🏷

Classifications

Harrison's Part

?

Discussions and Knowledge Gaps

2

Should an unruptured brain AVM be treated or observed? The only randomised trial, ARUBA, was stopped early for superiority of medical management, with death or symptomatic stroke in 10.1% on medical management versus 30.7% with intervention over 33 months. A 1,770-patient target-trial-emulation study of a national registry subsequently found the opposite over a longer horizon: 5-year haemorrhage-free survival of 96.2% with intervention versus 89.0% with conservative management. The two results are not obviously reconcilable, and the disagreement is not settled.

KNOWLEDGE GAP OPEN bavm-unruptured-management-controversy

Attached to

pathophysiology#Nidus Rupture and Intracranial Haemorrhage pathophysiology#Nidus Formation with Direct Arteriovenous Shunting

This is a mechanistically informative disagreement rather than a mere clinical dispute. The competing results are consistent with a lesion whose hazard is roughly constant and lifelong while the harm of intervention is front-loaded: a short-horizon trial then favours observation and a long-horizon analysis favours treatment, without either being wrong about what it measured. The registry analysis also found no benefit in high-grade or diffuse-nidus lesions, and only microsurgical resection carried a consistent benefit, which suggests the answer is conditional on lesion architecture rather than uniform. Curating a single management recommendation here would misrepresent the evidence, so the entry records both treatments with their own evidence and leaves the comparison open.

Proposed experiments

Long-horizon randomised comparison stratified by nidus architecture

bavm-long-horizon-randomisation

A randomised comparison of intervention versus observation in unruptured brain AVM with follow-up of at least ten years, prospectively stratified by Spetzler-Martin grade, nidus diffuseness and venous drainage pattern, powered to detect a crossing of the survival curves rather than only an early difference.

Readouts

Haemorrhage-free survival by lesion architecture stratum at 10 years

pathophysiology#Nidus Rupture and Intracranial Haemorrhage

Would support

pathophysiology#Nidus Rupture and Intracranial Haemorrhage

Show evidence (3 references)

PMID:24268105 SUPPORT Human Clinical

"The risk of death or stroke was significantly lower in the medical management group than in the interventional therapy group (hazard ratio 0.27, 95% CI 0.14-0.54)."

The randomised evidence favouring conservative management over the trial's follow-up period.

PMID:41231469 REFUTE Human Clinical

"the estimated 5-year hemorrhage-free survival was 96.23% (95% CI, 93.95%-97.65%) for the interventional treatment group and 89.00% (95% CI, 86.37%-91.24%) for the conservative management group"

Directly opposes the ARUBA conclusion over a longer horizon, which is why this is recorded as REFUTE against the conservative-management position rather than as concordant evidence.

PMID:41231469 SUPPORT Human Clinical

"Subgroup analyses indicated that the benefit of the intervention was not observed in patients with high-grade AVMs or diffuse nidus, for whom outcomes were similar to those of patients with conservative management."

Shows the answer is conditional on lesion architecture, which is the substantive content of the gap rather than a simple trial-versus-registry disagreement.

MEK and BRAF inhibition reverses the mutant endothelial phenotype in culture, suppresses arteriovenous shunting in zebrafish, and acts on patient-derived AVM endothelial cells, but no MEK inhibitor is established therapy for brain AVM. Does pathway inhibition modify an established human nidus, as opposed to preventing lesion formation or reversing it in a model organism?

HUMAN MODEL MISMATCH OPEN bavm-mek-inhibition-translation

Attached to

pathophysiology#MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability pathophysiology#Constitutive RAS-MAPK-ERK Signalling in Brain Endothelial Cells

The distinction matters because the model systems test a different question from the clinical one. Animal and zebrafish models express mutant KRAS and then observe lesion formation, so they largely test prevention; a patient presents with a nidus that has existed for years and is structurally remodelled, and it is not established that switching off the driving signal reverses established architecture. Two further obstacles are specific to this disease: variant allele frequencies are low and lesional tissue is hard to obtain, so patients cannot easily be genotyped before treatment, and the natural-history haemorrhage rate of roughly 1.3% per year in unruptured lesions means any trial of a drug with systemic toxicity needs a long horizon and a large cohort to show benefit.

Proposed experiments

MEK inhibition in established, genotyped brain AVM

bavm-mek-established-lesion-trial

An early-phase trial of MEK inhibition in patients with an angiographically stable brain AVM genotyped by endovascular or cell-free DNA sampling, with nidus volume and shunt flow on serial angiography as the primary readout, to test whether pathway inhibition remodels an established lesion rather than only preventing lesion formation.

Readouts

Change in nidus volume and shunt flow on serial angiography

pathophysiology#MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability

Would support

pathophysiology#MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability

Show evidence (2 references)

PMID:41708990 SUPPORT Model Organism

"Glycolysis inhibition also reversed arteriovenous shunts and potentiated the effect of MEK inhibition in a KRAS-mutant zebrafish model."

The shunt-reversal result on which the therapeutic hope rests is a zebrafish result - the model-to-human gap this discussion is about. Evidence source is MODEL_ORGANISM.

PMID:41999461 SUPPORT Other

"However, clinical translation remains challenging because of low variant allele frequencies and limited access to lesional tissue for genetic testing."

States the specific practical barriers to translating the target into genotype-guided human therapy. Evidence source is OTHER as this is a review article.

⚙

Pathophysiology

11

Somatic Activating KRAS or BRAF Mutation in Brain Endothelium

A post-zygotic activating mutation in KRAS, or less often BRAF, arises in a brain endothelial progenitor during development and is confined to the descendants of that cell. Because the mutation is mosaic and lesion-restricted - detectable in AVM tissue and in endothelial-enriched cultures derived from it, undetectable in the same patient's blood - the disease is focal and sporadic rather than inherited, and the lesion has a definite clonal boundary. This is the initiating event of sporadic brain AVM.

brain microvascular endothelial cell CL:2000044

Animal models

Endothelial BrafV600E mouse model of sporadic brain AVM Mouse recapitulates Endothelial-restricted expression of a BRAF activating mutation is the model's design and matches the human somatic mosaic lesion in cell type and gene.

Downstream

Constitutive RAS-MAPK-ERK Signalling in Brain Endothelial Cells An activating mutation in KRAS or BRAF locks the RAS-MAPK cascade into a ligand-independent active state.

Show evidence (1 reference)

PMID:29298116 SUPPORT Human Clinical

"We detected somatic activating KRAS mutations in tissue samples from 45 of the 72 patients and in none of the 21 paired blood samples."

Establishes the somatic, lesion-restricted mosaic mutation that initiates the disease, and the tissue-versus-blood contrast that defines it as post-zygotic.

Constitutive RAS-MAPK-ERK Signalling in Brain Endothelial Cells

Mutant KRAS raises ERK activity in brain endothelial cells and holds the MAPK-ERK cascade active independently of upstream growth-factor input. This is the rate-limiting node of the disease: it is common to the KRAS and BRAF genotypes, it is what endothelial-specific expression of either mutant reproduces in animal models, and pharmacological inhibition of MAPK-ERK reverses the downstream cellular phenotype. That reversibility is what makes the node causal rather than a bystander correlate, and it is the rationale for MEK-directed therapy.

brain microvascular endothelial cell CL:2000044

ERK1 and ERK2 cascade GO:0070371 ⇑ GAIN OF FUNCTION MAPK cascade GO:0000165 ↑ INCREASED

Downstream

Aberrant Angiogenic and Notch Programme with Loss of Arteriovenous Identity Sustained ERK activity drives the angiogenic and Notch transcriptional programme and the migratory phenotype of mutant endothelium.

Endothelial Glycolytic Reprogramming via GLUT1 and Hexokinase-2 Mutant KRAS signalling raises glucose uptake and glycolytic flux, supplying the metabolic demand of the angiogenic phenotype.

MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability Because the pathway is constitutively active and its inhibition reverses the cellular phenotype, it is the actionable drug target of the disease.

Show evidence (2 references)

PMID:29298116 SUPPORT In Vitro

"expression of mutant KRAS (KRASG12V) in endothelial cells in vitro induced increased ERK (extracellular signal-regulated kinase) activity"

Directly demonstrates that the mutation raises ERK activity in endothelial cells, the molecular content of this node.

PMID:29298116 SUPPORT Human Clinical

"We propose that these malformations develop as a result of KRAS-induced activation of the MAPK-ERK signaling pathway in brain endothelial cells."

States the paper's central causal proposal, which is exactly the claim of this node.

Endothelial Glycolytic Reprogramming via GLUT1 and Hexokinase-2

Mutant KRAS raises endothelial glucose uptake and glycolytic flux by increasing membrane localisation of glucose transporters such as GLUT1 and inducing hexokinase-2. Notably, HK2 appears to be the only glycolytic enzyme elevated, so this is a targeted metabolic rewiring rather than a generalised upregulation, and HK2 is elevated in human as well as mouse AVM tissue. The reprogramming is required for the angiogenic phenotype: inhibiting glycolytic flux or knocking down HK2 suppresses sprouting angiogenesis in mutant endothelial cells, and glycolysis inhibition potentiates MEK inhibition.

glycolytic process GO:0006096 ↑ INCREASED

Downstream

Aberrant Angiogenic and Notch Programme with Loss of Arteriovenous Identity Glycolytic flux supplies the biosynthetic and energetic demand of sprouting angiogenesis; blocking it suppresses sprouting in mutant endothelium.

Show evidence (2 references)

PMID:41708990 SUPPORT In Vitro

"We found that KRASG12V expression in the endothelium increased angiogenesis, which was accompanied by enhanced glucose uptake and glycolytic flux."

Establishes the metabolic shift downstream of mutant KRAS that this node represents.

PMID:41708990 SUPPORT In Vitro

"Critically, either pharmacologic inhibition of glycolytic flux or knockdown of HK2 suppressed sprouting angiogenesis in cultured KRASG12V endothelial cells."

Demonstrates that the glycolytic shift is required for the angiogenic phenotype, justifying the causal downstream edge rather than a mere association.

Aberrant Angiogenic and Notch Programme with Loss of Arteriovenous Identity

Mutant endothelium upregulates genes of angiogenesis and Notch signalling and becomes abnormally migratory. Single-cell transcriptomic profiling of human AVM endothelium additionally shows endothelial-to-mesenchymal-transition-like features and loss of the normal arterial-versus-venous identity programme. Losing arteriovenous identity is the specific cellular failure that matters here: it is why the malformed vessels cannot resolve into a properly specified arterial-capillary-venous sequence and instead form a direct shunt.

angiogenesis GO:0001525 ↑ INCREASED Notch signaling pathway GO:0007219 ↑ INCREASED cell migration GO:0016477 ↑ INCREASED

Downstream

Nidus Formation with Direct Arteriovenous Shunting Disordered angiogenesis without correct arteriovenous specification builds the tangle of shunting vessels that constitutes the nidus.

Deficient Mural Cell Coverage and Vessel Wall Instability Loss of Notch-dependent arteriovenous identity and the accompanying EndMT-like shift disrupt the endothelial signalling that recruits pericytes and smooth muscle cells to the vessel wall, so the malformed vessels are built without adequate mural investment.

Show evidence (3 references)

PMID:29298116 SUPPORT In Vitro

"increased expression of genes related to angiogenesis and Notch signaling, and enhanced migratory behavior. These processes were reversed by inhibition of MAPK (mitogen-activated protein kinase)-ERK signaling."

Establishes both the angiogenic/Notch/migratory programme of this node and its dependence on the upstream MAPK-ERK node, since inhibiting ERK reverses it.

PMID:41999461 SUPPORT Other

"including heightened angiogenic and inflammatory signaling, endothelial-to-mesenchymal transition-like features, and loss of normal arteriovenous identity"

Adds the loss of arteriovenous identity and EndMT-like phenotype seen in single-cell profiling of human AVM endothelium. Evidence source is OTHER as this is a review article.

PMID:35084939 SUPPORT Human Clinical

"we investigated cellular and molecular perturbations in brain arteriovenous malformations, which are a leading cause of stroke in young people, and identified pathologic endothelial transformations with abnormal vascular patterning and the ontology of vascularly derived inflammation"

Primary human single-cell evidence for pathologic endothelial transformation and abnormal vascular patterning in AVM tissue, upgrading this node from review-level to direct human-tissue support.

Nidus Formation with Direct Arteriovenous Shunting

The defining anatomical lesion: a focal tangle of dilated, tortuous vessels in which feeding arteries connect directly to draining veins with no intervening capillary bed. The absent capillary bed is the mechanistically important feature, because it is the capillary bed that normally drops arterial pressure before blood reaches the venous side. Without it, arterial pressure is transmitted directly into thin-walled venous structures, and the lesion behaves as a low-resistance high-flow shunt.

endothelial cell of artery CL:1000413 vein endothelial cell CL:0002543

Downstream

High-Flow Shunt Haemodynamics and Venous Hypertension A capillary-free arteriovenous connection transmits arterial pressure and flow directly into the venous outflow.

Vascular Inflammation and Blood-Brain Barrier Junction Loss The malformed nidus vessels are lined by mutant endothelium that is pro-inflammatory and barrier-deficient.

Show evidence (1 reference)

PMID:41999461 SUPPORT Other

"Brain arteriovenous malformations (bAVMs) are high-flow vascular lesions characterized by direct arteriovenous shunting without an intervening capillary bed."

Defines the nidus lesion in exactly the terms of this node - high flow, direct shunting, absent capillary bed. Evidence source is OTHER as this is a review article.

High-Flow Shunt Haemodynamics and Venous Hypertension

Arterial pressure delivered into the venous side produces venous hypertension, progressive dilation of draining veins, and mechanical loading of vessel walls that were never built to withstand it. The haemodynamic burden also perturbs perinidal cortex. This is the mechanical arm of rupture risk, and it is separable from - and additive to - the inflammatory arm. Deep venous drainage is among the anatomical features associated with higher haemorrhage risk in natural-history cohorts, which is consistent with an outflow-restriction mechanism.

Downstream

Nidus Rupture and Intracranial Haemorrhage Sustained pressure loading of thin-walled malformed vessels is the mechanical precipitant of rupture.

Perinidal Cortical Irritation and Seizure Generation Venous hypertension, perinidal gliosis and haemosiderin deposition render adjacent cortex epileptogenic.

Show evidence (1 reference)

PMID:25015366 SUPPORT Human Clinical

"Female sex (1.49, 95% CI 0.96-2.30) and exclusively deep venous drainage (1.60, 0.95-2.68, p = 0.02 in imputed dataset) may be additional predictors."

Deep venous drainage - an outflow-restriction feature - is a candidate haemorrhage predictor. Recorded as PARTIAL because the confidence interval crosses 1 in the primary analysis and the authors state it only "may be" a predictor.

Vascular Inflammation and Blood-Brain Barrier Junction Loss

Mutant endothelium is not merely misshapen, it is inflammatory. KRASG12V-expressing endothelial cells increase expression of pro-inflammatory mediators and reduce expression of blood-brain-barrier junction constituents, forming a leaky barrier in vitro. Their conditioned medium activates microglia, and medium from those primed microglia in turn degrades junction proteins in wild-type endothelial cells - a feed-forward loop in which the mutant clone recruits normal neighbouring tissue into the lesion. Blood-derived macrophages worsen the barrier defect further. This is the inflammatory arm of vascular instability, distinct from the haemodynamic arm.

brain microvascular endothelial cell CL:2000044 microglial cell CL:0000129 macrophage CL:0000235

Downstream

Nidus Rupture and Intracranial Haemorrhage Loss of junctional integrity and inflammatory degradation of the vessel wall destabilise malformed vessels and promote leakage and rupture.

Show evidence (3 references)

PMID:41696778 SUPPORT In Vitro

"Here we demonstrate that KRASG12V expressing cultured ECs (KRAS-G12V-EC) have increased expression of pro-inflammatory mediators and reduced expression of blood-brain-barrier (BBB) junction constituents."

Establishes the two components of this node - pro-inflammatory secretion and loss of barrier junction proteins - as direct consequences of the mutation in endothelium.

PMID:41696778 SUPPORT In Vitro

"The conditioned medium from KRAS-G12V-EC can activate BV2-microglia (BV2-MG) and conditioned media from this primed BV2-MG can compromise the expression of EC-junction constituents when added to wild-type ECs."

Documents the feed-forward loop by which mutant endothelium recruits microglia that then damage non-mutant endothelium, which is why the lesion destabilises beyond the mutant clone itself.

PMID:35084939 SUPPORT Human Clinical

"We illustrate the interplay between vascular and immune cells that contributes to brain hemorrhage and catalog opportunities for targeting angiogenic and inflammatory programs in vascular malformations."

Single-cell profiling of human brain AVM tissue ties vascular-immune crosstalk directly to haemorrhage, which is the human-tissue counterpart of the in-vitro endothelium-microglia loop and the reason this node is not merely a culture phenomenon.

Perinidal Cortical Irritation and Seizure Generation

Cortex adjacent to the nidus becomes epileptogenic through venous hypertension, gliosis, and haemosiderin deposition from prior microhaemorrhage. Seizure is the second commonest presentation of brain AVM after haemorrhage, and in surgical series of unruptured lesions it is the dominant presenting symptom.

Show evidence (1 reference)

PMID:40428814 SUPPORT Human Clinical

"Seizures are a common clinical manifestation in patients with brain AVMs, ranking as the second most frequent presentation."

Establishes seizure as the second commonest presentation of brain AVM, the clinical content of this node.

Nidus Rupture and Intracranial Haemorrhage

Rupture of malformed nidus or draining vessels produces intracerebral, intraventricular or subarachnoid haemorrhage - the presentation that makes brain AVM a leading cause of haemorrhagic stroke in the young. In an individual-patient meta-analysis of four untreated cohorts spanning 6,074 patient-years, the annual haemorrhage rate was 2.3% overall, and this figure is strongly conditioned on presentation: 4.8% per year in lesions that had already ruptured versus 1.3% per year in those that had not. Prior haemorrhage and increasing age were the independent predictors.

Show evidence (2 references)

PMID:25015366 SUPPORT Human Clinical

"A total of 141 hemorrhage events occurred during 6,074 patient-years of follow-up (annual rate of 2.3%, 95% confidence interval [CI] 2.0%-2.7%), higher for ruptured (4.8%, 3.9%-5.9%) than unruptured (1.3%, 1.0%-1.7%) AVMs at presentation."

Quantifies the natural-history haemorrhage rate and the ruptured-versus-unruptured split that governs management decisions.

PMID:25015366 SUPPORT Human Clinical

"This large, individual patient data meta-analysis identified hemorrhagic presentation and increasing age as independent predictors of hemorrhage during follow-up."

Identifies the independent predictors of subsequent haemorrhage in untreated disease.

MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability

Because the disease runs through a single constitutively active kinase cascade, inhibiting that cascade is a candidate disease-modifying therapy - the first such prospect for a lesion otherwise treated only by destroying or removing it. Inhibition of MAPK-ERK reverses the angiogenic, Notch and migratory phenotype of mutant endothelium in vitro; MEK and BRAF inhibitors show promise in preclinical models; and combined glycolysis and MEK inhibition suppresses angiogenesis in patient-derived AVM endothelial cells. This is explicitly a preclinical vulnerability: no MEK inhibitor is established therapy for brain AVM, and the `bavm-mek-inhibition-translation` discussion records what stands between the model data and clinical use. The clinical motivation is the haemorrhage risk documented on the rupture node: absent a medical option, the only alternative is a destructive intervention on an often asymptomatic lesion. That is a rationale for pursuing the target, not a causal relationship, so it is recorded here rather than encoded as a pathograph edge.

ERK1 and ERK2 cascade GO:0070371

Animal models

Endothelial BrafV600E mouse model of sporadic brain AVM Mouse partially recapitulates Dabrafenib slowed lesion progression when given early, but its efficacy against established lesions was explicitly left uncertain - which is precisely the prevention-versus-reversal distinction the `bavm-mek-inhibition-translation` discussion is about, now carried as a structured model claim rather than only as prose.

Show evidence (3 references)

PMID:29298116 SUPPORT In Vitro

"These processes were reversed by inhibition of MAPK (mitogen-activated protein kinase)-ERK signaling."

Demonstrates pharmacological reversibility of the mutant endothelial phenotype, the basis for treating this pathway as a therapeutic target.

PMID:41999461 SUPPORT Other

"MEK and BRAF inhibitors targeting the RAS/MAPK pathway have shown promising results in preclinical studies. However, clinical translation remains challenging because of low variant allele frequencies and limited access to lesional tissue for genetic testing."

Supports the target while explicitly stating that translation is unresolved. Recorded as PARTIAL because the same sentence that supports the vulnerability also qualifies its clinical readiness.

PMID:41708990 SUPPORT In Vitro

"Finally, combined glycolysis and MEK inhibition suppressed angiogenesis in patient-derived bAVM primary endothelial cells."

Shows the combination acting on endothelial cells derived from actual patient lesions, the closest available approach to human evidence for this target.

Deficient Mural Cell Coverage and Vessel Wall Instability

Nidus vessels are structurally defective, and a specific component of that defect is inadequate mural cell coverage. Pericytes and smooth muscle cells are what give a vessel wall its integrity, and changes in their number and coverage are implicated in bAVM; their recruitment during vascular remodelling is governed by PDGF-B/PDGFR-beta, EphrinB2/EphB4 and angiopoietin/Tie2 signalling. This sits between the malformed nidus and rupture as the structural reason those vessels are fragile - complementing, not duplicating, the inflammatory arm, which degrades endothelial junctions rather than mural coverage. Human bulk RNA-seq of bAVM tissue is consistent with it: extracellular-matrix and angiopoietin-TIE genes are among the downregulated set.

pericyte CL:0000669 smooth muscle cell CL:0000192

Animal models

Endothelial BrafV600E mouse model of sporadic brain AVM Mouse recapitulates The model's lesions show reduced vascular wall stability with spontaneous haemorrhage, the phenotype this node describes.

Downstream

Nidus Rupture and Intracranial Haemorrhage Vessels with deficient mural coverage lack the wall integrity to contain arterial pressure, which is the structural precondition for rupture.

Show evidence (2 references)

PMID:34541474 SUPPORT Other

"The vessel walls in bAVMs have structural defects, which impair vascular integrity. Mural cells are essential structural and functional components of blood vessels and play a critical role in maintaining vascular integrity. Changes in mural cell number and coverage have been implicated in bAVMs."

Establishes deficient mural cell coverage as a structural defect of bAVM vessels impairing vascular integrity. Evidence source is OTHER as this is a review article.

PMID:31795902 SUPPORT Human Clinical

"we found 498 genes downregulated including genes implicated in extracellular matrix composition, the binary angiopoietin-TIE system, and TGF (transforming growth factor)-beta signaling"

Human bAVM tissue RNA-seq showing downregulation of extracellular-matrix and angiopoietin-TIE programmes, the signalling system that governs mural cell recruitment.

⬡

Pathograph

Use the checkboxes to hide or show graph categories. Hover nodes for evidence and cross-linked metadata.

●

Phenotypes

5

Cardiovascular 1

Stroke HP:0001297

Temporal: ACUTE

Show evidence (1 reference)

PMID:24268105 SUPPORT Human Clinical

"The primary outcome is time to the composite endpoint of death or symptomatic stroke"

Stroke is the primary clinical outcome of brain AVM against which management strategies are judged.

Nervous System 1

Seizure HP:0001250

Show evidence (2 references)

PMID:40428814 SUPPORT Human Clinical

"Seizures are a common clinical manifestation in patients with brain AVMs, ranking as the second most frequent presentation."

