage AAV-delivered gene silencing program, SPK-miHTT, for Huntington’s. In non-human primate testing, they knocked down mutant Htt levels up to 90%. We had previously wondered how Roche would handle this program in light of tominersen, but despite tominersen’s last ditch resuscitation, Roche is surely accentuating its next-gen alternatives.
Affiris had two mHtt vaccine candidates in early preclinical stage testing, one of which is an active vaccine (‘bi-epitopic’), one is a mAb. Preliminary animal findings for mAB C6-17 showed reduced mHtt and improved functionality, but that program appears to have been shelved, Affiris has shifted its inhouse programming to hypercholesteremia.
CAG/MMR
A MGH group (Gusella et al) found CAGs are altered by other genes, with increased expansion impacting the age of symptom onset; such genes are associated with DNA maintenance and repair (MMR, DNA mismatch repair). This makes them potentially risky as targets for intervention, since off-target effects could have unforeseen, potentially disastrous results. There are companies that are seeking to mediating the expansion of CAG repeats via intervening at MMR, though their track record is mixed at best.
Promising work on MMR malfunction has been done by LoQus23 Therapeutics, funded by Dementia Discovery Fund and Novartis Venture Fund. LoQus23 has collaborated with HitGen in screening the latter’s library for small molecules that may impact the MMR process (via MSH3) and thereby ameliorate its impact on CAG expansions and symptomatic onset. Somewhat surprising, given Pfizer’s withdrawal from neuroscience, is the fact that a Pfizer group has also been screening MSH3 inhibitors. They report having identified MSH3 selective small molecules, and are now trying to optimize them.
A UC Irvine group has reported that the enzyme PIAS1 mediates the accumulation of toxic mHtt aggregates; knocking down PIAS1 reduces aggregation and improves the behavioral phenotype in animal models. The group also linked PIAS1 to DNA damage repair pathways via the enzyme PKNP, suggesting a dual function for PIAS1 knockdown. The question is how to accomplish PIAS1 reduction safely.
Triplet Therapeutics was focused on the upstream role of genes within the MMR pathway in spurring increases in CAG repeats in HD, and the use of ASOs and siRNA to achieve a partial knockdown of these genes. TTX-3360 was an ASO that modified genes governing repeat expansions (Rapid Expansion Destruction, or RED), with Huntington’s the first indication. Triplet raised $59 million, and ran preclinical testing in iPS cells and mouse models, as well as in nonhuman primates, achieving significant knockdown in the striatum and other target regions. Triplet had projected initiating clinical testing in 2021, but had trouble raising additional funding after initial studies indicated the ASO had some neurotoxicity. In late 2022, the investors pulled the plug and shut down the company.
Locana Therapeutics, using technology spun out of UCSD, was using CRISPR to target mRNA for genes mediating REDs in Huntington’s. This program never made it past early preclinical development before Locana went out of business.
Targeting mHtt
As a monogenic disorder, Huntington’s epitomizes a context where genetic manipulation might eventually provide an upstream route to disease-modification, even a cure. The latter is raised by the tantalizing prospect of intervening at the very root of the pathophysiological process. For all neurodegenerative diseases that involve a toxic protein motif, the eventual emergence of overt symptoms belies the decades of aberrant aggregation that preceded it. This is near certain in Huntington’s, where there is no doubt as to the identity of the lethal ‘weapon’. Even if mutant huntingtin production can be safely reduced, the question remains as to how one contends with the ‘toxic waste’ already embedded in critical brain regions. If gene therapies aimed at mHtt production were to hit their mark, the production of mHtt is very unlikely to be completely shut down, and the neutralization of existing mHtt could still be of clinical significance.
Having said that, this could be a time-limited, generational issue. Unlike other neurodegenerative disorders where identifying patients at risk, before they come anywhere near prodromal stage, is not yet feasible, Huntington’s risk can be delineated via genetic testing, albeit with a narrow band of uncertainty around certain parameters of CAG repeats and the potential for repeat expansion. Once a disease-modifier is validated, the dilemma that individuals have faced regarding the utility of testing when no treatment is available would largely vanish: Anyone with a familial history of HD would have genetic testing early on, positive results triggering the initiation of disease-arresting therapy that would prevent the accumulation of mutant huntingtin aggregates from ever approaching neurotoxic levels. But this exists only in theory at present, and targeting mutant huntingtin production has proven to be more challenging than might originally have been thought. Thus defusing and removing toxic elements continues to be pertinent.
Aggregation-inhibition
Vybion developed an antibody (INT41) that they claimed binds to huntingtin and prevents its aggregation, while inhibiting its binding to chromatin and cell membranes. In a mouse model, animals treated with AAV-delivered INT41 show no cognitive deterioration. Vybion had hoped to be in the clinic in 2015, using Convection Enhanced Delivery of viral vectors delivering the gene for INT41, but as of 2018, they were still in preclinical stage, and with no news for five years, their viability appears nil.
