98% of Drugs Cannot Cross the Blood Brain Barrier. Denali Just Got One Through.

The Blood Brain Barrier Problem

On March 25, 2026, the FDA granted accelerated approval to AVLAYAH (tividenofusp alfa eknm), an enzyme replacement therapy from Denali Therapeutics for neurologic manifestations of Hunter syndrome in pediatric patients. AVLAYAH is the first FDA approved biologic specifically designed to cross the blood brain barrier and reach the brain after intravenous administration. No biologic had previously achieved this.

Six days later, Eli Lilly announced it would acquire Centessa Pharmaceuticals for up to $7.8 billion, adding an orexin receptor agonist for narcolepsy to its CNS portfolio. In the same week, a rare disease biotech proved that a large molecule can be engineered to enter the brain, and the world's largest pharmaceutical company bet nearly $8 billion on a CNS drug candidate.

These events are connected. The blood brain barrier blocks 98% of small molecules and effectively 100% of large molecules from reaching the brain. CNS drugs carry a 6% clinical approval rate compared to 13% for non CNS drugs. Alzheimer's disease alone has consumed $42.5 billion in clinical R&D since 1995, with a 99.6% failure rate across that period. The barrier is the common factor.

AVLAYAH treats Hunter syndrome only. Its broader relevance is that it validates a principle applicable across neuroscience: the barrier can be engineered around using receptor mediated transcytosis, and a biologic using this approach now has FDA approval.

Why the Blood Brain Barrier Defeats Drugs

The blood brain barrier is a multi layered defense system, not a single membrane. Brain capillary endothelial cells are connected by tight junctions, protein complexes made of claudins, occludin, and junction adhesion molecules that seal intercellular gaps so completely that even small ions cannot pass between them. That physical seal alone would make drug delivery to the brain difficult. But the barrier adds active defenses on top of it.

Efflux transporters, most critically P glycoprotein (P gp) and breast cancer resistance protein (BCRP), sit on the blood facing surface of endothelial cells and function as molecular pumps. A drug may be lipophilic enough to diffuse into the cell, only to be ejected back into the bloodstream before it reaches brain tissue. P gp alone reduces brain exposure of its substrates by 10 to 25 fold. Metabolic enzymes within the endothelial cells degrade many compounds during transit. Pericytes and astrocyte end feet wrap around the capillaries, providing structural reinforcement and signaling that maintains barrier integrity.

The result is a system that allows oxygen, carbon dioxide, glucose, and certain amino acids to pass while excluding nearly everything a pharmaceutical chemist wants to deliver.

For small molecules, BBB penetration follows physicochemical rules that are stricter than standard oral bioavailability thresholds. The CNS adapted variant of Lipinski's rules requires lower molecular weight (below 400 Da versus 500 Da), fewer hydrogen bond donors, reduced polar surface area, and an optimal LogP near 2. Even within these parameters, only approximately 2% of small molecules achieve meaningful brain exposure. The molecule must be small enough, lipophilic enough to cross the membrane, but not so lipophilic that it becomes trapped in the membrane or triggers P gp efflux.

For biologics, passive diffusion is physically impossible. A monoclonal antibody weighs approximately 150,000 Da, more than 300 times the molecular weight threshold for CNS penetration. This is why pharmaceutical companies have spent decades treating the blood brain barrier as an absolute wall for large molecule therapeutics.

The economics of this failure rate explain why many pharmaceutical companies exited neuroscience entirely during the 2010s. CNS drug development takes roughly 12 years on average, nearly double the 6 to 7 year timeline for non CNS drugs. The clinical approval rate sits at 6%, less than half the 13% rate in other therapeutic areas. For Alzheimer's disease specifically, the period from 2002 to 2012 produced a 99.6% failure rate across all clinical candidates. The BBB is central to these numbers. Drugs that cannot reach the brain in sufficient concentration cannot produce the pharmacological effect needed to demonstrate efficacy in clinical trials.

The Transport Vehicle Platform

Denali Therapeutics was founded in 2015 with $215 million in first round financing, the largest initial biotech round at the time, around a single premise: engineering molecules to cross the blood brain barrier. The company's Transport Vehicle (TV) platform is the technology behind AVLAYAH, and it works by exploiting the brain's own supply chain.

The mechanism relies on receptor mediated transcytosis, the biological process the brain uses to import essential molecules like iron. Transferrin receptor 1 (TfR1) sits abundantly on brain capillary endothelial cells. Under normal physiology, iron loaded transferrin binds TfR1 on the blood facing side of the cell, the receptor ligand complex is pulled into vesicles, trafficked across the cell, and released on the brain facing side. Denali's Transport Vehicle is an engineered antibody Fc fragment designed to bind TfR1, hijacking this shuttle system.