Establishes seizure as the second commonest presentation of brain AVM.

PMID:41303829 SUPPORT Human Clinical

"Epileptic seizure was the most common presenting symptom (80.4%)."

Quantifies seizure as the dominant presentation in a surgical series of unruptured lesions. Not mapped to a frequency band because the cohort is restricted to unruptured, surgically treated AVMs.

Other 3

Cerebral arteriovenous malformation HP:0002408

Show evidence (1 reference)

PMID:41999461 SUPPORT Other

"Brain arteriovenous malformations (bAVMs) are high-flow vascular lesions characterized by direct arteriovenous shunting without an intervening capillary bed."

Defines the structural lesion. Evidence source is OTHER as this is a review article.

Intracranial haemorrhage Intracranial hemorrhage HP:0002170

Temporal: ACUTE

Show evidence (2 references)

PMID:29298116 SUPPORT Human Clinical

"are a leading cause of hemorrhagic stroke in young adults and children"

Establishes haemorrhagic stroke as the defining clinical danger of brain AVM.

PMID:25015366 SUPPORT Human Clinical

"A total of 141 hemorrhage events occurred during 6,074 patient-years of follow-up (annual rate of 2.3%, 95% confidence interval [CI] 2.0%-2.7%), higher for ruptured (4.8%, 3.9%-5.9%) than unruptured (1.3%, 1.0%-1.7%) AVMs at presentation."

Supplies the annual haemorrhage rate in untreated disease, stratified by whether the lesion had previously ruptured.

Focal neurological deficit

Show evidence (2 references)

PMID:41303829 SUPPORT Human Clinical

"In our cohort, epileptic seizures were the most frequent initial clinical manifestation, whereas headaches and focal neurological deficits were less commonly observed."

Names focal neurological deficits as a presenting manifestation and places their frequency below seizure in an unruptured surgical cohort. No frequency band is asserted because the source gives an ordering, not a proportion, and the cohort is surgically selected.

PMID:42138076 SUPPORT Other

"bAVMs present across a clinically heterogeneous spectrum with variable risk of hemorrhage, epilepsy, neurologic deficit, and systemic involvement"

Independently names neurologic deficit as one of the core clinical consequences of brain AVM. Evidence source is OTHER as this is a review article.

🧬

Genetic Associations

2

KRAS (Somatic activating mutation in lesional endothelium)

Gene: KRAS hgnc:6407 relationship_type: SOMATIC_DRIVER variant_origin: SOMATIC

Show evidence (2 references)

PMID:29298116 SUPPORT Human Clinical

"We detected somatic activating KRAS mutations in tissue samples from 45 of the 72 patients and in none of the 21 paired blood samples."

Establishes both the causal gene and its somatic, lesion-restricted mosaic nature - present in AVM tissue, absent from the same patients' blood.

PMID:29298116 SUPPORT Human Clinical

"We identified activating KRAS mutations in the majority of tissue samples of arteriovenous malformations of the brain that we analyzed."

States that activating KRAS mutation is the majority genotype of sporadic brain AVM, supporting CAUSAL rather than modifier status.

BRAF (Somatic activating mutation in lesional endothelium)

Gene: BRAF hgnc:1097 relationship_type: SOMATIC_DRIVER variant_origin: SOMATIC

Show evidence (2 references)

PMID:41999461 SUPPORT Other

"The identification of somatic KRAS and BRAF mutations in sporadic bAVM endothelial cells (ECs) has fundamentally reshaped current understanding of bAVM biology, indicating that activation of the RAS/MAPK pathway drives aberrant angiogenic programs."

Names BRAF alongside KRAS as a somatic driver in bAVM endothelium converging on RAS/MAPK. Evidence source is OTHER as this is a review article.

PMID:41999461 SUPPORT Model Organism

"Furthermore, animal models with EC-specific expression of mutant KRAS or BRAF exhibit bAVM-like lesions, which support the hypothesis that hyperactivation of the RAS/MAPK pathway is a key driver of lesion formation."

Endothelial-specific expression of either mutant gene reproduces the lesion in animals, supporting pathway-level rather than gene-specific causation. Evidence source is MODEL_ORGANISM as the claim rests on animal models.

💊

Medical Actions

4

Microsurgical resection

Action: Neurosurgical Procedure NCIT:C15656

Open surgical excision of the nidus, the only modality that removes the lesion immediately and completely and the only one consistently associated with reduced haemorrhage risk in sensitivity analyses of large registry data. Outcome depends steeply on lesion grade: in a 12-year single-centre series of unruptured lesions stratified by supplementary Spetzler-Martin grade, poor outcomes were rare in the low-risk group but common in the high-risk group.

Mechanism Target:

INHIBITS Nidus Formation with Direct Arteriovenous Shunting — Resection physically removes the shunting nidus, eliminating both the haemodynamic and the inflammatory arms of the lesion at once.

Show evidence (2 references)

PMID:41303829 SUPPORT Human Clinical

"In the low-risk group, there were no deaths, and poor outcomes were rare (6.8% at discharge; 2.3% at 9 months). Conversely, the high-risk group demonstrated significantly worse outcomes (62.5% poor outcome at discharge, 28.6% at 9 months)."

Supports resection in low-grade lesions while documenting substantial harm in high-grade lesions. Recorded as PARTIAL because the same result both supports and restricts the treatment.

PMID:41231469 SUPPORT Human Clinical

"Only microsurgical resection was consistently associated with reduced hemorrhage risk in sensitivity analyses."

Among interventional modalities, resection was the one with a consistent haemorrhage benefit in a 1,770-patient target-trial-emulation analysis.

Stereotactic radiosurgery

Action: Stereotactic Radiosurgery NCIT:C15358

Focused radiation inducing progressive endothelial proliferation and luminal closure of nidus vessels over two to three years. Obliteration is delayed and incomplete, so the patient remains at haemorrhage risk during the latency period, and radiation-induced change is an additional hazard - in a paediatric unruptured cohort, obliteration reached 64% at 5 years and 82% at 10 years while permanent radiation-induced change occurred in 8%.

Mechanism Target:

INHIBITS Nidus Formation with Direct Arteriovenous Shunting — Radiation-induced obliteration progressively closes nidus vessels, abolishing the shunt over a period of years rather than immediately.

Show evidence (2 references)

PMID:31942635 SUPPORT Human Clinical

"Estimated probabilities of AVM obliteration at 5 and 10 yr were 64% and 82%, respectively."

Quantifies delayed and incomplete obliteration. Recorded as PARTIAL because the same cohort shows the treatment leaves a substantial fraction unobliterated for years.

PMID:31942635 SUPPORT Human Clinical

"The primary endpoint occurred in 14%, comprising hemorrhagic stroke, death, and permanent radiation-induced changes in 6%, 3%, and 8%, respectively."

Quantifies the harm profile, including radiation-induced change, that has to be set against the obliteration benefit.

Conservative medical management

Action: Supportive Care NCIT:C15747

Observation with symptomatic treatment - antiepileptic drugs for seizures, analgesia - without attempting to eradicate the lesion. For unruptured lesions this is a genuinely competitive strategy rather than a default: the ARUBA randomised trial was halted early for superiority of medical management, with the composite of death or symptomatic stroke reached by 10.1% on medical management versus 30.7% with intervention. The advantage did not decay with longer observation: at a mean 50.4 months the incidence of death or symptomatic stroke remained lower with medical management alone (3.39 vs 12.32 per 100 patient-years; hazard ratio 0.31, 95% CI 0.17 to 0.56), with fewer adverse events. What remains contested is whether that generalises beyond the trial's enrolled population - see the `bavm-unruptured-management-controversy` discussion.

Mechanism Target:

INHIBITS Perinidal Cortical Irritation and Seizure Generation — Antiepileptic drugs suppress the seizures generated by perinidal cortex without altering the lesion itself. Modality is OTHER rather than BEHAVIORAL because the content is observation plus antiepileptic pharmacotherapy, not a behavioural, dietary or lifestyle intervention.

Show evidence (3 references)

PMID:24268105 SUPPORT Human Clinical

"The ARUBA trial showed that medical management alone is superior to medical management with interventional therapy for the prevention of death or stroke in patients with unruptured brain arteriovenous malformations followed up for 33 months."

The randomised trial's conclusion supporting conservative management for unruptured lesions, with its follow-up duration stated.

PMID:24268105 SUPPORT Human Clinical

"The primary endpoint had been reached by 11 (10.1%) patients in the medical management group compared with 35 (30.7%) in the interventional therapy group."

Quantifies the difference in death or symptomatic stroke that led the data and safety monitoring board to halt randomisation.

PMID:32562682 SUPPORT Human Clinical

"After extended follow-up, ARUBA showed that medical management alone remained superior to interventional therapy for the prevention of death or symptomatic stroke in patients with an unruptured brain arteriovenous malformation."

The final follow-up analysis, which supersedes the 2014 interim horizon and shows the effect did not decay with longer observation.

Endovascular embolization

Action: Embolization Therapy NCIT:C15230

Catheter-delivered occlusion of the nidus or its feeding arteries, usually with a liquid embolic (EVOH/Onyx or n-BCA). In practice it is most often adjunctive - devascularising a lesion before resection or radiosurgery - rather than curative on its own, and it was one of the three interventional modalities randomised in ARUBA. Notably, in a 1,770-patient registry emulation only microsurgical resection carried a consistent haemorrhage benefit in sensitivity analyses, so embolization should not be assumed to inherit resection's result.

Mechanism Target:

INHIBITS Nidus Formation with Direct Arteriovenous Shunting — Embolic occlusion of nidus compartments and feeding arteries removes shunting channels, either as definitive treatment or to reduce flow before resection or radiosurgery.

Show evidence (2 references)

PMID:24268105 SUPPORT Human Clinical

"medical management with interventional therapy (ie, neurosurgery, embolisation, or stereotactic radiotherapy, alone or in combination)"

Establishes embolization as one of the three standard interventional modalities, as randomised in ARUBA.

PMID:41231469 SUPPORT Human Clinical

"Only microsurgical resection was consistently associated with reduced hemorrhage risk in sensitivity analyses."

Recorded as PARTIAL because it bounds rather than supports this modality: among interventional treatments only resection showed a consistent benefit, so embolization cannot be assumed to share it.

🔬

Diagnosis

4

Digital subtraction angiography

Catheter DSA is the reference standard for brain AVM, defining the feeding arteries, nidus architecture and venous drainage pattern in the temporal resolution that lower-tech modalities cannot match. It is what the angioarchitectural grading schemes are scored against, and what determines whether a lesion is treatable.

digital subtraction angiography NCIT:C190556

Results: Demonstration of a nidus with arteriovenous shunting, with feeding-artery and draining-vein anatomy sufficient to assign a Spetzler-Martin grade.

Show evidence (1 reference)

PMID:39125532 SUPPORT Human Clinical

"using digital subtraction angiography (DSA) as the reference standard"

Confirms DSA as the reference standard against which other modalities are measured.

Non-contrast-enhanced time-resolved MR angiography

Non-contrast 4D MR angiography detects the shunt and characterises angioarchitecture with accuracy comparable to contrast-enhanced 4D-MRA, and grades concordantly with DSA on the Spetzler-Martin and Buffalo scales. Its value is specific to this disease: brain AVM is a lifelong lesion requiring repeated imaging, so avoiding cumulative gadolinium exposure over decades of follow-up matters more here than for a one-off diagnostic question.

magnetic resonance angiography NCIT:C190557

Results: Detection of arteriovenous shunting and grading concordant with DSA, without gadolinium-based contrast.

Show evidence (2 references)

PMID:39125532 SUPPORT Human Clinical

"Our results demonstrated that NCE-4D-MRA had a higher accuracy and specificity compared to CE-4D-MRA (0.85 vs. 0.83 and 95% vs. 85%, respectively) and similar agreement, with DSA detecting shunts in bAVMs or residuals."

Quantifies non-contrast 4D-MRA accuracy against the contrast-enhanced comparator and DSA.

PMID:39125532 SUPPORT Human Clinical

"The safety profile of imaging techniques is a significant concern in the long-term follow up of bAVMs"

States the contrast-avoidance rationale that makes this modality specifically valuable in a lifelong lesion requiring repeated imaging.

Somatic genotyping from lesion tissue or cell-free DNA

This is the diagnostic consequence of the entry's central mechanistic claim, and it is counterintuitive enough to be worth stating plainly: because the causative KRAS or BRAF mutation is a post-zygotic mosaic event confined to lesional endothelium, it is undetectable in blood. A conventional germline blood panel will be negative in a patient whose disease is entirely genetically explained. Genotyping therefore requires resected or endovascularly biopsied lesion tissue, or ultra-sensitive cell-free DNA methods, and the low variant allele fraction is the main obstacle to genotype-guided therapy.

molecular genetic testing NCIT:C15709

Results: Detection of a somatic activating KRAS or BRAF variant in lesional tissue; a negative blood test does not exclude the diagnosis.

Show evidence (2 references)

PMID:29298116 SUPPORT Human Clinical

"We detected somatic activating KRAS mutations in tissue samples from 45 of the 72 patients and in none of the 21 paired blood samples."

The tissue-positive, blood-negative contrast that makes lesional sampling necessary and a germline blood panel uninformative.

PMID:41999461 SUPPORT Other

"Future approaches combining minimally invasive sampling methods, such as endovascular biopsy and peripheral blood cell-free DNA analysis, with ultra-sensitive detection technologies are expected to help overcome these limitations."

Names the emerging sampling routes that address the tissue-access problem. Evidence source is OTHER as this is a review article.

📊

Prevalence

2

New York islands, United States (prospective population-based surveillance)

Annual Incidence 1.34 per 100,000 (1.18–1.49) 1–9 per 100,000

Average annual detection rate of brain AVM, from 284 prospectively ascertained patients over 21,216,467 person-years in a defined population of 9.4 million. This is a detection rate for a congenital lesion rather than a true incidence of lesion formation, since the malformation predates its discovery.

Show evidence (1 reference)

PMID:12690217 SUPPORT Human Clinical

"leading to an average annual AVM detection rate of 1.34 per 100,000 person-years (95% CI, 1.18 to 1.49)"

Prospective population-based detection rate, the basis for the normalised rate recorded here.

Scotland, adults aged 16 and over (prospective population-based study)

Annual Incidence 1.12 per 100,000 (0.9–1.37) 1–9 per 100,000

Crude first-ever-diagnosis detection rate for brain AVM among Scottish adults, from the Scottish Intracranial Vascular Malformation Study. Independently reproduces the New York figure in a different health system, which is why both are recorded.

Show evidence (1 reference)

PMID:12702837 SUPPORT Human Clinical

"1.12 (95% CI, 0.90 to 1.37) for brain AVMs"

Crude annual detection rate per 100,000 adults for brain AVM in a prospective population-based study.

🔬

Clinical Trials

2

NCT00389181 PHASE_III COMPLETED

ARUBA - A Randomised trial of Unruptured Brain Arteriovenous malformations. Compared medical management alone against medical management plus interventional therapy (neurosurgery, embolisation or stereotactic radiotherapy) in adults with unruptured brain AVM, across 39 sites in nine countries. Randomisation was halted early in 2013 for superiority of medical management. It remains the only completed randomised trial in this disease and the anchor of the management controversy recorded in the `bavm-unruptured-management-controversy` discussion.

Target Phenotypes: Stroke HP:0001297 Intracranial hemorrhage HP:0002170

Show evidence (1 reference)

PMID:24268105 SUPPORT Human Clinical

"This trial is registered with ClinicalTrials.gov, number NCT00389181."

Ties the published trial report to its ClinicalTrials.gov registration, which is the identifier this record is keyed on.

NCT02098252 NOT_APPLICABLE RECRUITING

TOBAS (Treatment of Brain AVMs Study), a randomised controlled trial plus registry designed to address the question ARUBA left open, and structured to be less vulnerable to the generalisability criticism levelled at it: it enrols ruptured as well as unruptured lesions, and pairs randomisation with a registry for patients who cannot be randomised. Directly relevant to the `bavm-unruptured-management-controversy` discussion.

Target Phenotypes: Stroke HP:0001297 Intracranial hemorrhage HP:0002170

Show evidence (1 reference)

clinicaltrials:NCT02098252 SUPPORT Human Clinical

"The trial has been designed to test a) whether medical management or interventional therapy will reduce the risk of death or debilitating stroke"

States the trial's primary question, which is the unresolved comparison this entry's management-controversy discussion records.

🐁

Animal Models

1

Endothelial BrafV600E mouse model of sporadic brain AVM

A mouse model in which BrafV600E is expressed in brain endothelial cells, developed specifically because earlier models produced heterogeneous and atypical lesions. It reproduces the human lesion closely, across ages and brain regions, with dilated vessels, reduced vascular wall stability, spontaneous haemorrhage and neuroinflammation.

Species

Mouse

Genotype

Brain endothelial-cell-specific BrafV600E expression

Publication

PMID:38700584

Pathograph links

Somatic Activating KRAS or BRAF Mutation in Brain Endothelium recapitulates fidelity: high Endothelial-restricted expression of a BRAF activating mutation is the model's design and matches the human somatic mosaic lesion in cell type and gene.

Limitations: The mutation is engineered and lineage-wide within the targeted endothelium rather than arising as a spontaneous focal clone, so the model does not reproduce the mosaic clone-boundary architecture of human disease.

Readouts

Formation of AVM lesions resembling human bAVM

Direction: INCREASED

Interpretation: Lesion formation on introducing the mutation is the causal demonstration that the endothelial mutation is sufficient.

Show evidence (1 reference)

PMID:38700584 SUPPORT Model Organism

"Mice with BrafV600E mutations in brain ECs developed BAVM closely resembled that of human lesions."

Reports lesion formation resembling human bAVM after endothelial BrafV600E expression.

Show evidence (1 reference)

PMID:38700584 SUPPORT Model Organism

"Here we developed a novel mouse model of sporadic BAVM that is consistent with clinical manifestations in humans."

Supports treating this model as informative for the somatic-mutation node.

Deficient Mural Cell Coverage and Vessel Wall Instability recapitulates fidelity: moderate The model's lesions show reduced vascular wall stability with spontaneous haemorrhage, the phenotype this node describes.

Limitations: Wall instability is reported as a pathological description rather than a quantified mural cell coverage measurement, so the model supports the phenotype more strongly than it supports the specific mural-cell mechanism.

Readouts

Vascular wall stability and spontaneous haemorrhage

Direction: DECREASED

Interpretation: Reduced wall stability with spontaneous bleeding is the in vivo correlate of the structural instability node.

Show evidence (1 reference)

PMID:38700584 SUPPORT Model Organism

"Pathological features of BAVM were primarily dilated blood vessels with reduced vascular wall stability, accompanied by spontaneous hemorrhage and neuroinflammation."

Reports reduced vascular wall stability with spontaneous haemorrhage in the model.

Show evidence (1 reference)

PMID:38700584 SUPPORT Model Organism

"Pathological features of BAVM were primarily dilated blood vessels with reduced vascular wall stability, accompanied by spontaneous hemorrhage and neuroinflammation."

Supports treating this model as informative for vessel wall instability.

MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability partially recapitulates fidelity: moderate Dabrafenib slowed lesion progression when given early, but its efficacy against established lesions was explicitly left uncertain - which is precisely the prevention-versus-reversal distinction the `bavm-mek-inhibition-translation` discussion is about, now carried as a structured model claim rather than only as prose.

Limitations: The therapeutic result is prevention of progression in a model where treatment can begin at lesion induction. Human patients present with lesions established for years, and the source states efficacy against established lesions remained uncertain.

Readouts

Lesion progression under early dabrafenib

Direction: DECREASED

Interpretation: Slowed progression on early BRAF inhibition supports the pathway as a target while leaving established-lesion reversal untested.

Show evidence (1 reference)

PMID:38700584 SUPPORT Model Organism

"Early administration of Dabrafenib was found to be effective in slowing the progression of BAVMs; however, its efficacy in treating established BAVM lesions remained uncertain."

Reports the prevention benefit and states explicitly that established-lesion efficacy is unresolved. Recorded as PARTIAL for that reason.

Show evidence (1 reference)

PMID:38700584 SUPPORT Model Organism

"Early administration of Dabrafenib was found to be effective in slowing the progression of BAVMs; however, its efficacy in treating established BAVM lesions remained uncertain."

The model informs the drug-target node while bounding what it can show, which is the substance of the model-mismatch discussion.

{ }

Source YAML

click to show

name: Brain arteriovenous malformation
creation_date: '2026-08-19T23:20:00Z'
category: Vascular Disease
description: >-
  A brain arteriovenous malformation (bAVM) is a focal tangle - the nidus - of
  dilated arterioles draining directly into veins with no intervening capillary bed.
  It is a leading cause of haemorrhagic stroke in children and young adults, and
  presents with intracranial haemorrhage, seizures, or focal neurological deficit.

  For most of the twentieth century bAVM was understood as a static congenital
  anatomical accident. That view is obsolete. Somatic activating KRAS mutations were
  found in the endothelium of the majority of sporadic lesions, present in AVM tissue
  and absent from the same patients' blood - a mosaic, lesion-restricted mutation,
  which is why the disease is focal, sporadic, and not inherited. Mutant KRAS drives
  constitutive MAPK-ERK signalling in brain endothelial cells, which in turn drives an
  aberrant angiogenic and Notch transcriptional programme, endothelial migration, loss
  of normal arteriovenous identity, and a glycolytic shift; the result is a nidus that
  shunts arterial blood directly into veins. Inhibiting MAPK-ERK reverses the cellular
  phenotype, which is what makes the pathway causal rather than correlative.

  The lesion then becomes clinically dangerous by two partly separable routes, both
  curated here: the haemodynamic route, in which high-flow shunting and venous
  hypertension load thin-walled vessels, and the inflammatory route, in which mutant
  endothelium secretes pro-inflammatory mediators, loses blood-brain-barrier junction
  proteins, and recruits microglia and macrophages that further degrade vessel
  integrity. This entry is deliberately about the sporadic, somatic-mosaic disease; the
  germline syndromic AVMs are separate entries (`Hereditary_Hemorrhagic_Telangiectasia`,
  `Capillary_Malformation-Arteriovenous_Malformation_Syndrome`), as is
  `Cerebral_Proliferative_Angiopathy`.
disease_term:
  preferred_term: arteriovenous malformations of the brain
  term:
    id: MONDO:0007154
    label: arteriovenous malformations of the brain
synonyms:
- Cerebral arteriovenous malformation
- Intracranial arteriovenous malformation
- bAVM
- Brain AVM
- Arteriovenous malformation of the brain, somatic
parents:
- Arteriovenous malformation
- Cerebrovascular disease
- Vascular malformation
classifications:
  harrisons_chapter:
  - classification_value: NEUROLOGIC
    evidence:
    - reference: PMID:29298116
      reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
      supports: SUPPORT
      evidence_source: HUMAN_CLINICAL
      snippet: >-
        Sporadic arteriovenous malformations of the brain, which are morphologically
        abnormal connections between arteries and veins in the brain vasculature, are a
        leading cause of hemorrhagic stroke in young adults and children.
      explanation: >-
        Brain AVM is a cerebrovascular lesion presenting as haemorrhagic stroke and
        seizure, placing it in Harrison's Neurologic Disorders Part.

genetic:
- name: KRAS
  notes: >-
    Somatic activating KRAS mutations - most often the canonical hotspot substitutions
    at codon 12 - are the predominant genetic cause of sporadic brain AVM. They are
    detectable in lesional tissue and endothelial-enriched cultures from that tissue but
    not in paired blood, establishing them as post-zygotic mosaic events confined to the
    lesion rather than inherited variants. This is the reason bAVM is focal and
    sporadic, and the reason genetic testing requires lesional tissue or ultra-sensitive
    cell-free DNA methods rather than a standard blood panel.
  gene_term:
    preferred_term: KRAS
    term:
      id: hgnc:6407
      label: KRAS
  association: Somatic activating mutation in lesional endothelium
  relationship_type: SOMATIC_DRIVER
  variant_origin: SOMATIC
  evidence:
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      We detected somatic activating KRAS mutations in tissue samples from 45 of the 72
      patients and in none of the 21 paired blood samples.
    explanation: >-
      Establishes both the causal gene and its somatic, lesion-restricted mosaic nature -
      present in AVM tissue, absent from the same patients' blood.
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      We identified activating KRAS mutations in the majority of tissue samples of
      arteriovenous malformations of the brain that we analyzed.
    explanation: >-
      States that activating KRAS mutation is the majority genotype of sporadic brain
      AVM, supporting CAUSAL rather than modifier status.

