An opportune time for targeted brain AVM therapy

Brain arteriovenous malformations (bAVMs) - high-flow, developmentally anomalous connections between cerebral arteries and veins without the intervening buffer of a capillary network - are a leading cause of hemorrhagic stroke that can lead to death or severe, life-long disability.¹ Treatment options include open neurosurgical resection, cerebral angiography with embolization, and stereotactic radiosurgery (SRS), or a combination thereof. However, many bAVMs are inoperable (due to their eloquent location and associated high risk of morbidity and mortality) and others can fail to respond to SRS.¹ Moreover, bAVMs can be in active growth or static phases. Given their frequent diagnosis in childhood and an annual rupture rate of ~4% compounded over one’s lifetime,¹ many untreated bAVMs are “ticking time bombs” that necessitate the urgent development of disease-modifying therapeutics.

Somatic mutations – i.e., DNA changes that occur after conception to cells other than the egg and sperm – have been increasingly implicated in neurologic and neurosurgical diseases.² Nikolaev et al. first described the role of somatic mutations in the pathogenesis of bAVMs,³ and the role of genetics in bAVMs is well documented.^4–9 Gain-of-function KRAS mutations (p.Gly12Asp, p.Gly12Val, and p.Gln61His), causing MAPK-ERK hyperactivation, were identified in 45/72 (63%) of surgically-resected bAVMs.³ Additional human cohorts report somatic KRAS mutations (p.Gly12Asp and p.Gly12Val) in 8/16 (50%)¹⁰ and 16/21 (76%, p.Gly12Asp, p.Gly12Val, p.Gly12Ala, p.Gly12Cys, and p.S65_A66insDS)¹¹ of bAVMs. KRAS mutations are present at a low allele frequency (0.5 – 4%), enriched in endothelial cell (EC) populations, and hyperactivate MAPK-ERK signaling.³ In mice and zebrafish harboring the KRAS p.Gly12Cys mutation, Fish et al. demonstrate that KRAS expression in ECs is sufficient to cause bAVMs and is dependent on MEK activation.¹² Inhibition of MEK - but not PI3K - can also ameliorate VEGF-mediated angiogenesis in patient-derived, KRAS-mutant ECs,³ and the MEK inhibitor trametinib has been shown to reduce the flow and size of KRAS-mutated intramedullary spinal AVMs.¹³ Moreover, administration of sotorasib (a targeted therapy against the KRAS p.Gly12Cys mutation) cured two patients with extracranial AVMs associated with somatic KRAS (p.Gly12Cys) mutations.¹⁴ Adagrasib, another targeted therapy against the KRAS (p.Gly12Cys) mutation that has been successfully used to treat non-small cell lung cancer, may also be an agent that could be deployed. Additional pharmacologic strategies to develop ‘pan-KRAS’ inhibitors could also offer a ‘one-size-fits-most’ approach to targeted therapy for bAVMs.¹⁵ KRAS inhibition may therefore be an attractive strategy for targeted bAVM therapy.

KRAS is one of the most commonly mutated genes in cancer and significant efforts have been made to develop strategies to block its activity.^15,16 However, many KRAS inhibitors have failed in clinical testing due to drug-associated toxicity, intratumoral genetic heterogeneity, and downstream redundancy in RAS signaling. These obstacles could potentially be overcome by taking advantage of the genetic homogeneity of bAVMs with targeted, endovascularly-enabled strategies for diagnosis and therapeutic delivery (such as sotorasib or gene editing technologies) to cerebral ECs (the most accessible layer of cells for blood-based drug delivery).¹⁷

How would such a strategy work in human patients? We envisage bAVMs deemed inoperable (large, eloquent location, etc.), untreatable with SRS, persistent despite SRS, actively growing on surveillance imaging, or harbor genetic risk factors associated with hemorrhage¹⁸ may be candidates. Assembly of a multidisciplinary adjudication committee guiding patient selection for in situ genetic correction therapy will be paramount. However, rational selection of molecular therapies for bAVMs necessitates a minimally invasive method for the detection of somatic pathogenic variants. “Endoluminal biopsy” - a new method that enables the isolation of lesional ECs from retrieved endovascular devices - has been successfully employed for bAVM,¹⁹ cerebral aneurysms,²⁰ and vein of Galen aneurysmal malformations (VOGMs).²¹ Briefly, endoluminal biopsy involves deploying and resting a coil in the desired vessel segment for ~1–2 minutes before the coil is completely withdrawn, explanted, rinsed, and lesional ECs are isolated by flow cytometry. Additionally, various “liquid biopsy” techniques can enable detection of pathogenic mutations from circulating cell-free DNA. We imagine a clinical workflow integrating: 1) bAVM sampling enabled by endoluminal biopsy and/or liquid biopsy of perilesional or peripheral blood (Figure 1A); 2) trio-based whole exome sequencing (WES) for germline mutation detection in sporadic and inherited bAVM cases (Figure 1B); 3) genetic detection of pathogenic KRAS (or other) somatic driver mutations from this patient material (e.g., by digital droplet polymerase chain reaction) that may guide selection of molecular targeted therapies (Figure 1C), or enable 4) bespoke gene editing strategies to correct bAVM pathogenic mutations in situ (Figure 1D).

