Crossing the blood–brain barrier: emerging therapeutic strategies for neurological disease

Abstract

The blood–brain barrier is a physiological barrier that can prevent both small and complex drugs from reaching the brain to exert a pharmacological effect. For treatment of neurological diseases, drug concentrations at the target site are a fundamental parameter for therapeutic effect; thus, the blood–brain barrier is a major obstacle to be overcome. Novel strategies have been developed to circumvent the blood–brain barrier, including CSF delivery, intracranial delivery, ultrasound-based methods, membrane transporters, receptor-mediated transcytosis, and nanotherapeutics. These approaches each have their advantages and disadvantages. CSF delivery and intracranial delivery are direct but invasive techniques that have not yet shown efficacy in clinical trials, although development of novel delivery devices might improve these approaches. Ultrasound-based disruption has shown some efficacy in clinical trials, but in some cases still requires invasive procedures. Approaches using membrane transporters and receptor-mediated transcytosis are less invasive than other techniques but can have off-target effects. Nanotherapeutics have shown early promise, but these strategies are in their infancy. This Personal View will clarify which strategies may be most appropriate in various contexts across divergent neurological diseases and what is required, in the futurefor clinical studies, to allow strategies to become established. ·

Introduction

The blood–brain barrier is a protective semipermeable border between the CNS and the circulatory system, which prevents substances in the blood from reaching the brain. The presence of this barrier poses a challenge for the delivery of drugs to the brain for the treatment of neurological disorders.

Early strategies to transport drugs across the blood–brain barrier included intrathecal or intracranial injection, but in the past five years, approaches have evolved that incorporate unique pharmacological agents and innovative devices. Based on robust preclinical data, drug-delivery strategies using ultrasound, nanotherapeutics, and pharmacological targeting of membrane transporters and receptor-mediated transcytosis are now entering clinical trials.

In this Personal View, we discuss the difficulties surrounding therapeutic delivery to the brain and the challenges of developing new drugs for neurological conditions, particularly brain tumours and neurodegenerative diseases. We describe novel strategies to enhance blood–brain barrier penetration focused on those that have been developed in the past 5 years and have advanced into clinical trials. The order of strategies presented is from the oldest first to the newest last. Finally, we provide insights into future implementation of these innovative approaches into clinical practice. Some potential strategies have been excluded due to lack of clinical evidence. This includes intranasal delivery, as the translational aspects have not been viable thus far given the marked anatomical differences of olfaction between mouse and humans.

Structure of the blood–brain barrier and challenges for treatment

The neurovascular unit, which comprises endothelial cells, astrocytes, pericytes, neurons, and microglia cells, forms a dynamic barrier to block most agents from having a pharmacological effect on the brain (figure 1). It is speculated that the blood-brain barrier prevents more than 98% of small compound drugs and nearly 100% of large-molecule therapeutics from penetrating the brain sufficiently to have a pharmacological effect¹. Passive diffusion of lipid-soluble drugs is possible via tight junctions, formed by the non-fenestrated brain endothelial cells. Membrane transporters on brain endothelial cells permit active or carrier-mediated movement of selected molecules across the blood–brain barrier. The presence of tight junctions, low permeability, and expression of various transporters means that brain endothelial cells are physiologically distinct from peripheral endothelial cells

Figure 1: Structure of the neurovascular unit and blood–brain barrier.

Non-fenestrated brain endothelial cells form a strict selective interface of tight junctions along the basement membrane, comprising the blood–brain barrier2^1–4. Pericytes, astrocytes, microglia, and neurons wrap around this barrier, creating the neurovascular unit. Components of the neurovascular unit can help to maintain the integrity and protective capacity of the blood–brain barrier (astrocytes and pericytes), ^1,2,5,6 or might release factors to increase the permeability of the barrier (microglia) ⁷. Between the brain endothelial cells are tight junctions that comprise various proteins to provide structural stability and signalling, including junctional adhesion molecules (JAM) and membrane-spanning occludin and claudin proteins. These tight junctions restrict paracellular transport, only allowing passive diffusion of lipid-soluble drugs.

The complex structure and selective nature of the blood–brain barrier imposes clinical challenges for the treatment of neurological and neurodegenerative disorders. Various disease states also have differential effects on the blood–brain barrier (Panel 1) ^8–22, which include alterations in permeability, integrity, transporter expression, infiltration of inflammatory cells, and vascular wall components. Changes in the blood–brain barrier during disease progression are not well understood or characterised and may change the effectiveness of therapies at different stages of the disease. Some therapeutics (e.g. antibodies) can have extremely low concentrations in the brain (0·01–0·1% of circulating levels) that limit their pharmacological effects. Understanding the cellular architecture, transport mechanisms, and potential off-target interactions in the blood–brain barrier is required to attempt novel drug design and delivery.

Panel 1. Blood–brain barrier disruption in neurological disease Alzheimer’s disease^8–10

• Permeability of the blood–brain barrier is increased

• Brain endothelial cells express genes associated with Alzheimer’s disease susceptibility

• Increased amyloid β impairs the blood–brain barrier function through disruption of tight junctions, increasing permeability, and promotes neoangiogenesis of abnormal and leaky vessels

• Inflammatory activation is increased

• Neurovascular dysfunction is observed with down regulation of Pgp transporter in brain endothelial cells

Amyotrophic lateral sclerosis^11–13

• The ArpC3 subunit of the actin-relation protein complex is downregulated in motor cortex brain endothelial cells, leading to mislocalisation of tight junctions

• Reduced HLA-E protein expression leads to natural killer cell-mediated blood–brain barrier breakdown

• Permeability of the blood–brain barrier is increased

Brain metastasis^14,15

• Proteins associated with vascular permeability are enriched

• Astrocytes open gap junctions in the presence of tumour cells, facilitating disease progression

Glioblastoma^16,17

• Expression of transporters is deregulated (eg, upregulation of solute carrier family transporters)

• The blood–brain barrier has heterogeneous areas of permeability

• Expression is upregulated of laminins, collagens, nidogens, and integrins, as well as genes encoding collagen in blood vessels surrounding the tumour

• Various subtypes of brain endothelial cells show activation, or breakdown due to abnormal angiogenesis, resulting in blood–brain barrier impairment and dysregulation

Multiple sclerosis¹⁸

• Damage to blood–brain barrier from inflammation and immune cell infiltration

• Dysfunction of brain endothelial cells reduces surface expression of tight junction proteins