- name: BRAF
  notes: >-
    Somatic activating BRAF mutations account for a minority of sporadic brain AVMs and
    converge on the same RAS/MAPK output as KRAS. Endothelial-specific expression of
    mutant BRAF produces AVM-like lesions in animal models, as does mutant KRAS, which is
    the strongest evidence that pathway activation rather than the specific gene is the
    driver.
  gene_term:
    preferred_term: BRAF
    term:
      id: hgnc:1097
      label: BRAF
  association: Somatic activating mutation in lesional endothelium
  relationship_type: SOMATIC_DRIVER
  variant_origin: SOMATIC
  evidence:
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      The identification of somatic KRAS and BRAF mutations in sporadic bAVM endothelial
      cells (ECs) has fundamentally reshaped current understanding of bAVM biology,
      indicating that activation of the RAS/MAPK pathway drives aberrant angiogenic
      programs.
    explanation: >-
      Names BRAF alongside KRAS as a somatic driver in bAVM endothelium converging on
      RAS/MAPK. Evidence source is OTHER as this is a review article.
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: MODEL_ORGANISM
    snippet: >-
      Furthermore, animal models with EC-specific expression of mutant KRAS or BRAF
      exhibit bAVM-like lesions, which support the hypothesis that hyperactivation of the
      RAS/MAPK pathway is a key driver of lesion formation.
    explanation: >-
      Endothelial-specific expression of either mutant gene reproduces the lesion in
      animals, supporting pathway-level rather than gene-specific causation. Evidence
      source is MODEL_ORGANISM as the claim rests on animal models.

pathophysiology:
- name: Somatic Activating KRAS or BRAF Mutation in Brain Endothelium
  biological_scale: MOLECULAR
  role: trigger
  description: >-
    A post-zygotic activating mutation in KRAS, or less often BRAF, arises in a brain
    endothelial progenitor during development and is confined to the descendants of that
    cell. Because the mutation is mosaic and lesion-restricted - detectable in AVM tissue
    and in endothelial-enriched cultures derived from it, undetectable in the same
    patient's blood - the disease is focal and sporadic rather than inherited, and the
    lesion has a definite clonal boundary. This is the initiating event of sporadic brain
    AVM.
  cell_types:
  - preferred_term: brain microvascular endothelial cell
    term:
      id: CL:2000044
      label: brain microvascular endothelial cell
  downstream:
  - target: Constitutive RAS-MAPK-ERK Signalling in Brain Endothelial Cells
    causal_link_type: DIRECT
    description: >-
      An activating mutation in KRAS or BRAF locks the RAS-MAPK cascade into a
      ligand-independent active state.
  evidence:
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      We detected somatic activating KRAS mutations in tissue samples from 45 of the 72
      patients and in none of the 21 paired blood samples.
    explanation: >-
      Establishes the somatic, lesion-restricted mosaic mutation that initiates the
      disease, and the tissue-versus-blood contrast that defines it as post-zygotic.

- name: Constitutive RAS-MAPK-ERK Signalling in Brain Endothelial Cells
  biological_scale: MOLECULAR
  role: central_effector
  description: >-
    Mutant KRAS raises ERK activity in brain endothelial cells and holds the MAPK-ERK
    cascade active independently of upstream growth-factor input. This is the
    rate-limiting node of the disease: it is common to the KRAS and BRAF genotypes, it is
    what endothelial-specific expression of either mutant reproduces in animal models,
    and pharmacological inhibition of MAPK-ERK reverses the downstream cellular
    phenotype. That reversibility is what makes the node causal rather than a bystander
    correlate, and it is the rationale for MEK-directed therapy.
  biological_processes:
  - preferred_term: ERK1 and ERK2 cascade
    modifier: GAIN_OF_FUNCTION
    term:
      id: GO:0070371
      label: ERK1 and ERK2 cascade
  - preferred_term: MAPK cascade
    modifier: INCREASED
    term:
      id: GO:0000165
      label: MAPK cascade
  cell_types:
  - preferred_term: brain microvascular endothelial cell
    term:
      id: CL:2000044
      label: brain microvascular endothelial cell
  downstream:
  - target: Aberrant Angiogenic and Notch Programme with Loss of Arteriovenous Identity
    causal_link_type: DIRECT
    description: >-
      Sustained ERK activity drives the angiogenic and Notch transcriptional programme
      and the migratory phenotype of mutant endothelium.
  - target: Endothelial Glycolytic Reprogramming via GLUT1 and Hexokinase-2
    causal_link_type: DIRECT
    description: >-
      Mutant KRAS signalling raises glucose uptake and glycolytic flux, supplying the
      metabolic demand of the angiogenic phenotype.
  - target: MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability
    causal_link_type: DIRECT
    description: >-
      Because the pathway is constitutively active and its inhibition reverses the
      cellular phenotype, it is the actionable drug target of the disease.
  evidence:
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      expression of mutant KRAS (KRASG12V) in endothelial cells in vitro induced increased
      ERK (extracellular signal-regulated kinase) activity
    explanation: >-
      Directly demonstrates that the mutation raises ERK activity in endothelial cells,
      the molecular content of this node.
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      We propose that these malformations develop as a result of KRAS-induced activation
      of the MAPK-ERK signaling pathway in brain endothelial cells.
    explanation: >-
      States the paper's central causal proposal, which is exactly the claim of this node.

- name: Endothelial Glycolytic Reprogramming via GLUT1 and Hexokinase-2
  biological_scale: MOLECULAR
  role: modifier
  description: >-
    Mutant KRAS raises endothelial glucose uptake and glycolytic flux by increasing
    membrane localisation of glucose transporters such as GLUT1 and inducing
    hexokinase-2. Notably, HK2 appears to be the only glycolytic enzyme elevated, so this
    is a targeted metabolic rewiring rather than a generalised upregulation, and HK2 is
    elevated in human as well as mouse AVM tissue. The reprogramming is required for the
    angiogenic phenotype: inhibiting glycolytic flux or knocking down HK2 suppresses
    sprouting angiogenesis in mutant endothelial cells, and glycolysis inhibition
    potentiates MEK inhibition.
  biological_processes:
  - preferred_term: glycolytic process
    modifier: INCREASED
    term:
      id: GO:0006096
      label: glycolytic process
  downstream:
  - target: Aberrant Angiogenic and Notch Programme with Loss of Arteriovenous Identity
    causal_link_type: DIRECT
    description: >-
      Glycolytic flux supplies the biosynthetic and energetic demand of sprouting
      angiogenesis; blocking it suppresses sprouting in mutant endothelium.
  evidence:
  - reference: PMID:41708990
    reference_title: >-
      KRAS-dependent glycolytic reprogramming of endothelial cells in sporadic
      arteriovenous malformations.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      We found that KRASG12V expression in the endothelium increased angiogenesis, which
      was accompanied by enhanced glucose uptake and glycolytic flux.
    explanation: >-
      Establishes the metabolic shift downstream of mutant KRAS that this node represents.
  - reference: PMID:41708990
    reference_title: >-
      KRAS-dependent glycolytic reprogramming of endothelial cells in sporadic
      arteriovenous malformations.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      Critically, either pharmacologic inhibition of glycolytic flux or knockdown of HK2
      suppressed sprouting angiogenesis in cultured KRASG12V endothelial cells.
    explanation: >-
      Demonstrates that the glycolytic shift is required for the angiogenic phenotype,
      justifying the causal downstream edge rather than a mere association.

- name: Aberrant Angiogenic and Notch Programme with Loss of Arteriovenous Identity
  biological_scale: CELLULAR
  role: central_effector
  description: >-
    Mutant endothelium upregulates genes of angiogenesis and Notch signalling and becomes
    abnormally migratory. Single-cell transcriptomic profiling of human AVM endothelium
    additionally shows endothelial-to-mesenchymal-transition-like features and loss of the
    normal arterial-versus-venous identity programme. Losing arteriovenous identity is the
    specific cellular failure that matters here: it is why the malformed vessels cannot
    resolve into a properly specified arterial-capillary-venous sequence and instead form a
    direct shunt.
  biological_processes:
  - preferred_term: angiogenesis
    modifier: INCREASED
    term:
      id: GO:0001525
      label: angiogenesis
  - preferred_term: Notch signaling pathway
    modifier: INCREASED
    term:
      id: GO:0007219
      label: Notch signaling pathway
  - preferred_term: cell migration
    modifier: INCREASED
    term:
      id: GO:0016477
      label: cell migration
  downstream:
  - target: Nidus Formation with Direct Arteriovenous Shunting
    causal_link_type: DIRECT
    description: >-
      Disordered angiogenesis without correct arteriovenous specification builds the
      tangle of shunting vessels that constitutes the nidus.
  - target: Deficient Mural Cell Coverage and Vessel Wall Instability
    causal_link_type: DIRECT
    description: >-
      Loss of Notch-dependent arteriovenous identity and the accompanying EndMT-like shift
      disrupt the endothelial signalling that recruits pericytes and smooth muscle cells to the
      vessel wall, so the malformed vessels are built without adequate mural investment.
  evidence:
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      increased expression of genes related to angiogenesis and Notch signaling, and
      enhanced migratory behavior. These processes were reversed by inhibition of MAPK
      (mitogen-activated protein kinase)-ERK signaling.
    explanation: >-
      Establishes both the angiogenic/Notch/migratory programme of this node and its
      dependence on the upstream MAPK-ERK node, since inhibiting ERK reverses it.
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      including heightened angiogenic and inflammatory signaling, endothelial-to-mesenchymal
      transition-like features, and loss of normal arteriovenous identity
    explanation: >-
      Adds the loss of arteriovenous identity and EndMT-like phenotype seen in single-cell
      profiling of human AVM endothelium. Evidence source is OTHER as this is a review
      article.
  - reference: PMID:35084939
    reference_title: A single-cell atlas of the normal and malformed human brain vasculature.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      we investigated cellular and molecular perturbations in brain arteriovenous
      malformations, which are a leading cause of stroke in young people, and identified
      pathologic endothelial transformations with abnormal vascular patterning and the
      ontology of vascularly derived inflammation
    explanation: >-
      Primary human single-cell evidence for pathologic endothelial transformation and
      abnormal vascular patterning in AVM tissue, upgrading this node from review-level to
      direct human-tissue support.

- name: Nidus Formation with Direct Arteriovenous Shunting
  biological_scale: TISSUE
  role: central_effector
  description: >-
    The defining anatomical lesion: a focal tangle of dilated, tortuous vessels in which
    feeding arteries connect directly to draining veins with no intervening capillary bed.
    The absent capillary bed is the mechanistically important feature, because it is the
    capillary bed that normally drops arterial pressure before blood reaches the venous
    side. Without it, arterial pressure is transmitted directly into thin-walled venous
    structures, and the lesion behaves as a low-resistance high-flow shunt.
  cell_types:
  - preferred_term: endothelial cell of artery
    term:
      id: CL:1000413
      label: endothelial cell of artery
  - preferred_term: vein endothelial cell
    term:
      id: CL:0002543
      label: vein endothelial cell
  downstream:
  - target: High-Flow Shunt Haemodynamics and Venous Hypertension
    causal_link_type: DIRECT
    description: >-
      A capillary-free arteriovenous connection transmits arterial pressure and flow
      directly into the venous outflow.
  - target: Vascular Inflammation and Blood-Brain Barrier Junction Loss
    causal_link_type: DIRECT
    description: >-
      The malformed nidus vessels are lined by mutant endothelium that is
      pro-inflammatory and barrier-deficient.
  evidence:
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      Brain arteriovenous malformations (bAVMs) are high-flow vascular lesions
      characterized by direct arteriovenous shunting without an intervening capillary bed.
    explanation: >-
      Defines the nidus lesion in exactly the terms of this node - high flow, direct
      shunting, absent capillary bed. Evidence source is OTHER as this is a review article.

- name: High-Flow Shunt Haemodynamics and Venous Hypertension
  biological_scale: TISSUE
  role: consequence
  description: >-
    Arterial pressure delivered into the venous side produces venous hypertension,
    progressive dilation of draining veins, and mechanical loading of vessel walls that
    were never built to withstand it. The haemodynamic burden also perturbs perinidal
    cortex. This is the mechanical arm of rupture risk, and it is separable from - and
    additive to - the inflammatory arm. Deep venous drainage is among the anatomical
    features associated with higher haemorrhage risk in natural-history cohorts, which is
    consistent with an outflow-restriction mechanism.
  downstream:
  - target: Nidus Rupture and Intracranial Haemorrhage
    causal_link_type: DIRECT
    description: >-
      Sustained pressure loading of thin-walled malformed vessels is the mechanical
      precipitant of rupture.
  - target: Perinidal Cortical Irritation and Seizure Generation
    causal_link_type: INDIRECT_KNOWN_INTERMEDIATES
    description: >-
      Venous hypertension, perinidal gliosis and haemosiderin deposition render adjacent
      cortex epileptogenic.
  evidence:
  - reference: PMID:25015366
    reference_title: >-
      Untreated brain arteriovenous malformation: patient-level meta-analysis of hemorrhage
      predictors.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Female sex (1.49, 95% CI 0.96-2.30) and exclusively deep venous drainage (1.60,
      0.95-2.68, p = 0.02 in imputed dataset) may be additional predictors.
    explanation: >-
      Deep venous drainage - an outflow-restriction feature - is a candidate haemorrhage
      predictor. Recorded as PARTIAL because the confidence interval crosses 1 in the
      primary analysis and the authors state it only "may be" a predictor.

- name: Vascular Inflammation and Blood-Brain Barrier Junction Loss
  biological_scale: CELLULAR
  role: central_effector
  description: >-
    Mutant endothelium is not merely misshapen, it is inflammatory. KRASG12V-expressing
    endothelial cells increase expression of pro-inflammatory mediators and reduce
    expression of blood-brain-barrier junction constituents, forming a leaky barrier in
    vitro. Their conditioned medium activates microglia, and medium from those primed
    microglia in turn degrades junction proteins in wild-type endothelial cells - a
    feed-forward loop in which the mutant clone recruits normal neighbouring tissue into
    the lesion. Blood-derived macrophages worsen the barrier defect further. This is the
    inflammatory arm of vascular instability, distinct from the haemodynamic arm.
  cell_types:
  - preferred_term: brain microvascular endothelial cell
    term:
      id: CL:2000044
      label: brain microvascular endothelial cell
  - preferred_term: microglial cell
    term:
      id: CL:0000129
      label: microglial cell
  - preferred_term: macrophage
    term:
      id: CL:0000235
      label: macrophage
  downstream:
  - target: Nidus Rupture and Intracranial Haemorrhage
    causal_link_type: DIRECT
    description: >-
      Loss of junctional integrity and inflammatory degradation of the vessel wall
      destabilise malformed vessels and promote leakage and rupture.
  evidence:
  - reference: PMID:41696778
    reference_title: >-
      Mutant KRAS in brain endothelial cells promotes vascular inflammation and impairs
      vascular integrity in brain arteriovenous malformation.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      Here we demonstrate that KRASG12V expressing cultured ECs (KRAS-G12V-EC) have
      increased expression of pro-inflammatory mediators and reduced expression of
      blood-brain-barrier (BBB) junction constituents.
    explanation: >-
      Establishes the two components of this node - pro-inflammatory secretion and loss of
      barrier junction proteins - as direct consequences of the mutation in endothelium.
  - reference: PMID:41696778
    reference_title: >-
      Mutant KRAS in brain endothelial cells promotes vascular inflammation and impairs
      vascular integrity in brain arteriovenous malformation.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      The conditioned medium from KRAS-G12V-EC can activate BV2-microglia (BV2-MG) and
      conditioned media from this primed BV2-MG can compromise the expression of
      EC-junction constituents when added to wild-type ECs.
    explanation: >-
      Documents the feed-forward loop by which mutant endothelium recruits microglia that
      then damage non-mutant endothelium, which is why the lesion destabilises beyond the
      mutant clone itself.
  - reference: PMID:35084939
    reference_title: A single-cell atlas of the normal and malformed human brain vasculature.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      We illustrate the interplay between vascular and immune cells that contributes to
      brain hemorrhage and catalog opportunities for targeting angiogenic and inflammatory
      programs in vascular malformations.
    explanation: >-
      Single-cell profiling of human brain AVM tissue ties vascular-immune crosstalk
      directly to haemorrhage, which is the human-tissue counterpart of the in-vitro
      endothelium-microglia loop and the reason this node is not merely a culture
      phenomenon.

- name: Perinidal Cortical Irritation and Seizure Generation
  biological_scale: TISSUE
  role: consequence
  description: >-
    Cortex adjacent to the nidus becomes epileptogenic through venous hypertension,
    gliosis, and haemosiderin deposition from prior microhaemorrhage. Seizure is the
    second commonest presentation of brain AVM after haemorrhage, and in surgical series
    of unruptured lesions it is the dominant presenting symptom.
  evidence:
  - reference: PMID:40428814
    reference_title: >-
      The Dynamics of Seizures After Microsurgical Treatment of Brain AVMs in Patients with
      Symptomatic Epilepsy: A Single-Center Experience over 10 Years.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Seizures are a common clinical manifestation in patients with brain AVMs, ranking as
      the second most frequent presentation.
    explanation: >-
      Establishes seizure as the second commonest presentation of brain AVM, the clinical
      content of this node.

- name: Nidus Rupture and Intracranial Haemorrhage
  biological_scale: ORGANISM
  role: consequence
  description: >-
    Rupture of malformed nidus or draining vessels produces intracerebral, intraventricular
    or subarachnoid haemorrhage - the presentation that makes brain AVM a leading cause of
    haemorrhagic stroke in the young. In an individual-patient meta-analysis of four
    untreated cohorts spanning 6,074 patient-years, the annual haemorrhage rate was 2.3%
    overall, and this figure is strongly conditioned on presentation: 4.8% per year in
    lesions that had already ruptured versus 1.3% per year in those that had not. Prior
    haemorrhage and increasing age were the independent predictors.
  evidence:
  - reference: PMID:25015366
    reference_title: >-
      Untreated brain arteriovenous malformation: patient-level meta-analysis of hemorrhage
      predictors.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      A total of 141 hemorrhage events occurred during 6,074 patient-years of follow-up
      (annual rate of 2.3%, 95% confidence interval [CI] 2.0%-2.7%), higher for ruptured
      (4.8%, 3.9%-5.9%) than unruptured (1.3%, 1.0%-1.7%) AVMs at presentation.
    explanation: >-
      Quantifies the natural-history haemorrhage rate and the ruptured-versus-unruptured
      split that governs management decisions.
  - reference: PMID:25015366
    reference_title: >-
      Untreated brain arteriovenous malformation: patient-level meta-analysis of hemorrhage
      predictors.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      This large, individual patient data meta-analysis identified hemorrhagic presentation
      and increasing age as independent predictors of hemorrhage during follow-up.
    explanation: >-
      Identifies the independent predictors of subsequent haemorrhage in untreated disease.

- name: MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability
  biological_scale: MOLECULAR
  role: therapeutic_vulnerability
  description: >-
    Because the disease runs through a single constitutively active kinase cascade,
    inhibiting that cascade is a candidate disease-modifying therapy - the first such
    prospect for a lesion otherwise treated only by destroying or removing it. Inhibition
    of MAPK-ERK reverses the angiogenic, Notch and migratory phenotype of mutant endothelium
    in vitro; MEK and BRAF inhibitors show promise in preclinical models; and combined
    glycolysis and MEK inhibition suppresses angiogenesis in patient-derived AVM endothelial
    cells. This is explicitly a preclinical vulnerability: no MEK inhibitor is established
    therapy for brain AVM, and the `bavm-mek-inhibition-translation` discussion records what
    stands between the model data and clinical use. The clinical motivation is the haemorrhage
    risk documented on the rupture node: absent a medical option, the only alternative is a
    destructive intervention on an often asymptomatic lesion. That is a rationale for pursuing
    the target, not a causal relationship, so it is recorded here rather than encoded as a
    pathograph edge.
  biological_processes:
  - preferred_term: ERK1 and ERK2 cascade
    term:
      id: GO:0070371
      label: ERK1 and ERK2 cascade
  evidence:
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      These processes were reversed by inhibition of MAPK (mitogen-activated protein
      kinase)-ERK signaling.
    explanation: >-
      Demonstrates pharmacological reversibility of the mutant endothelial phenotype, the
      basis for treating this pathway as a therapeutic target.
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      MEK and BRAF inhibitors targeting the RAS/MAPK pathway have shown promising results in
      preclinical studies. However, clinical translation remains challenging because of low
      variant allele frequencies and limited access to lesional tissue for genetic testing.
    explanation: >-
      Supports the target while explicitly stating that translation is unresolved. Recorded
      as PARTIAL because the same sentence that supports the vulnerability also qualifies
      its clinical readiness.
  - reference: PMID:41708990
    reference_title: >-
      KRAS-dependent glycolytic reprogramming of endothelial cells in sporadic arteriovenous
      malformations.
    supports: SUPPORT
    evidence_source: IN_VITRO
    snippet: >-
      Finally, combined glycolysis and MEK inhibition suppressed angiogenesis in
      patient-derived bAVM primary endothelial cells.
    explanation: >-
      Shows the combination acting on endothelial cells derived from actual patient lesions,
      the closest available approach to human evidence for this target.