Work done at Stanford using the chaperone TRiC showed that TRiC prevents mHtt aggregation. Similarly, a UCSD group reported (in PNAS) in vitro data utilizing a novel model mimicking cortical-striatal interaction, wherein a subunit of TRIC inhibited mHtt aggregation and the resultant impairment of BDNF transport to the striatum. However, the TRIC premise appears to have lost its momentum, perhaps because recent work has suggested TRiC may play a role in uncontrolled proliferation, raising concerns regarding oncogenic effects.
Misfolded Protein Removal
Toxic huntingtin, in soluble and/or insoluble form, can be removed to prevent destructive cellular effects. As is the case with Alzheimer’s and Parkinson’s, toxicity results when such aggregates cannot be eliminated via cellular waste disposal systems, both ubiquitin-proteasome activity and autophagy. Improving either or both of those processes would be an avenue to reducing the toxic ‘load’ placed upon neurons by mHtt.
Ubiquitin-Proteasome
Fostering the functioning of the ubiquitin-proteasome system could be a route to disassembling and removing mHtt aggregates. It has been noted that the proteasome activator PA28 improves cell viability, improving proteasome functioning may promote the removal of mutant huntingtin.
The ubiquitin ligase UBR5 (the gene for which also plays a role in mediating CAG repeats) appears to be a key role to promoting aggregate-clearance. In a c.elegans model, silencing UBR5 production leads to a rapid increase in toxic aggregates. A U. Cologne group has used iPS cells derived from HD patients to confirm that UBR5 depletion is associated with a dramatic increase in toxic huntingtin aggregates. However, those researchers have been working with a silencing model to show the negative effects of deleting UBR5–they have yet to identify a means of upregulating UBR5 levels to foster neuroprotection.
Working from a concept that was initially pursued by CHDI in developing CHDI-709, Arvinas has a orally-bioavailable PROTAC protein degradation program that links soluble mHtt to an E3 ligase in order to tag it for ubiquitination and clearance, before it can begin to aggregate. The clearance spares wild-type Htt. They have shown BBB-penetrance for their prototypes. They have in vitro data suggesting that clearing soluble mHtt also has an indirect impact on the level of insoluble aggregates. This is much earlier in development than their AD/tau program, but the HD PROTAC has shown adequate pk characteristics in mouse models.
Autophagy
PAQ Therapeutics’ ‘ATTEC’ technology promotes autophagy by linking the autophagy regulating factor MAP1LC3B to a neurodegenerative protein. Earlier work pointed to the likelihood that targeting mHtt in Huntington’s could be a first emphasis, but after announcing a $30 million round and a collaboration with Insilico in 2021, there has been no news in two years.
Casma Therapeutics also had an autophagy program against huntingtin, but has become increasingly vague over time, disclosing only an ‘exploratory’ program in ‘Inflammation’.
Rapamycin has been reported to increase autophagy and decreases mHtt accumulation in a HD model, which could be an alternate route to reducing mutant huntingtin, though it appears to only be useful if introduced early in the disease process, and its immunosuppressant effects complicate chronic utilization.
Unfolded Protein Response/UPR
A Spanish group (Lopez-Hurtado et al) reported in 2016 that NovoNordisk’s diabetes drug repaglinide delayed symptom onset and reduced atrophy in a mouse model of HD. Their hypothesis was that repaglinide inhibits a toxic interaction between DREAM (Downstream Regulatory Element Antagonistic Modulator) and a transcription factor (ATF6) associated with UPR. They reported that DREAM-knockout increased survival, though somewhat confusingly, they also reported DREAM levels were lower in HD striatal tissue and in transgenic mice–which they attributed to an endogenous neuroprotective response. In 2019, they reported having developed a better inhibitory DREAM ligand, IQM-PC330, with superior potency and duration of effect. Both IQM-PC330 and PC332 have also been shown to be kv.4 potassium channel blockers. The group has not obtained an industry partnership for this work, and the originators appear to be focused on potential analgesic applications.
Other Disease-Modifying Tactics
Complement Inhibition
The complement system has been implicated in Alzheimer’s and Huntington’s, disrupting the tagging of potentially toxic aggregates, impairing their removal by phagocytosis. Annexon Bioscience has focused on Clq as a key element in facilitating the clearance of damaged cells/components; C1q tags healthy synapses for removal when hyperactivated by stressors, resulting in excessive synaptic pruning and neurodegeneration. Annexon is developing Clq inhibitors that are intended to not impact other components of the complement pathway, allowing those immunological elements to remain functional. ANX005 completed a 28pt, open-label PhII pilot study back in 2022. In this trial, patients with manifest or potential HD received ANX005 IV for twenty-two weeks. Patients appeared to stabilize over the nine months of the study, in contrast to a historical reference group wherein deterioration occurred. The improvement was most salient in those patients with high levels of the complement marker, C4a, which supported Annexon’s targeting premise. However, five patients out of the 28 had dropped out of the study, which raised concerns regarding tolerability. But in the three for whom treatment seemed linked to their adverse events, it turned out that all had a previously undiagnosed autoimmune condition. Future testing will include screening for that biomarker. Annexon states they plan to begin PhIII in 2024, their resources are currently focused on their Geographic Atrophy program. Allyx Therapeutics has not pursued HD with their repurposed complement modulator.