• Reduced integrity of the blood–brain barrier

Parkinson’s disease^19,20

• High contrast agent transfer coefficient (Ktrans) values compared with controls indicate increased diffusion and potential leakage of the blood–brain barrier, although choline PET and gadolinium diffusion do not show integrity changes

• The blood–brain barrier shows pathological changes compared with age-matched controls

Epilepsy^21,22

• Increased permeability due to inflammation, oxidative stress and tight junction alterations

• Increased expression of efflux transporters, particularly Pgp and BCRP transporters

Strategies to overcome the blood–brain barrier

The heterogeneity of blood–brain barrier permeability, particularly changes to endothelial cells across different neurological disorders, highlights the challenges of drug delivery to the CNS. In the past 5 years, advances in therapeutic strategies to cross the blood–brain barrier have been made (table 1) and can be categorised into CSF delivery, intracranial delivery, ultrasound-based methods, membrane transporters, receptor-mediated transcytosis, and nanotherapeutics.

Table 1:

Completed clinical studies identified through PubMed or registry at ClinicalTrials.gov in the last five years that involve strategies that cross or open the blood–brain barrier.

NCT id and/or lead author name and reference	Study design and phase	Inclusion criteria	Intervention	Outcomes	Study status and results
CSF delivery
NCT02623699 26	Phase 3, three part study (randomise, quadruple masked, parallel assignment) across a 12 week period	50 patients (adult (<18, all sexes) with amyotrophic lateral sclerosis due to SOD1 mutation	Antisense oligonucleotide targeting SOD1	Incidence of adverse events and serious adverse events, vital sign/physical and neurological abnormalities, pharmacokinetic analysis of ASO	Completed 2021–07-16 Decreased CSF SOD1 concentrations but had no effect on clinical endpoints and was associated with adverse events
NCT03186989 27,28	Randomized, double-blind, placebo-controlled study, across 100 weeks Phase 1/2	464 patients (50–74, all sexes) with mild Alzheimer’s disease	Antisense oligonucleotide targeting MAPT	Incidence and severity of adverse events that are related to treatment CSF total-tau protein concentration	Completed 2022–05-13 Reduced key tau biomarkers associated with cognitive decline
NCT03225846 29	A Multicenter, Randomized, Double-blind, Placebo-controlled, Phase 1b/2a Study	88 patients (25–65, all sexes) with early manifestation Huntington’s disease	Antisense oligonucleotide targeting a SNP on the same allele as the pathogenic CAG expansion	Safety and tolerability of Clinical Effects: Total Functional Capacity	Discontinued due to lack of efficacy 2021–05-10
NCT03225833 29	A Multicenter, Randomized, Double-blind, Placebo-controlled, Phase 1b/2a Study	61 patients (25–65, all sexes) with early manifestations of Huntington’s disease	Antisense oligonucleotide targeting a SNP on the same allele as the pathogenic CAG expansion	Safety and tolerability Clinical Effects: Total Functional Capacity	Discontinued due to lack of efficacy 2021–05-11
NCT02899611 31	Phase 1, 2 interventional study	6 patients (18–65, all sexes), with medically refractory focal epilepsy	Valproate, ICV administration (escalated stepwise from 3mg/day to 60mg/day, up to day 64)	Maximum tolerated dose and safety, determined through the incidence of adverse events, changes in the number of seizures	Completed 2021–06-01 Direct intraventricular administration of valproic acid reduced seizure activity by >50% in 4 subjects
NCT04153175	Phase 2, interventional study	30 patients (18–70, all sexes) with focal seizures	CV delivery of CT-010 (reformulation of valproate) or a placebo via an implantable pump and a cranial port and dual lumen catheter (CIC)	Frequency of total monthly seizures compared to baseline	Terminated due to lack of efficacy 2024–03-01
Direct intracranial injection
NCT04802733	Phase 1, interventional study	12 patients (50–78, all sexes) with a diagnosis of Parkinson’s disease	MSK-DA01, with Cell Delivery Device	Safety and tolerability, measured through the incidence of serious adverse events at 1-year post transplant. Evidence of cell survival measured through change in 18F-DOPA uptake using PET. Changes in Motor Function	Completed 2024–05 Well-tolerated and exploratory clinical endpoints show improvement
NCT00390299 34	Phase 1, interventional study	23 patients (18+ all sexes) with recurrent glioblastoma multiforme	Carcinoembryonic Antigen-Expressing Measles Virus	Assess the safety and toxicity of intratumoural and resection cavity administration, and maximum tolerated dose	Completed 2019–11-30 Well tolerated with no dose-limiting toxicity
NCT01811992 35	Phase 1, Interventional study	19 patients (18–75, all sexes) with primary malignant glioma	Dose Escalation of Ad-hCMV-TK and Ad-hCMV-Flt3L	Maximum tolerated dose, measured by adverse events	Completed 2021–01 Safe and shows combining two vectors was feasible
NCT02026271 36	Phase 1, interventional study	40 patients (18–75, all sexes) with recurrent/progressive glioblastoma or grade III malignant glioma	an Inducible Adenoviral Vector Engineered to Express hIL-12 in the Presence of the Activator Ligand Veledimex	Safety and tolerability of varying doses of the intervention, measured by incidence and severity of adverse events	Completed 2019–08 Safe with possible immunological antitumour effect
Convection enhanced delivery
Whone et al. 2019 43	Randomised trial	6 pilot-stage patients, 35 primary stage patients (35–75, all sexes) with Parkinson’s disease for 5+ years	Glial cell-derived neurotrophic factor vs placebo (dilutant artificial CSF)	Percentage change from baseline in motor score after 40 weeks (based on 6 pilot-stage patients)	Study reported 2019 Putamen-wide sustained delivery of GDNF
NCT03566199 46	Interventional, Phase 1 study	7 patients (≥2–21, all sexes) with diffuse midline glioma	Panobinostat (MTX110)	Tolerability and efficacy (measured via proportion of grade 3 or higher, treatment-related adverse events)	Completed 2021–03-31 Repeated administration was tolerable and showed a median overall survival of 26 months
NCT04264143	Phase 1, Interventional study	9 patients (3–18, all sexes) with newly diagnosed diffuse midline gliomas	Infusate with MTX110 and gadolinium,	Incidence of adverse effects, maximum tolerated dose	Completed 2023–11-22 Treatment was well tolerated
NCT01502917 47	Phase 1	46 patients (2–17 years old, all sexes) with Non-Progressive Diffuse Pontine Gliomas	Radioactive iodine-labelled monoclonal antibody omburtamab	Maximum tolerated dose, safety, and toxicity over two years	Completed 2022–01 Treatment considered safe
Focused ultrasound
NCT03321487 24	First-in-human	Four patients (56–70) with amyotrophic lateral sclerosis	Transcranial magnetic resonance-guided focused ultrasound (MRgFUS) combined with intravenous ultrasound contrast (perflutren lipid microbubbles)	Feasibility and safety of transient BBB permeabilization (measured through gadolinium leakage) & safety	Completed 2022–12-30 Transient blood–brain barrier opening in the primary motor cortex Treatment well tolerated
NCT03714243 59	Prospective phase 1, single-arm open-labelled study	Four patients (31–56, F) with Her2-positive breast cancer and brain metastases	MRgFUS plus trastuzumab-based therapy	Characterization of treatment-related adverse events, through clinical neurologic exams and neuroimaging studies	Completed 2022–03 Increased entrance and concentration of monoclonal antibodies in the brain
Rezai et al. 2024 60	Prospective, open-label, single-group, single-institution, proof-of-concept trial	Three individuals (50–85, M/all sexes) with Alzheimer’s disease (mild cognitive impairment)	Aducanumab (1 mg per kilogram of body weight for 2 months, followed by 3 mg per kilogram for 2 months and 6 mg per kilogram for 2 months) infusion	Amyloid removal (measured by standard uptake value ratio)	Completed 2022–05 Amyloid β levels reduced by 32% Few adverse events
NCT03608553 61	Pilot study – part of a prospective, single-arm, non-randomised phase I clinical trial	Three participants with Parkinson’s disease	Magnetic resonance-guided focused ultrasound	Safety and BBB opening efficacy measured by 18F-choline-PET uptake	Completed 2021–12-31 Enhanced 18F-choline-PET uptake was observed in the targeted brain regions No serious adverse events
Low-intensity pulsed ultrasound with microbubbles
Sonabend et al 2023 56	Phase 1, dose-escalation study	17 patients	Albumin-bound paclitaxel	Pharmacokinetics	Completed 2023 3.7x increase in the mean brain parenchymal concentrations of albumin-bound paclitaxel
NCT03744026 55	Phase 1/2	33 patients, ≥ 18 years with recurrent de novo glioblastoma	Carboplatin	Dose-limiting toxicity, safety, and efficacy	Completed 2022–06-30 A 5.9-fold increase in parenchymal concentrations of carboplatin
NCT03119961 63	Phase 1/2	10 people (50–85, all sexes) with mild Alzheimer’s disease	1-MHz ultrasound only	Safety, efficacy of BBB opening and clearance of amyloid and tau	Completed 2020–10-07 Safe and effective BBB disruption
Membrane transporters
NCT04430842	Dose-escalation trial, Phase 1	15 patients, ≥ 18 years with advanced or metastatic cancers with high LAT1 signature	QBS10072S (QBS72S) a cytotoxic compound that is transported by LAT1	Determination of the maximum tolerated dose as indicated by the incidence of adverse events and their severity	Completed 2022–12-22 Safe and has led to further trials
NCT04268784 75	Phase 1	96 healthy volunteers (18–50, all sexes)	DNL343 and placebo Single and repeating oral dose(s). Not a substrate of Pgp transporter	Safety (incidence of adverse events), tolerability, pharmacokinetics and pharmacodynamics	Completed 2021–08-03 DNL343 safe and shows CSF penetrance
NCT05006352 75	Phase 1b	30 patients (18–80, all sexes), with a diagnosis of sporadic or familial ALS	DNL343 and placebo (oral repeating dose). Not a substrate of Pgp transporter	Safety (incidence of adverse events), pharmacokinetics and pharmacodynamics	Completed 2023–12 DNL343 safe and shows CSF distribution
Receptor-mediated transcytosis
Kumthekar et al.79	Phase 2	72 patients with recurrent brain metastases from breast cancer; 28 patients with leptomeningeal carcinomatosis (26–76 years, female)	ANG1005 (paclitaxel linked to angiopep-2), 600mg/m2 by IV every 3 weeks	Determined of intracranial response rate, measured by central independent radiology facility review	Completed 2019 CNS and systemic treatment effect observed
Nanotherapeutics
NCT03020017 94	First-in-human, phase 0	Eight participants (≥ 18 years, all sexes) with recurrent glioblastoma	RNA interference–based spherical nucleic acid with a gold nanoparticle core, intravenously administered	Assess safety (incidence of adverse events)	Completed 2020–08-19 Nanoparticle passed through the blood-brain barrier and accumulated in the tumour