- name: Deficient Mural Cell Coverage and Vessel Wall Instability
  biological_scale: TISSUE
  role: central_effector
  description: >-
    Nidus vessels are structurally defective, and a specific component of that defect is
    inadequate mural cell coverage. Pericytes and smooth muscle cells are what give a vessel
    wall its integrity, and changes in their number and coverage are implicated in bAVM; their
    recruitment during vascular remodelling is governed by PDGF-B/PDGFR-beta, EphrinB2/EphB4
    and angiopoietin/Tie2 signalling. This sits between the malformed nidus and rupture as the
    structural reason those vessels are fragile - complementing, not duplicating, the
    inflammatory arm, which degrades endothelial junctions rather than mural coverage. Human
    bulk RNA-seq of bAVM tissue is consistent with it: extracellular-matrix and
    angiopoietin-TIE genes are among the downregulated set.
  cell_types:
  - preferred_term: pericyte
    term:
      id: CL:0000669
      label: pericyte
  - preferred_term: smooth muscle cell
    term:
      id: CL:0000192
      label: smooth muscle cell
  downstream:
  - target: Nidus Rupture and Intracranial Haemorrhage
    causal_link_type: DIRECT
    description: >-
      Vessels with deficient mural coverage lack the wall integrity to contain arterial
      pressure, which is the structural precondition for rupture.
  evidence:
  - reference: PMID:34541474
    reference_title: The role of mural cells in hemorrhage of brain arteriovenous malformation.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      The vessel walls in bAVMs have structural defects, which impair vascular integrity. Mural
      cells are essential structural and functional components of blood vessels and play a
      critical role in maintaining vascular integrity. Changes in mural cell number and
      coverage have been implicated in bAVMs.
    explanation: >-
      Establishes deficient mural cell coverage as a structural defect of bAVM vessels
      impairing vascular integrity. Evidence source is OTHER as this is a review article.
  - reference: PMID:31795902
    reference_title: >-
      RNA-Sequencing Highlights Inflammation and Impaired Integrity of the Vascular Wall in
      Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      we found 498 genes downregulated including genes implicated in extracellular matrix
      composition, the binary angiopoietin-TIE system, and TGF (transforming growth factor)-beta
      signaling
    explanation: >-
      Human bAVM tissue RNA-seq showing downregulation of extracellular-matrix and
      angiopoietin-TIE programmes, the signalling system that governs mural cell recruitment.

phenotypes:
- category: Structural
  name: Cerebral arteriovenous malformation
  description: >-
    The defining structural lesion: a nidus of dilated vessels shunting arterial blood
    directly into veins without an intervening capillary bed, demonstrated on catheter
    angiography, CT angiography or MR angiography.
  phenotype_term:
    preferred_term: Cerebral arteriovenous malformation
    term:
      id: HP:0002408
      label: Cerebral arteriovenous malformation
  evidence:
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      Brain arteriovenous malformations (bAVMs) are high-flow vascular lesions
      characterized by direct arteriovenous shunting without an intervening capillary bed.
    explanation: >-
      Defines the structural lesion. Evidence source is OTHER as this is a review article.

- category: Clinical
  name: Intracranial haemorrhage
  description: >-
    Rupture producing intracerebral, intraventricular or subarachnoid blood, and the
    presentation that makes brain AVM a leading cause of haemorrhagic stroke in children
    and young adults. No frequency band is asserted: haemorrhage is an incident event with
    an annual rate (2.3% per year overall in untreated disease) rather than a
    cross-sectional proportion of patients, so mapping it to FrequencyEnum would misstate
    it.
  phenotype_term:
    preferred_term: Intracranial hemorrhage
    term:
      id: HP:0002170
      label: Intracranial hemorrhage
    temporality: ACUTE
  evidence:
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      are a leading cause of hemorrhagic stroke in young adults and children
    explanation: >-
      Establishes haemorrhagic stroke as the defining clinical danger of brain AVM.
  - reference: PMID:25015366
    reference_title: >-
      Untreated brain arteriovenous malformation: patient-level meta-analysis of hemorrhage
      predictors.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      A total of 141 hemorrhage events occurred during 6,074 patient-years of follow-up
      (annual rate of 2.3%, 95% confidence interval [CI] 2.0%-2.7%), higher for ruptured
      (4.8%, 3.9%-5.9%) than unruptured (1.3%, 1.0%-1.7%) AVMs at presentation.
    explanation: >-
      Supplies the annual haemorrhage rate in untreated disease, stratified by whether the
      lesion had previously ruptured.

- category: Clinical
  name: Seizure
  description: >-
    Seizures arising from cortex adjacent to the nidus, and the second commonest
    presentation of brain AVM overall. In surgical series restricted to unruptured
    lesions, seizure dominates the presenting picture - it was the presenting symptom in
    80.4% of one 52-patient unruptured cohort - but that figure describes a
    surgically-selected unruptured population and is not a frequency for the disease as a
    whole, so no band is asserted.
  phenotype_term:
    preferred_term: Seizure
    term:
      id: HP:0001250
      label: Seizure
  evidence:
  - reference: PMID:40428814
    reference_title: >-
      The Dynamics of Seizures After Microsurgical Treatment of Brain AVMs in Patients with
      Symptomatic Epilepsy: A Single-Center Experience over 10 Years.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Seizures are a common clinical manifestation in patients with brain AVMs, ranking as
      the second most frequent presentation.
    explanation: >-
      Establishes seizure as the second commonest presentation of brain AVM.
  - reference: PMID:41303829
    reference_title: >-
      Microsurgical Resection of Unruptured Brain Arteriovenous Malformations: A 12-Year
      Single-Center Experience.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Epileptic seizure was the most common presenting symptom (80.4%).
    explanation: >-
      Quantifies seizure as the dominant presentation in a surgical series of unruptured
      lesions. Not mapped to a frequency band because the cohort is restricted to
      unruptured, surgically treated AVMs.

- category: Clinical
  name: Stroke
  description: >-
    Haemorrhagic stroke from nidus rupture, and less often ischaemic events. Stroke is the
    outcome that dominates management decisions, and it was the composite endpoint of the
    randomised trial of unruptured lesions.
  phenotype_term:
    preferred_term: Stroke
    term:
      id: HP:0001297
      label: Stroke
    temporality: ACUTE
  evidence:
  - reference: PMID:24268105
    reference_title: >-
      Medical management with or without interventional therapy for unruptured brain
      arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      The primary outcome is time to the composite endpoint of death or symptomatic stroke
    explanation: >-
      Stroke is the primary clinical outcome of brain AVM against which management
      strategies are judged.

- category: Clinical
  name: Focal neurological deficit
  description: >-
    Focal deficits - weakness, sensory disturbance, speech difficulty, visual field loss -
    arising from haemorrhage into eloquent territory, mass effect of the nidus, or perinidal
    ischaemia. In surgical series of unruptured lesions they are a less common presentation
    than seizure. No ontology term is bound: HPO has no general "focal neurological deficit"
    concept, and the available evidence names the deficits collectively rather than
    identifying hemiparesis specifically, so binding to HP:0001269 Hemiparesis would assert a
    particular deficit the sources do not single out. ECTO-style substitution is not available
    here either; the broad alternative, HP:0012638 Abnormal nervous system physiology, is too
    coarse to carry meaning.
  phenotype_term:
    preferred_term: focal neurological deficit
  evidence:
  - reference: PMID:41303829
    reference_title: >-
      Microsurgical Resection of Unruptured Brain Arteriovenous Malformations: A 12-Year
      Single-Center Experience.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      In our cohort, epileptic seizures were the most frequent initial clinical manifestation,
      whereas headaches and focal neurological deficits were less commonly observed.
    explanation: >-
      Names focal neurological deficits as a presenting manifestation and places their
      frequency below seizure in an unruptured surgical cohort. No frequency band is asserted
      because the source gives an ordering, not a proportion, and the cohort is
      surgically selected.
  - reference: PMID:42138076
    reference_title: Towards precision medicine for brain arteriovenous malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      bAVMs present across a clinically heterogeneous spectrum with variable risk of
      hemorrhage, epilepsy, neurologic deficit, and systemic involvement
    explanation: >-
      Independently names neurologic deficit as one of the core clinical consequences of brain
      AVM. Evidence source is OTHER as this is a review article.

prevalence:
- population: New York islands, United States (prospective population-based surveillance)
  measure_type: ANNUAL_INCIDENCE
  prevalence_class: BAND_1_9_PER_100000
  rate_per_100000: 1.34
  rate_low: 1.18
  rate_high: 1.49
  notes: >-
    Average annual detection rate of brain AVM, from 284 prospectively ascertained
    patients over 21,216,467 person-years in a defined population of 9.4 million. This is a
    detection rate for a congenital lesion rather than a true incidence of lesion
    formation, since the malformation predates its discovery.
  evidence:
  - reference: PMID:12690217
    reference_title: 'The New York Islands AVM Study: design, study progress, and initial results.'
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      leading to an average annual AVM detection rate of 1.34 per 100,000 person-years (95%
      CI, 1.18 to 1.49)
    explanation: >-
      Prospective population-based detection rate, the basis for the normalised rate
      recorded here.
- population: Scotland, adults aged 16 and over (prospective population-based study)
  measure_type: ANNUAL_INCIDENCE
  prevalence_class: BAND_1_9_PER_100000
  rate_per_100000: 1.12
  rate_low: 0.90
  rate_high: 1.37
  notes: >-
    Crude first-ever-diagnosis detection rate for brain AVM among Scottish adults, from the
    Scottish Intracranial Vascular Malformation Study. Independently reproduces the New York
    figure in a different health system, which is why both are recorded.
  evidence:
  - reference: PMID:12702837
    reference_title: >-
      Prospective, population-based detection of intracranial vascular malformations in
      adults: the Scottish Intracranial Vascular Malformation Study (SIVMS).
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      1.12 (95% CI, 0.90 to 1.37) for brain AVMs
    explanation: >-
      Crude annual detection rate per 100,000 adults for brain AVM in a prospective
      population-based study.

diagnosis:
- name: Digital subtraction angiography
  description: >-
    Catheter DSA is the reference standard for brain AVM, defining the feeding arteries, nidus
    architecture and venous drainage pattern in the temporal resolution that lower-tech
    modalities cannot match. It is what the angioarchitectural grading schemes are scored
    against, and what determines whether a lesion is treatable.
  diagnosis_term:
    preferred_term: digital subtraction angiography
    term:
      id: NCIT:C190556
      label: Angiography
  results: >-
    Demonstration of a nidus with arteriovenous shunting, with feeding-artery and
    draining-vein anatomy sufficient to assign a Spetzler-Martin grade.
  evidence:
  - reference: PMID:39125532
    reference_title: >-
      Diagnostic Accuracy of Non-Contrast-Enhanced Time-Resolved MR Angiography to Assess
      Angioarchitectural Classification Features of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      using digital subtraction angiography (DSA) as the reference standard
    explanation: >-
      Confirms DSA as the reference standard against which other modalities are measured.

- name: Non-contrast-enhanced time-resolved MR angiography
  description: >-
    Non-contrast 4D MR angiography detects the shunt and characterises angioarchitecture with
    accuracy comparable to contrast-enhanced 4D-MRA, and grades concordantly with DSA on the
    Spetzler-Martin and Buffalo scales. Its value is specific to this disease: brain AVM is a
    lifelong lesion requiring repeated imaging, so avoiding cumulative gadolinium exposure over
    decades of follow-up matters more here than for a one-off diagnostic question.
  diagnosis_term:
    preferred_term: magnetic resonance angiography
    term:
      id: NCIT:C190557
      label: Magnetic Resonance Angiography
  results: >-
    Detection of arteriovenous shunting and grading concordant with DSA, without
    gadolinium-based contrast.
  evidence:
  - reference: PMID:39125532
    reference_title: >-
      Diagnostic Accuracy of Non-Contrast-Enhanced Time-Resolved MR Angiography to Assess
      Angioarchitectural Classification Features of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Our results demonstrated that NCE-4D-MRA had a higher accuracy and specificity compared to
      CE-4D-MRA (0.85 vs. 0.83 and 95% vs. 85%, respectively) and similar agreement, with DSA
      detecting shunts in bAVMs or residuals.
    explanation: >-
      Quantifies non-contrast 4D-MRA accuracy against the contrast-enhanced comparator and DSA.
  - reference: PMID:39125532
    reference_title: >-
      Diagnostic Accuracy of Non-Contrast-Enhanced Time-Resolved MR Angiography to Assess
      Angioarchitectural Classification Features of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      The safety profile of imaging techniques is a significant concern in the long-term follow
      up of bAVMs
    explanation: >-
      States the contrast-avoidance rationale that makes this modality specifically valuable in
      a lifelong lesion requiring repeated imaging.

- name: Angioarchitectural grading (Spetzler-Martin and related scales)
  description: >-
    Grading by nidus size, eloquence of adjacent brain and venous drainage pattern is not a
    descriptive exercise but the operative determinant of whether intervention helps or harms:
    in a microsurgical series stratified by supplementary Spetzler-Martin grade, low-risk
    patients had rare poor outcomes while high-risk patients had poor outcomes in the majority.
    Grading is therefore curated here as a diagnostic step rather than buried in treatment
    prose.
  diagnosis_term:
    preferred_term: disease grading
    term:
      id: NCIT:C18020
      label: Diagnostic Procedure
  results: >-
    A Spetzler-Martin or supplementary Spetzler-Martin grade that stratifies operative risk.
  evidence:
  - reference: PMID:41303829
    reference_title: >-
      Microsurgical Resection of Unruptured Brain Arteriovenous Malformations: A 12-Year
      Single-Center Experience.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      All patients were classified according to the supplementary Spetzler-Martin (suppl-SM)
      grading system and stratified into low-risk (suppl-SM 6) and high-risk (suppl-SM > 6)
      groups.
    explanation: >-
      Documents the grading system in operational use to stratify surgical risk.
  - reference: PMID:39125532
    reference_title: >-
      Diagnostic Accuracy of Non-Contrast-Enhanced Time-Resolved MR Angiography to Assess
      Angioarchitectural Classification Features of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Concordance in grading bAVMs was substantial between NCE-4D-MRA and DSA, particularly for
      the Spetzler-Martin and Buffalo scales
    explanation: >-
      Shows the grading scales can be applied on non-invasive imaging with substantial
      concordance against DSA.

- name: Somatic genotyping from lesion tissue or cell-free DNA
  description: >-
    This is the diagnostic consequence of the entry's central mechanistic claim, and it is
    counterintuitive enough to be worth stating plainly: because the causative KRAS or BRAF
    mutation is a post-zygotic mosaic event confined to lesional endothelium, it is
    undetectable in blood. A conventional germline blood panel will be negative in a patient
    whose disease is entirely genetically explained. Genotyping therefore requires resected or
    endovascularly biopsied lesion tissue, or ultra-sensitive cell-free DNA methods, and the
    low variant allele fraction is the main obstacle to genotype-guided therapy.
  diagnosis_term:
    preferred_term: molecular genetic testing
    term:
      id: NCIT:C15709
      label: Genetic Testing
  results: >-
    Detection of a somatic activating KRAS or BRAF variant in lesional tissue; a negative blood
    test does not exclude the diagnosis.
  evidence:
  - reference: PMID:29298116
    reference_title: Somatic Activating KRAS Mutations in Arteriovenous Malformations of the Brain.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      We detected somatic activating KRAS mutations in tissue samples from 45 of the 72 patients
      and in none of the 21 paired blood samples.
    explanation: >-
      The tissue-positive, blood-negative contrast that makes lesional sampling necessary and a
      germline blood panel uninformative.
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      Future approaches combining minimally invasive sampling methods, such as endovascular
      biopsy and peripheral blood cell-free DNA analysis, with ultra-sensitive detection
      technologies are expected to help overcome these limitations.
    explanation: >-
      Names the emerging sampling routes that address the tissue-access problem. Evidence source
      is OTHER as this is a review article.

treatments:
- name: Microsurgical resection
  description: >-
    Open surgical excision of the nidus, the only modality that removes the lesion
    immediately and completely and the only one consistently associated with reduced
    haemorrhage risk in sensitivity analyses of large registry data. Outcome depends
    steeply on lesion grade: in a 12-year single-centre series of unruptured lesions
    stratified by supplementary Spetzler-Martin grade, poor outcomes were rare in the
    low-risk group but common in the high-risk group.
  treatment_term:
    preferred_term: Neurosurgical Procedure
    term:
      id: NCIT:C15656
      label: Neurosurgical Procedure
  therapeutic_modality: SURGERY
  target_mechanisms:
  - target: Nidus Formation with Direct Arteriovenous Shunting
    treatment_effect: INHIBITS
    description: >-
      Resection physically removes the shunting nidus, eliminating both the haemodynamic
      and the inflammatory arms of the lesion at once.
  evidence:
  - reference: PMID:41303829
    reference_title: >-
      Microsurgical Resection of Unruptured Brain Arteriovenous Malformations: A 12-Year
      Single-Center Experience.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      In the low-risk group, there were no deaths, and poor outcomes were rare (6.8% at
      discharge; 2.3% at 9 months). Conversely, the high-risk group demonstrated
      significantly worse outcomes (62.5% poor outcome at discharge, 28.6% at 9 months).
    explanation: >-
      Supports resection in low-grade lesions while documenting substantial harm in
      high-grade lesions. Recorded as PARTIAL because the same result both supports and
      restricts the treatment.
  - reference: PMID:41231469
    reference_title: >-
      Interventional Treatment vs Conservative Management of Unruptured Brain Arteriovenous
      Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Only microsurgical resection was consistently associated with reduced hemorrhage risk
      in sensitivity analyses.
    explanation: >-
      Among interventional modalities, resection was the one with a consistent haemorrhage
      benefit in a 1,770-patient target-trial-emulation analysis.

- name: Stereotactic radiosurgery
  description: >-
    Focused radiation inducing progressive endothelial proliferation and luminal closure of
    nidus vessels over two to three years. Obliteration is delayed and incomplete, so the
    patient remains at haemorrhage risk during the latency period, and radiation-induced
    change is an additional hazard - in a paediatric unruptured cohort, obliteration reached
    64% at 5 years and 82% at 10 years while permanent radiation-induced change occurred in
    8%.
  treatment_term:
    preferred_term: Stereotactic Radiosurgery
    term:
      id: NCIT:C15358
      label: Stereotactic Radiosurgery
  therapeutic_modality: RADIOTHERAPY
  target_mechanisms:
  - target: Nidus Formation with Direct Arteriovenous Shunting
    treatment_effect: INHIBITS
    description: >-
      Radiation-induced obliteration progressively closes nidus vessels, abolishing the
      shunt over a period of years rather than immediately.
  evidence:
  - reference: PMID:31942635
    reference_title: "Radiosurgery for Unruptured Intervention-Naïve Pediatric Brain Arteriovenous Malformations."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Estimated probabilities of AVM obliteration at 5 and 10 yr were 64% and 82%,
      respectively.
    explanation: >-
      Quantifies delayed and incomplete obliteration. Recorded as PARTIAL because the same
      cohort shows the treatment leaves a substantial fraction unobliterated for years.
  - reference: PMID:31942635
    reference_title: "Radiosurgery for Unruptured Intervention-Naïve Pediatric Brain Arteriovenous Malformations."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      The primary endpoint occurred in 14%, comprising hemorrhagic stroke, death, and
      permanent radiation-induced changes in 6%, 3%, and 8%, respectively.
    explanation: >-
      Quantifies the harm profile, including radiation-induced change, that has to be set
      against the obliteration benefit.

- name: Conservative medical management
  description: >-
    Observation with symptomatic treatment - antiepileptic drugs for seizures, analgesia -
    without attempting to eradicate the lesion. For unruptured lesions this is a genuinely
    competitive strategy rather than a default: the ARUBA randomised trial was halted early
    for superiority of medical management, with the composite of death or symptomatic
    stroke reached by 10.1% on medical management versus 30.7% with intervention. The
    advantage did not decay with longer observation: at a mean 50.4 months the incidence of
    death or symptomatic stroke remained lower with medical management alone (3.39 vs 12.32
    per 100 patient-years; hazard ratio 0.31, 95% CI 0.17 to 0.56), with fewer adverse events.
    What remains contested is whether that generalises beyond the trial's enrolled population -
    see the `bavm-unruptured-management-controversy` discussion.
  treatment_term:
    preferred_term: Supportive Care
    term:
      id: NCIT:C15747
      label: Supportive Care
  therapeutic_modality: OTHER
  target_mechanisms:
  - target: Perinidal Cortical Irritation and Seizure Generation
    treatment_effect: INHIBITS
    description: >-
      Antiepileptic drugs suppress the seizures generated by perinidal cortex without altering
      the lesion itself. Modality is OTHER rather than BEHAVIORAL because the content is
      observation plus antiepileptic pharmacotherapy, not a behavioural, dietary or lifestyle
      intervention.
  evidence:
  - reference: PMID:24268105
    reference_title: >-
      Medical management with or without interventional therapy for unruptured brain
      arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      The ARUBA trial showed that medical management alone is superior to medical management
      with interventional therapy for the prevention of death or stroke in patients with
      unruptured brain arteriovenous malformations followed up for 33 months.
    explanation: >-
      The randomised trial's conclusion supporting conservative management for unruptured
      lesions, with its follow-up duration stated.
  - reference: PMID:24268105
    reference_title: >-
      Medical management with or without interventional therapy for unruptured brain
      arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      The primary endpoint had been reached by 11 (10.1%) patients in the medical management
      group compared with 35 (30.7%) in the interventional therapy group.
    explanation: >-
      Quantifies the difference in death or symptomatic stroke that led the data and safety
      monitoring board to halt randomisation.
  - reference: PMID:32562682
    reference_title: >-
      Medical management with interventional therapy versus medical management alone for
      unruptured brain arteriovenous malformations (ARUBA): final follow-up of a multicentre,
      non-blinded, randomised controlled trial.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      After extended follow-up, ARUBA showed that medical management alone remained superior to
      interventional therapy for the prevention of death or symptomatic stroke in patients with
      an unruptured brain arteriovenous malformation.
    explanation: >-
      The final follow-up analysis, which supersedes the 2014 interim horizon and shows the
      effect did not decay with longer observation.

- name: Endovascular embolization
  description: >-
    Catheter-delivered occlusion of the nidus or its feeding arteries, usually with a liquid
    embolic (EVOH/Onyx or n-BCA). In practice it is most often adjunctive - devascularising a
    lesion before resection or radiosurgery - rather than curative on its own, and it was one
    of the three interventional modalities randomised in ARUBA. Notably, in a 1,770-patient
    registry emulation only microsurgical resection carried a consistent haemorrhage benefit
    in sensitivity analyses, so embolization should not be assumed to inherit resection's
    result.
  treatment_term:
    preferred_term: Embolization Therapy
    term:
      id: NCIT:C15230
      label: Embolization Therapy
  therapeutic_modality: DEVICE
  target_mechanisms:
  - target: Nidus Formation with Direct Arteriovenous Shunting
    treatment_effect: INHIBITS
    description: >-
      Embolic occlusion of nidus compartments and feeding arteries removes shunting channels,
      either as definitive treatment or to reduce flow before resection or radiosurgery.
  evidence:
  - reference: PMID:24268105
    reference_title: >-
      Medical management with or without interventional therapy for unruptured brain
      arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      medical management with interventional therapy (ie, neurosurgery, embolisation, or
      stereotactic radiotherapy, alone or in combination)
    explanation: >-
      Establishes embolization as one of the three standard interventional modalities, as
      randomised in ARUBA.
  - reference: PMID:41231469
    reference_title: >-
      Interventional Treatment vs Conservative Management of Unruptured Brain Arteriovenous
      Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Only microsurgical resection was consistently associated with reduced hemorrhage risk in
      sensitivity analyses.
    explanation: >-
      Recorded as PARTIAL because it bounds rather than supports this modality: among
      interventional treatments only resection showed a consistent benefit, so embolization
      cannot be assumed to share it.

clinical_trials:
- name: NCT00389181
  phase: PHASE_III
  status: COMPLETED
  description: >-
    ARUBA - A Randomised trial of Unruptured Brain Arteriovenous malformations. Compared
    medical management alone against medical management plus interventional therapy
    (neurosurgery, embolisation or stereotactic radiotherapy) in adults with unruptured
    brain AVM, across 39 sites in nine countries. Randomisation was halted early in 2013 for
    superiority of medical management. It remains the only completed randomised trial in
    this disease and the anchor of the management controversy recorded in the
    `bavm-unruptured-management-controversy` discussion.
  target_phenotypes:
  - preferred_term: Stroke
    term:
      id: HP:0001297
      label: Stroke
  - preferred_term: Intracranial hemorrhage
    term:
      id: HP:0002170
      label: Intracranial hemorrhage
  evidence:
  - reference: PMID:24268105
    reference_title: >-
      Medical management with or without interventional therapy for unruptured brain
      arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      This trial is registered with ClinicalTrials.gov, number NCT00389181.
    explanation: >-
      Ties the published trial report to its ClinicalTrials.gov registration, which is the
      identifier this record is keyed on.

