Familial Cerebral Cavernous Malformations

Autosomal dominant vascular disease of the central nervous system caused by loss of function of KRIT1, CCM2 or PDCD10, with multiple cerebral and spinal cavernous malformations, risk of intracranial haemorrhage, epileptic seizures and focal neurological deficits, and incomplete, age-dependent penetrance.

Literature and regulatory information updated: 9 October 2026.

0. Initial Reference Table

1. Introduction

1.1 Definition and scope

Cerebral cavernous malformations (CCM, from cerebral cavernous malformations), also called cavernomas, cavernous angiomas or cavernous haemangiomas, are low-flow vascular lesions of the central nervous system (CNS) formed by abnormally dilated, permeable capillaries grouped into caverns lined by a single endothelial layer, without mature vascular wall elements or intervening brain parenchyma [1, 2]. Each lesion may measure from a few millimetres to several centimetres and contains blood at different degradation stages, intracavernous thrombosis and a haemosiderin rim reflecting repeated microhaemorrhages [1]. Because they are poorly or not at all visible on catheter angiography, they were called “angiographically occult vascular lesions” for decades [1].

The familial form (FCCM, from familial cerebral cavernous malformations; termed FCM, familial cavernous malformation, in the 2025 guideline) is an autosomal dominant disorder caused by heterozygous germline loss-of-function variants in one of three genes: KRIT1 (CCM1 locus, 7q21.2), CCM2 (CCM2 locus, 7p13) and PDCD10 (CCM3 locus, 3q26.1) [1, 2]. The corresponding Online Mendelian Inheritance in Man (OMIM) entities are 116860 (CCM1), 603284 (CCM2) and 603285 (CCM3); in Orphanet the familial form has code ORPHA:221061 within group ORPHA:164, which also includes sporadic forms [1, 7, 8]. OMIM entry 116860 also groups under the same number the historical designations of CNS and retinal cavernous malformations and hyperkeratotic cutaneous capillary-venous malformations associated with cerebral capillary malformations, corresponding to extraneural manifestations of the same KRIT1 disorder [3].

This article addresses FCCM in its three gene-defined forms. Sporadic cavernous malformations (generally a single lesion associated with a developmental venous anomaly, DVA), forms associated with somatic PIK3CA or MAP3K3 variants and radiation-induced lesions are addressed as pathophysiological context and differential diagnoses. OMIM entity 619538, sometimes called CCM4, corresponds to cavernous malformations with activating somatic PIK3CA variants rather than a fourth Mendelian gene; Orphanet associates somatic PIK3CA with the same ORPHA:221061 entity, but this association describes a lesional mechanism rather than a hereditary form [5, 7]. Spinal cavernous malformations belong to the familial spectrum and are addressed in the clinical and therapeutic sections.

Operationally, a clinical diagnosis of familial disease is established in either of two scenarios: multiple non-contiguous cerebral cavernous malformations generally not clustered around a DVA; or a cavernous malformation in an individual with at least one relative with a cavernous malformation [1, 2]. Molecular diagnosis requires a heterozygous germline pathogenic or likely pathogenic variant in KRIT1, CCM2 or PDCD10 [1, 2].

1.2 Historical overview

The hereditary nature of cavernous malformations became apparent with magnetic resonance imaging (MRI). In 1988, Rigamonti and colleagues described 24 patients with histologically confirmed cavernous malformation in the New England Journal of Medicine: 11 without a hereditary trait and 13 from six unrelated Mexican-American families [15]. Of 64 first- and second-degree relatives examined, 83% were asymptomatic and 11% had epileptic seizures; MRI, performed in 16 relatives (5 asymptomatic), showed cavernous malformations in 14, including 11 with multiple lesions [15]. The authors concluded that lesions were more prevalent than assumed, familial disease was more common than expected, with multiple lesions and clustering in Mexican-American families, and MRI was the technique of choice, far more accurate than computed tomography (CT) and angiography [15].

In 1994, Zabramski and colleagues published the natural history study of cavernous malformation families introducing the four-type imaging classification still used to describe lesions and their haemorrhagic risk [1, 16]. The GeneReviews reference list assigns PubMed identifier (PMID) 28113854 to this article; the correct identifier, confirmed in Europe PMC, is 8113854 [16].