The critical engineering challenge was affinity tuning. The Fc domain must bind TfR1 strongly enough to trigger internalization but weakly enough to release its cargo on the other side. As Denali's chief medical officer Carole Ho described it: "You want it to bind but then come off. We've fine tuned this affinity." Denali used directed evolution to optimize this balance, and crystal structures confirmed that the TV binding site is separate from both the natural transferrin binding site and the FcRn binding site, preserving normal pharmacokinetic properties of the Fc domain.

In AVLAYAH, the TV Fc is fused to iduronate 2 sulfatase (IDS), the enzyme missing in Hunter syndrome. Once across the blood brain barrier, the enzyme is internalized by neurons via the mannose 6 phosphate receptor and trafficked to lysosomes, where it degrades the accumulated heparan sulfate that drives neurodegeneration. The result is a single weekly intravenous infusion that delivers functional enzyme to both the body and the brain.

This approval reflects the FDA's collaborative engagement to incorporate biomarker evidence to help accelerate development of urgently needed treatments.

Ryan Watts, CEO, Denali Therapeutics, March 2026

The clinical evidence supporting FDA drug approval came from a Phase 1/2 study published in the New England Journal of Medicine in January 2026, enrolling 47 male participants aged 3 months to 13 years.

Biomarker results at Week 24 showed a 91% average decrease in cerebrospinal fluid heparan sulfate from baseline, with 93% of patients (41 of 44) reaching CSF heparan sulfate levels within the range of unaffected children. These reductions were sustained through Week 153 at 92%. Peripheral outcomes included an 88% decline in urine heparan sulfate at Week 24 and normalization of liver volume. Serum neurofilament light chain, a marker of active neuronal injury, showed a 21% reduction at Week 49 that deepened to a 76% reduction by Week 153, with 85% of patients reaching normal ranges.

The FDA granted accelerated approval based on the surrogate endpoint of CSF heparan sulfate reduction. AVLAYAH received Breakthrough Therapy, Fast Track, Orphan Drug, and Rare Pediatric Disease designations. The drug is priced at $5,200 per 150 mg vial, with estimated annual maintenance costs between $270,000 and $811,000 depending on patient weight. Continued approval depends on the ongoing global Phase 2/3 COMPASS study comparing tividenofusp alfa against idursulfase.

Competing Strategies for BBB Penetration

Denali is not alone in pursuing blood brain barrier crossing. Multiple pharmaceutical companies and biotech companies are developing competing approaches, each with distinct engineering trade offs relevant to pharmaceutical product development and formulation.

Receptor mediated transcytosis: the leading approach. The transferrin receptor pathway that Denali exploits is the most clinically advanced strategy. Roche and Genentech have developed Brain Shuttle technology, using monovalent anti TfR binding fused to therapeutic antibodies, now in clinical development for CNS targets. JCR Pharmaceuticals received approval in Japan in 2021 for pabinafusp alfa (JR 141), an anti TfR antibody fused to human iduronate 2 sulfatase for MPS II, making it the first TfR based therapy to reach any market, though it uses a different engineering approach than Denali's TV. Aliada Therapeutics is pursuing dual targeting of TfR1 and CD98hc for enhanced brain delivery.

Beyond the transferrin receptor, ArmaGen Technologies developed AGT 181, an anti insulin receptor antibody linked to laronidase for Hurler syndrome (MPS I). The LRP1 pathway has produced Angiopep 2, a peptide conjugated to chemotherapeutics that demonstrated high BBB influx rates in preclinical studies.

Focused ultrasound: breaking the barrier physically. An entirely different approach uses sound waves to temporarily open the blood brain barrier. Insightec's Exablate Neuro system combines transcranial MR guided focused ultrasound with intravenous microbubbles. The microbubbles oscillate under ultrasound energy, mechanically loosening tight junctions in a targeted brain region for a window of hours. The system holds FDA Breakthrough Device designation for glioblastoma BBB disruption, and a Lancet Oncology published trial demonstrated safety with possible survival benefit. Carthera's SonoCloud 9, an implantable nine emitter ultrasound device placed during tumor resection, has completed Phase 1/2 in recurrent glioblastoma: 33 patients received 90 total sonications with carboplatin without dose limiting toxicity. Clinical trials of focused ultrasound BBB opening in Alzheimer's patients are now underway, with a follow on trial planned for August 2026.