CSF delivery

CSF is routinely accessed in neurological and neurosurgical clinical practice for diagnosis, and to treat hydrocephalus and raised intracranial pressure. Intraventricular and intrathecal routes of administration, including with small implantable devices, are well-established routes for direct drug delivery. The blood-brain barrier and the blood-CSF barrier consist of distinct membranes. The blood-CSF barrier is composed of choroid plexus epithelial cells and tight junctions and does not allow access to the inner brain parenchyma. Although other CNS barriers might allow crossing from the CSF to the brain, the in vivo, physiologically relevant evidence for this crossing is limited - perhaps explaining the lack of success with the CSF delivery approach²³. CSF delivery has been proposed as a strategy for tumours that involve the ependyma or reside in the ventricles²⁴. Although this may be relevant for leptomeningeal metastases from systemic cancers, most primary tumours are present in the brain parenchyma, and it is unlikely that sufficient drug concentrations will be achieved by CSF delivery ²³.

Antisense oligonucleotides (ASOs) are short, synthetic, single-stranded oligodeoxynucleotides that can alter RNA and reduce, restore, or modify protein expression through several distinct mechanisms²⁵. These large molecules appear to have wide uptake from CSF, and early clinical trials in both amyotrophic lateral sclerosis (NCT02623699) to target the SOD1 gene (figure 2A)²⁶ and Alzheimer’s disease (AD) (NCT03186989), to target the MAPT gene^27,28, showed promise. In the Alzheimer’s trial, the ASO therapy reduced key tau biomarkers associated with cognitive decline and has been advanced to a phase 2 trial. However, an ASO for Huntington’s disease was discontinued (NCT03225846, and NCT03225833) because of adverse events and lack of efficacy²⁹.