Genetic mapping advanced in 1995. Dubovsky and colleagues linked a cavernous malformation gene to the long arm of chromosome 7 (7q11–q22) in a large Hispanic family, with a lod score of 4.2 at marker D7S804 and a shared 33 cM haplotype [17]. Günel and colleagues independently localised the gene to 7q11.2–q21 in two extended families (lod score 6.88 at D7S669), in a 7 cM region between ELN and D7S802 [18]. Johnson and colleagues refined the CCM1 locus location in the same year [19].

In 1999, two groups identified KRIT1 as the CCM1 gene. Sahoo and colleagues, by positional cloning, described seven mutations in 23 CCM1-linked families and an identical mutation in 16 of 21 Mexican-American families, demonstrating the Hispanic founder effect and opening the possibility of presymptomatic diagnosis in this population [20]. Laberge-le Couteulx and colleagues, in a French series, demonstrated that truncating KRIT1 mutations cause hereditary cavernous angiomas, estimated cavernous malformation prevalence at approximately 0.5% and recognised three loci (7q, 7p and 3q) [21].

The CCM2 gene was identified in 2003 by Liquori and colleagues, who found eight mutations in nine families in MGC4607, encoding a protein with a phosphotyrosine-binding (PTB) domain similar to ICAP1α, a known KRIT1 partner, linking the disease to integrin signalling [22]. In 2004, Denier and colleagues confirmed MGC4607 as CCM2 in a French series, estimating by genetic linkage that 40%, 20% and 40% of families were distributed across the 7q, 7p and 3q loci, and describing two large deletions removing exon 1 and eight point variants [23]. In 2005, Bergametti and colleagues identified PDCD10 (programmed cell death 10, also TFAR15) as the CCM3 gene from a 4 Mb de novo deletion at 3q26–27 and six deleterious variants in seven families, three de novo [24].

In 2009, Akers and colleagues and Pagenstecher and colleagues demonstrated that lesions in all three hereditary forms contain, in addition to the germline variant, a second somatic variant in the other allele of the same gene, present only in a subset of endothelial cells lining the caverns, establishing the two-hit mechanism and the need for complete loss of function of a CCM gene [25, 26]. From 2021, activating somatic PIK3CA and MAP3K3 variants were described in sporadic lesions and lesions in patients with a germline variant, acting as a third hit or an alternative lesion cause [1, 27].

Clinically, the Angioma Alliance (now Alliance to Cure Cavernous Malformation) published the first management guideline synopsis in 2017, reviewing literature through October 2014 [9]. The current guideline, published online on 21 May 2025 and in volume 98 of Neurosurgery in 2026, derives from a systematic review of 2672 publications (1 October 2014 to 15 March 2023), of which 234 were selected for mandatory consideration, and contains 53 recommendations approved by unanimous consensus through a Delphi process [2]. Recommendations use the American Heart Association/American Stroke Association (AHA/ASA) class and level system: 18 (34%) are class 1, 31 (58%) class 2 and 4 (8%) class 3; 3 (6%) have evidence level A, 19 (36%) level B and 31 (58%) level C, without formal assessment of study risk of bias [2]. The update adds explicit diagnostic criteria and specific recommendations for spinal lesions, children and older adults [2].

1.3 Nomenclature, recommendation classification and conventions

Gene nomenclature follows the HUGO Gene Nomenclature Committee (HGNC): KRIT1 (HGNC:1573; synonyms CCM1, CAM), CCM2 (HGNC:21708; synonyms MGC4607, OSM, C7orf22; the protein is also called malcavernin) and PDCD10 (HGNC:8761; synonyms CCM3, TFAR15) [3–5]. CCM1, CCM2 and CCM3 loci remain used to designate gene-defined clinical forms. Variants are described according to the Human Genome Variation Society (HGVS) on MANE Select (Matched Annotation from NCBI and EMBL-EBI) transcripts: NM_194454.3 for KRIT1, NM_031443.4 for CCM2 and NM_007217.4 for PDCD10 [6]. GeneReviews uses earlier transcripts (NM_194456.1 for KRIT1 and NM_031443.3 for CCM2); for variants cited in this article, KRIT1 c. numbering is identical in both transcripts [1, 6].