Neuroinflammation/Immune System Dysfunction
Laquinimod, Teva/Active Biotech’s immunomodulator, had been reported to reduce cytokine production and to restore motor function and striatal volume in HD mouse models. Teva ran a 400pt Phase II that came up empty on the primary endpoint after one year of treatment. Active Biotech reported that the trial did reveal changes in markers of neuroinflammation and atrophy, but acknowledged there was no linkage to functional measures.
The kynurenine pathway is disrupted in HD, and it has been hypothesized that inhibiting KMO, an enzyme that produces metabolites, including quinolinic acid, could be useful in HD. Psychogenics developed a mouse model that displays 120 CAG repeats, both cortical and striatal atrophy, and proinflammatory changes in kynurenine metabolism. CHDI (in conjunction with Psychogenics and Evotec) sponsored mouse studies of a KMO-inhibitor, CHDI-340246, and while the inhibitor normalized some electrophysiological measures, no impact on behavior or the rate of progression was seen. Lundbeck developed KMO blockers that target the kynurenine signalling pathway, relevant to inflammation in HD, based on animal models. In 2018, Lundbeck outlicensed the KMO-blocker program to MindImmune Therapeutics, but MindImmune’s focus has been on a different Alzheimer’s program. Kynexis is now developing an old Mitsubishi KAT2 inhibitor, but not for HD.
Excitotoxicity
Calcium influx and excitotoxicity intersect in Huntington’s, and calcium dysregulation can affect the mitochondria permeability transition, leading to cell destabilization. There are also reports that NMDA receptor activity is increased, which could lead to glutamatergic excitotoxicity. It has been suggested that extrasynaptic NMDA receptors are more prevalent in HD, and that glutamatergic activation of these receptors triggers apoptosis.
Axonis has animal model data suggesting that its KCC2 (chloride transporter) targeting program may provide an alternate route to reducing excitotoxicity in HD, but that is very early-stage; their first focus is SCI–PhI readiness is expected a year from now.
A Cornell group had worked on the role of the enzyme transglutaminase-2 in a HD model, noting that overactivity of that enzyme renders cells more vulnerable to oxidative stress. Reducing transglutaminase-2, either via pharmacological or RNAi means, produced neuroprotection in vitro. One program that sought to utilize this pathway in HD came from Raptor Pharmaceuticals, with Procysbi/cysteamine. A French academic group ran a 96pt, 18-month duration Phase II trial in HD. In early 2014, an analysis of the first 89 patients showed a slowing of deterioration of motor function was reported. The effect was more pronounced in patients not also taking tetrabenazine–one hypothesis for this was that tetrabenazine had attenuated visible motor symptoms, providing less ‘bandwidth’ within which Procysbi’s impact could be detected.
Raptor tried to convert its open-label extension into something more definitive by initiating treatment in additional patients, setting up a delayed-start comparison, claiming the results showed a statistical trend towards a slowing of disease-progression. They cited a more robust trend for completers, though this would be skewed by the fact nonresponders would be much less likely to be completers. Raptor also claimed trends on the UHDRS Functional Scale, nothing was said about cognition. At the time, Raptor said that they would proceed into a pivotal trial, but when Horizon Pharma acquired Raptor in late 2016, development in HD ended.
As was noted on p.80, a Cornell group reported that mutant huntingtin triggered mitochondrial fragmentation, both in vitro and in vivo, and that the pathway by which this damage occurs is via an interaction between mutant huntingtin and an enzyme involved in mitochondrial fission, DRP1, work subsequently supported by a University of Barcelona team.
Mitoconix was a Stanford spinout that was working with a dynamin-1 inhibitor they believed inhibited mitochondrial fission. The lead compound, MTC-1203, showed efficacy in mouse models of HD and PD. They were able to raise $20 million in 2017, but the program turned out to not be viable, and Mitoconix shut down.
Metal-Binding
Alterity (formerly Prana Biotechnology’s metal-binding tactic had been proposed to have anti-excitotoxic effects and to keep metal ions from exacerbating huntingtin aggregation. They ran a 109pt Phase II for PBT2 in HD, which yielded only a fluke finding on a single neuropsych measure, and they shifted their corporate focus to MSA.