Fig. 2: Methods and measurement strategies for overcoming the BBB.

A) CSF Delivery. Delivery through the CSF-brain barrier has been targeted using antisense oligonucleotide (ASO) therapy in amyotrophic lateral sclerosis (ALS). The SOD1 protein misfolding is responsible for neuronal degradation in ALS. The ASO will bind the mRNA of SOD1 and initiate the mRNA degradation processes thereby leading to an overall reduction in SOD1 PROTEIN^26,27. B) Intracranial delivery. Convection-enhanced delivery (CED) can be adapted to deliver a range of therapies^43,45,64. A catheter-pump system creates a positive pressure gradient and allows the diffusion of larger molecules to tissues through bulk flow. Intracranial injection will deliver therapies, such as the oncolytic virus, directly to the target site. C) Fluorescent microscopy image of the brain of a glioblastoma patient who underwent intraoperative sonication with intravenous administration of microbubbles, chemotherapy, and fluorescein as part of pharmacokinetic study to study the effect of sonication on the concentration of drugs in the peri-tumoural brain⁵⁶. Cortex regions that underwent sonication exhibit increased fluorescence intensity compared to the surrounding brain. D) Mechanism of action of ultrasound with concomitant intravenous microbubbles. An ultrasound device is used to cavitate circulating microbubbles and increase the permeability of the BBB. This technique uses an implanted ultrasound device to deliver low-intensity pulsed ultrasound for glioblastoma treatment⁵⁶ or for the removal of amyloid plaques in Alzheimer’s⁶⁰.

CSF administration has been used for delivery of anti-seizure drugs³⁰, (NCT02899611)³¹; however, a planned phase 2 study was terminated due to lack of efficacy (NCT04153175). Therapeutic strategies that utilise CSF delivery should be developed on a case-by-case basis for neurological disorders and brain malignancies.

Intracranial delivery

Direct intracranial injection

Direct injection of agents into a focal target in the brain bypasses the blood–brain barrier and minimises the risk of systemic toxicity. However, intracranial injection is ideally suited to single treatment protocol, for agents that can diffuse to the target site or that has a discrete target that can be accessed surgically or stereotactically.

In Parkinson’s disease, direct injection into the putamen of dopamine-producing stem cells (NCT04802733) has progressed to an open-label study in 12 patients and a phase 2 trial is expected. Surgical access for deep brain stimulation is well established for this target so logically this should be possible to do safely, but proof of engraftment and persistent efficacy are needed.

Direct intracranial injection may be appropriate for lesional diseases such as tumours. Viral therapies such as the herpes simplex virus can be engineered to replicate in glioblastoma cells³². These viruses are injected into the tumour or peritumoural region, and because viruses are self-replicating, multi dosing is not a limitation (figure 2B). In a phase 1 trial of 41 patients with recurrent glioblastoma, CAN-3110 (a variant of the herpes simplex virus) was shown to enhance anti-cancer immune responses (NCT03152318)³³. The therapeutic effect of this viral therapy is mediated partly through the immune system, and therefore efforts have been focused on enhancing these anti-tumour-mediated effects. This includes inducing the expression of an immunologic target such as a carcinoembryonic antigen (NCT00390299)³⁴ or providing immune-stimulatory components such as Flt3L (NCT01811992) ³⁵ or IL-12 (NCT02026271)³⁶. Robust preclinical animal work is the basis for future trials to investigate enhancing the anti-tumour immune responses by various means with these viruses^37–40.

Convection enhanced delivery

For therapeutics that require wider distribution, convection-enhanced delivery can be used instead of a single stereotactic injection (figure 2B). Convection-enhanced delivery relies on a positive pressure gradient generated through a pump that enables large molecules that would otherwise struggle to reach the area of interest—due to their size and subsequent diffusion rate—to be administered⁴¹. Other advantages of this strategy include a more even and larger distribution area, relative to diffusion-based treatments. The lack of reliance on a steep diffusion gradient also allows a consistent concentration of drug to be delivered⁴².

Convection-enhanced delivery of glial cell-derived neurotrophic factor (GDNF) as a neurorestorative and neuroprotective therapy has been investigated in Parkinson’s disease⁴³. Initial trials used intraventricular delivery of GDNF, but this approach led to off-target effects, therefore, convection-enhanced delivery was tried ⁴⁴. The trial randomised 41 people with late-stage Parkinson’s disease to putamen-wide sustained delivery of GDNF versus placebo and reported that convection-enhanced delivery was an acceptable route for drug delivery⁴³. However, the lack of improvement in motor function or quality of life indicates that either the premise for the role of the growth factor is flawed or that subjects with early-stage disease in which innervation of the striatum with dopaminergic neurons is maintained need to be enrolled.

Convection-enhanced delivery has also been used in brain tumours clinical trials. MTX-110 is a water-soluble formulation of a histone deacetylase (HDAC) inhibitor (aqueous Panobinostat). Treatment with MTX-100 was considered for diffuse midline glioma after it showed efficacy in a rodent xenograft model⁴⁵, which led to a phase 1 clinical trial (NCT03566199)⁴⁶. Seven patients received a total of 48 infusions, with three participants experiencing dose-limited toxicities. Repeated administration by convection-enhanced delivery was tolerable and the median overall survival of 26 months that compared favourably to historical data. The combined infusions for each patient resulted in tumour coverage of 35–81%⁴⁶. Convection-enhanced delivery of MTX110 is currently being evaluated for the treatment of recurrent glioblastoma (NCT05324501), diffuse midline glioma (NCT04264143), and medulloblastoma (NCT04315064). Strategies that implant multiple catheters to increase coverage are forthcoming, but a single flexible catheter product is available commercially (NCT01502917)⁴⁷.

Convection-enhanced delivery is most suitable for brain disorders that already require surgical intervention, but the approach has several disadvantages, including a slow infusion rate, geometrical and anatomical constraints that limit delivery, an increased risk of implant infection, and a surgical recovery period.