In the 2025 guideline, class 1 indicates a strong recommendation (benefit far outweighing risk), class 2 (2a and 2b) lesser or more uncertain benefit, and class 3 absence of benefit or risk similar to benefit; level A rests on multiple randomised trials or meta-analyses, level B on randomised (B-R) or non-randomised (B-NR) trials and level C on limited data (C-LD) or expert opinion (C-EO) [2]. The 2017 synopsis used earlier Roman numeral notation (class I to III, level A to C) [9]. Since 2008, symptomatic intracranial haemorrhage (ICH) has been defined as acute or subacute symptoms (headache, epileptic seizure, altered consciousness or a new or worsened focal neurological deficit, FND, referable to lesion location) accompanied by radiological, pathological, surgical or, rarely, CSF-only evidence of recent extra- or intralesional haemorrhage; simple lesion diameter increase without other evidence of recent haemorrhage and an isolated haemosiderin rim do not meet this definition [2, 9].

2. Molecular Genetics and Pathophysiology

2.1 Genes and proteins

The three genes encode proteins without intrinsic enzymatic activity that function as adaptors and molecular scaffolds of a ternary complex (CCM complex), essential for vascular development and maintenance of endothelial cell junctions [1, 2]. The table summarises each gene's characteristics.

KRIT1 functions as an anchor regulating cell structure and shear stress response through integrins, reactive oxygen species homeostasis, intercellular junction stability and vascular permeability in an inflammatory context, as well as activating angiogenesis through Delta-Notch signalling [1]. Its loss makes endothelial cells more stem-like, proliferative and invasive [1]. CCM2 contains a PTB domain binding KRIT1 and regulating its location, and scaffolds p38 MAPK and Rho-kinase signalling [1, 4]. PDCD10, whose overexpression activates caspase 3 and apoptosis, forms a complex with KRIT1 and CCM2 and participates in vascular endothelial growth factor (VEGF)-dependent vascular development and DLL4–Notch-mediated angiogenesis [1].

2.2 Molecular mechanism: biallelic loss, signalling and third hit

The disease mechanism is biallelic loss of function in lesion endothelial cells. The germline variant inactivates one allele in all cells; in each lesion, a second somatic variant inactivates the other allele of the same gene in a subset of endothelial cells lining the caverns, not interstitial cells [1, 25, 26]. Next-generation sequencing (NGS) of familial lesional tissue confirmed both mutated copies, and in some sporadic cases (not all) two somatic variants inactivate both copies of the same gene [1, 27]. A contribution from isolated haploinsufficiency is not excluded [1]. The disease's dynamic nature, with new lesions appearing throughout life, follows from this requirement for a second local somatic event.

Loss of the CCM complex removes regulation of MAP3K3 (MEKK3)-dependent signalling and leads to overactivation of RhoA and Rho-kinase (ROCK) in endothelial cells, disrupting intercellular junctions, increasing permeability and inducing a senescence-associated secretory phenotype with inflammation, migration and invasiveness [4, 28]. Specific (NRL-1049) or non-specific (fasudil, statins) ROCK inhibition reduces lesion burden in mouse models [28]. Barrier dysfunction is accompanied by iron deposition, inflammation and lesion growth, and circulating biomarkers such as inflammatory cytokines and plasma microRNAs associated with mutational status are under investigation [29].

Activating somatic PIK3CA variants have been found in sporadic cavernous malformations and lesions in patients with a germline variant in any of the three genes; phosphoinositide 3-kinase (PI3K) activation accelerates lesion growth when combined with CCM complex loss, acting as a third hit [1, 27]. Somatic MAP3K3 variants occur in sporadic lesions, in which overt haemorrhage appears less frequent than in lesions associated with familial disease genes [1]. In an Australian series, a somatic second hit in PIK3CA was identified in 4 of 7 lesions with available tissue, all from patients with a germline KRIT1 variant, and 8 of 10 sporadic lesions had recurrent somatic variants in PIK3CA, MAP3K3 or CCM2 [30]. These somatic variants exist only in the lesion, are not inherited, are absent from leucocytes and cannot be used for germline diagnosis [1].