- name: NCT02098252
  phase: NOT_APPLICABLE
  status: RECRUITING
  description: >-
    TOBAS (Treatment of Brain AVMs Study), a randomised controlled trial plus registry designed
    to address the question ARUBA left open, and structured to be less vulnerable to the
    generalisability criticism levelled at it: it enrols ruptured as well as unruptured
    lesions, and pairs randomisation with a registry for patients who cannot be randomised.
    Directly relevant to the `bavm-unruptured-management-controversy` discussion.
  target_phenotypes:
  - preferred_term: Stroke
    term:
      id: HP:0001297
      label: Stroke
  - preferred_term: Intracranial hemorrhage
    term:
      id: HP:0002170
      label: Intracranial hemorrhage
  evidence:
  - reference: clinicaltrials:NCT02098252
    reference_title: "Treatment of Brain AVMs (TOBAS) Study: A Randomized Controlled Trial and Registry"
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      The trial has been designed to test a) whether medical management or interventional
      therapy will reduce the risk of death or debilitating stroke
    explanation: >-
      States the trial's primary question, which is the unresolved comparison this entry's
      management-controversy discussion records.


animal_models:
- name: Endothelial BrafV600E mouse model of sporadic brain AVM
  species: Mouse
  genotype: Brain endothelial-cell-specific BrafV600E expression
  publication: PMID:38700584
  description: >-
    A mouse model in which BrafV600E is expressed in brain endothelial cells, developed
    specifically because earlier models produced heterogeneous and atypical lesions. It
    reproduces the human lesion closely, across ages and brain regions, with dilated vessels,
    reduced vascular wall stability, spontaneous haemorrhage and neuroinflammation.
  modeled_mechanisms:
  - target: Somatic Activating KRAS or BRAF Mutation in Brain Endothelium
    relationship: RECAPITULATES
    fidelity: HIGH
    description: >-
      Endothelial-restricted expression of a BRAF activating mutation is the model's design and
      matches the human somatic mosaic lesion in cell type and gene.
    limitations: >-
      The mutation is engineered and lineage-wide within the targeted endothelium rather than
      arising as a spontaneous focal clone, so the model does not reproduce the mosaic
      clone-boundary architecture of human disease.
    readouts:
    - name: Formation of AVM lesions resembling human bAVM
      target: Somatic Activating KRAS or BRAF Mutation in Brain Endothelium
      direction: INCREASED
      interpretation: >-
        Lesion formation on introducing the mutation is the causal demonstration that the
        endothelial mutation is sufficient.
      evidence:
      - reference: PMID:38700584
        reference_title: "Somatic Braf(V600E) mutation in the cerebral endothelium induces brain arteriovenous malformations."
        supports: SUPPORT
        evidence_source: MODEL_ORGANISM
        snippet: >-
          Mice with BrafV600E mutations in brain ECs developed BAVM closely resembled that of
          human lesions.
        explanation: >-
          Reports lesion formation resembling human bAVM after endothelial BrafV600E expression.
    evidence:
    - reference: PMID:38700584
      reference_title: "Somatic Braf(V600E) mutation in the cerebral endothelium induces brain arteriovenous malformations."
      supports: SUPPORT
      evidence_source: MODEL_ORGANISM
      snippet: >-
        Here we developed a novel mouse model of sporadic BAVM that is consistent with clinical
        manifestations in humans.
      explanation: >-
        Supports treating this model as informative for the somatic-mutation node.
  - target: Deficient Mural Cell Coverage and Vessel Wall Instability
    relationship: RECAPITULATES
    fidelity: MODERATE
    description: >-
      The model's lesions show reduced vascular wall stability with spontaneous haemorrhage,
      the phenotype this node describes.
    limitations: >-
      Wall instability is reported as a pathological description rather than a quantified mural
      cell coverage measurement, so the model supports the phenotype more strongly than it
      supports the specific mural-cell mechanism.
    readouts:
    - name: Vascular wall stability and spontaneous haemorrhage
      target: Deficient Mural Cell Coverage and Vessel Wall Instability
      direction: DECREASED
      interpretation: >-
        Reduced wall stability with spontaneous bleeding is the in vivo correlate of the
        structural instability node.
      evidence:
      - reference: PMID:38700584
        reference_title: "Somatic Braf(V600E) mutation in the cerebral endothelium induces brain arteriovenous malformations."
        supports: SUPPORT
        evidence_source: MODEL_ORGANISM
        snippet: >-
          Pathological features of BAVM were primarily dilated blood vessels with reduced
          vascular wall stability, accompanied by spontaneous hemorrhage and neuroinflammation.
        explanation: >-
          Reports reduced vascular wall stability with spontaneous haemorrhage in the model.
    evidence:
    - reference: PMID:38700584
      reference_title: "Somatic Braf(V600E) mutation in the cerebral endothelium induces brain arteriovenous malformations."
      supports: SUPPORT
      evidence_source: MODEL_ORGANISM
      snippet: >-
        Pathological features of BAVM were primarily dilated blood vessels with reduced vascular
        wall stability, accompanied by spontaneous hemorrhage and neuroinflammation.
      explanation: >-
        Supports treating this model as informative for vessel wall instability.
  - target: MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability
    relationship: PARTIALLY_RECAPITULATES
    fidelity: MODERATE
    description: >-
      Dabrafenib slowed lesion progression when given early, but its efficacy against
      established lesions was explicitly left uncertain - which is precisely the
      prevention-versus-reversal distinction the `bavm-mek-inhibition-translation` discussion
      is about, now carried as a structured model claim rather than only as prose.
    limitations: >-
      The therapeutic result is prevention of progression in a model where treatment can begin
      at lesion induction. Human patients present with lesions established for years, and the
      source states efficacy against established lesions remained uncertain.
    readouts:
    - name: Lesion progression under early dabrafenib
      target: MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability
      direction: DECREASED
      interpretation: >-
        Slowed progression on early BRAF inhibition supports the pathway as a target while
        leaving established-lesion reversal untested.
      evidence:
      - reference: PMID:38700584
        reference_title: "Somatic Braf(V600E) mutation in the cerebral endothelium induces brain arteriovenous malformations."
        supports: SUPPORT
        evidence_source: MODEL_ORGANISM
        snippet: >-
          Early administration of Dabrafenib was found to be effective in slowing the progression
          of BAVMs; however, its efficacy in treating established BAVM lesions remained uncertain.
        explanation: >-
          Reports the prevention benefit and states explicitly that established-lesion efficacy
          is unresolved. Recorded as PARTIAL for that reason.
    evidence:
    - reference: PMID:38700584
      reference_title: "Somatic Braf(V600E) mutation in the cerebral endothelium induces brain arteriovenous malformations."
      supports: SUPPORT
      evidence_source: MODEL_ORGANISM
      snippet: >-
        Early administration of Dabrafenib was found to be effective in slowing the progression
        of BAVMs; however, its efficacy in treating established BAVM lesions remained uncertain.
      explanation: >-
        The model informs the drug-target node while bounding what it can show, which is the
        substance of the model-mismatch discussion.

discussions:
- discussion_id: bavm-unruptured-management-controversy
  kind: KNOWLEDGE_GAP
  status: OPEN
  attaches_to:
  - pathophysiology#Nidus Rupture and Intracranial Haemorrhage
  - pathophysiology#Nidus Formation with Direct Arteriovenous Shunting
  prompt: >-
    Should an unruptured brain AVM be treated or observed? The only randomised trial,
    ARUBA, was stopped early for superiority of medical management, with death or
    symptomatic stroke in 10.1% on medical management versus 30.7% with intervention over
    33 months. A 1,770-patient target-trial-emulation study of a national registry
    subsequently found the opposite over a longer horizon: 5-year haemorrhage-free survival
    of 96.2% with intervention versus 89.0% with conservative management. The two results
    are not obviously reconcilable, and the disagreement is not settled.
  rationale: >-
    This is a mechanistically informative disagreement rather than a mere clinical dispute.
    The competing results are consistent with a lesion whose hazard is roughly constant and
    lifelong while the harm of intervention is front-loaded: a short-horizon trial then
    favours observation and a long-horizon analysis favours treatment, without either being
    wrong about what it measured. The registry analysis also found no benefit in high-grade
    or diffuse-nidus lesions, and only microsurgical resection carried a consistent
    benefit, which suggests the answer is conditional on lesion architecture rather than
    uniform. Curating a single management recommendation here would misrepresent the
    evidence, so the entry records both treatments with their own evidence and leaves the
    comparison open.
  evidence:
  - reference: PMID:24268105
    reference_title: >-
      Medical management with or without interventional therapy for unruptured brain
      arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      The risk of death or stroke was significantly lower in the medical management group
      than in the interventional therapy group (hazard ratio 0.27, 95% CI 0.14-0.54).
    explanation: >-
      The randomised evidence favouring conservative management over the trial's follow-up
      period.
  - reference: PMID:41231469
    reference_title: >-
      Interventional Treatment vs Conservative Management of Unruptured Brain Arteriovenous
      Malformations.
    supports: REFUTE
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      the estimated 5-year hemorrhage-free survival was 96.23% (95% CI, 93.95%-97.65%) for
      the interventional treatment group and 89.00% (95% CI, 86.37%-91.24%) for the
      conservative management group
    explanation: >-
      Directly opposes the ARUBA conclusion over a longer horizon, which is why this is
      recorded as REFUTE against the conservative-management position rather than as
      concordant evidence.
  - reference: PMID:41231469
    reference_title: >-
      Interventional Treatment vs Conservative Management of Unruptured Brain Arteriovenous
      Malformations.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    snippet: >-
      Subgroup analyses indicated that the benefit of the intervention was not observed in
      patients with high-grade AVMs or diffuse nidus, for whom outcomes were similar to
      those of patients with conservative management.
    explanation: >-
      Shows the answer is conditional on lesion architecture, which is the substantive
      content of the gap rather than a simple trial-versus-registry disagreement.
  proposed_experiments:
  - experiment_id: bavm-long-horizon-randomisation
    name: Long-horizon randomised comparison stratified by nidus architecture
    description: >-
      A randomised comparison of intervention versus observation in unruptured brain AVM
      with follow-up of at least ten years, prospectively stratified by Spetzler-Martin
      grade, nidus diffuseness and venous drainage pattern, powered to detect a crossing of
      the survival curves rather than only an early difference.
    readouts:
    - name: Haemorrhage-free survival by lesion architecture stratum at 10 years
      target: pathophysiology#Nidus Rupture and Intracranial Haemorrhage
    would_support:
    - pathophysiology#Nidus Rupture and Intracranial Haemorrhage

- discussion_id: bavm-mek-inhibition-translation
  kind: HUMAN_MODEL_MISMATCH
  status: OPEN
  attaches_to:
  - pathophysiology#MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability
  - pathophysiology#Constitutive RAS-MAPK-ERK Signalling in Brain Endothelial Cells
  prompt: >-
    MEK and BRAF inhibition reverses the mutant endothelial phenotype in culture, suppresses
    arteriovenous shunting in zebrafish, and acts on patient-derived AVM endothelial cells,
    but no MEK inhibitor is established therapy for brain AVM. Does pathway inhibition
    modify an established human nidus, as opposed to preventing lesion formation or
    reversing it in a model organism?
  rationale: >-
    The distinction matters because the model systems test a different question from the
    clinical one. Animal and zebrafish models express mutant KRAS and then observe lesion
    formation, so they largely test prevention; a patient presents with a nidus that has
    existed for years and is structurally remodelled, and it is not established that
    switching off the driving signal reverses established architecture. Two further
    obstacles are specific to this disease: variant allele frequencies are low and lesional
    tissue is hard to obtain, so patients cannot easily be genotyped before treatment, and
    the natural-history haemorrhage rate of roughly 1.3% per year in unruptured lesions
    means any trial of a drug with systemic toxicity needs a long horizon and a large
    cohort to show benefit.
  evidence:
  - reference: PMID:41708990
    reference_title: >-
      KRAS-dependent glycolytic reprogramming of endothelial cells in sporadic arteriovenous
      malformations.
    supports: SUPPORT
    evidence_source: MODEL_ORGANISM
    snippet: >-
      Glycolysis inhibition also reversed arteriovenous shunts and potentiated the effect of
      MEK inhibition in a KRAS-mutant zebrafish model.
    explanation: >-
      The shunt-reversal result on which the therapeutic hope rests is a zebrafish result -
      the model-to-human gap this discussion is about. Evidence source is MODEL_ORGANISM.
  - reference: PMID:41999461
    reference_title: Advances in the Genetics and Molecular Biology of Brain Arteriovenous Malformations.
    supports: SUPPORT
    evidence_source: OTHER
    snippet: >-
      However, clinical translation remains challenging because of low variant allele
      frequencies and limited access to lesional tissue for genetic testing.
    explanation: >-
      States the specific practical barriers to translating the target into genotype-guided
      human therapy. Evidence source is OTHER as this is a review article.
  proposed_experiments:
  - experiment_id: bavm-mek-established-lesion-trial
    name: MEK inhibition in established, genotyped brain AVM
    description: >-
      An early-phase trial of MEK inhibition in patients with an angiographically stable
      brain AVM genotyped by endovascular or cell-free DNA sampling, with nidus volume and
      shunt flow on serial angiography as the primary readout, to test whether pathway
      inhibition remodels an established lesion rather than only preventing lesion formation.
    readouts:
    - name: Change in nidus volume and shunt flow on serial angiography
      target: pathophysiology#MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability
    would_support:
    - pathophysiology#MAPK-ERK Pathway Inhibition as Therapeutic Vulnerability

notes: >-
  Scope: this entry covers the sporadic, somatic-mosaic brain AVM keyed to MONDO:0007154.
  The germline syndromic causes of brain AVM are curated separately
  (`Hereditary_Hemorrhagic_Telangiectasia`,
  `Capillary_Malformation-Arteriovenous_Malformation_Syndrome`), as is the distinct entity
  `Cerebral_Proliferative_Angiopathy`. MONDO carries this term under `hereditary disease`,
  which is misleading for the sporadic somatic disease the term actually denotes; the
  mutations are post-zygotic and absent from blood, and the entry is curated accordingly.

  Module conformance: none declared. The obvious candidate,
  `sustaining_proliferative_signaling`, is explicitly scoped as the first hallmark of
  cancer, running from an "Oncogenic Growth-Signal Lesion" to
  "Growth-Factor-Independent Proliferation". Brain AVM shares the constitutive RAS-MAPK
  node but is not a neoplasm: there is no uncontrolled proliferative expansion and no tumour,
  and the pathological output is a malformed vascular architecture. Declaring conformance
  would import a neoplastic frame the disease does not have. If a
  non-neoplastic RASopathy or mosaic-vascular-malformation module is created later, this
  entry is a natural conformer and the RAS-MAPK node is the attachment point.

  Reference scoping. The `DOI_*.md` files committed with this entry were written by the deep
  research run's own reference-validation pass, not deliberately fetched by a curator - `just
  research-disorder` resolves the report's citations as a side effect - so an uncited one
  reflects what the report cited, not a curatorial judgement of relevance. Five of them are
  metadata stubs with `content_type: unavailable` and zero body text, including the ARUBA final
  follow-up, so they cannot be quoted at all. Where such a stub carried content worth having,
  the DOI was resolved to its PMID and the real abstract fetched and cited instead: ARUBA final
  follow-up (PMID:32562682), mural cell review (PMID:34541474), human bAVM RNA-seq
  (PMID:31795902), non-contrast 4D-MRA accuracy (PMID:39125532), and the endothelial BrafV600E
  mouse (PMID:38700584). Deliberately not consumed: the single-cell atlases beyond
  PMID:35084939 (they report findings shared across several vascular-dependent CNS diseases
  rather than bAVM-specific ones), the endothelial methylome, the CasRx knockdown study, and the
  VALE risk score - the last because a prognostic score is not a mechanism and this entry
  already carries natural-history risk from a four-cohort individual-patient meta-analysis.

  Modifier genes deliberately omitted. Candidate haemorrhage-risk modifiers (IL6 -174G>C,
  TNF-alpha -238G>A, APOE e2) are surfaced by the deep-research report, but the cache file
  for the source it cites is a metadata stub with no body text, and the PMID that DOI
  resolves to is a natural-history review whose abstract does not mention modifiers at all.
  There is no quotable evidence, so they are not curated. Recorded here so the omission is
  auditable rather than reading as under-consumption: this is an evidence gap, not a scoping
  decision, and the genes should be added if a source with a quotable abstract is found.

  Frequency discipline: no phenotype carries a `frequency:` band. Haemorrhage is an incident
  event with an annual rate rather than a cross-sectional proportion; the 80.4% seizure
  figure comes from a surgical series restricted to unruptured lesions and is not a
  disease-wide frequency. Both numbers are quoted in the relevant descriptions with their
  cohort context instead.

📚

References & Deep Research

Deep Research

1

Falcon ▸

Disease Characteristics Research Template

Edison Scientific Literature 33 citations 2026-08-19T23:06:12.762912

Question: You are an expert researcher providing comprehensive, well-cited information.

Provide detailed information focusing on: 1. Key concepts and definitions with current understanding 2. Recent developments and latest research (prioritize 2023-2024 sources) 3. Current applications and real-world implementations 4. Expert opinions and analysis from authoritative sources 5. Relevant statistics and data from recent studies

Format as a comprehensive research report with proper citations. Include URLs and publication dates where available. Always prioritize recent, authoritative sources and provide specific citations for all major claims.

Disease Characteristics Research Template

Target Disease

Disease Name: Brain arteriovenous malformation
MONDO ID: (if available)
Category: Vascular Disease

Research Objectives

Please provide a comprehensive research report on Brain arteriovenous malformation covering all of the disease characteristics listed below. This report will be used to populate a disease knowledge base entry. Be thorough and cite primary literature (PMID preferred) for all claims.

For each section, suggested databases/resources are listed. These are the first places you should search for information on each topic.

1. Disease Information

Search first: OMIM, Orphanet, ICD-10/ICD-11, MeSH, PubMed

What is the disease? Provide a concise overview.
What are the key identifiers? (OMIM, Orphanet, ICD-10/ICD-11, MeSH, Mondo)
What are the common synonyms and alternative names?
Is the information derived from individual patients (e.g., EHR) or aggregated disease-level resources?

2. Etiology

Disease Causal Factors: What are the primary causes? (genetic, environmental, infectious, mechanistic)
Risk Factors:

Search first: PubMed, Cochrane Library, UpToDate, clinical guidelines, ClinVar, ClinGen, GWAS Catalog, PheGenI, CTD, CDC, WHO, epidemiological databases
Genetic risk factors (causal variants, susceptibility loci, modifier genes)
Environmental risk factors (toxins, lifestyle, occupational exposures, age, sex, family history)
Protective Factors:

Search first: PubMed, Cochrane Library, clinical trial databases, GWAS Catalog, gnomAD, WHO, CDC, nutrition databases
Genetic protective factors (protective variants, modifier alleles)
Environmental protective factors (diet, lifestyle, exposures that reduce risk)
Gene-Environment Interactions: How do genetic and environmental factors interact to influence disease?

Search first: CTD, PubMed, PheGenI, GxE databases

3. Phenotypes

Search first: HPO (Human Phenotype Ontology), OMIM, Orphanet, PubMed, clinicaltrials.gov, MedDRA, SNOMED CT, DECIPHER, LOINC

For each phenotype, provide: - Phenotype type: symptoms, clinical signs, physical manifestations, behavioral changes, or laboratory abnormalities

For symptoms/signs: HPO, OMIM, Orphanet, PubMed For behavioral changes: HPO, DSM, RDoC (Research Domain Criteria), PubMed For laboratory abnormalities: LOINC, SNOMED CT, LabTests Online, PubMed - Phenotype characteristics: Search first: OMIM, Orphanet, HPO, PubMed - Age of symptom onset (neonatal, childhood, adult-onset, late-onset) - Symptom severity (mild, moderate, severe, variable) - Symptom progression (stable, progressive, episodic, fluctuating) - Frequency among affected individuals (percentage or qualitative) - Quality of life impact: Effects on daily functioning and well-being (per-phenotype when possible) Search first: EQ-5D database, SF-36, WHO QOL databases, PubMed - Suggest HPO (Human Phenotype Ontology) terms for each phenotype

4. Genetic/Molecular Information

Causal Genes: Gene mutations or chromosomal abnormalities responsible for disease (gene symbols, OMIM IDs)

Search first: OMIM, ClinVar, HGMD, Ensembl, NCBI Gene
Pathogenic Variants:
Affected genes (gene symbols, HGNC IDs) > Search first: OMIM, NCBI Gene, Ensembl, HGNC, UniProt, GeneCards
Variant classification (pathogenic, likely pathogenic, VUS per ACMG/AMP guidelines) > Search first: ClinVar, ClinGen, ACMG/AMP guidelines, VarSome
Variant type/class (missense, frameshift, nonsense, splice-site, structural)
Allele frequency in population databases > Search first: gnomAD, 1000 Genomes, ExAC, TOPMed, dbSNP
Somatic vs germline origin > Search first: COSMIC (somatic), ClinVar, ICGC, TCGA
Functional consequences (loss of function, gain of function, dominant negative)
Modifier Genes: Genes that modify disease severity or expression
Epigenetic Information: DNA methylation, histone modifications, chromatin changes affecting disease

Search first: ENCODE, Roadmap Epigenomics, MethBase, DiseaseMeth
Chromosomal Abnormalities: Large-scale genetic changes (aneuploidy, translocations, inversions)

Search first: DECIPHER, ClinVar, ECARUCA, UCSC Genome Browser

5. Environmental Information

Environmental Factors: Non-genetic contributing factors (toxins, radiation, pollution, occupational exposure)

Search first: CTD (Comparative Toxicogenomics Database), TOXNET, PubMed, EPA databases
Lifestyle Factors: Behavioral factors (smoking, diet, exercise, alcohol consumption)

Search first: CDC databases, WHO, PubMed, NHANES
Infectious Agents: If applicable, pathogens causing or triggering disease (bacteria, viruses, fungi, parasites)

Search first: NCBI Taxonomy, ViPR, BV-BRC, MicrobeDB, GIDEON

6. Mechanism / Pathophysiology

Molecular Pathways: Specific signaling cascades or biochemical pathways involved (Wnt, MAPK, mTOR, PI3K-AKT, etc.)

Search first: KEGG, Reactome, WikiPathways, PathBank, BioCyc
Cellular Processes: Cell-level mechanisms (apoptosis, autophagy, cell cycle dysregulation, inflammation, etc.)