Epigenetics
Acetylation is suppressed by mutant huntingtin, which in turn leads to the decreased transcription of proteins necessary for cellular survival/stability, including CREB and BDNF. A drosophila model showed that inhibiting deacetylase slows degeneration and cell death, perhaps via enhancing autophagic removal of protein aggregates. Thus HDAC inhibitors could be applicable to Huntington’s, though most existing HDAC inhibitors have BBB access and/or side effect issues. Alkermes has not cited HD as an interest for their preclinical work, nor has Eikonizo Therapeutics, with their preclinical, brain-penetrant HDAC6 inhibitors. Otsuka’s partnership with Eikonizo only mentions ALS in specific, but does cite ‘rare diseases’ as a more inclusive area of interest..
Another approach to epigenetic modifications is Oryzon Genomics’ LSD1 inhibitor vafidemstat. in clinical studies across several therapeutic domains. They have said they have HD animal-model data suggesting that vafidemstat could have utility in HD, but HD is not currently on their expansive menu for development.
OtherTargets
Vaccinex has long been developing pepinemab/VX15, an antibody against the guidance factor semaphorin-4D, which they believe contributes to HD damage via proinflammatory/apoptotic effects. Vaccinex initiated a 116pt Phase II back in 2015, with the antibody administered monthly. However, after the first 36 patients were run, the protocol was amended and another 265 patients eventually enrolled, including both prodromal and early HD disease, receiving monthly IV infusions of pepinemab for eighteen months. Vaccinex reported in 2020 that, while the primary endpoints were missed, the pepinemab group (179 patients) with more advanced, manifest disease showed a trend towards improvement in cognition on two components from the HD-CAB composite battery, However, the scores covered a range from plus or minus just over 0.1 point on the HD-CAB, a minuscule range of no clinical significance. A trend towards benefit in manifest, choreiform disease was also reported on the CGI-C, a measure of general functioning, and on MRI assessment of ventricular expansion and caudate atrophy. However, this was generally seen as a failure. Vaccinex has taken pepinemab into a PhI/II in Alzheimer’s, and cites it as Phase III ‘ready’ in manifest HD. The investigators subsequently published their results in Nature Medicine, unconvincingly claiming that the results justified going into PhIII for HD, but they have backed off, stating that they are evaluating their strategy, in discussions with the FDA about a PhIII. However, they had just $1.9 million in cash mid-2023; without a partner they have no path forward, and none is likely to emerge.
A group at University of Cambridge reported a QC inhibitor decreased mHtt aggregation and cell death in drosophila. Vivoryon/Probiodrug has been developing a QC inhibitor (PQ912) for AD, but has not cited any interest in HD.
Other
A University of Padua group reported in mid-2023 that boosting the level of MTF1, a transcription factor, counteracts the toxic effects of mHtt, including oxidative stress. In transgenic mice, MTF1 improves motor function and reduces mHtt aggregation.
A UCLA group had reported back in 2014 that ATM (ataxia-telegiectsia-mutated) kinase activity is elevated in HD mouse model cells and appeared to contribute to mHtt-triggered cytotoxicity. Developing selective AMT-kinase inhibitors has been challenging, due to the need to avoid a kinase essential to autophagy. But in 2021, a CHDI/Charles River consortium reported success, producing more selective ATM inhibitors that are orally bioavailable and have superior pk characteristics.
Durably Under-the-Radar
CHDI has had a very longterm collaboration with Evotec assessing targets and screening small molecules for HD. This partnership began in 2006, and was most recently extended from 2018 to 2023. There has been no announcement yet as to another extension, and little externally discernible progress. The collaboration has yet to bring anything into the clinic, and Evotec continues to state that “With access to these resources, CHDI has all the tools in place to rapidly discover novel drugs against Huntington’s disease targets and further optimise them to the point of clinical development.” To this point, the alliance is a prime example of patience, with the first tangible payoff the development of a PET-ligand for mHtt (Compound (11C)-2) that is selective for mHtt over amyloid or tau aggregates, and has performed well in nonhuman primates. The next step is clinical validation.
CHDI is nothing if not patient: They have also had a collaboration with Charles River Laboratories since 2005, which has yielded some novel targets, as well as animal models and molecular screens. The collaboration recently reported on the development of ATM (ataxia telangiectasia-mutated) kinase inhibitors as therapeutic candidates in HD. The first generation of molecules risked downregulating autophagy, they have now devised optimized candidates that may be selective enough to avoid those off-target effects. CHDI has also long collaborated with Psychogenics on the development of HD mouse models and the screening of candidate compounds.