Ultrasound-based methods

Ultrasound is an emerging strategy to enhance blood–brain barrier penetration. This approach uses sound waves to resonate intravenously administered microbubbles that are co-administered with a drug. The microbubbles open the endothelial junctions around the brain capillaries, enabling penetration of concomitantly administered drug across the blood–brain barrier. Success of this technique, using a range of chemotherapeutic and immunotherapeutic agents, has been reported in preclinical animal models^48–51. Translation into the clinic requires that sound waves either penetrate the human skull (which is considerably denser than in rodent models) or bypass the bone^52,53. High-energy transcranial ultrasound devices have been developed to penetrate the human skull, incorporating stereotactic guidance and modelling of skull attenuation of sound waves. Alternatively, bypassing the skull is done with an implantable ultrasound device, in which an array of emitters are positioned epidurally in a skull window. This approach uses low-energy sound waves and is suitable for multidosing^54–56. Over the past few years, both transcranial and skull-implantable ultrasound devices have proven to be safe and feasible as a means of repeated blood–brain barrier opening.

Focused ultrasound

Transcranial magnetic resonance-guided focused ultrasound paired with microbubbles can open the blood–brain barrier in small regions deep within the brain (figure 2D). This strategy was evaluated in a first-in-human study of four patients with amyotrophic lateral sclerosis without drug (NCT03321487)²⁴. Transient blood–brain barrier opening was demonstrated in the primary motor cortex, and no major adverse events were reported. Accessing the primary motor cortex has traditionally impeded the development of effective disease-modifying treatments for amyotrophic lateral sclerosis. A non-invasive strategy such as magnetic resonance-guided focused ultrasound has the potential to enable therapeutic access to affected neurons. However, disadvantages include issues with acoustic pressure, appropriate sonication power, and vessel damage ⁵⁷. These technical safety aspects, as well as the ability of the patient to communicate symptoms, must be carefully considered for future studies. To date, no clinical trials are open for this indication. Likewise in Alzheimer’s Disease (NCT03739905) blood-brain barrier opening has been shown to be focally achieved on PET studies but no therapeutic agents were delivered and opening in and of itself in the so-called “default mode network” areas did not affect cognitive scores or disease biomarkers⁵⁸.

Studies in brain metastases (NCT03714243)⁵⁹ and Alzheimer’s disease⁶⁰ have shown that focused ultrasound facilitates increased entrance and concentrations of therapeutic monoclonal antibodies in the brain. In a study of three individuals with Alzheimer’s disease, focused ultrasound was applied to one hemisphere of the brain alongside aducanumab infusions once a month for 6 months, with the objective of enhancing amyloid removal⁶⁰. The transient blood–brain barrier disruptions were safe and reduced amyloid β levels by 32% (measured by standard uptake value ratio) in the region that received ultrasound compared with the untreated hemisphere of the brain. However, this study did not quantify drug penetration, so it cannot be definitively ascertained if focused ultrasound directly enhanced delivery, which should be confirmed in future trials. Another monoclonal antibody, lecanemab, is already being similarly applied in an ongoing phase 0 trial (NCT05469009). In a brain metastases study, 20 infusions of trastuzumab were delivered to 4 patients with her-2 positive breast cancer indicating repeated treatment is feasible.

In the specific context of therapeutic drug delivery using this technology, clinical studies are directed at the delivery of doxorubicin in the indication of paediatric DIPG (NCT05630209, NCT05615623). Because magnetic resonance-guided focused ultrasound requires the patient to be placed into a stereotactic frame for each treatment, therapeutics that require frequent dosing would not be suitable from the perspective of patient tolerance and cost. In Parkinson’s Disease, focused ablation with ultrasound is becoming well established but magnetic resonance-guided focused ultrasound has also been used to open the blood–brain barrier in the nigrostrial region (NCT03608553)⁶¹. In a pilot study with three participants, no adverse events were reported and enhanced ¹⁸F-choline-PET uptake was observed in the targeted brain regions. A phase 1/2 study is underway for bilateral putamenal delivery of recombinant glucocerebrosidase in patients with Parkinson’s Disease (NCT05565443).

Low-intensity pulsed ultrasound

Brain tumours have been a rich source of progress in low-intensity skull-bypassing ultrasound development as surgery to remove a cranial window is performed at the point of resection so the risks of open operation can be justified for that reason alone. For example, paclitaxel is a potent chemotherapy drug, but it does not cross the blood–brain barrier ⁶². In a phase 1 trial, an ultrasound device (composed of nine 1 MHz ultrasound emitters) was directly implanted in 17 patients with recurrent glioblastoma through a cranial window in the skull. Albumin-bound paclitaxel was administered immediately after ultrasound. Biopsy specimens of sonicated and non-sonicated peritumoral brain tissue were obtained, and pharmacokinetic analysis showed a 3.7-fold increase in brain parenchymal paclitaxel when compared with non-sonicated samples⁵⁶. A phase 1/2 clinical trial evaluated the safety of the same implantable ultrasound device for delivery of carboplatin in 33 patients with glioblastoma (NCT03744026)⁵⁵, showing enhanced drug delivery across the blood–brain barrier, with a 5.9-fold increase in parenchymal concentrations of carboplatin in sonicated brain regions (figure 2C).

Low-intensity ultrasound has also been tested in a pilot study of blood–brain barrier disruption to aid in the clearance of amyloid and tau aggregates in people with Alzheimer’s disease (NCT03119961) ⁶³. A 1-MHz ultrasound device was implanted in the skull of ten people with mild Alzheimer’s disease over the left supramarginal gyrus and was well tolerated. Clearly the risk of open surgical implantation for an older population is not trivial – this differs clinically from the risk profile for a patient with an incurable brain tumour like glioblastoma. Hence the other four open clinical trials to use this approach include treating recurrent glioblastoma patients with carboplatin (NCT05902169); newly diagnosed glioblastoma patients with balstilimab, botensilimab and liposomal doxorubicin (NCT05864534); recurrent glioblastoma patients with albumin-bound paclitaxel and carboplatin (NCT04528680), and in paediatric malignant brain tumour patients with carboplatin (NCT05293197). A general advantage of this strategy is the large area that can be covered (e.g. in the dominant hemisphere for a disease like Alzheimer’s) compared with focused ultrasound and future devices may even have adjustable direction and coverage.

Membrane transporters

Two types of membrane transporters on brain endothelial cells could be targeted for CNS drug delivery: solute carrier transporters such as LAT1 (SLC7A5) which is a large neutral amino acid transporter; and the ATP-binding cassette (ABC) family of efflux transporters (figure 3). The complexity of membrane transporters concerning substrate specificity and number has previously made this area challenging for therapeutic approaches, but there is now renewed interest. Safety considerations and a restriction to small molecules are outstanding areas of challenge for using membrane transporters to enhance drug delivery to the brain^65,66.

Fig. 3: Transport mechanisms across the blood-brain barrier.