Figure 1.

Personalized medicine approaches for identifying clinically relevant and targetable pathogenic mutations causing bAVM. (A) Somatic tissue acquisition via endoluminal or liquid biopsy of perilesional or peripheral blood for detection of somatic KRAS (or other pathogenic) driver mutations. (B) Detection of pathogenic somatic driver mutations in sporadic bAVM and Mendelian, pedigree analysis of bAVM patient and family where ‘second hit’ mutations occur in bAVMs associated with hereditary hemorrhagic telangiectasia (HHT) and capillary malformation AVM (CM-AVM) syndrome. (C) KRAS, BRAF, and MEK signaling pathway in bAVMs with annotation of commercially available pharmacologic modulators. (D) Design and delivery of bespoke gene editing tools to correct bAVM pathogenic mutations in situ.

While genetically guided pharmacological strategies are attractive, gene-editing approaches could provide further precision and durability.²² The ability to alter genomic sequences has been dramatically simplified by the discovery that CRISPR nucleases can be readily directed to target and edit DNA using a reprogrammable guide RNA (gRNA).^23–25 Beyond traditional nuclease-based editing approaches that are effective at modifying genes to knock-out function, recently developed base editors can introduce single nucleotide changes,^26–29 and polymerization-based technologies, including prime editors³⁰ or DNA-dependent DNA polymerase-based editors,^31,32 can create custom DNA changes into the genome using RNA or DNA templates, respectively. Further, engineered CRISPR-Cas proteins with modified targeting ranges and minimized off-target editing can allow for safe, efficient editing.^25,33–36 Adeno-associated viral (AAV)-delivery of base editors to EC and smooth muscle cells have already been employed to treat severe monogenic vasculopathies caused by gain-of-function missense mutations.^37,38 For example, Koblan et al. used in vivo base editing to rescue accelerated vascular pathology in a mouse model of LMNA-mutated Hutchinson-Gilford progeria syndrome (HGPS).³⁸ In addition, our group recently coupled a mutation-specific CRISPR-Cas9 editor with a vascular-tropic AAV capsid (AAV-PR)³⁹ in mice to restore both cerebral and systemic vascular homeostasis and prolong survival in mice with severe ACTA2-mutation driven smooth muscle vasculopathy,³⁷ a gene also mutated in Moyamoya disease.³⁹

Based on the confluence of advances in endoluminal and liquid biopsy techniques, next-generation DNA sequencing, multi-omics biomarkers, KRAS pharmacology, and gene editing technology, along with our deepening knowledge of the human genetics and biology of bAVMs, we believe it is a propitious and opportune moment (Gr., kairos) to move towards implementing this paradigm. However, important considerations and limitations include 1) patient selection, 2) rigor and reproducibility of detecting low frequency somatic driver mutations by endoluminal and/or liquid biopsy, 3) potential off-target effects of gene-editing therapies in vivo, 4) strategies for therapeutic delivery (i.e., vascular-specific AAVs, systemic versus endovascular-mediated deployment of packaged genetic medicines, etc.), and 4) regulatory hurdles (e.g., Food and Drug Administration approval). Forthcoming studies by our group to develop and validate molecular diagnostic technology for somatic mutation detection from bAVMs using endoluminal and liquid biopsy, design of gene-editing strategies in human cells, proof-of-concept studies in genetically modified mice that develop sporadic bAVMs, and optimization of therapeutic delivery platforms are ongoing. Other critical challenges to widespread implementation of gene-editing therapies for bAVMs include cost⁴⁰ and availability restricted to select centers, among others. Nonetheless, we believe this strategy could benefit patients with both sporadically-occurring, KRAS-mutant bAVMs (comprising ~50–70% of patients), as well as bAVMs that commonly occur in rare Mendelian syndromes such as hereditary hemorrhagic telangiectasia (HHT, caused by mutations in ACVRL1, ENG, GDF, or SMAD4)⁴¹ and capillary malformation-AVM syndrome (CM-AVM, caused by mutations in RASA1 and EPHB4),⁴² where somatic “second-hit” loss-of-function mutations determine bAVM spatio-temporal pathogenesis.