Neurotrophins for Neuroprotection
BDNF is an important contributor to neural development and neuroprotection, and normal huntingtin upregulates BDNF. The loss of wild-type huntingtin means less BDNF is available, and the brain is left more vulnerable to degeneration. It should be noted that this is not universally accepted: A Northwestern U. group suggested that it is not BDNF deficiency that underlies decreased neuroplasticity in HD, but that it is instead the dysregulation of p75 neurotrophin receptors that impairs plasticity. But for those focused on the BDNF angle, BDNF could in theory be introduced via viral vectors or engineered cells, albeit with the difficulties attendant to those technologies. One study used AAV to introduce BDNF expression in the subventricular zone, and in a mouse model, increased striatal neuron survival. INSERM used AAV to deliver genes expressing the neurotrophic factor CNTF, and in a rat model, provided neuroprotection to the striatum. The late Ceregene sponsored a rodent trial where AAV delivery of neurturin to the striatum produced neuronal protection and functional improvement, but funding issues shelved that work. Another Ceregene study, delivering GDNF via AAV, improved behavioral function in mice.
A UC Davis group has used mesenchymal stem cells to produce BDNF in mouse models of HD, while a Korean group (CHA University) reported that BDNF-overexpressing neural stem cells, implanted in a HD rat model, interwired and produced functional improvement.
A U. Penn group reported in 2014 that mTORC1 activity is impaired in HD, and upregulating it by introducing Rheb in mouse models reduced atrophy. A Kyungpook University group (Nam et al) has reported that AAV delivery of the gene for Rheb upregulates neurotrophic factors and is neuroprotective, including in Huntington’s.
Cell Therapy
Cell replacement therapies have not received anywhere near the attention in Huntington’s that they have in Parkinson’s. Most cell-implant concepts for HD have focused on the striatum, which would be primarily relevant to motor symptoms, not the cognitive or psychiatric deterioration also seen in HD. In a review of the HD cell therapy literature published in 2021, Bachoud-Levi reported that of the 70 HD patients treated with fetal cell therapy in open-label studies since 1990, four showed durable improvement lasting over six years, and this includes a major study (MIG-HD) that enrolled 45 patients. There were several methodological issues that complicated the interpretation of these results, and the perennial questions regarding stem cell therapy in all CNS contexts apply here: What are the right cells, at what level of differentiation, implanted how and where in patients at which level of disease-progression, that provide a better likelihood of positive outcome? A 2023 publication of mouse model findings from a University of Rochester group reported that implanting ‘young’ glial cells provided replacement of pathogenic cells delayed disease progression. But that is a long way from testability in humans.
California’s CIRM has provided over $27 million in grants for HD cell therapy work, including $12 million to a UC Irvine group that reported their human neural stem cells reduce mHtt accumulation in mice, and improve functioning. As was noted above, a CHA University group reported functional benefit from the implantation of BDNF-overexpressing neural stem cells in a rodent model of HD, though that would seem a neurotrophin-delivery method, not a means of achieving structural interwiring.
The question is also begged as to whether these cells can survive. A study of fetal cell implants reported that grafts in the putamen and caudate survived but did not interwire with host tissue, and did not provide any clinical benefit. Another trial reported post-mortem data on three patients, implanted with fetal grafts a decade or more before. Some interwiring was noted, but the grafted tissue then degenerated more rapidly than did the host cells around it, suggesting that microglial activation produced a hostile environment within which the cells could not survive and function. A review by Gogel and Gubernator concluded that HD cell implantation offers temporary symptomatic improvement, but no lasting benefit, due to a lack of longterm interwiring and survival, the latter attributed to microglial activation and excitotoxicity.
An Emory group combined gene therapy (RNAi-induced lowering of mHtt) and cell therapy (neural progenitor cells engineered to differentiate into GABAergic neurons), and reported that HD-model mice showed ‘significant improvement in motor functions’ and in lifespan when they received both–more than when they received the cell implants alone. The implanted cells in fact differentiated into multiple neuronal types, thus they were sufficiently integrated into the ‘locale’ that their development was reflective of local signals.
Programs to Watch
Symptomatic
Sage’s SAGE-718 has promise as a cognitive enhancer in HD; two PhII trials should have results coming next year.
Prilenia Therapeutics has postulated disease-modifying effects for pridopidine, but based on the compound’s lengthy history, one can argue that they would be fortunate to find any symptomatic benefits. Their large PhIII with top-line results failed earlier this year, Prilenia has not fully disclosed the post-hoc signals they claim supports pridopidine in patients not receiving neuroleptic or anti-chorea drugs. It must be wondered how long the funding leash will be for Prilenia.
Disease Modifiers
Our optimism regarding nearterm disease-modification in Huntington’s vaporized in the wake of tominersen’s meltdown, and Roche’s decision to go into another PhIII with early-stage patients strikes us as an attempt to traverse an unbridgeable gap between efficacy and safety–we will not be surprised if the trial ends up terminated early.