In a Ccm3 mouse model without Pik3ca gain of function, rapamycin (an mTORC1 inhibitor) may have prevented exceptionally large lesions but increased mean lesion burden and animal mortality without altering haemorrhage, arguing against extrapolation of mTOR inhibition to familial disease without activated PIK3CA [31]. Current interpretation integrates these data into a genetic taxonomy of structural neurovascular lesions, in which somatic, frequently oncogenic variants constitute potential pharmacological targets [32].

2.3 Gut–brain axis and modifiers

In mice with endothelial CCM gene inactivation, increased intestinal permeability allows passage of Gram-negative bacteria and lipopolysaccharide, activating toll-like receptor 4 (TLR4) in brain endothelium and accelerating lesion formation; an emulsifier-rich diet, thinning the intestinal mucus barrier, increased lesion numbers [1, 2]. Intestinal epithelial PDCD10 loss reduces mucin production and facilitates bacterial translocation, potentially contributing to greater CCM3 severity [1]. In humans, faecal microbiome differences have been described between patients and controls and according to severity, without intervention studies [2]. Polymorphisms in vitamin D metabolism genes and low 25-hydroxyvitamin D levels were associated with more aggressive disease, and vitamin D3 restored endothelial integrity in CCM2-deficient cells in vitro [1, 2]. Genotype does not explain all clinical variability, and genetic and environmental modifiers are under study [9].

2.4 Inheritance pattern, penetrance and expressivity

Inheritance is autosomal dominant with incomplete, age-dependent penetrance, higher in adults and when susceptibility-sensitive sequences and higher magnetic fields are used [2]. Asymptomatic lesion frequency may conceal a family's dominant pattern [1]. In a French series of 64 families with a KRIT1 variant (202 heterozygotes), 62% were symptomatic, and 58% of those aged 50 or older had symptoms; of 53 asymptomatic heterozygotes, 45 had MRI lesions (three with type 4 lesions only) and 5 had neither clinical nor imaging findings, in a study performed without susceptibility-weighted imaging (SWI), underestimating imaging penetrance [1, 33]. PDCD10 penetrance may be lower than KRIT1, and CCM2 penetrance is undetermined [1, 34]. Expressivity varies within the same family, from asymptomatic carriers to fatal haemorrhage, as illustrated by the case in section 10 [35].

2.5 Gene proportions and variant spectrum

Distribution of identified variants and each method's yield are shown in the following table, based on European series and HGMD (Human Gene Mutation Database) [1, 36].

Most variants cause loss of function through nonsense, frameshift or splice variants introducing a premature termination codon and producing unstable messenger RNA, alongside deletions or duplications of one or more exons or the entire gene [9]. Variation in deletion contribution to CCM2 reflects the high prevalence of a founder exon 2–10 deletion in the United States (USA), rare in an Italian population; in cohorts without a sequencing-detectable variant, (multi)exon KRIT1 deletions were found in 5% in the USA, 4% in France and 50% in Italy [1, 37]. An Australian series of nine families with several affected individuals, with low standard analysis yield, identified germline KRIT1 or CCM2 variants in all nine after investigating small copy number variants (CNV) and deep intronic variants; in 3 of 9 these were KRIT1 exon deletions or splice variants, supporting inclusion of these variant types in panels [30]. In a large international cohort with genetic testing in 47.7% of 1592 patients, variants were distributed across KRIT1 in 31.0%, CCM2 in 4.8% and PDCD10 in 1.9% of the total, proportions influenced by centre composition [38].

2.6 Founder and recurrent variants

The table presents founder variants normalised using VariantValidator and classified in ClinVar (accessed October 2026), with historical nomenclature.

The KRIT1 c.1363C>T variant, known as the “Common Hispanic Mutation”, was found in 16 of 21 Mexican-American families in the original description [20] and explains most FCCM cases in Hispanic people from families with approximately 400 years of presence in the south-western USA and northern Mexico; GeneReviews indicates a frequency of 70% in this population [1, 2]. The CCM2 c.30+5_30+6delinsTT variant alters messenger RNA splicing and is recurrent in US Ashkenazi Jews, with cases in many metropolitan areas [2, 41]. The 2025 guideline writes it as “c.30+5_30_6delinsTT”, a typographical error corrected here. The CCM2 exon 2–10 deletion was described in at least 50 unrelated white families with ancestry in the eighteenth-century south-eastern USA, still concentrated in the Deep South; genetic genealogy reconstructed its founder origin, and GeneReviews indicates that up to 22% of individuals with FCCM in North America may carry this deletion, undetectable by sequencing [1, 2, 42]. ClinVar also lists CCM2 NM_031443.4:c.30+5G>T (VCV002759819) as pathogenic or likely pathogenic [39].