Three different transport systems can be targeted: receptor-mediated transcytosis, the amino acid transporter LAT1, and the ABC transporter (Pgp substrate system). Receptor-mediated transport uses the vesicular trafficking system within the brain endothelial cells to allow transcytosis. Ligand-receptor complexes facilitate this without disruption to the barrier. Amino acid transporters, such as the LAT1 transporter, make use of the expression on both the abluminal and luminal sides of the membrane. ABC transporters, such as the P-gp transporters, are efflux transporters existing on the luminal side of the membrane.

Amino acid transporters

LAT1 is an amino acid transporter that is widely expressed across the luminal and abluminal membranes of the blood–brain barrier and has a large transport capacity^{67, 68}. Since LAT1 is present on either side of the blood–brain barrier, transport across the membrane is possible without toxicity to endothelial cells or tight junctions.

A chemotherapeutic agent, designated QBS10072S, has been designed to bypass DNA repair mechanisms, and has a chemical moiety that makes it a substrate of LAT1 to enhance blood–brain barrier crossing^69,70. A dose-escalation trial of QBS10072S has been completed in 15 patients with advanced/metastatic cancers with a high LAT1 expression (NCT04430842). The safety, tolerability, and dose profile has led to the agent now being evaluated for glioblastoma (NCT02977780) and brain metastases (NCT05305365).

4-chlorokynurenine is a prodrug of an NMDA receptor antagonist and is in clinical development for various CNS disorders including neuropathic pain, major depressive disorder, and levodopa-induced dyskinesia⁷¹. Preclinical studies indicate that 4-chlorokynurenine crosses the blood–brain barrier via LAT1, after which the active metabolite 7-chlorokynurenic acid leaves the brain extracellular fluid via probenecid-sensitive organic anion transporters. Probenecid could be used to boost the bioavailability of 7-chlorokynurenic acid in the prefrontal cortex by blocking the activity of probenecid-sensitive transporters, which would otherwise have pumped the active metabolite out of the brain^72,73. Co-administration of probenecid and 4-chlorokynurenineis is being evaluated in a phase 1 trial in healthy volunteers to determine if this boosting strategy can increase the CNS concentration of 7-chlorokynurenic acid (NCT05280054).

ABC transporters

ABC transporters undermine the potential intracranial efficacy of many agents due to Pgp efflux transporter expression at the blood-brain barrier that pumps substrates out of the brain.

In neurodegenerative diseases, activation of eIF2B modulates the integrated stress response that controls protein synthesis, as well as responses to cellular insult, and is a proposed drug target. During early work on eIF2B activators, compounds were assessed for interactions with Pgp. DNL343 was developed; it is not a substrate for the Pgp transporter and retains selective eIF2B activation function. DNL343 is CNS penetrant, (NCT04268784, NCT05006352)^74,75 and it’s efficacy is currently under investigation in the Healey platform trial in amyotrophic lateral sclerosis (NCT05842941).

Some highly effective systemic anticancer drugs, such as paclitaxel and docetaxel, are substrates of the Pgp transporter. Inhibiting these ABC membrane transporters is problematic as they are expressed elsewhere in the body, leading to toxicity in organs such as the liver. Instead, molecules designed to overcome the efflux system have shown promise. An example is lorlatinib, a third-generation tyrosine kinase inhibitor designed to avoid being a substrate of Pgp transporter. In a randomised phase 3 trial (NCT03052608) for advanced ALK-positive non-small cell lung cancer, 71% of the patients with brain metastases who received lorlatinib had an intracranial complete response⁷⁶. Further studies will clarify the rates of CNS progression but these preliminary results are highly significant as brain metastases patients are usually excluded from such trials due to such poor intracranial efficacy⁷⁷.

Receptor-mediated transcytosis

Receptor-mediated transcytosis is a non-invasive strategy to cross the blood–brain barrier that entails binding of a ligand to a receptor on the luminal membrane of the blood–brain barrier (figure 3). Vesicle-mediated endocytosis and subsequent intracellular trafficking to the abluminal blood–brain barrier membrane allows the crossing of the ligand to the brain parenchyma⁷⁸. An important consideration is selection of the receptor, because expression can be altered in disease or during aging. Also, the drug to be transported must avoid lysosomal compartments to prevent degradation.

The LRP1-mediated endocytosis mechanism has been exploited to treat brain metastases from breast cancer⁷⁹. LRP1 is expressed at high concentrations on the blood–brain barrier and on tumour cells, making it an ideal transport system for chemotherapeutic molecules. LRP1 protects the structure of the blood–brain barrier, regulating angiogenesis, clearing toxins, and acting as a diverse endocytic receptor⁸⁰. Expression of LRP on tumours induces migration and invasion, inhibits apoptosis, and contributes to metastasis⁸¹. A synthetic peptide called angiopep-2 can cross the blood–brain barrier using LRP1-mediated endocytosis. A peptide–drug conjugate called ANG1005, which consists of three paclitaxel molecules linked to angiopep-2, can cross the blood–brain barrier. The paclitaxel is later cleaved from the peptide via lysosomal esterases. A phase 2 study demonstrated benefit for patients with leptomeningeal disease (median overall survival of 8.0 months (95% CI, 5.4–9.4)⁷⁹. A phase 3 trial is underway for leptomeningeal disease and brain metastases from breast cancer (NCT03613181).

Iron transport mechanisms in the brain have been targeted via receptor-mediated transcytosis to facilitate blood–brain barrier penetration for the treatment of mucopolysaccharidosis type II (also known as Hunter syndrome)⁸². This lysosomal storage disorder stems from a deficiency in IDS, an enzyme that impairs various cellular functions. Mucopolysaccharidosis type II causes an accumulation of gangliosides in the brain, which activate microglia and an inflammatory response that triggers neuronal death⁸³. DNL310 is a molecule consisting of IDS linked to an antibody fragment that binds the transferrin receptor TfR1. This receptor mediates the uptake of iron-loaded transferrin, which allows the transfer of iron to the brain. Targeting TfR1 for blood–brain barrier transport has previously been limited by off-target binding, resulting in anaemia and other downstream effects, and sparse delivery into the brain parenchyma⁸⁴. Therefore, the antibody must display a low affinity to TfR1 to allow transcytosis and prevent too strong of binding to the endothelial cells. DNL310 has shown activity in mice⁸², and early-stage clinical trials in patients with mucopolysaccharidosis type II are underway (NCT04251026)⁸⁵.

