Investing in Neuroscience and CNS: Why the Hardest Field in Medicine Is Becoming Investable

Why does the central nervous system matter as an investment area?

The central nervous system, which for an investor means neurodegeneration such as Alzheimer’s, Parkinson’s, and amyotrophic lateral sclerosis, alongside broader neurology, psychiatry, and chronic pain, sits over one of the largest and most durable pools of unmet need in medicine. Public-health data has long shown that dementia and other neurodegenerative conditions rise steeply with age, and the populations of most developed economies are aging in a way that is already largely fixed, because the relevant cohorts have already been born. Depression and chronic pain, meanwhile, are among the most prevalent and disabling conditions worldwide, and they cut across every demographic rather than concentrating in the old.

What makes CNS distinctive is not only the size of the need but how little of it is well served. Most neurodegenerative diseases still have no treatment that meaningfully changes their course. Much of psychiatry continues to rely on mechanisms discovered decades ago and refined incrementally. Pain management has been dominated by opioids, with the harm that has followed. For a pre-seed and seed investor, that combination, enormous need paired with a shallow standard of care, is the signature of a field where a single mechanistic insight can create a category rather than take share within one.

Why has CNS drug development failed so often in the clinic?

Central nervous system programs have historically carried among the lowest probabilities of clinical success of any therapeutic area, and Alzheimer’s in particular became a byword for late-stage failure. The reasons were structural rather than accidental, and understanding them is the key to underwriting the field intelligently.

First, targets were often correlative rather than causal. A protein that accumulates in a diseased brain may be a driver, a bystander, or a consequence, and for years developers could not reliably tell the difference, so substantial capital was spent modulating biology that turned out to be downstream of the actual disease. Second, the animal and cellular models were poor mirrors of human neurobiology. A compound that rescued a transgenic mouse frequently told you little about a human patient, because the mouse did not have the human disease so much as a caricature of one gene’s effect. Third, and most fundamental, the brain was effectively a closed box. Developers often could not directly measure whether a drug reached its target, engaged it, or altered the disease, so trials read out only years later on clinical endpoints, by which point the money and the patients were already committed. Layered on top of all of this was the blood-brain barrier, which excludes the large majority of drug-like molecules from the tissue they are meant to treat.

Why is the failure rate starting to change?

Three shifts, each maturing over roughly the past decade, are converting CNS from a field that fails for unknowable reasons into one that can fail, and succeed, for understood ones.

The first is human genetics. Large-scale sequencing and genome-wide association studies have produced targets with genetic evidence that they contribute to disease in people, not just in models. Early-onset familial Alzheimer’s implicates the amyloid pathway through mutations in APP and the presenilin genes. Parkinson’s has causal and risk-conferring variants in genes such as SNCA, LRRK2, and GBA. ALS has been organized in part around genes including SOD1, C9orf72, TARDBP, and FUS. A target with human genetic validation is not a guarantee, but it is a fundamentally different starting point, because the biology has already been implicated in the only organism that ultimately counts.

The second is biomarkers. Fluid measures such as cerebrospinal fluid and, increasingly, plasma phosphorylated tau, amyloid ratios, and neurofilament light chain, together with amyloid and tau PET imaging, let developers observe whether a drug is engaging its target and whether the disease is bending, in some settings before a clinical endpoint would declare a result. This shortens feedback loops, lets small early trials carry real information, and allows enrollment of patients defined by their biology rather than by a syndromic label alone. The third is modality. Antisense oligonucleotides, small interfering RNA, gene therapy, and engineered antibodies have given developers ways to act on targets that small molecules could never reach, and several have now shown that the CNS can be druggable when the mechanism and the delivery are right.

What has actually worked, and what does it teach?

The proof points are recent but real. An antisense oligonucleotide delivered into the spinal fluid helped transform spinal muscular atrophy from a frequently fatal infantile disease into a treatable one, and an ASO directed at SOD1 has been approved for that genetic form of ALS. Amyloid-targeting antibodies, after a long history of failure, have produced the first therapies shown to slow the rate of cognitive decline in early Alzheimer’s, lending support to a form of the amyloid hypothesis and to the biomarker-defined trial design that tested it more rigorously. In psychiatry, rapid-acting mechanisms such as ketamine and its S-enantiomer esketamine, and a neuroactive steroid approach to postpartum depression, interrupted a long drought of genuinely new mechanisms, and a muscarinic approach to schizophrenia has shown that antipsychotic efficacy is achievable without relying on direct dopamine-receptor blockade and its associated side-effect burden. In pain, selective sodium-channel inhibitors aimed at Nav1.7 and Nav1.8 have pursued analgesia through peripheral mechanisms intended to avoid opioid addiction liability.