2.7 Genotype–phenotype correlations

CCM3 (PDCD10) is the most aggressive form. It presents earlier, frequently before 15 years and with haemorrhage, has a greater lesion burden, more de novo variants and is associated with skin lesions, scoliosis, spinal lesions, cognitive impairment and benign tumours (meningioma, vestibular schwannoma, astrocytoma) [9, 11, 34]. Mean ages at first symptomatic cerebral haemorrhage were 12.9 ± 11.6 years for PDCD10, 22.9 ± 13.9 years for KRIT1 and 38.5 ± 21.9 years for CCM2 [1, 34]. In a series of 18 patients with CCM3, the ICH rate was 20% per patient-year (95% confidence interval, 95% CI, 14–28%) from symptom onset and 24% (16–35%) for recurrence, but annual risk per lesion, 0.3% (0.2–0.4%), was similar to other genotypes, attributing higher individual risk to greater lesion burden [2]. In the international cohort of 1592 patients, CCM3 was associated with higher event risk than CCM1 (hazard ratio, HR, 1.799; 95% CI 1.008–3.208) [38]. In Shenkar's series of 18 patients with PDCD10, 39% had scoliosis and 5 had brain tumours: 2 meningiomas, 2 acoustic neuromas and 1 cerebellar astrocytoma [1, 11].

CCM1 (KRIT1) probably has the least severe course [9]; it is associated with spinal lesions in up to approximately 70% of carriers assessed with targeted imaging [3, 43], more skin lesions than CCM2 and PDCD10 [1] and, in a small study, more new lesion formation [1]. In carriers of the Hispanic c.1363C>T variant, cutaneous vascular malformation prevalence is 9–20% and the predominant morphological subtype differs from that observed with other variants [1, 44]. For CCM2, children with a CCM2 variant and positive family history had higher haemorrhage risk in a paediatric cohort of 41 patients [1, 45].

2.8 Allelic diseases and related entities

No other phenotypes associated with germline KRIT1, CCM2 or PDCD10 variants are known [1]. Cutaneous lesions (hyperkeratotic cutaneous capillary-venous malformations, punctate capillary malformations and deep blue nodules) and retinal lesions belong to the same spectrum and do not constitute distinct allelic disorders [1, 3]. The ClinGen actionability framework (AC116, cerebral cavernous malformations 1, 2 and 3) classifies actionability as limited in adult and paediatric contexts [3].

3. Clinical Spectrum and Manifestations by System

3.1 Lesions: number, size, location and histopathology

In familial disease, lesions are multiple, ranging from one or two to hundreds (typically 6 to 20), and counts depend on age and imaging technique; numbers tend to increase over time and each lesion may grow, shrink or change signal [1]. New lesions form at a rate of 0.2 to 1 lesion per person-year according to GeneReviews, and 0.4 to 2.7 new lesions per patient-year according to the 2017 synopsis, with the highest values in CCM3; interstudy differences partly reflect differences in sequence (SWI or gradient echo, GRE), slice thickness and field strength [1, 9]. Between 70% and 86% of lesions are supratentorial and 16–24% infratentorial, with almost half of the latter in the brainstem [1]. Young people with FCCM have a higher prevalence of white matter hyperintensities than patients with sporadic lesions, of unknown significance [1].

Histologically, caverns measure 2 to 55 mm (mean 8 mm), are lined by a single endothelial layer without mature vascular wall elements or intervening parenchyma, and show thrombosis and intra- and extralesional haemorrhage, with perilesional oedema when haemorrhage is recent [1]. The Zabramski classification, described in section 5, relates MRI appearance to the lesion's evolutionary stage [1, 16].