A new version of the anti-amyloid antibody gantenerumab (RO7126209) for treatment of Alzheimer’s disease exploits TfR1 for enhanced blood–brain barrier crossing⁸⁶. Based on data from a non-human primate model, this receptor-mediated transcytosis approach is predicted to increase brain exposure by 4–18-fold in humans which could be particularly impactful for a disease such as Alzheimer’s due to its diffuse nature⁸⁷. A phase 1b/2a clinical trial is in progress in people with mild-to-moderate Alzheimer’s disease to ascertain safety, pharmacokinetics, and pharmacodynamics (NCT04639050).

Nanotherapeutics

Nanosystems can encapsulate, carry, and deliver a variety of therapeutic agents, including drugs and nucleic acids to the central nervous system. Nanoparticles are sized between 1–100 nm in diameter and fall into two distinct categories: organic (e.g. lipid, polymeric) and inorganic (e.g. metals)⁸⁸. No guidelines are available regarding the use of nanoparticles for drug delivery but overall the production cost and safety appears to have thus far restricted the licensed agents that have reached clinical trials for neurological disease^89–91.

RNA molecules can be engineered to silence genes in genetic disorders such as epilepsy syndromes, Parkinson’s disease^92,93 or oncogenes in cancer²¹. These RNA interference molecules can be attached to an oligonucleotide carrier for therapeutic delivery⁹⁴. In addition to the challenges posed by the blood-brain barrier, unmodified oligonucleotides have a short in vivo half-life, might trigger an immune response, and are not efficient at targeting specific cell populations. Nanoparticles could house small therapeutics such as small interfering RNA (siRNA) and be designed to overcome these issues. Brain-penetrant RNA interference–based spherical nucleic acids (SNAs) which consist of gold nanoparticle cores covalently conjugated with radially oriented and densely packed siRNA oligonucleotides, were used to target the oncogene BCL2L12 in glioblastoma patients⁹⁴ (figure 4). In a phase 0 first-in-human trial, eight participants were enrolled to determine the safety, pharmacodynamics, and accumulation of these siRNA nanostructures for the treatment of glioblastoma (NCT03020017). Intravenously administered microdoses of the nanoparticle were recorded in endothelial, immune, and tumour cells, with subsequent reduction of target protein expression. These results demonstrated proof-of-concept of nanoparticle delivery past the blood–brain barrier, but glioblastoma oncogene heterogeneity is a confounder.

Fig 4. Therapeutic delivery of drugs using nanoparticles.

A nanoparticle containing a gold-core, conjugated with a spherical nucleic acid (SNA-NP) will target oncogenes upregulated in glioblastoma. The SNA-NP are administered intravenously (I.V) and will cross the BBB through paracellular pathways.

Overall nanoparticles have had far less impact than envisaged due to issues highlighted by these studies – accumulation, aggregation concerns, and unstable, impractical agents to handle and prepare for clinical use. Future studies will need to be directed toward the pharmacokinetics and clearance of nanotherapeutics to inform appropriate safety considerations and schedules.

Towards clinical implementation

These novel treatment approaches are in their infancy and require robust clinical trial data to establish their role in routine clinical practise. Each mechanism will require different implementation strategies and resources for education, training, and integrated care models. Moreover, it is important to consider that high-risk invasive strategies will be less favoured for older patients or those with comorbidities. When applied to some neurodegenerative disorders, use of these novel strategies might encounter hesitancy from both patients and providers, because for some disorders (e.g. Parkinson’s disease), existing therapies are available that, although not disease-modifying, improve quality of life. Ongoing clinical trials of strategies to enhance blood–brain barrier crossing (table 2) are notable for being conducted predominantly in neuro-oncology and for neurological conditions with poor prognosis. High treatment risks might be more acceptable to patients with diseases with a poor prognosis and few alternative treatments.

Table 2:

Ongoing clinical trials of blood–brain barrier circumventing strategies identified through PubMed or registry at ClinicalTrials.gov.