Nanoparticle and peptide based delivery. Surface modified nanoparticles conjugated with specific ligands, including peptides, antibodies, and surfactants, represent a growing area of pharmaceutical biotechnology for BBB crossing. Increasing ligand density improves polyvalency and avidity at the BBB surface. Cell penetrating peptides (CPPs) such as TAT peptide (derived from HIV), penetratin, and synthetic sequences facilitate transport across biological membranes. Bispecific antibodies serve as molecular Trojan horses, with one arm targeting a BBB receptor for transcytosis and the other carrying the therapeutic payload.

Each of these strategies in pharmaceutical research introduces distinct formulation challenges. Receptor mediated transcytosis requires protein engineering, affinity optimization, and management of fusion protein stability. Focused ultrasound demands precise microbubble formulation and temporal coordination with drug administration. Nanoparticle approaches must solve particle size control, surface functionalization, and reproducible manufacturing at scale. Each approach depends on formulation science to reach clinical viability.

Big Pharma Bets Billions on the Brain

The Lilly Centessa acquisition fits a broader pattern of pharmaceutical R&D capital flowing back into neuroscience after years of industry retreat.

Centessa's lead asset, cleminorexton (formerly ORX750), is a highly selective oral orexin receptor 2 agonist, a once daily pill for narcolepsy and sleep wake disorders. The science behind it is what analysts call precision neurology. Narcolepsy type 1 results from destruction of orexin producing neurons in the hypothalamus. Rather than managing symptoms with stimulants, orexin agonists directly replace the missing neuropeptide signaling. Cleminorexton showed a potential best in class profile in Phase 2a studies across narcolepsy type 1, narcolepsy type 2, and idiopathic hypersomnia, with superior wakefulness maintenance and no liver toxicity, an issue that derailed earlier orexin agonists. Lilly paid a 44% premium ($38.00 per share in cash) plus a contingent value right of up to $9.00 per share tied to FDA drug approval for specific indications before 2030.

The orexin receptor space alone has become intensely competitive. Takeda's oveporexton (TAK 861), a twice daily orexin agonist, has an NDA under Priority Review with a PDUFA date in Q3 2026 after Phase 3 met all primary and secondary endpoints. Alkermes has alixorexton in clinical development; its stock rose 18% on the Lilly Centessa news. Jazz Pharmaceuticals, the current narcolepsy market leader with Xyrem and Xywav, faces disruption from the orexin class. The global narcolepsy treatment market is projected to exceed $4.5 billion by 2030.

CNS/neurology M&A captured $30.7 billion in 2025 deal value, overtaking oncology's $23.5 billion for the first time. Johnson & Johnson acquired Intra Cellular Therapies for $14.6 billion for Caplyta (schizophrenia, bipolar depression), projected for approximately $5 billion in annual sales. Bristol Myers Squibb paid $14 billion for Karuna Therapeutics in 2024 for KarXT in schizophrenia. Novartis committed $12 billion to the Avidity neuroscience deal. Alkermes acquired Avadel Pharmaceuticals for $2.1 billion in October 2025 to gain an approved sleep disorder therapy.

CNS has become a major growth area, and pharmaceutical companies are paying premium valuations to build positions. Lilly is using the cash generated by its dominant GLP 1 franchise (tirzepatide) to expand aggressively into neuroscience while the obesity and diabetes markets absorb increasing competition. The CNS therapeutics market is projected to reach $285 billion by 2035, growing at approximately 9% annually. The BBB drug delivery market alone is expected to grow from $1.93 billion in 2024 to $2.66 billion by 2030, with some projections reaching $55 billion by 2035 as platform technologies like Denali's Transport Vehicle enable entirely new therapeutic categories.

Predicting Brain Penetration Before Synthesis

The difficulty of BBB penetration has made it a natural target for computational pharmaceutical research. Several converging approaches in machine learning and pharmacokinetic modeling are changing how drug metabolism and brain exposure are predicted before a molecule ever enters a patient.

A comprehensive 2025 review in Molecular Informatics evaluated the current state of ML models for BBB permeability prediction. The field has advanced from simple binary classifiers (BBB permeable versus BBB impermeable) to quantitative models that predict brain to plasma concentration ratios. Random Forest models demonstrate the best balance between accuracy and generalizability, outperforming more complex gradient boosting methods that tend to overfit on training data. XGBoost classifiers combined with transformer based SMILES encoders such as MegaMolBART are producing improved molecular representation learning. The cumulative Read Across Structure Activity Relationship (c RASAR) approach combines structural similarity with machine learning for BBB permeability classification. Recent work has also applied large language models to BBB permeability prediction, using pre trained chemical language representations.