Selectivity has not been a panacea, given the outcomes for the first two Wave Life Sciences oligonucleotides, we do not expect anything better from the third.
UniQure’s AMT-130 program has generated some impact on the rate of progression, and it may be that the mHtt elevation seen in the higher-dose cohort is a fluke. But even at best, AMT-130 would be a difficult product to administer, and would not be scalable even to the limited size of the HD population. Voyager’s revised HD program combines mHtt and MSH3 targets, which in theory could augment its impact, but it is very early in development.
Slowing disease progression by modifying DNA mismatch repair (MMR) is an alternate strategy, one that has the advantage of being small-molecule based. LoQus23 and (surprisingly) Pfizer are both active therein, but neither has yet reached the clinic.
Annexon saw enough in PhII that they plan to start PhIII next year with the c1q inhibitor ANX005, and Arvinas has an oral PROTAC selective for mHtt, but is very early in its development. Allyx Therapeutics has thus far been focused solely on Alzheimer’s.
CHDI has been supporting up to 75 FTEs at Evotec for the past seventeen years, eventually we will see what that investment has produced and if it has value.
Prospects
Huntington’s offers more mechanistic clarity than most of its neurodegenerative peers; we have certainty as to the antecedent genetic flaw that leads to the pathology, and there is a level of diagnostic definitiveness that greatly simplifies the task of trial enrollment. But recent failures illustrate the daunting fact that even this level of granular clarity does not guarantee success.
The size of the prodromal and manifest Huntington’s population is in the 150-200,000 range in the US, making this far more than a minor market for any company that can develop a disease-slowing treatment. Such a drug would be administered early, and outside of the one-shot gene therapies, chronically.
In the wake of the success achieved in Spinal Muscular Atrophy, we had said for a few years that Huntington’s would be the next neurodegeneration success story. That is no longer the case. With all due respect to the biomarker work-around in ALS-SOD1, and the perenially disappointing amyloid mAbs, slowing FTD-GRN is more likely to be the next clearcut achievement in CNS disease-modification, before the HD puzzle is solved.
From NP Winter Issue 2023
Lecanemab’s Risk-Benefit Conundrum:
Anti-amyloid mAbs have been the repository of more wishful thinking than any other category of neurotherapeutics. Which accounts for this quote from the CMO for Alzheimer’s UK, who said in reference to lecanemab: “It’s just a first step, but this new Alzheimer’s drug could be a huge breakthrough.” Which to our eyes seems semi-detached from reality, given that there is ample debate as to whether lecanemab’s treatment effect, barely visible to the empirical naked eye, will be perceptible to patients and families beyond hopeful expectancy. Less aspirational and more to the practical point were two critical quotes to be found in a November 27 piece in Science titled ‘Second Death Linked to Potential Antibody Treatment for Alzheimer’s Disease‘, quotes that bring into stark profile the dilemma that prescribers and families are going to face when it comes to lecanemab and, most likely, donanemab as well:
“Even if that (antibody treatment) only means 6 or 12 more months of knowing who their kids are…it’s meaningful to dementia patients and their families.” (University ofKentucky neuroscientist Donna Wilcox)
s“As soon as they put it (tPA) in her, it was like her body was on fire, She was screaming, and it took like eight people to hold herdown.” (the husband of a patient who died)
The conundrum? There may be no more heart-wrenching an anecdotal image of the toll taken by AD than a parent who not longer recognizes their own children; evoking that nightmare scenario is a potent gambit for those advocating the use of an anti-amyloid mAb. But the calculation of acceptable risk is transformed when contemplating the even-more-awful image of an elderly parent ‘screaming…like her body was on fire.’ Many families (and physicians) who might otherwise consider worthwhile any chance of extending autonomy and awareness will recalculate the risk-benefit analysis if the risk includes even a small possibility of such horrific suffering.
The context of the death was this: A woman who had received lecanemab (believed to have been initiated in the open-label extension) developed stroke symptoms and was administered tPA. The thrombolytic triggered extensive cerebral bleeding which led to her death a few days later. The autopsy found signs of cerebral amyloid angiopathy (CAA), which perhaps had weakened the blood vessels, a weakening exacerbated by the impact of the mAb on vascular amyloid, so that tPA then precipitated a massive rupture nd hemorrhage. CAA is another manifestation of amyloid pathology, and is believed to be present in up to half of AD patients, though it is difficult to diagnose other than post-mortem. There was another report of a lecanemab-linked death where a patient had also received an anticoagulant, for atrial fibrillation, and a third death, details unavailable. A subsequent (1/4/23) case report in NEJM suggested that the stroke reflected a causal role of lecanemab in the onset of stroke.