Search first: Gene Ontology (GO), Reactome, KEGG, PubMed
Protein Dysfunction: How protein structure or function is altered (misfolding, aggregation, loss of function, gain of function)

Search first: UniProt, PDB (Protein Data Bank), InterPro, Pfam, AlphaFold
Metabolic Changes: Alterations in metabolic processes (energy metabolism, lipid metabolism, amino acid metabolism)

Search first: KEGG, BioCyc, HMDB (Human Metabolome Database), BRENDA
Immune System Involvement: Role of immune response (autoimmunity, immunodeficiency, chronic inflammation)

Search first: ImmPort, Immunome Database, IEDB, Gene Ontology
Tissue Damage Mechanisms: How tissues/ are injured (oxidative stress, ischemia, fibrosis, necrosis)

Search first: PubMed, Gene Ontology, Reactome
Biochemical Abnormalities: Specific molecular defects (enzyme deficiencies, receptor dysfunction, ion channel defects)

Search first: BRENDA, UniProt, KEGG, OMIM, PubMed
Epigenetic Changes: DNA methylation, histone modifications affecting gene expression in disease

Search first: ENCODE, Roadmap Epigenomics, MethBase, DiseaseMeth
Molecular Profiling (if available):
Transcriptomics/gene expression changes > Search first: GEO (Gene Expression Omnibus), ArrayExpress, GTEx, Human Cell Atlas, SRA
Proteomics findings > Search first: PRIDE, ProteomeXchange, Human Protein Atlas, STRING, BioGRID
Metabolomics signatures > Search first: MetaboLights, Metabolomics Workbench, HMDB, METLIN
Lipidomics alterations > Search first: LIPID MAPS, SwissLipids, LipidHome, Metabolomics Workbench
Genomic structural features > Search first: UCSC Genome Browser, Ensembl, NCBI, dbVar, DGV
Advanced Technologies (if applicable):
Single-cell analysis findings (cell-type specific mechanisms, cellular heterogeneity) > Search first: Human Cell Atlas, Single Cell Portal, GEO, CELLxGENE
Spatial transcriptomics findings > Search first: GEO, Spatial Research, Vizgen, 10x Genomics data
Multi-omics integration results > Search first: TCGA, ICGC, cBioPortal, LinkedOmics, PubMed
Functional genomics screens (CRISPR, RNAi) > Search first: DepMap, GenomeRNAi, PubMed, BioGRID ORCS

For each mechanism, describe: - The causal chain from initial trigger to clinical manifestation - Which mechanisms are upstream vs downstream - What cell types and biological processes are involved - Suggest GO terms for biological processes and CL terms for cell types

7. Anatomical Structures Affected

Organ Level:
Primary organs directly affected
Secondary organ involvement (complications, secondary effects)
Body systems involved (cardiovascular, nervous, digestive, respiratory, endocrine, etc.)

Search first: Uberon, FMA (Foundational Model of Anatomy), OMIM, HPO, ICD-11, MeSH, SNOMED CT
Tissue and Cell Level:
Specific tissue types affected (epithelial, connective, muscle, nervous)
Specific cell populations targeted (with Cell Ontology terms)

Search first: Uberon, Human Protein Atlas, Cell Ontology, Human Cell Atlas, CellMarker, PanglaoDB
Subcellular Level:
Cellular compartments involved (mitochondria, nucleus, ER, lysosomes) (with GO Cellular Component terms)

Search first: Gene Ontology (Cellular Component), UniProt, Human Protein Atlas
Localization:
Specific anatomical sites (with UBERON terms) > Search first: FMA, Uberon, NeuroNames (for brain), SNOMED CT
Lateralization (unilateral, bilateral, asymmetric) > Search first: HPO, clinical literature, imaging databases

8. Temporal Development

Onset:
Typical age of onset (congenital, pediatric, adult, geriatric)
Onset pattern (acute, subacute, chronic, insidious)

Search first: OMIM, Orphanet, HPO, PubMed
Progression:
Disease stages (early, intermediate, advanced, end-stage) > Search first: Cancer Staging Manual (AJCC), WHO classifications, PubMed
Progression rate (rapid, slow, variable)
Disease course pattern (episodic, relapsing-remitting, progressive, stable)
Disease duration (self-limited, chronic lifelong)

Search first: Disease registries, longitudinal cohort databases, natural history studies, PubMed, Orphanet, OMIM
Patterns:
Remission patterns (spontaneous, treatment-induced) > Search first: Clinical trial databases, disease registries, PubMed
Critical periods (time windows of vulnerability or opportunity for intervention) > Search first: PubMed, developmental biology databases, clinical guidelines

9. Inheritance and Population

Epidemiology:
Prevalence (cases per 100,000 at given time)
Incidence (new cases per 100,000 per year)

Search first: Orphanet, CDC, WHO, GBD (Global Burden of Disease), national registries, SEER, disease registries
For Genetic Etiology:
Inheritance pattern (AD, AR, X-linked, mitochondrial, multifactorial, polygenic) > Search first: OMIM, Orphanet, ClinVar, GTR (Genetic Testing Registry)
Penetrance (complete, incomplete, age-dependent) > Search first: ClinVar, OMIM, PubMed, ClinGen
Expressivity (variable, consistent) > Search first: OMIM, ClinVar, PubMed
Genetic anticipation (increasing severity in successive generations) > Search first: OMIM, PubMed (especially for repeat expansion disorders)
Germline mosaicism > Search first: ClinVar, OMIM, genetic counseling literature, PubMed
Founder effects (population-specific mutations) > Search first: gnomAD, population genetics databases, PubMed
Consanguinity role > Search first: OMIM, population studies, genetic counseling resources
Carrier frequency > Search first: gnomAD, carrier screening databases, GeneReviews, GTR
Population Demographics:
Affected populations (ethnic or demographic groups with higher prevalence) > Search first: gnomAD, 1000 Genomes, PAGE Study, PubMed, population registries
Geographic distribution (endemic areas, regional variation) > Search first: WHO, CDC, GBD, Orphanet, geographic epidemiology databases
Geographic distribution of specific variants
Sex ratio (male:female) > Search first: Disease registries, OMIM, PubMed, epidemiological databases
Age distribution of affected individuals > Search first: CDC, disease registries, SEER, Orphanet

10. Diagnostics

Clinical Tests:
Laboratory tests (blood, urine, tissue chemistry, specific enzyme assays) > Search first: LOINC, LabTests Online, PubMed
Biomarkers (proteins, metabolites, genetic markers, circulating biomarkers) > Search first: FDA Biomarker List, BEST (Biomarkers, EndpointS, and other Tools), PubMed
Imaging studies (X-ray, CT, MRI, PET, ultrasound) > Search first: RadLex, DICOM, Radiopaedia, imaging databases
Functional tests (pulmonary function, cardiac stress tests) > Search first: LOINC, clinical guidelines, PubMed
Electrophysiology (EEG, EMG, ECG, nerve conduction studies) > Search first: LOINC, clinical neurophysiology databases, PubMed
Biopsy findings (histopathology, immunohistochemistry) > Search first: SNOMED CT, College of American Pathologists resources, PubMed
Pathology findings (microscopic examination) > Search first: SNOMED CT, Digital Pathology databases, PubMed
Genetic Testing:

Search first: GTR (Genetic Testing Registry), GeneReviews, ClinGen
Overview of recommended genetic testing approach
Whole genome sequencing (WGS) utility > Search first: GTR, ClinVar, GEL (Genomics England), gnomAD
Whole exome sequencing (WES) utility > Search first: GTR, ClinVar, OMIM, GeneMatcher
Gene panels (which panels, which genes) > Search first: GTR, ClinVar, laboratory-specific databases
Single gene testing > Search first: GTR, ClinVar, OMIM, GeneReviews
Chromosomal microarray (CMA) > Search first: DECIPHER, ClinVar, dbVar, ECARUCA
Karyotyping > Search first: Chromosome Abnormality Database, ClinVar, cytogenetics resources
FISH > Search first: ClinVar, cytogenetics databases, PubMed
Mitochondrial DNA testing > Search first: MITOMAP, MSeqDR, ClinVar, GTR
Repeat expansion testing > Search first: GTR, ClinVar, repeat expansion databases, PubMed
Omics-Based Diagnostics (if applicable):
RNA sequencing / transcriptomics > Search first: GEO, ArrayExpress, GTEx, RNA-seq databases
Proteomics > Search first: PRIDE, ProteomeXchange, FDA Biomarker database
Metabolomics > Search first: MetaboLights, Metabolomics Workbench, HMDB
Epigenomics > Search first: GEO, ENCODE, Roadmap Epigenomics, MethBase
Liquid biopsy > Search first: COSMIC, ClinVar, liquid biopsy databases, PubMed
Clinical Criteria:
Standardized diagnostic criteria (DSM, ICD, society guidelines) > Search first: DSM-5, ICD-11, clinical society guidelines, UpToDate
Differential diagnosis (other conditions to rule out, with distinguishing features) > Search first: DynaMed, UpToDate, clinical decision support systems
Screening:
Screening methods for asymptomatic individuals (newborn screening, carrier screening, cascade screening) > Search first: ACMG recommendations, CDC newborn screening, GTR

11. Outcome/Prognosis

Survival and Mortality:
Survival rate (5-year, 10-year, overall) > Search first: SEER, cancer registries, disease-specific registries, PubMed
Life expectancy (with and without treatment if applicable) > Search first: Orphanet, disease registries, actuarial databases, PubMed
Mortality rate > Search first: CDC, WHO, GBD, national mortality databases
Disease-specific mortality (deaths directly attributable to disease) > Search first: Disease registries, CDC Wonder, GBD, PubMed
Morbidity and Function:
Morbidity (disease-related disability and health impacts) > Search first: GBD, WHO, disability databases, PubMed
Disability outcomes (long-term functional impairments) > Search first: ICF (International Classification of Functioning), disability registries
Quality of life measures (EQ-5D, SF-36, PROMIS, disease-specific tools) > Search first: EQ-5D database, SF-36, PROMIS, PubMed
Disease Course:
Complications (secondary problems: infections, organ failure, etc.) > Search first: ICD codes, disease registries, clinical databases, PubMed
Recovery potential (likelihood and extent of recovery, with vs without treatment) > Search first: Natural history studies, rehabilitation databases, PubMed
Prediction:
Prognostic factors (age, disease severity, biomarkers, treatment response) > Search first: Prognostic models databases, clinical calculators, PubMed
Prognostic biomarkers (molecular markers predicting disease course) > Search first: FDA Biomarker database, PubMed, cancer prognostic databases

12. Treatment

Pharmacotherapy:
Pharmacological treatments (drug names, drug classes, mechanisms of action) > Search first: DrugBank, RxNorm, ATC classification, DailyMed, FDA databases
Pharmacogenomics (how genetic variants affect drug metabolism, efficacy, toxicity) > Search first: PharmGKB, CPIC (Clinical Pharmacogenetics), FDA Table of PGx Biomarkers
Advanced Therapeutics:
Gene therapy (viral vectors, CRISPR, gene replacement, gene editing) > Search first: ClinicalTrials.gov, FDA gene therapy database, ASGCT resources
Cell therapy (stem cell transplant, CAR-T, cellular therapeutics) > Search first: ClinicalTrials.gov, FDA cell therapy database, FACT standards
RNA-based therapies (ASOs, siRNA, mRNA therapies) > Search first: ClinicalTrials.gov, FDA approvals, PubMed
Targeted therapies (treatments directed at specific molecular targets) > Search first: My Cancer Genome, OncoKB, ClinicalTrials.gov, FDA approvals
Immunotherapies (checkpoint inhibitors, monoclonal antibodies) > Search first: Cancer Immunotherapy Database, FDA approvals, ClinicalTrials.gov
Surgical and Interventional:
Surgical interventions (types of surgery, timing, outcomes) > Search first: CPT codes, surgical registries, clinical guidelines, PubMed
Supportive and Rehabilitative:
Supportive care (symptom management, pain control, nutrition) > Search first: Clinical guidelines, Cochrane Library, PubMed
Rehabilitation (physical therapy, occupational therapy, speech therapy) > Search first: Rehabilitation medicine databases, clinical guidelines, PubMed
Experimental:
Experimental treatments in clinical trials (with NCT identifiers if available) > Search first: ClinicalTrials.gov, EU Clinical Trials Register, WHO ICTRP
Treatment Outcomes:
Treatment response rates > Search first: Clinical trial databases, FDA reviews, systematic reviews, PubMed
Side effects and adverse events > Search first: FDA Adverse Event Reporting System (FAERS), MedWatch, PubMed
Treatment Strategy:
Treatment algorithms (clinical pathways, decision trees) > Search first: Clinical practice guidelines, NCCN Guidelines, UpToDate
Combination therapies > Search first: ClinicalTrials.gov, treatment guidelines, PubMed
Personalized medicine approaches (genotype-guided treatment) > Search first: My Cancer Genome, CIViC, PharmGKB, precision medicine databases

For each treatment, suggest NCIT (NCI Thesaurus) clinical-intervention terms where applicable.

13. Prevention

Prevention Levels:
Primary prevention (preventing disease occurrence: vaccination, risk factor modification) > Search first: CDC, WHO, USPSTF recommendations, Cochrane Library
Secondary prevention (early detection and treatment: screening programs, early intervention) > Search first: USPSTF, CDC screening guidelines, WHO
Tertiary prevention (preventing complications in those with disease) > Search first: Clinical guidelines, disease management protocols, PubMed
Immunization: Vaccine strategies (if applicable)

Search first: CDC vaccine schedules, WHO immunization, FDA vaccine database
Screening and Early Detection:
Screening programs (population-based: newborn screening, cancer screening) > Search first: CDC screening programs, USPSTF, cancer screening databases
Genetic screening (carrier screening, preimplantation genetic diagnosis, prenatal testing) > Search first: ACMG recommendations, ACOG guidelines, GTR
Risk stratification (identifying high-risk individuals for targeted prevention) > Search first: Risk prediction models, clinical calculators, PubMed
Behavioral Interventions: Lifestyle modifications to reduce risk

Search first: CDC, WHO, behavioral intervention databases, Cochrane Library
Counseling: Genetic counseling (risk assessment, family planning guidance)

Search first: NSGC resources, ACMG guidelines, GeneReviews
Public Health:
Public health interventions (sanitation, vector control, health education) > Search first: CDC, WHO, public health databases, PubMed
Environmental interventions (reducing environmental risk factors) > Search first: EPA databases, WHO environmental health, PubMed
Prophylaxis: Preventive medications or procedures

Search first: Clinical guidelines, FDA approvals, PubMed

14. Other Species / Natural Disease

Taxonomy: Species affected (with NCBI Taxon identifiers)

Search first: NCBI Taxonomy
Breed: Specific breeds affected (with VBO identifiers if applicable)

Search first: VBO (Vertebrate Breed Ontology)
Gene: Orthologous genes in other species (with NCBI Gene IDs)

Search first: NCBI Gene
Natural Disease:
Naturally occurring disease in other species (companion animals, wildlife) > Search first: OMIA (Online Mendelian Inheritance in Animals), VetCompass, PubMed
Veterinary relevance and importance in animal health > Search first: OMIA, veterinary databases, PubMed
Comparative Biology:
Comparative pathology (similarities and differences across species) > Search first: OMIA, comparative pathology databases, PubMed
Evolutionary conservation of disease mechanisms > Search first: HomoloGene, OrthoMCL, Alliance of Genome Resources
Transmission (if applicable):
Zoonotic potential > Search first: CDC zoonotic diseases, WHO zoonoses, GIDEON
Cross-species susceptibility > Search first: NCBI Taxonomy, veterinary databases, PubMed

15. Model Organisms

Model Types:
Model organism type (mammalian, invertebrate, cellular, in vitro) > Search first: Alliance of Genome Resources, model organism databases
Specific model systems (mouse, rat, zebrafish, Drosophila, C. elegans, yeast, cell lines, organoids, iPSCs) > Search first: MGI, RGD, ZFIN, FlyBase, WormBase, SGD, ATCC, Cellosaurus
Induced models (drug treatment, surgical intervention, environmental manipulation) > Search first: MGI, model organism databases, PubMed
Genetic Models:
Types available (knockout, knock-in, transgenic, conditional, humanized) > Search first: MGI, IMPC, KOMP, EuMMCR, IMSR
Model Characteristics:
Phenotype recapitulation (how well model reproduces human disease features) > Search first: Model organism databases, comparative studies, PubMed
Model limitations (aspects of human disease not captured) > Search first: Model organism databases, PubMed, review articles
Applications:
Research applications (what aspects of disease can be studied) > Search first: Model organism databases, PubMed
Resources:
Model databases > Search first: MGI, RGD, ZFIN, FlyBase, WormBase, IMSR, EMMA, MMRRC

Citation Requirements

Cite primary literature (PMID preferred) for all mechanistic and clinical claims
Prioritize recent reviews and landmark papers
Include direct quotes from abstracts where possible to support key statements
Distinguish evidence source types: human clinical, model organism, in vitro, computational

Output Format

Structure your response as a comprehensive narrative organized by the sections above. For each section, provide: - Factual content with specific details (numbers, percentages, gene names, variant nomenclature) - Ontology term suggestions (HPO, GO, CL, UBERON, CHEBI, NCIT, MONDO) where applicable - Evidence citations with PMIDs - Direct quotes from abstracts to support key claims - Clear indication when information is not available or not applicable for this disease

This report will be used to populate a disease knowledge base entry with: - Pathophysiology descriptions with causal chains - Gene/protein annotations (HGNC, GO terms) - Phenotype associations (HP terms) with frequencies - Cell type involvement (CL terms) - Anatomical locations (UBERON terms) - Chemical entities (CHEBI terms) - Treatment annotations (NCIT terms) - Evidence items with PMIDs and exact abstract quotes - Epidemiology, prognosis, diagnostic, and prevention information - Animal model descriptions with phenotype recapitulation details

Brain arteriovenous malformation: disease-characteristics report

Scope. This report summarizes disease-level evidence rather than individual EHR records. Evidence is labeled as human clinical, human tissue/omics, animal, or expert synthesis. PMID values are supplied where available in the retrieved evidence; DOI links are used otherwise. Most recent evidence was prioritized, but landmark older studies remain necessary for natural history and randomized treatment evidence.

Executive summary

Brain arteriovenous malformation (bAVM) is a rare, high-flow cerebrovascular lesion in which a vascular nidus connects arteries directly to veins without an intervening capillary bed. The shunt exposes veins to arterial pressure and creates risks of intracranial hemorrhage, seizures, focal deficits, headache, and death or disability. Modern evidence no longer supports a purely static congenital-lesion model: most bAVMs are sporadic, endothelial mosaic disorders commonly involving activating KRAS–RAF–MEK–ERK signaling, with ongoing angiogenesis, mural-cell instability, inflammation, and vascular remodeling. Familial disease is uncommon and is principally associated with hereditary hemorrhagic telangiectasia (HHT) or capillary-malformation–AVM syndromes. (prado2019recentadvancesin pages 1-3, chen2023developmentandvalidation pages 1-2, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

No medication is proven to eradicate sporadic bAVM. Current care combines observation and symptom treatment with microsurgical resection, endovascular embolization, stereotactic radiosurgery (SRS), or multimodal therapy. Selection should be individualized by a multidisciplinary cerebrovascular team because the delayed natural-history risk must be balanced against immediate procedural risk. ARUBA found better approximately four-year outcomes with medical management than intervention for trial-eligible unruptured AVMs, but longer-term and lesion-specific uncertainty remains; TOBAS is designed to address that uncertainty. (mohr2020medicalmanagementwith pages 1-2, NCT02098252 chunk 1)

Domain	Core fact	Suggested ontology terms/IDs	Evidence type
Definition/identifiers	Brain arteriovenous malformation (brain AVM, bAVM) is a high-flow vascular malformation with direct artery-to-vein shunting and no intervening capillary bed; MONDO and OMIM identifiers are available. (OpenTargets Search: brain arteriovenous malformation, chen2023developmentandvalidation pages 1-2, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)	MONDO:0007154; OMIM:108010; MeSH: Arteriovenous Malformations	Aggregated disease ontology + human clinical
Synonyms	Common names include brain AVM, cerebral arteriovenous malformation, intracranial arteriovenous malformation, bAVM. (prado2019recentadvancesin pages 1-3, NCT02098252 chunk 1)	Labels only	Aggregated disease-level resources + human clinical
Epidemiology	bAVM is uncommon; symptomatic discovery incidence is about 1.1 per 100,000 population, and prevalence estimates around 10 per 100,000 are cited in reviews. (NCT02098252 chunk 1, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)	Label only: rare vascular disease	Human clinical/epidemiology
Major phenotype	Intracranial hemorrhage is a major presentation and complication; natural-history rupture risk is commonly estimated around 1%–4% per year depending on cohort and lesion features. (rutledge2014hemorrhageratesand pages 4-6, chen2023developmentandvalidation pages 1-2, NCT02098252 chunk 1)	HPO: HP:0002170 intracranial hemorrhage	Human clinical
Major phenotype	Seizures are a common presenting manifestation. (prado2019recentadvancesin pages 1-3, NCT02098252 chunk 1)	HPO: HP:0001250 seizures	Human clinical
Major phenotype	Headache is a recognized presenting symptom. (NCT02098252 chunk 1)	HPO: HP:0002315 headache	Human clinical
Major phenotype	Focal neurologic deficit can occur at presentation or after hemorrhage/treatment. (prado2019recentadvancesin pages 1-3, NCT02098252 chunk 1)	HPO label: focal neurologic deficit	Human clinical
Anatomy	Primary affected structure is the brain vasculature, with a nidus connecting cerebral arteries and veins. (chen2023developmentandvalidation pages 1-2, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)	UBERON:0000955 brain; UBERON label: cerebral blood vessel	Human clinical + human tissue
Key cell type	Endothelial cells are the principal disease-driving cell type in both sporadic and familial forms. (saito2024crisprcasrxsuppresseskrasinduced pages 1-2, winkler2022asinglecellatlas pages 1-3, walchli2024singlecellatlasof pages 1-2)	CL:0000115 endothelial cell	Human tissue/omics + animal model
Key cell type	Pericytes participate in neurovascular-unit dysfunction and vessel instability. (winkler2022asinglecellatlas pages 1-3, walchli2024singlecellatlasof pages 1-2)	CL:0000669 pericyte	Human tissue/omics
Key cell type	Vascular smooth muscle cells/mural cells contribute to vessel maturation failure and hemorrhagic vulnerability. (pan2021theroleof pages 5-6, scimone2024methylomeanalysisof pages 1-2)	Cell label: vascular smooth muscle cell	Human tissue + review
Key cell type	Monocyte/macrophage infiltration is part of the inflammatory lesion microenvironment and may contribute to rupture biology. (winkler2022asinglecellatlas pages 1-3, hauer2020rnasequencinghighlightsinflammation pages 1-2)	Cell label: monocyte/macrophage	Human tissue/omics
Mechanism	Pathologic angiogenesis and loss of vascular quiescence are central features. (hauer2020rnasequencinghighlightsinflammation pages 1-2, winkler2022asinglecellatlas pages 1-3)	GO:0001525 angiogenesis	Human tissue/omics
Mechanism	Somatic RAS/MAPK activation is a key upstream driver in sporadic bAVM. (saito2024crisprcasrxsuppresseskrasinduced pages 1-2, tu2024somaticbrafv600emutation pages 1-6)	GO:0000165 MAPK cascade	Human tissue genetics + animal model
Mechanism	Inflammatory signaling and immune-cell cross-talk are enriched in lesional tissue. (winkler2022asinglecellatlas pages 1-3, hauer2020rnasequencinghighlightsinflammation pages 1-2)	GO:0006954 inflammatory response	Human tissue/omics
Mechanism	Endothelial migration/cytoskeletal remodeling is dysregulated in bAVM tissue. (hauer2020rnasequencinghighlightsinflammation pages 1-2)	GO label: endothelial cell migration	Human bulk RNA-seq
Mechanism	Abnormal arteriovenous specification/differentiation is a recurrent theme across transcriptomic and developmental studies. (walchli2024singlecellatlasof pages 1-2, scimone2024methylomeanalysisof pages 1-2)	GO label: arteriovenous specification	Human tissue/omics
Somatic genes	Sporadic bAVMs frequently harbor activating somatic mutations in KRAS; BRAF is also implicated. (saito2024crisprcasrxsuppresseskrasinduced pages 1-2, tu2024somaticbrafv600emutation pages 1-6, prado2019recentadvancesin pages 1-3)	HGNC labels: KRAS; BRAF	Human lesion genetics + animal model
Germline genes	Familial/syndromic AVM predisposition includes HHT genes ENG, ACVRL1, SMAD4, GDF2 and CM-AVM genes RASA1, EPHB4. (saito2024crisprcasrxsuppresseskrasinduced pages 1-2, prado2019recentadvancesin pages 1-3, OpenTargets Search: brain arteriovenous malformation)	HGNC labels: ENG; ACVRL1; SMAD4; GDF2; RASA1; EPHB4	Human inherited disease genetics
Epigenetics	Endothelial-cell methylome studies implicate differential methylation in KRAS, RBPJ, EPHB1 and pathways involving EC–VSMC crosstalk. (scimone2024methylomeanalysisof pages 1-2)	Label only: DNA methylation	Human tissue/epigenomics
Transcriptomics	Bulk RNA-seq of resected bAVMs showed 736 upregulated and 498 downregulated genes, highlighting inflammation, cytoskeletal remodeling, reduced ECM integrity, angiopoietin-TIE, and TGF-β signaling changes. (hauer2020rnasequencinghighlightsinflammation pages 1-2)	GO labels: extracellular matrix organization; cell migration	Human bulk RNA-seq
Single-cell profiling	Single-cell atlases identified abnormal endothelial states, altered arteriovenous zonation, immune crosstalk, and atypical endothelial MHC class II upregulation in brain vascular malformations. (winkler2022asinglecellatlas pages 1-3, walchli2024singlecellatlasof pages 1-2)	CL:0000115 endothelial cell; GO label: antigen presentation	Human single-cell RNA-seq
Diagnostics	Digital subtraction angiography remains the reference standard for diagnosis and angioarchitectural characterization. (chauvet2024diagnosticaccuracyof pages 1-2, chen2023developmentandvalidation pages 2-3)	Procedure label: digital subtraction angiography	Human clinical imaging
Diagnostics	MRI/MRA are widely used; 2024 data support non-contrast 4D MRA for follow-up/characterization with accuracy comparable to contrast-enhanced 4D MRA in one retrospective cohort. (chauvet2024diagnosticaccuracyof pages 1-2, chauvet2024diagnosticaccuracyof pages 2-5)	Procedure labels: MRI; MRA	Human clinical imaging
Prognostic tools	Imaging-based hemorrhage stratification includes the VALE score: ventricular system involvement, venous aneurysm, deep location, exclusively deep drainage. (chen2023developmentandvalidation pages 1-2)	Labels only: ventricular involvement; venous aneurysm; deep drainage	Human clinical prognostic model
Interventions	Standard interventions include microsurgical resection, endovascular embolization, stereotactic radiosurgery, or multimodality therapy; conservative management remains important for selected unruptured lesions. (NCT02098252 chunk 1, mohr2020medicalmanagementwith pages 1-2, NCT04572568 chunk 1)	NCIT labels: surgical resection; embolization; stereotactic radiosurgery; conservative management	Human clinical + registry/trial
Current trials/implementation	Ongoing real-world and interventional studies include TOBAS (NCT02098252), MATCH registry (NCT04572568), and a liquid embolic agent randomized trial (PARTNER; NCT07314047). (NCT02098252 chunk 1, NCT04572568 chunk 1, NCT07314047 chunk 1)	ClinicalTrials.gov: NCT02098252; NCT04572568; NCT07314047	Clinical trials/registry
Experimental therapeutics	Preclinical targeted approaches include MEK/ERK-pathway inhibition, BRAF inhibition, and CRISPR/CasRx knockdown of mutant KRAS in endothelial-driven mouse models. (saito2024crisprcasrxsuppresseskrasinduced pages 2-4, tu2024somaticbrafv600emutation pages 1-6, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)	Intervention labels: MEK inhibitor; BRAF inhibitor; CRISPR/CasRx	Animal model/preclinical
Prevention/screening	No established primary prevention exists for sporadic bAVM; secondary prevention is focused on risk stratification, imaging surveillance, and genetic/syndromic evaluation when HHT or CM-AVM is suspected. (NCT02098252 chunk 1, prado2019recentadvancesin pages 1-3, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)	Label only: genetic counseling; surveillance imaging	Human clinical + inherited disease context