The common thread is instructive for an early-stage investor. Each success rested on a target with meaningful human validation, a way to reach the relevant tissue, and, increasingly, a biomarker or genetically defined population that let the trial ask a precise question. Where any one of those legs was missing, the program tended to fail regardless of how elegant the underlying science was.

How much does the blood-brain barrier still matter?

The blood-brain barrier remains the defining physical constraint of the field, and any serious CNS company has an answer to it. The barrier is formed by tight junctions between the endothelial cells of brain capillaries and reinforced by efflux transporters such as P-glycoprotein that actively pump many molecules back out. It excludes the large majority of small molecules and essentially all unmodified large molecules, which is one reason so many pharmacologically valid drugs never worked in the brain.

The routes across it are now a genuine design space rather than a blank wall. Some programs bypass the barrier with intrathecal or intracerebroventricular delivery, administering directly into the cerebrospinal fluid, which is how the approved CNS antisense oligonucleotides reach their target. Others exploit receptor-mediated transcytosis, engineering a therapeutic to cross on a natural shuttle such as the transferrin receptor, the so-called molecular Trojan horse approach now being built into brain-penetrant antibodies and enzymes. Gene therapy work has focused in part on engineering viral capsids that cross the barrier after a systemic dose, and focused ultrasound can transiently and reversibly open the barrier at a targeted location. For an investor, the question is never only whether a molecule is potent in a dish. It is whether the team has a credible, ideally validated, plan to deliver an active dose to the right cells in the brain, and how much of the company’s risk sits in that delivery step.

Why are CNS trials so long and endpoint-heavy?

Even with better targets and delivery, CNS trials remain among the longest and most expensive in medicine, and the reasons are intrinsic to the biology. Neurodegeneration progresses over years, so a disease-modifying claim often requires long follow-up to detect a change in the slope of decline. The endpoints that regulators and payers accept are frequently clinical rating scales that are noisy, subject to placebo effects, and slow to move, which forces large sample sizes and long durations. Psychiatry carries its own version of this, with high placebo response rates that can obscure a real effect and endpoints that depend on subjective report.

This is precisely where biomarkers change the economics rather than just the science. A program that can show target engagement in the fluid, then an effect on a progression marker such as neurofilament light or a tau species, can generate a meaningful signal in a smaller, faster study, and can enrich its trial with patients whose biology matches the mechanism. It does not eliminate the eventual need for a clinical endpoint, and regulators generally still require one for full approval, but it moves the first moment of truth earlier and cheaper, which is what makes a capital-efficient early-stage plan possible. When Sonnerie looks at a CNS spinout, the biomarker strategy is not a scientific nicety. It is central to whether the company can reach an inflection on an amount of money that a seed syndicate can realistically assemble.

What does a fundable pre-seed CNS spinout look like?

A CNS company we can underwrite at the first institutional check tends to combine a small number of specific things. It starts from a target with human genetic or strong human biological validation, so that the central bet is on execution and delivery rather than on whether the biology is real at all. It has a modality matched to that target, whether small molecule, ASO, gene therapy, antibody, or otherwise, chosen because it fits the biology rather than because it is the founder’s favorite tool.

It has a delivery answer for the blood-brain barrier that is explicit and, ideally, already supported by data. It has a biomarker and patient-selection strategy that lets the first clinical study read out on something faster than a multi-year cognitive endpoint. And it usually has intellectual property anchored in a university lab, with a clean or cleanable license, which is the origin point Sonnerie is built around. The most common failure mode we decline is not weak science. It is strong science with no credible path to the brain and no way to see an early signal, which together can guarantee a long, expensive trek to a binary readout that a pre-seed company cannot finance.

How does Sonnerie evaluate a CNS opportunity?

We evaluate CNS spinouts through the same lens we bring to the rest of healthcare, sharpened for the specific realities of the brain. We look first at the strength of target validation, weighting human genetic evidence heavily and treating model-only rescue with caution. We then ask how the company crosses the blood-brain barrier and how much of the total program risk lives in that step, because a delivery problem discovered in the clinic is among the most expensive kinds of surprise in this field.

We scrutinize the biomarker and endpoint plan, because it shapes both the scientific quality and the capital efficiency of the first trials, and we assess the founding scientist and the operating team against the translational distance still to travel from bench to a clinical asset. Because we write the first institutional check into university spinouts, our edge is being close enough to the source to read the science before it becomes a deck, and operator-led enough to help a founding scientist build the company around it. In a field this hard, the signal is loudest early, in the genetics, the delivery data, and the biomarker design, well before a priced round arrives to tell everyone what the market has already decided to believe.