This reminds us that, when discussing the emergence of ARIA-E (edema) or ARIA-H (microhemorrhages), sponsors tend to emphasize that 98-99% of ARIA patients ‘fully resolve’ after the mAb is stopped and treatment is provided. We have always wondered about the other 1-2%, those that do not ‘fully resolve.’ 1-2% sounds rare, but when talking about a disease affecting 6.5 million patients in the US alone, that represents a potentially substantial subgroup. In the case of the lecanemab CLARITY trial, 2.8% of lecanemab patients developed symptomatic ARIA-E, 81% of whom were described as fully resolved at four months, which begs the question of what happened with the other 19%, who comprised 0.53% of the active treatment group; 21 patients whose longterm AE sequelae were not disclosed. This blind spot is from a rigorously executed and monitored study, where the level of clinical oversight surely exceeded what can generally be expected in clinical practice.
It has been suggested that the risk of potentially dangerous CAA would be reduced if amyloid mAB therapy was restricted to patients with mild AD, but according to the Science report, this woman had only retired from a professional position a year before, so this was relatively early, and her CAA was not flagged by Eisai’s screening. APOE4+ patients appear to carry more risk of ARIA-E, but with that genetic loading found in up to 70% of the AD population, that is not a viable exclusionary criterion. Thus it may not be possible to parse out the at-risk population, leaving families and prescribers to contend with the fear that a well-intentioned intervention might have deadly results.
With all due respect to the claims that lecanemab is a potentially “huge breakthrough“, a .45 point impact on the CDR-SB falls well short of the one point difference commonly cited as the Minimum Clinically Important Difference for that measure. Even ICER seemed befuddled: “In aggregate, the net health benefits of lecanemab in patients with early AD may be small or even substantial, but there remains a possibility of net harm from ARIA.” Which covers the full gamut of possible outcomes.
Unlike Biogen with Aduhelm, Eisai did leave the microphones in the CTAD auditorium, though the questions lobbed from the floor were softballs that sounded like pre-specified plants. Eisai pushed back on the safety issue raised by the patient deaths, though they acknowledged that these risk factors would have to be weighed in a treatment decision.
We have expected that a significant subset of Alzheimer’s patients would be willing to try a therapy that could be even modestly useful in extending functional independence, but this presumes that the downside is not overly daunting. The phrase ‘like her body was on fire‘ will reverberate in the Alzheimer’s community, reducing the portion that believes the limited upside gains are worth such uncommon but devastating risks, and deterring physicians leery of taking them on. Even Eisai cited a target market penetrance of just 100,000 patients by year three of the launch, which, given the size of the AD population, is strikingly tiny. Perhaps that explains their pricing Leqembi at $26,500 per year, portraying this as an $11,100 discount to the ‘real”value of the drug, well beyond the $8500-20,600 range estimated by ICER. Eisai seems to expect very limited adoption; if they truly believe in its value, they might have considered pricing it lower and expanding its range.
‘Psychedelics Update’
2022 was a year where the psychedelic therapies movement featured an odd mélange of the credible and the absurd, reflecting the still-early developmental stage of this area. A few examples of each:
Credible: The New England Journal of Medicine, which epitomizes traditional medical practice as much as does any publication, published Compass Pathways‘ detailed findings from their ground-breaking PhII trial of psilocybin in TRD. The detailed findings support our initial comment that ‘COMP360’ is a “tool, not a panacea.” 29% of the high-dose group were in remission after three weeks, not an unimpressive accomplishment in a refractory population, but a panacea would not leave 71% of the treatment population behind. Compass is about to launch a PhIII study that will enroll ‘almost 1000’ TRD patients, adding a second dosing three weeks after the first. Given the complexity of the undertaking, it was reassuring that Compass hired an Otsuka drug executive, Kabir Nath, to become their CEO, an essential addition of industry expertise to a Company that had lacked the organizational coherency required for this developmental stage. The question remains open as to whether Compass’ IP strategy is viable, or whether they are blazing a path that eases the way for numerous competitors as well.
Incredible: The National Park Service issued a Public Service Announcement reminding readers that they should eschew licking the Sonoran desert toad which is a source of the psychedelic molecule 5-MeO-DMT, along with other, potentially less appetizing alkaloids. We suspect that the numbers of toad-lickers is relatively small, but it is a reminder that, unlike any other area of drug development, even as the biopharm industry attempts to systematize and medicalize the testing and eventual deployment of these substances, there is an extensive parallel movement that is geared towards the sacramental, spiritual, and relatively unregulated use of psychedelics. This is a movement that reflects thousands of years of ethnobotanical tradition that intermittently percolates into broader public awareness. NIR is sympathetic to that tradition and its aspirations, but that is not the focus of our coverage, aimed at exploring the possible expansion of the quasi-spiritual into a broader utilization in the treatment of psychiatric distress and illness. We occasionally trigger an ‘OK Boomer’ reflex response from those who believe that the lessons of the sixties and seventies are not relevant to this psychedelic renaissance, and/or that attempts to medicalize psychedelics are greed-driven and antithetical to their consciousness-raising potential. While we disagree, there may be no nearterm route to bridging that divide. But surely we can at least agree on this: Sonoran toads should only be licked if they have provided informed consent and chosen a ‘safe word.’