• ML models consistently identify the same critical molecular features for BBB penetration: reduced hydrogen bond donor and heteroatom counts, topological polar surface area below 80 A squared, molecular weight below 400 Da, and optimal lipophilicity near LogP 2.
• P glycoprotein substrate prediction is becoming computationally tractable. In silico P gp efflux models, hierarchical support vector regression for quantitative efflux ratio prediction, and combined passive permeability plus active efflux models now provide integrated views of brain exposure potential.
• Digital twin frameworks for CNS pharmacokinetics are emerging. SpatialCNS PBPK, a 9 compartment permeability limited model, enables mechanistic prediction of spatial drug distribution across different brain regions, CSF compartments, and brain tumors.
• 3D non human primate digital models now mirror in vivo CNS physiology for predicting intra CSF drug dispersion. Multi scale modeling integrates atomistic molecular dynamics simulations of drug BBB interactions with macroscopic PBPK models of systemic distribution.

P glycoprotein substrate prediction is a particularly consequential area for pharmaceutical chemistry. If computational models can reliably identify whether a drug candidate will be expelled by P gp efflux before synthesis, the hit rate for CNS drug candidates improves substantially. Hierarchical support vector regression methods now predict quantitative efflux ratios from molecular structure, and combined passive permeability plus active efflux models provide a more complete picture of brain exposure potential than either alone.

Digital twins for CNS pharmacokinetics are the next stage of this computational work. SpatialCNS PBPK is a 9 compartment permeability limited model that simulates spatial drug distribution across different brain regions, cerebrospinal fluid compartments, and brain tumors. Three dimensional non human primate digital models can mirror in vivo CNS physiology for predicting drug dispersion within the CSF. These tools integrate complex physiological factors including glymphatic flow and CSF dynamics into predictive frameworks. For pharmaceutical product development in CNS, the ability to predict BBB penetration, P gp efflux liability, and regional brain distribution before synthesis could compress the drug development stages that have driven the 12 year average timeline for CNS therapeutics.

The Formulation Challenge Has Changed

For decades, CNS formulation science focused on optimizing physicochemical properties within the narrow window that allows passive BBB diffusion: controlling lipophilicity, minimizing polar surface area, reducing molecular weight. AVLAYAH's approval signals that the problem has expanded. The question is no longer limited to "how do we make a molecule small enough and lipophilic enough to diffuse across the barrier?" It now includes "how do we engineer delivery systems that exploit the brain's own transport machinery?"

This changes the toolkit for formulation scientists working in pharmaceutical product development for CNS indications.

• Receptor mediated transcytosis engineering: Understanding TfR1, insulin receptor, and LRP1 binding kinetics as formulation design parameters. The affinity and release balance that Denali spent years optimizing is fundamentally a formulation problem.
• Biologic stability and delivery: AVLAYAH is an enzyme Fc fusion protein administered by weekly IV infusion. Protein formulation challenges, including aggregation, deamidation, oxidation, and cold chain management, become central to BBB crossing biologics.
• Multi compartment pharmacokinetics: CNS targeted biologics must achieve therapeutic levels in brain parenchyma, CSF, and peripheral tissues simultaneously. PK modeling must account for BBB transcytosis rates, brain interstitial fluid turnover, and glymphatic clearance.
• Combination delivery strategies: Focused ultrasound BBB opening combined with systemically administered therapeutics, and nanoparticle RMT ligand conjugates, require formulation expertise that spans modalities.

Every major pharmaceutical company is rebuilding its CNS pipeline. The M&A data shows $30.7 billion in CNS deals in 2025 alone. The biotech companies developing BBB crossing platforms, and the formulation scientists who understand how to engineer, stabilize, and deliver these complex molecules, are well positioned as CNS investment accelerates.

The computational tools described in this briefing, from ML based BBB permeability prediction to digital twin CNS PK models, are directly relevant to the formulation workflow. Predicting whether a molecule will cross the barrier, modeling its distribution once it arrives, and optimizing the delivery vehicle's physicochemical properties are formulation science problems that benefit from the same AI driven approaches now being applied to discovery. DeepC's platform, built for formulation intelligence and grounded in regulatory data from FDA, EMA, and ICH sources, addresses the part of the pipeline these CNS programs will depend on as they advance from proof of concept to commercial manufacturing.