Table: This table condenses ontology-ready facts for brain arteriovenous malformation, spanning identifiers, phenotypes, cells, mechanisms, diagnostics, and interventions. It is designed to support direct knowledge-base population while keeping uncertain ontology IDs as labels only.

1. Disease information

Definition and identifiers

A bAVM is a tangle of abnormally dilated cerebral vessels with direct arteriovenous shunting, high flow, low resistance, and no normal capillary interface. The central tangle is the nidus; feeding arteries, the nidus, and draining veins form the lesion’s angioarchitecture. (chen2023developmentandvalidation pages 2-3, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

MONDO: MONDO:0007154, arteriovenous malformations of the brain.
OMIM: 108010, Arteriovenous malformations of brain.
MeSH: the broad indexed concept is Arteriovenous Malformations; “brain/cerebral AVM” supplies anatomical qualification.
ICD: coding varies by jurisdiction and whether congenital status, rupture, or hemorrhage is being represented. ICD-10-CM commonly uses Q28.2 for arteriovenous malformation of cerebral vessels; associated intracranial hemorrhage should be coded separately. ICD-11 mapping should be verified against the deployment’s current browser/version rather than inferred from ICD-10.
Synonyms: brain AVM, bAVM, cerebral AVM, cerebral arteriovenous malformation, intracranial AVM, pial AVM. “Dural arteriovenous fistula,” “cavernous malformation,” and “vein of Galen malformation” are distinct entities and should not be merged.

Open Targets identifies ENG as a disease-associated target for MONDO:0007154 and separately links ENG and ACVRL1 to HHT1 and HHT2, respectively. (OpenTargets Search: brain arteriovenous malformation)

2. Etiology, risk factors, protective factors, and gene–environment interaction

Causal architecture

Sporadic bAVM (>95%). The strongest current model is a postzygotic, endothelial mosaic activating variant—most often KRAS, less often BRAF or another RAS/MAPK component—arising in a permissive vascular-development or remodeling context. The mutant clone perturbs endothelial proliferation, identity, migration, and interaction with mural and immune cells. Reported somatic-variant detection varies substantially, approximately 28%–87%, because tissue sampling, endothelial enrichment, sequencing depth, and variant allele fraction differ. (prado2019recentadvancesin pages 1-3, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

Familial/syndromic bAVM (<5%). The major predisposition is HHT, an autosomal-dominant disorder caused mainly by loss-of-function variants in ENG or ACVRL1, and less often SMAD4 or GDF2. CM-AVM syndromes involve RASA1 or EPHB4. HHT-related endothelial BMP9/10–ENG–ACVRL1–SMAD1/5/8 insufficiency interacts with angiogenic and hemodynamic cues; a local “second hit” or mosaic event may help determine where a lesion forms. (prado2019recentadvancesin pages 1-3, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

Risk factors

Established or repeatedly supported hemorrhage-risk factors include previous hemorrhage—the most reproducible predictor—deep location, exclusively deep venous drainage, ventricular-system involvement, and venous aneurysm. Silent microhemorrhage and tissue hemosiderin have also been associated with subsequent hemorrhage. The 2023 VALE model formalized ventricular involvement, venous aneurysm, deep location, and exclusively deep drainage. (rutledge2014hemorrhageratesand pages 4-6, chen2023developmentandvalidation pages 1-2)

Candidate modifiers include inflammatory polymorphisms in IL6 and TNF, APOE ε2, and EPHB4 variants, but these associations require independent replication before clinical genetic prediction. Reported effect estimates include OR 2.4 for IL6 −174G>C, HR 4.0 for TNF −238G>A, and HR 5.1 for APOE ε2. (rutledge2014hemorrhageratesand pages 4-6)

Age is not a simple causal exposure: lesions can exist from development, form or enlarge postnatally, and present at any age, but symptomatic diagnosis is concentrated in children and young-to-middle-aged adults. Sex differences are inconsistent across cohorts; a large Chinese series was 58.3% male, but that should not be treated as a universal biological ratio. (chen2023developmentandvalidation pages 1-2, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

Environmental/lifestyle factors and protection

No toxin, pollutant, diet, infection, occupation, smoking pattern, alcohol exposure, or exercise level is established as a primary cause of sporadic bAVM. Pregnancy-related hemodynamic and hormonal effects remain clinically debated rather than proven causes of lesion formation. Hypertension may worsen consequences of hemorrhage and is treated as part of general vascular care, but it is not an established origin of the malformation.

No validated genetic protective allele or lifestyle intervention prevents bAVM formation. Practical “protective” management therefore means blood-pressure control, avoidance of illicit sympathomimetics, appropriate seizure treatment, and individualized avoidance of activities or medications judged to create unacceptable bleeding consequences—not proven lesion-prevention strategies.

Gene–environment interaction

The best-supported interaction is gene × angiogenic/hemodynamic context: endothelial RAS/MAPK activation or HHT-pathway loss alters responses to VEGF, shear stress, flow direction, and vessel injury. High-flow shunting then becomes a feed-forward stimulus for remodeling, venous hypertension, inflammation, and wall failure. Direct human evidence for conventional environmental G×E interactions is limited.

3. Phenotypes

Intracranial hemorrhage—HP:0002170. Usually acute and potentially severe; it may be intraparenchymal, intraventricular, or subarachnoid. More than half of symptomatic historical cohorts presented with hemorrhage, although incidentally detected lesions are increasing. First-hemorrhage mortality has been reported at 10%–30%; approximately 10%–20% of survivors may have long-term disability, with estimates dependent on cohort and era. (NCT02098252 chunk 1)
Seizures—HP:0001250. Episodic and variable in severity; approximately 20%–25% present with seizures. Some respond to antiseizure medication, whereas others become drug resistant. Seizure restrictions affect driving, education, employment, and independence. (NCT02098252 chunk 1)
Headache—HP:0002315. Episodic or chronic, often nonspecific; phenotype frequency and causal attribution vary. Sudden severe headache requires emergency assessment for hemorrhage.
Focal neurologic deficit. Weakness, sensory loss, aphasia, visual-field deficit, ataxia, or cranial-nerve dysfunction may reflect hemorrhage, venous congestion, ischemic steal, mass effect, seizure/postictal change, or treatment injury. Suggested HPO concepts include hemiparesis (HP:0001269), aphasia (HP:0002381), visual-field defect, ataxia (HP:0001251), and cognitive impairment (HP:0100543), selected according to the actual manifestation.
Incidental/asymptomatic lesion. Increasingly identified by MRI performed for another reason; not equivalent to biologically inactive disease.
Pediatric presentation. bAVM is the dominant vascular cause of spontaneous pediatric intracranial hemorrhage: one meta-analysis found AVMs constituted 68.3% of detailed vascular causes; another systematic review found 1,226 AVMs, 70.99% of reported vascular causes. These are proportions among hemorrhagic pediatric cohorts, not population prevalence.

Quality-of-life loss is driven by fear of rupture, headache and epilepsy, neurologic disability, treatment recovery, and uncertainty. Standard instruments include EQ-5D, SF-36, PROMIS, modified Rankin Scale (mRS), seizure-specific measures, and neuropsychological testing. High-quality phenotype-specific QOL percentages remain limited.

4. Genetic and molecular information

Genes and variant classes

KRAS: activating somatic missense variants, commonly codon 12 substitutions such as p.Gly12Val and p.Gly12Asp; constitutive GTP-bound signaling activates RAF–MEK–ERK and can also affect PI3K–AKT–mTOR. Lesional variant allele fractions are often low because the mutation is mosaic and endothelial restricted; population frequency is therefore not meaningfully represented by germline gnomAD frequency. (saito2024crisprcasrxsuppresseskrasinduced pages 1-2)
BRAF: somatic activating variants, including p.Val600Glu, occur in a minority. Endothelial BrafV600E was sufficient to produce AVM-like lesions, hemorrhage, seizures, and motor deficits in mice. (tu2024somaticbrafv600emutation pages 1-6)
ENG, ACVRL1, SMAD4, GDF2: heterozygous germline loss-of-function variants causing HHT-related predisposition; inheritance is autosomal dominant with age-dependent and variable expression.
RASA1, EPHB4: heterozygous germline loss-of-function variants causing CM-AVM spectrum; EPHB4 is especially relevant to arterial–venous identity.
Other proposed genes and susceptibility variants should be considered research-level unless supported by a recognizable syndrome and clinically curated variant evidence.

Variant interpretation must distinguish (1) germline constitutional testing from blood/saliva and (2) low-frequency somatic testing of resected or endovascularly sampled lesion material. A negative blood panel does not exclude a lesion-restricted driver. ACMG/AMP classification applies most directly to germline findings; somatic variants require disease-specific functional and mosaic evidence. No recurrent large chromosomal abnormality, repeat expansion, mitochondrial variant, or aneuploidy defines isolated bAVM.

Modifier and epigenetic evidence

Human endothelial methylome analysis reported differential methylation of RBPJ, KRAS, EPHB1, adhesion/cytoskeletal loci, EC–vascular-smooth-muscle crosstalk genes, and long noncoding RNAs. The authors also implicated non-CpG CHG methylation in neurovascular-development pathways. This is discovery-level evidence from a small tissue study, not a validated diagnostic signature. (scimone2024methylomeanalysisof pages 1-2)

5. Environmental information

There is no established infectious agent and no zoonotic or transmissible mechanism. No environmental toxicant or radiation exposure has been established as causal. Radiation is a treatment modality, not a known cause of ordinary bAVM. Lifestyle exposures do not currently support causal annotation in CTD-like knowledge bases. General cerebrovascular risk reduction remains advisable because comorbidity can worsen treatment and hemorrhage outcomes.

6. Mechanism and pathophysiology

Causal chain

Upstream trigger: endothelial somatic RAS/MAPK activation, or germline BMP/TGF-β/RASA1–EPHB4 pathway insufficiency with local permissive events.
Endothelial transformation: ERK-driven proliferation and EndMT-like change, disturbed arterial–venous zonation, altered Notch/ephrin signaling, loss of BBB/CNS-specific properties, and abnormal response to flow.
Malformed shunt: capillary specification fails or regresses, producing direct arterial-to-venous channels and a nidus.
Neurovascular-unit failure: defective endothelial–pericyte–smooth-muscle crosstalk, reduced vessel maturation, extracellular-matrix disruption, and inadequate mural coverage.
Feed-forward remodeling: high flow and venous hypertension stimulate VEGF/angiogenesis, cytoskeletal remodeling, and inflammatory recruitment.
Tissue injury: fragile vessels leak or rupture; hemorrhage causes mechanical injury, edema, inflammation, ischemia, neuronal loss, seizures, and focal deficits.

Human transcriptomic and single-cell evidence

Bulk RNA sequencing of 12 bAVMs versus 16 control intracranial arteries found 736 upregulated genes, including cytoskeletal, migration, inflammatory-cytokine, neutrophil, and macrophage programs, and 498 downregulated genes, including extracellular-matrix, angiopoietin–TIE, and TGF-β programs. Forty-seven GO terms were enriched, supporting inflammation, loss of vascular quiescence, and impaired wall integrity. (hauer2020rnasequencinghighlightsinflammation pages 1-2)

The 2022 human cerebrovascular atlas profiled 181,388 cells and found loss of normal endothelial arteriovenous zonation, a nidus-associated angiogenic state, and vascular–immune crosstalk. Its abstract states: “We illustrated an interplay between vascular and immune cells contributory to brain hemorrhage.” GPNMB-positive monocytes were associated with depletion of stabilizing smooth-muscle cells in bAVMs that had bled. (winkler2022asinglecellatlas pages 1-3)

A larger 2024 Nature atlas analyzed 606,380 cells from 117 samples and 68 human fetuses/adult patients. Diseased vasculature showed altered arteriovenous differentiation, reactivated fetal programs, loss of CNS-specific endothelial properties, MHC-class-II upregulation, and immune/angiogenic endothelial-to-perivascular signaling. These findings are shared across several vascular-dependent CNS diseases and are not all bAVM-specific. (walchli2024singlecellatlasof pages 1-2)

Suggested mechanistic annotations

GO: angiogenesis (GO:0001525), MAPK cascade (GO:0000165), inflammatory response (GO:0006954), endothelial-cell migration, extracellular-matrix organization, blood-vessel remodeling, regulation of vascular permeability, response to fluid shear stress, and endothelial-to-mesenchymal transition.
CL: endothelial cell (CL:0000115), pericyte (CL:0000669), vascular smooth-muscle cell, perivascular fibroblast, monocyte, macrophage, neutrophil, astrocyte, neuron, and microglial cell.
Subcellular: plasma membrane receptor complexes, cytosol and nucleus for MAPK signaling, adherens/tight junctions, actin cytoskeleton, extracellular matrix, and endoplasmic reticulum. No defining mitochondrial, lysosomal, or protein-aggregation defect is known.
Metabolism/proteomics/lipidomics: inflammatory and hypoxic metabolic remodeling is plausible, but no clinically validated metabolomic, proteomic, or lipidomic signature exists.

7. Anatomical structures affected

The primary organ is the brain (UBERON:0000955), particularly cerebral arteries, arterioles, veins, and intervening parenchyma. Lesions may be cortical/lobar, deep (basal ganglia, thalamus, corpus callosum, insula), brainstem, or cerebellar; they may be superficial or deep and drain superficially, deeply, or both. Lateralization is usually unilateral/asymmetric, although multiple or bilateral lesions occur, especially in HHT. Secondary structures include ventricles and subarachnoid spaces when hemorrhage extends, and remote cortex involved in epileptic networks.

At tissue level, affected components are vascular endothelium, basement membrane/extracellular matrix, pericytes, smooth muscle, perivascular fibroblasts, and adjacent neural/glial tissue. The defining subcellular systems are endothelial junctions, cytoskeleton, receptor signaling complexes, and transcriptional machinery.

8. Temporal development

The lesion may originate during vascular development, but de novo postnatal formation and interval growth are documented; therefore “congenital” should not be interpreted as invariably complete and static at birth. Presentation may occur from infancy through old age, most often before age 40 in symptomatic cohorts. (NCT02098252 chunk 1, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

Clinical onset is frequently acute with hemorrhage or seizure; incidental detection is asymptomatic/insidious. Untreated disease is chronic and lifelong, with a persistent annual hemorrhage hazard commonly estimated around 1%–3% for unruptured lesions and 2%–4% overall, modified substantially by prior rupture and angioarchitecture. Spontaneous complete obliteration is rare. SRS produces delayed involution over years, during which hemorrhage risk persists; surgery can provide immediate cure if complete, while embolization may be curative in selected anatomy or adjunctive/staged. (rutledge2014hemorrhageratesand pages 4-6, chen2023developmentandvalidation pages 2-3, NCT02098252 chunk 1)

Critical windows include acute hematoma management, post-hemorrhage stabilization and angiographic reassessment, the latency interval after SRS, and long-term surveillance after apparent cure—especially in children, where recurrence is more concerning.

9. Inheritance and population epidemiology

Population prevalence is approximately 10 per 100,000 in a recent mechanistic review; an older synthesis cited 0.05%, illustrating methodological variability. Symptomatic discovery incidence is approximately 1.1 per 100,000 person-years. bAVMs cause roughly 1%–2% of all strokes but a disproportionate fraction of hemorrhagic stroke in children and younger adults. (prado2019recentadvancesin pages 1-3, NCT02098252 chunk 1, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

Most isolated cases are non-Mendelian, lesion-restricted mosaics and have low sibling/offspring recurrence risk. HHT and CM-AVM are autosomal dominant, with incomplete/age-dependent penetrance and variable expressivity. HHT prevalence is approximately 1 in 5,000; one review reported brain AVM in 13.4% of HHT1 versus 2.4% of HHT2. Genetic anticipation, repeat expansion, consanguinity effects, and a defined carrier frequency are not features of sporadic bAVM. Founder variants exist in some HHT populations but are syndrome- and ancestry-specific. (prado2019recentadvancesin pages 1-3)

No consistent high-risk ethnicity or endemic region is established after accounting for ascertainment. The median age in the 3,962-patient Chinese VALE cohort was 26.1 years and 58.3% were male, but external demographic generalization requires caution. (chen2023developmentandvalidation pages 1-2)

10. Diagnostics

Clinical and imaging work-up

Acute suspected hemorrhage is assessed with noncontrast CT, often followed by CT angiography. MRI characterizes nidus location, prior hemorrhage, hemosiderin, edema, eloquent structures, and associated parenchymal injury; MRA evaluates flow noninvasively. Catheter digital-subtraction angiography (DSA) remains the reference standard for feeding arteries, nidus, high-flow fistulas, associated aneurysms, venous drainage, stenosis, and treatment planning. (chauvet2024diagnosticaccuracyof pages 2-5, chauvet2024diagnosticaccuracyof pages 1-2)

In a retrospective 2024 study of 54 MRA pairs from 43 patients, noncontrast 4D-MRA had accuracy 0.85 and specificity 95%, versus 0.83 and 85% for contrast-enhanced 4D-MRA, with DSA as reference. The authors concluded that noncontrast 4D-MRA may support repeated follow-up while avoiding gadolinium, but prospective validation is required. Publication: 31 July 2024; DOI: https://doi.org/10.3390/diagnostics14151656. (chauvet2024diagnosticaccuracyof pages 2-5, chauvet2024diagnosticaccuracyof pages 1-2)

EEG is used for seizure classification, not AVM diagnosis. Routine blood/urine chemistry has no diagnostic signature. Histology shows malformed arterialized and venous vessels with variable wall thickness, gliosis, hemosiderin, inflammation, and absent normal capillary organization, but biopsy solely for diagnosis is generally inappropriate because of bleeding risk.

Classification and differential diagnosis

Spetzler–Martin grade uses nidus size, eloquence, and deep venous drainage to estimate surgical risk; supplementary surgical, radiosurgical, and embolization scales may be added. Differential diagnoses include dural AV fistula, developmental venous anomaly, cavernous malformation, capillary telangiectasia, aneurysm, moyamoya-associated collaterals, hemorrhagic tumor, and vein of Galen malformation.

Genetic testing

Routine germline testing is not indicated for every solitary sporadic bAVM. Test when there are multiple AVMs, mucocutaneous telangiectases, recurrent epistaxis, pulmonary/hepatic AVMs, capillary malformations, limb overgrowth, or family history. A panel should include ENG, ACVRL1, SMAD4, GDF2, RASA1, and EPHB4, with deletion/duplication analysis. WES/WGS can be used for unresolved syndromic cases. CMA, karyotype, FISH, mitochondrial sequencing, and repeat-expansion testing have no routine role.

Somatic testing requires affected tissue or validated endovascular sampling with deep sequencing/digital PCR for low-VAF KRAS/BRAF/MAP2K1-pathway variants. It is currently research-oriented and may become relevant for targeted therapy. Blood-based “liquid biopsy,” RNA-seq, methylomics, proteomics, and metabolomics are not validated clinical diagnostics.