The Role of Psychotherapeutic Support/Integration:
The degree to which psychedelic therapies rely upon psychotherapeutic support for maximum treatment impact is a critical component of the clinical development process, with adherents on both sides of the argument. Much of the anecdotal literature regarding psychedelics/empathogens from the 1950s and 1960s came from psychotherapists who utilized psychedelics as adjuncts to ongoing therapy, not as stand-alone interventions. This tradition was recapitulated by the first two pioneers to make it into highly controlled clinical testing: MAPS‘ work with MDMA in PTSD, Compass‘ psilocybin trial in TRD. Both of these trials emphasize preparatory and integration work as integral to the therapeutic experience, but this comes at considerable cost: The MAPS MDMA protocol utilized more than forty(!) therapist hours, the Compass PhII trial required at least fifteen hours, which will increase with their PhIII, since that utilizes two dosing sessions. Such therapeutic packages require extensive training to ensure that the therapists follow the protocol, with varying allowances for skilled improvisation. Such protocols, if commercialized, would carry a hefty pricetag, well into five figures. Companies like Small Pharma and Beckley Psytech have also adopted psychotherapeutic support as core elements of their clinical trial structure.
In contrast, MindMed is running its pilot study of LSD in Generalized Anxiety Disorder without any psychotherapeutic prep or integration: A ‘monitor’ sits with the patient during the twelve hour dosing for the sake of safety, but that is it. Somewhat to our surprise, MindMed’s CMO is an eloquent advocate of psychodynamic psychotherapy, but he explains the MindMed approach as zeroing in on the purely pharmacological impact of the LSD session, without conflation with psychotherapy. It is reassuring that this decision was not based on a devaluation of psychotherapy per se, but it is ironic that, despite the embrace of the concept that medication and psychotherapy are best delivered in concert, the program would eschew it in testing this legacy psychedelic. One can make the logical case that this protocol could set the stage for a relatively more affordable psychedelic treatment program, though we question the practicality of such a long-duration agent. Somewhat tongue-in- cheek, we would note that the use of LSD without psychotherapy as anxiolysis does not enjoy extensive anecdotal support, there were no reports of a ‘Great Tranquillity’ during the years 1966-69. Our bias is that, given the generally acknowledged salience of ‘set and setting’, that this sacrifices the potential value promoted by incorporating psychological preparation into that ‘set’ from the outset. Synergy is a terrible thing to waste.
Even as some companies in the US experiment with reducing the psychotherapeutic component, Health Canada has now mandated two therapists be ‘in-the-room’, while setting standards for preparation and integration work. This was partially due to the MAPS PhII scandal where two therapists were grossly inappropriate with a patient, though that episode showed the risks even with a therapist pair. Based on these guidelines, MindMed’s LSD trial would be not be permitted in Canada.
The Most Critical Question in Psychedelics:
Are hallucinations a bug or a feature? Do the erstwhile benefits of compounds aggregated under the umbrella of ‘psychedelics’ reflect an impact upon neuroplasticity, and if so, does such psychoplastogenicity require a psychedelic, hallucinatory experience? While the initial consensus leaned towards the widely held belief that an efficacious therapeutic journey requires a ‘trip’, companies like Delix, Cognesy, Onsero, Psilera, and BetterLife are among those holding that a hallucinatory experience is a nonessential side effect. From a traditional pharma viewpoint, relevant to partnering potential, eliminating the ‘trip’ and turning these molecules into at-home, self-administered pills would be far more familiar, and appealing framework, than figuring out how to commercialize a therapeutic package that must be administered in a clinic setting.
For NIR’s part, we find intuitive resonance in the idea that some cognitive-behavioral patterns are so structurally embedded that they must be disassembled in order to promulgate fresh neural iterations. Our suspicion is that some behaviorally entrenched disorders, particularly addictions that have ‘hijacked’ the neural reward system, may require full-blown ‘demo and renovation’ for lasting change. Indeed, the very promising Johns Hopkins work on smoking cessation and NYU‘s alcohol abuse trials use exceptionally high doses of psilocybin, ranging up to more than 60% above Compass‘ high dose. On the other hand, the dramatic but transient disarray of hallucinatory experience may not be as crucial in anxiety disorders, for example. But this is purely speculation, the proof will be in the data yet to come.
But while we await solidly grounded, empirical evidence, it is hard to shake the feeling that this class of psychoactive drugs is somehow different. Even Onsero‘s founder, the eminent and highly rational neuropharmacologist Bryan Roth, said this at ACNP: “There may be something magical about psychedelics per se.”