3.2 Frequency of manifestations

Disease manifests mainly between the second and fifth decades, incidentally or with epileptic seizures, FND, headaches or haemorrhage [1, 2]. The two overview sources give different percentages because they measure different populations and time points. The table presents both.

The internal GeneReviews discrepancy regarding focal deficits (25–30% in the table and 35–50% in the text) is not explained by the source; the table range agrees with the 2025 guideline [1, 2]. Up to half of heterozygotes are initially asymptomatic, but at least half of these have lesions on imaging [1]. In a genetically confirmed series, 20% of symptomatic patients presented before 15 years [1, 34].

3.3 Intracranial haemorrhage: natural history

Natural history data mainly derive from cohorts mixing sporadic and familial forms, and interpretation requires attention to population, denominator, outcome and follow-up period.

In Horne and colleagues' individual patient data meta-analysis (1620 people aged 16 or older, seven cohorts from six studies, sporadic and familial forms), 204 ICH occurred in 5197 person-years, and 5-year ICH risk was 15.8% (95% CI 13.7–17.9) [46]. Independent predictors were presentation with ICH or FND without recent haemorrhage (HR 5.6; 95% CI 3.2–9.7) and brainstem location (HR 4.4; 2.3–8.6); age, sex and multiplicity had no independent prognostic value [46].

In the Scottish population-based audit of adults diagnosed in 1999–2003 (139 patients, 1177 person-years, 97% completeness), 5-year risk of a first ICH was 2.4% (0.0–5.7) and recurrent ICH 29.5% (4.1–55.0); for the composite ICH or FND outcome, 9.3% (3.1–15.4) and 42.4% (26.8–58.0), respectively [47]. Annual recurrence risk fell from 19.8% (6.1–33.4) in year one to 5.0% (0.0–14.8) in year five and was greater in women [47]. With prolonged follow-up of the same audit (300 patients, diagnosis in 1999–2003 or 2006–2010, follow-up to 31 December 2023, 4779 person-years, median 15 years without treatment), 40 patients (13%) had ICH or FND, 72 (24%) became dependent (Oxford Handicap Scale, OHS, 3–5) and 53 (18%) died, 7 (2%) from lesion-related causes; recurrence risk was far higher than first-event risk (HR 8.66; 4.44–16.90), but declined markedly 5 years after the first event, leading the authors to recommend a 5-year horizon rather than extrapolation of annual rates throughout life [48]. In a Dutch single-centre cohort (Amsterdam, 2000–2023, conservative management, 265 adults), 19% had haemorrhage over a median 58 months, with an annual rate of 6.0% in those presenting with haemorrhage and 1.5% in the remainder; 7% died (0.8% from a lesion-attributable cause) and 77% remained independent [49].

The 2025 guideline systematic review (studies with at least 20 patients and 1 year of follow-up) found annual ICH rates of 0.7–7.5% per person-year overall, 0.2–4.9% for first ICH and 1.7–29.5% for second ICH; in a series of 42 patients with recurrent ICH, 5-year third-ICH risk was 66.7% (50–80%) [2]. A meta-analysis of seven studies (over 5081 person-years) estimated an annual rate of 2.5% (1.3–5.1%), and annual risk decreases over the first 5 follow-up years [2]. In incidental lesions, first-ICH risk is 0.08% per person-year [2]; in Scotland, the probability of an asymptomatic lesion becoming symptomatic was 0.25% annually (95% CI 0.22–0.29), approximately 1 in 400 annually [50].

In familial disease, annual ICH rates per patient are higher than for sporadic lesions: 4.3–6.5% in the 2017 synopsis and 4.3–16.5% in the 2025 guideline; per lesion-year, 0.6–1.1%, similar to sporadic forms [2, 9]. Historical estimates cited by GeneReviews range from 2.8% to 16.5% annually depending on prior haemorrhage, and large natural history studies showed no difference between familial and sporadic forms, with limited follow-up and without assessing lesion number and type [1, 16]. In the Brain Vascular Malformation Consortium (BVMC) project, 386 patients with familial disease and follow-up had a symptomatic ICH rate of 2.8 per 100 patient-years (95% CI 1.9–4.1): 4.5 (2.6–8.1) in those with ICH before enrolment and 2.0 (1.3–3.5) in the remainder; total lesion count was associated with subsequent ICH (HR 1.37 per doubling of count; 1.10–1.71), adjusted for prior ICH, age, sex and family structure [51]. GeneReviews reports this study with n=321 [1]; the article abstract indicates 386 patients with follow-up data [51].