11. Outcome and prognosis

The major adverse outcome is rupture. Overall untreated hemorrhage risk is approximately 2%–4% annually and around 1%–3% for unruptured lesions; prior hemorrhage raises subsequent risk. (rutledge2014hemorrhageratesand pages 4-6, chen2023developmentandvalidation pages 2-3, NCT02098252 chunk 1)

The 2023 VALE study included 3,962 patients. In 1,028 conservatively managed patients, 36 hemorrhages occurred over median 4.2 years. AUCs were 0.77 in derivation, 0.85 in external validation, and 0.73 in conservative validation. Ten-year hemorrhage-free survival was 95.5% low risk, 92.8% moderate risk, and 75.8% high risk. This is promising but should be externally tested across health systems and ancestries. Publication: 1 March 2023; DOI: https://doi.org/10.1001/jamanetworkopen.2023.1070. (chen2023developmentandvalidation pages 1-2)

Functional outcome depends on rupture severity, initial neurologic status, lesion location, complete obliteration, treatment complications, seizures, and rehabilitation. There is no single meaningful five- or ten-year survival figure analogous to cancer survival. Morbidity includes epilepsy, motor/language/cognitive deficits, visual loss, chronic headache, anxiety, educational/employment disruption, and treatment-related stroke or radiation injury.

12. Treatment

Strategy

Management should be decided by a multidisciplinary team including vascular neurosurgery, interventional neuroradiology, radiosurgery/radiation oncology, stroke neurology, epilepsy care, and rehabilitation. The goal of definitive treatment is complete shunt obliteration; partial treatment does not reliably remove lifetime hemorrhage risk.

Conservative/medical management: observation with MRI/angiographic surveillance as appropriate; antihypertensives for comorbid hypertension; antiseizure drugs for epilepsy; headache treatment; and rehabilitation. There is no approved disease-modifying drug.
Microsurgical resection: immediate cure when complete; favored for selected small, superficial, surgically accessible low-grade lesions, particularly after rupture or with an evacuable hematoma. Risks include hemorrhage, ischemia, focal deficit, infection, and death.
Endovascular embolization: liquid embolic agents such as ethylene-vinyl alcohol copolymer (Onyx) or n-butyl cyanoacrylate are delivered through microcatheters. Uses include targeted treatment of aneurysms/high-flow fistulas, flow or volume reduction before surgery/SRS, and cure in selected compact lesions. Risks include hemorrhage, ischemic stroke, catheter complications, and incomplete occlusion.
SRS: focused radiation for small/deep or surgically high-risk lesions. Obliteration is delayed, and radiation edema, necrosis, cyst formation, neurologic deficit, and latency-period hemorrhage can occur.
Multimodal therapy: staged embolization plus resection or SRS is common for complex anatomy, but every added procedure adds risk.

Suggested NCIT labels are Surgical Resection, Endovascular Embolization, Stereotactic Radiosurgery, Radiation Therapy, Anticonvulsant Therapy, and Rehabilitation Therapy. Relevant chemicals include cyanoacrylate and ethylene-vinyl alcohol copolymer; CHEBI mapping should be performed against the exact product/compound.

Randomized evidence and expert interpretation

ARUBA randomized 226 adults with unruptured, treatment-eligible AVMs. At mean 50-month follow-up, medical management remained superior to medical management plus intervention for stroke or death (HR 0.31, 95% CI 0.17–0.56). The investigators advise informing patients of the absolute and relative early risks of intervention, while explicitly noting that outcomes beyond five years remain uncertain. (mohr2020medicalmanagementwith pages 1-2)

This does not mean all unruptured bAVMs should never be treated. Expert criticism concerns selection, heterogeneous interventions, limited representation of optimal microsurgical candidates, and insufficient follow-up to capture lifetime hemorrhage prevention. The defensible conclusion is narrower: routine prophylactic intervention for every unruptured lesion is unsupported; treatment should be lesion-specific and, when uncertainty is material, trial-based.

Trials and real-world implementation

TOBAS—NCT02098252: recruiting randomized care trial and registry, target n=1,000. It compares conservative versus interventional management and nests randomization of embolization before surgery/SRS. The primary outcome is death or disabling stroke; planned follow-up extends to 2035–2036. https://clinicaltrials.gov/study/NCT02098252 (NCT02098252 chunk 1)
MATCH—NCT04572568: recruiting Chinese multicenter prospective registry, n=2,000, with real-world multidisciplinary pathways and outcomes including mRS, obliteration, hemorrhage, complications, epilepsy, headache, and neurologic function. https://clinicaltrials.gov/study/NCT04572568 (NCT04572568 chunk 1)
PARTNER—NCT07314047: randomized, open-label, noninferiority device trial, n=116, comparing a MicroPort liquid embolic agent with Onyx; primary endpoint is ≥50% target-AVM embolization. Registry posting occurred in 2026, so it is a current development rather than a 2023–2024 source. https://clinicaltrials.gov/study/NCT07314047 (NCT07314047 chunk 1)

Experimental precision therapy

MEK/ERK inhibitors suppress mutant-KRAS endothelial phenotypes in vitro and reduce malformation burden in fish/mice, but toxicity, blood–brain delivery, mosaic target detection, treatment duration, and rebound remain unresolved. In 2024, endothelial BrafV600E mice improved neurologically with dabrafenib, and CRISPR/CasRx knockdown of mutant KRAS suppressed lesion development in another mouse model. Neither constitutes human efficacy evidence. (tu2024somaticbrafv600emutation pages 1-6, saito2024crisprcasrxsuppresseskrasinduced pages 2-4, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

No validated pharmacogenomic algorithm, gene therapy, cell therapy, RNA therapeutic, or immunotherapy is approved for bAVM.

13. Prevention

Primary prevention: none is established for sporadic mosaic bAVM; vaccination and antimicrobial prophylaxis are not applicable. In familial disease, genetic counseling can inform reproductive options, including prenatal or preimplantation testing for a known pathogenic germline variant, but this prevents transmission rather than treating an existing lesion.

Secondary prevention: no population screening is recommended because prevalence is low and intervention has nontrivial risk. Targeted screening is appropriate in HHT/CM-AVM families according to syndrome-specific guidance. Early imaging after suggestive neurologic symptoms and structured rupture-risk assessment are central.

Tertiary prevention: seizure control, blood-pressure management, definitive treatment when benefit exceeds risk, surveillance for residual/recurrent shunting, stroke rehabilitation, fall/driving counseling, and management of depression/anxiety. Antibiotic prophylaxis is not routinely indicated merely because a cerebral AVM exists.

14. Other species and natural disease

Naturally occurring intracranial AVMs are reported sporadically in dogs and other animals, but evidence is mainly isolated veterinary case reports; there is no well-established breed association, VBO term, carrier frequency, or comparable population natural history. The condition is noninfectious and nonzoonotic. Orthologous RAS/MAPK, ENG–ACVRL1–SMAD, RASA1–EPHB4, VEGF, Notch, and ephrin pathways are strongly conserved across vertebrates.

Relevant taxa are Homo sapiens (NCBI Taxon 9606), Mus musculus (10090), and Danio rerio (7955). Exact orthologous NCBI Gene IDs should be resolved programmatically from current NCBI/Alliance releases rather than inferred here.

15. Model organisms

Mouse

Conditional endothelial KrasG12D models, using Cdh5-CreERT2 or brain-endothelium-directed AAV, reproduce dilated plexiform arteriovenous networks, high flow, hemorrhage, ERK activation, and endothelial proliferation. The 2024 JCI Insight model used AAV plus Cdh5-CreERT2, three-dimensional cleared-tissue imaging and scRNA-seq, and demonstrated suppression with mutant-selective CRISPR/CasRx. Exact abstract quote: “CRISPR/CasRx to knock down mutant KRAS expression … efficiently suppressed bAVM development.” Publication: 22 November 2024; DOI: https://doi.org/10.1172/jci.insight.179729. (saito2024crisprcasrxsuppresseskrasinduced pages 2-4, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

Endothelial BrafV600E mice developed MRI-visible lesions, brain hemorrhage, seizures, motor/balance deficits, and death; dabrafenib improved behavioral measures and neuronal preservation in small preclinical groups. DOI: https://doi.org/10.1007/s10456-024-09918-8. (tu2024somaticbrafv600emutation pages 1-6)

HHT models delete Eng or Acvrl1 in endothelium, often combined with angiogenic stimulation or injury, and establish abnormal endothelial responses to VEGF and blood flow. Their limitation is that they model syndromic pathway loss rather than the common sporadic KRAS mosaic disease.

Zebrafish and cellular systems

Zebrafish RAS/MAPK and HHT-pathway perturbations permit live vascular imaging, rapid developmental phenotyping, and drug screening; MEK inhibition can reduce abnormal vascular tangles. Their cerebrovascular anatomy and developmental timing differ from humans. Human bAVM endothelial cultures, iPSC-derived endothelium, organ-on-chip systems, and explants allow pathway and flow studies but do not reproduce full neurovascular-unit architecture or lifetime rupture.

Across models, key limitations are supraphysiologic or widespread recombination, severe early lethality, variant levels exceeding human mosaic VAFs, incomplete representation of lesion heterogeneity, and inability to model decades of human remodeling. Nevertheless, the 2024 studies provide strong causal evidence that endothelial RAS/MAPK activation is sufficient for AVM-like disease and establish tractable platforms for targeted-therapy testing. (tu2024somaticbrafv600emutation pages 1-6, saito2024crisprcasrxsuppresseskrasinduced pages 1-2)

Evidence gaps and knowledge-base cautions

Somatic-driver prevalence depends heavily on lesional sampling and assay sensitivity; absence of a detected variant is not evidence of a non-genetic lesion.
No circulating, proteomic, metabolomic, lipidomic, or epigenetic biomarker is validated for diagnosis or rupture prediction.
Natural-history estimates are averages and should not replace lesion-specific assessment.
ARUBA supports conservative management over heterogeneous intervention in its enrolled unruptured population over approximately four years, not a universal lifetime prohibition against treatment.
Targeted RAS/MAPK and gene-editing treatments remain preclinical.
Several ontology mappings requested—particularly exact ICD-11, SNOMED, LOINC, NCIT, CL, and UBERON descendants—should be validated against the knowledge base’s licensed/current release before ingestion.

References

(prado2019recentadvancesin pages 1-3): Leandro Barbosa Do Prado, Chul Han, S. Paul Oh, and Hua Su. Recent advances in basic research for brain arteriovenous malformation. International Journal of Molecular Sciences, 20:5324, Oct 2019. URL: https://doi.org/10.3390/ijms20215324, doi:10.3390/ijms20215324. This article has 62 citations.
(chen2023developmentandvalidation pages 1-2): Yu Chen, Heze Han, Xiangyu Meng, Hengwei Jin, Dezhi Gao, Li Ma, Ruinan Li, Zhipeng Li, Debin Yan, Haibin Zhang, Kexin Yuan, Ke Wang, Yukun Zhang, Yang Zhao, Weitao Jin, Runting Li, Fa Lin, Xiaofeng Chao, Zhengfeng Lin, Qiang Hao, Hao Wang, Xun Ye, Shuai Kang, Youxiang Li, Shibin Sun, Ali Liu, Shuo Wang, Yuanli Zhao, and Xiaolin Chen. Development and validation of a scoring system for hemorrhage risk in brain arteriovenous malformations. JAMA Network Open, 6:e231070, Mar 2023. URL: https://doi.org/10.1001/jamanetworkopen.2023.1070, doi:10.1001/jamanetworkopen.2023.1070. This article has 69 citations and is from a peer-reviewed journal.
(saito2024crisprcasrxsuppresseskrasinduced pages 1-2): Shoji Saito, Yuka Nakamura, Satoshi Miyashita, Tokiharu Sato, Kana Hoshina, Masayasu Okada, Hitoshi Hasegawa, Makoto Oishi, Yukihiko Fujii, Jakob Körbelin, Yoshiaki Kubota, Kazuki Tainaka, Manabu Natsumeda, and Masaki Ueno. Crispr/casrx suppresses kras-induced brain arteriovenous malformation developed in postnatal brain endothelial cells in mice. JCI Insight, Nov 2024. URL: https://doi.org/10.1172/jci.insight.179729, doi:10.1172/jci.insight.179729. This article has 11 citations and is from a domain leading peer-reviewed journal.
(mohr2020medicalmanagementwith pages 1-2): Jay P Mohr, Jessica R Overbey, Andreas Hartmann, Rüdiger von Kummer, Rustam Al-Shahi Salman, Helen Kim, H Bart van der Worp, Michael K Parides, Marco A Stefani, Emmanuel Houdart, Richard Libman, John Pile-Spellman, Kirsty Harkness, Charlotte Cordonnier, Ellen Moquete, Alessandra Biondi, Catharina J M Klijn, Christian Stapf, and Alan J Moskowitz. Medical management with interventional therapy versus medical management alone for unruptured brain arteriovenous malformations (aruba): final follow-up of a multicentre, non-blinded, randomised controlled trial. The Lancet Neurology, 19:573-581, Jul 2020. URL: https://doi.org/10.1016/s1474-4422(20)30181-2, doi:10.1016/s1474-4422(20)30181-2. This article has 215 citations and is from a highest quality peer-reviewed journal.
(NCT02098252 chunk 1): Treatment of Brain AVMs (TOBAS) Study. Centre hospitalier de l'Université de Montréal (CHUM). 2014. ClinicalTrials.gov Identifier: NCT02098252
(OpenTargets Search: brain arteriovenous malformation): Open Targets Query (brain arteriovenous malformation, 10 results). Buniello, A. et al. (2025). Open Targets Platform: facilitating therapeutic hypotheses building in drug discovery. Nucleic Acids Research.
(rutledge2014hemorrhageratesand pages 4-6): W. Caleb Rutledge, Nerissa U. Ko, Michael T. Lawton, and Helen Kim. Hemorrhage rates and risk factors in the natural history course of brain arteriovenous malformations. Translational Stroke Research, 5:538-542, Jun 2014. URL: https://doi.org/10.1007/s12975-014-0351-0, doi:10.1007/s12975-014-0351-0. This article has 144 citations and is from a peer-reviewed journal.
(winkler2022asinglecellatlas pages 1-3): Ethan A. Winkler, Chang N. Kim, Jayden M. Ross, Joseph H. Garcia, Eugene Gil, Irene Oh, Lindsay Q. Chen, David Wu, Joshua S. Catapano, Kunal Raygor, Kazim Narsinh, Helen Kim, Shantel Weinsheimer, Daniel L. Cooke, Brian P. Walcott, Michael T. Lawton, Nalin Gupta, Berislav V. Zlokovic, Edward F. Chang, Adib A. Abla, Daniel A. Lim, and Tomasz J. Nowakowski. A single-cell atlas of the normal and malformed human brain vasculature. Mar 2022. URL: https://doi.org/10.1126/science.abi7377, doi:10.1126/science.abi7377. This article has 333 citations and is from a highest quality peer-reviewed journal.
(walchli2024singlecellatlasof pages 1-2): Thomas Wälchli, Moheb Ghobrial, Marc Schwab, Shigeki Takada, Hang Zhong, Samuel Suntharalingham, Sandra Vetiska, Daymé Rodrigues Gonzalez, Ruilin Wu, Hubert Rehrauer, Anuroopa Dinesh, Kai Yu, Edward L. Y. Chen, Jeroen Bisschop, Fiona Farnhammer, Ann Mansur, Joanna Kalucka, Itay Tirosh, Luca Regli, Karl Schaller, Karl Frei, Troy Ketela, Mark Bernstein, Paul Kongkham, Peter Carmeliet, Taufik Valiante, Peter B. Dirks, Mario L. Suva, Gelareh Zadeh, Viviane Tabar, Ralph Schlapbach, Hartland W. Jackson, Katrien De Bock, Jason E. Fish, Philippe P. Monnier, Gary D. Bader, and Ivan Radovanovic. Single-cell atlas of the human brain vasculature across development, adulthood and disease. Nature, 632:603-613, Jul 2024. URL: https://doi.org/10.1038/s41586-024-07493-y, doi:10.1038/s41586-024-07493-y. This article has 197 citations and is from a highest quality peer-reviewed journal.
(pan2021theroleof pages 5-6): Peipei Pan, Sonali S Shaligram, Leandro Barbosa Do Prado, Liangliang He, and Hua Su. The role of mural cells in hemorrhage of brain arteriovenous malformation. Brain Hemorrhages, 2:49-56, Mar 2021. URL: https://doi.org/10.1016/j.hest.2020.10.005, doi:10.1016/j.hest.2020.10.005. This article has 20 citations.
(scimone2024methylomeanalysisof pages 1-2): Concetta Scimone, Luigi Donato, Simona Alibrandi, Alfredo Conti, Carlo Bortolotti, Antonino Germanò, Concetta Alafaci, Sergio Lucio Vinci, Rosalia D'Angelo, and Antonina Sidoti. Methylome analysis of endothelial cells suggests new insights on sporadic brain arteriovenous malformation. Heliyon, 10:e35126, Aug 2024. URL: https://doi.org/10.1016/j.heliyon.2024.e35126, doi:10.1016/j.heliyon.2024.e35126. This article has 11 citations.
(hauer2020rnasequencinghighlightsinflammation pages 1-2): Allard J. Hauer, Rachel Kleinloog, Fabrizio Giuliani, Gabriël J.E. Rinkel, Gerard A. de Kort, Jan Willem Berkelbach van der Sprenkel, Albert van der Zwan, Peter H. Gosselaar, Peter C. van Rijen, Jelkje J. de Boer-Bergsma, Patrick Deelen, Morris A. Swertz, Louis De Muynck, Philip Van Damme, Jan H. Veldink, Ynte M. Ruigrok, and Catharina J.M. Klijn. Rna-sequencing highlights inflammation and impaired integrity of the vascular wall in brain arteriovenous malformations. Stroke, 51:268-274, Jan 2020. URL: https://doi.org/10.1161/strokeaha.119.025657, doi:10.1161/strokeaha.119.025657. This article has 38 citations and is from a highest quality peer-reviewed journal.
(tu2024somaticbrafv600emutation pages 1-6): Tianqi Tu, Jiaxing Yu, Chendan Jiang, Shikun Zhang, Jingwei Li, Jian Ren, Shiju Zhang, Yuan Zhou, Ziwei Cui, Haohan Lu, Xiaosheng Meng, Zhanjing Wang, Dong Xing, Hongqi Zhang, and Tao Hong. Somatic brafv600e mutation in the cerebral endothelium induces brain arteriovenous malformations. Angiogenesis, 27:441-460, May 2024. URL: https://doi.org/10.1007/s10456-024-09918-8, doi:10.1007/s10456-024-09918-8. This article has 17 citations and is from a domain leading peer-reviewed journal.
(chauvet2024diagnosticaccuracyof pages 1-2): Grégoire Chauvet, Mourad Cheddad El Aouni, Elsa Magro, Ophélie Sabardu, Douraied Ben Salem, Jean-Christophe Gentric, and Julien Ognard. Diagnostic accuracy of non-contrast-enhanced time-resolved mr angiography to assess angioarchitectural classification features of brain arteriovenous malformations. Jul 2024. URL: https://doi.org/10.3390/diagnostics14151656, doi:10.3390/diagnostics14151656. This article has 8 citations.
(chen2023developmentandvalidation pages 2-3): Yu Chen, Heze Han, Xiangyu Meng, Hengwei Jin, Dezhi Gao, Li Ma, Ruinan Li, Zhipeng Li, Debin Yan, Haibin Zhang, Kexin Yuan, Ke Wang, Yukun Zhang, Yang Zhao, Weitao Jin, Runting Li, Fa Lin, Xiaofeng Chao, Zhengfeng Lin, Qiang Hao, Hao Wang, Xun Ye, Shuai Kang, Youxiang Li, Shibin Sun, Ali Liu, Shuo Wang, Yuanli Zhao, and Xiaolin Chen. Development and validation of a scoring system for hemorrhage risk in brain arteriovenous malformations. JAMA Network Open, 6:e231070, Mar 2023. URL: https://doi.org/10.1001/jamanetworkopen.2023.1070, doi:10.1001/jamanetworkopen.2023.1070. This article has 69 citations and is from a peer-reviewed journal.
(chauvet2024diagnosticaccuracyof pages 2-5): Grégoire Chauvet, Mourad Cheddad El Aouni, Elsa Magro, Ophélie Sabardu, Douraied Ben Salem, Jean-Christophe Gentric, and Julien Ognard. Diagnostic accuracy of non-contrast-enhanced time-resolved mr angiography to assess angioarchitectural classification features of brain arteriovenous malformations. Jul 2024. URL: https://doi.org/10.3390/diagnostics14151656, doi:10.3390/diagnostics14151656. This article has 8 citations.
(NCT04572568 chunk 1): Registry of Multimodality Treatment for Brain Arteriovenous Malformation in Mainland China. Beijing Tiantan Hospital. 2011. ClinicalTrials.gov Identifier: NCT04572568
(NCT07314047 chunk 1): The Liquid Embolic Agent for the Treatment of Brain Arteriovenous Malformation. MicroPort NeuroTech Co., Ltd.. 2024. ClinicalTrials.gov Identifier: NCT07314047
(saito2024crisprcasrxsuppresseskrasinduced pages 2-4): Shoji Saito, Yuka Nakamura, Satoshi Miyashita, Tokiharu Sato, Kana Hoshina, Masayasu Okada, Hitoshi Hasegawa, Makoto Oishi, Yukihiko Fujii, Jakob Körbelin, Yoshiaki Kubota, Kazuki Tainaka, Manabu Natsumeda, and Masaki Ueno. Crispr/casrx suppresses kras-induced brain arteriovenous malformation developed in postnatal brain endothelial cells in mice. JCI Insight, Nov 2024. URL: https://doi.org/10.1172/jci.insight.179729, doi:10.1172/jci.insight.179729. This article has 11 citations and is from a domain leading peer-reviewed journal.

Artifacts

Edison artifact artifact-00

Reference Validation

Checked with linkml-reference-validator 0.2.1.

Outcome	Count
References checked	13
Resolved	12
Unresolved (possible confabulation)	0
Unverifiable	1
References weighed for topical relevance	12
On topic	5
Off topic	0

12 of 13 references resolved; the rest could not be looked up either way.

Ask OpenScientist

OpenScientist Report

Classifications

Discussions and Knowledge Gaps

Pathophysiology

Pathograph

Phenotypes

Genetic Associations

Medical Actions

Diagnosis

Prevalence

Clinical Trials

Animal Models

Source YAML

References & Deep Research

Deep Research

Disease Characteristics Research Template

Target Disease

Research Objectives

1. Disease Information

2. Etiology

3. Phenotypes

4. Genetic/Molecular Information

5. Environmental Information

6. Mechanism / Pathophysiology

7. Anatomical Structures Affected

8. Temporal Development

9. Inheritance and Population

10. Diagnostics

11. Outcome/Prognosis

12. Treatment

13. Prevention

14. Other Species / Natural Disease

15. Model Organisms

Citation Requirements

Output Format

Brain arteriovenous malformation: disease-characteristics report

Executive summary

1. Disease information

Definition and identifiers

2. Etiology, risk factors, protective factors, and gene–environment interaction

Causal architecture

Risk factors

Environmental/lifestyle factors and protection

Gene–environment interaction

3. Phenotypes

4. Genetic and molecular information

Genes and variant classes

Modifier and epigenetic evidence

5. Environmental information

6. Mechanism and pathophysiology

Causal chain

Human transcriptomic and single-cell evidence

Suggested mechanistic annotations

7. Anatomical structures affected

8. Temporal development

9. Inheritance and population epidemiology

10. Diagnostics

Clinical and imaging work-up

Classification and differential diagnosis

Genetic testing

11. Outcome and prognosis

12. Treatment

Strategy

Randomized evidence and expert interpretation

Trials and real-world implementation

Experimental precision therapy

13. Prevention

14. Other species and natural disease

15. Model organisms

Mouse

Zebrafish and cellular systems

Evidence gaps and knowledge-base cautions

Artifacts

Reference Validation