NCT id and/or lead author name and reference	Study design and phase	Inclusion criteria	Intervention	Outcomes	Study status and results
Direct intracranial injection
NCT03152318 33	Interventional, allocated phase 1 study	62 patients (estimated), 18+, all sexes	Genetically engineered HSV-1 virus, (CAN-3110 oncolytic viral vector)	Maximum tolerated dose	Estimated completion 12/2025
Convection enhanced delivery
NCT05324501	Phase 1, Interventional	36 patients (18+, all sexes, with recurrent glioblastoma	Programmable pump and catheter system with MXT110	Safety and recommended dose, as measured by frequency of serious adverse events and dose-limiting toxicities	Estimated completion 2028–08-31
NCT04315064	Early phase 1, Interventional	5 patients (1–80, all sexes) with recurrent medulloblastoma	Infusion of MTX110 into the fourth ventricle/tumour resection cavity	Neurological adverse-events (grade 3–5) related to the study drug	Estimated completion 2025–12-30
Low-intensity pulsed ultrasound with microbubbles
NCT05293197	Phase 1, Interventional	24 patients (5–17 years, all sexes) with refractory malignant brain tumours	Sonication using a novel transducer device	Dose-limiting toxicity of ultrasound emissions from the device (measured by clinical and radiological evaluation)	Estimated completion 2026–10
NCT05864534	Phase 2, Interventional	25 patients (18+, all sexes) with newly diagnosed glioblastoma	Balstilimab,or Botensilimab or Liposomal Doxorubicin alongside sonication using transducer device	Toxicity rate, landmark survival analysis up to 18 months	Estimated completion 2026–08
NCT04528680	Phase 1, Phase 2, Interventional	57 patients with recurrent glioblastoma	Chemotherapy (albumin-bound paclitaxel –and in phase 2 carboplatin) alongside sonication using a transducer device	Dose-limiting toxicity (phase 1), 1-year survival rate (phase 2), overall survival	Estimated completion 2025–09
NCT05902169	Phase 3, Interventional	560 patients (18+, all sexes) undergoing planned resection for first recurrence glioblastoma	Carboplatin, Lomustine or Temozolomide alongside sonication using a transducer device	Overall survival measured over 24 months, tumour growth rate, progression-free survival	Estimated completion 2028–06-30
Focused ultrasound
NCT03739905	Interventional	30 patients (50–85, all sexes), with probable Alzheimer’s disease	Transcranial MR-guided focused ultrasound	Device and procedure-related adverse events, BBB disruption and closure	Estimated completion 2024–12
NCT05630209	Phase 1, Phase 2 Interventional	10 patients (5–21, all sexes) with diffuse intrinsic pontine gliomas	MR-guided focused ultrasound in combination with doxorubicin	Adverse events, blood-brain barrier disruption (measured by comparative MR images)	Estimated completion 2026–01
NCT05615623	Phase 1, Phase 2 Interventional	3 patients, (5–18, all sexes) with paediatric diffuse intrinsic pontine gliomas	MR-guided focused ultrasound with doxorubicin	Safety and feasibility, and preliminary efficacy of blood-brain barrier disruption	Estimated completion 2025–07-04
NCT05565443	Phase I, Phase 2, Interventional	14 patients (35–75, all sexes) with Parkinson’s disease	MR-guided focused ultrasound in combination with recombinant glucocerebrosidase)	Incidence of adverse events, feasibility of BBB opening for brain delivery	Estimated completion 2024–12-31
NCT05469009	Early phase 1, Interventional	15 patients (50–85, all sexes) with mild cognitive impairment or mild Alzheimer’s Disease	Acucaumab, or Lecanemab, alongside blood-brain barrier opening with a focused ultrasound device	Treatment intervention related adverse events, up to 5 years post last treatment	Estimated completion 2029–07
Membrane transporter
NCT02977780	Phase 2, Interventional	460 patients (18+, all sexes) with glioblastoma	Abemaciclib, Temozolomide, Neratinib, CC115, QBS10072S (cytotoxic LAT1 substrate)	Overall survival in experimental arms compared with standard therapy	Ongoing, estimated completion 2025-12-31
NCT05305365	Phase 2, Interventional	40 patients (18+, all sexes) with breast cancer, who have developed brain metastases	QBS10072S (cytotoxic LAT1 substrate) (18mg/m2) IV injection once a month	Overall response against intracranial tumour lesions, as assessed by mRANO-BM and RECIST 1.1. criteria	Ongoing, estimated completion 2026–08
NCT05280054	Phase 1, Interventional	24 healthy subjects (18–55, all sexes)	4-chlorokynurenine alone, or in combination with probenecid	Plasma and CSF concentrations of 7-chlorokynurenic acid and 4-chlorokynure nine (with and without probenecid	Estimated completion 2022-11-01
NCT05842941	Phase 2, Phase 3, Interventional	240 patients (18–100, all sexes) with amyotrophic lateral sclerosis	DNL343 and matching placebo, administered orally once daily per day for 24 weeks. Not a substrate of Pgp transporter	Disease progression as measure by change in disease severity against amyotrophic lateral sclerosis	Estimated completion 2025–08
				functional rating scale-revised and survival
NCT03052608 76	Phase 3, Interventional	296 anaplastic lymphoma kinase positive non-small cell lung cancer patients (18+, all sexes	Lorlatinib (Not a substrate of Pgp transporter), Crizotinib	Progression free survival based on blinded independent central review	Estimated completion 2028–12-31 Preliminary results show robust intracranial response
Receptor mediated transcytosis
NCT03613181	Phase 3, open-label, Interventional	150 patients (18, all sexes) with HER2-negative breast cancer, with newly diagnosed leptomeningeal carcinomatosis	ANG1005 binds to LRP1 to cross the blood–brain barrier	Overall survival, assessed for up to two years	estimated completion 2024–12
NCT04251026 85	Phase 1, Phase 2, Interventional	47 paediatric patients (up to 18, male) with mucopolysaccharidosis type II (hunter syndrome)	DNL310 (Tividenofusp alfa), targets transferrin receptor TfR1 to cross the blood–brain barrier	Incidence and severity of treatment emergent adverse events and infusion related reactions, change in baseline in urine total glycosamino	Estimated completion 2027–07
				glycan concentrations, and cocomitant medications
NCT04639050	Phase 1, Phase 2, Interventional	285 patients (50–85, all sexes) with prodromal, mild or moderate Alzheimer’s disease	RO7126209 exploits TfR1 for enhanced blood–brain barrier crossing	Percentage of participants with adverse events, change from baseline in brain amyloid load as measured by PET scan	Estimated completion 2028–12-31

Conclusions and future directions

An appreciation that the blood–brain barrier is one of the largest challenges to drug efficacy has prompted development of various novel techniques to overcome this barrier, ranging from direct intracranial approaches to nanotherapeutics. Advances in therapeutic strategies to cross the blood–brain barrier have, to date, been made mostly in the areas of neurodegeneration and neuro-oncology’s. No singular method of crossing the blood–brain barrier will likely be appropriate for the various neurological disorders, or even for individuals with the same disease. For example, the most effective treatment for a disease as heterogeneous as glioblastoma will differ between cases and possibly even in different regions within the tumour. In neurodegenerative disorders such as Alzheimer’s disease, for which surgical intervention is not part of the current standard of care, focused ultrasound to open the blood–brain barrier paired with a therapeutic agent is an exciting therapeutic prospect.

Each of the novel strategies described has promise for development into an established standard for enabling blood–brain barrier penetration—an exciting step forward for a previously unmet need. But first, appropriately designed and powered clinical studies are needed with a focus on the timing of treatment, demographic and genetic considerations, head-to-head comparison with other treatment strategies (rather than a placebo), and relevant primary and secondary outcome measures. These will include imaging of drug delivery, disease modification and clinical measures of efficacy such as cognition, survival but must also encompass patient and carer-determined parameters such as quality of life, ability to drive and work.

Looking forward, various preclinical studies show promise but have not progressed to clinical trials. This includes the use of nanoparticles to apply deep brain stimulation for the treatment of Parkinson’s disease, currently showing promise in vivo.

Future preclinical studies should be directed towards characterising disease alterations of the blood–brain barrier, the development of complex in vitro models of the blood–brain barrier for rapid screening of strategies, and expansion on the mechanistic understanding of how these strategies enable the blood–brain barrier to be crossed, to more fully optimise their clinical use for different groups of patients.

Search strategy and selection criteria

We searched Pubmed, ClinicalTrials.gov, and Google Scholar using the key search terms “blood-brain barrier AND therapeutics”, “blood-brain barrier AND clinical trial”, and “blood-brain barrier AND crossing”. We restricted our search to the past 5 years (from March 2019 to August 2024) and for publications that were in English. References were selected with respect to originality, impact and scope. Studies were prioritised if clinical studies had taken place and were primary papers.