Dammann and colleagues' retrospective international cohort (16 tertiary centres and a patient organisation, 1592 patients with confirmed or suspected familial disease, mean age 37.6 years, 55.7% women, follow-up from birth with median 42 years and 64,146 person-years) estimated a lifetime first symptomatic haemorrhage risk of approximately 80%, with a constant rate after 20 years, and lifetime first-seizure risk of approximately 45%; 775 patients (48.7%) had at least one haemorrhage and 447 (28.1%) at least one seizure, and both events were independently associated with worse disability on the modified Rankin Scale (mRS) [38]. The authors emphasise that the cohort is not population-based and tertiary-centre recruitment and retrospective collection tend to overestimate risk [38]. In contrast, in a French national reference centre cohort of 79 patients with familial disease asymptomatic at diagnosis (median age 41.7 years; 649.6 patient-years; median 4.0 years), 23 (29%) experienced a first clinical event, with event-free survival of approximately 97% at 1 year, 79% at 5 years and 69% at 10 years; incidence of any first event was 35.4 per 1000 person-years (FND 23.1; seizure 15.4), and male sex was associated with higher risk (HR 4.20; 1.60–11.04) [52].

In children with familial disease (41 patients), haemorrhage risk over 5 years was 5% per person-year, higher with a CCM2 variant, positive family history and prior symptomatic haemorrhage [1, 45]. In 18 children with exclusively type 4 lesions (KRIT1 or PDCD10, inherited or de novo), all asymptomatic, the mean haemorrhage rate was 1.3% annually over 5.5 years [1, 53]. In a series of 56 cases aged 25 or younger (14,936 records, 12 years), the overall rate was 1.6% per patient-year, 8.0% in symptomatic and 0.2% in incidental cases, without increased annual risk at young ages, although cumulative lifetime risk is probably higher because of longer life expectancy [9].

3.4 Epileptic seizures and epilepsy

Seizures are the most common symptom and are attributed to symptomatic haemorrhage or repeated microhaemorrhages, with haemosiderin deposition, gliosis and inflammation in adjacent cortex [2]. Cavernoma-related epilepsy (CRE, from cavernoma-related epilepsy) is more frequent with supratentorial cortical lesions, particularly temporal lesions, and risk may relate to the volume of displaced grey matter [2]. The 2017 synopsis adopted Rosenow's classification, reproduced in the table.

In the prospective cohort of 479 patients with familial disease, cumulative first-seizure incidence was 20% at 18 years (95% CI 17–23%) and 60% at 80 years (54–66%; 60.4% in the GeneReviews account); previous seizures were associated with more hospitalisations, and risk was higher with more lesions, larger lesions and, possibly, a PDCD10 variant [1, 2, 54]. In patients not presenting with seizures, prospective risk is approximately 1.2–1.5% per person-year, and a five-study meta-analysis found 1.5% per patient-year (1.1–2.2%), similar in familial and sporadic forms [1, 2]. In Josephson's prospective population-based study (139 adults), 5-year first-seizure risk was 6% (0–14%) in 38 patients presenting with ICH or FND and 4% (0–10%) in 57 with incidental presentation; among adults with epilepsy, 47% (27–67%) achieved 2 seizure-free years within 5 years [2]. In incidental lesions, seizure risk is below 1% annually [9].

In the Scottish audit (300 patients, median follow-up 15 years without intervention), 98 (33%) had seizures after any presentation, associated with younger age (adjusted HR 0.82 per 10 years) and seizure presentation (adjusted HR 12.94) [55]. Four in five patients with a first unprovoked seizure had recurrence within 10 years, justifying CRE diagnosis after a first seizure; of 110 patients with CRE, 82% (74–91%) had been seizure-free for 2 years and 62% (51–72%) for 5 years at 10-year follow-up, mainly on monotherapy [55, 56]. After CRE diagnosis, 50–60% become seizure-free with medication and approximately 20–25% develop drug-resistant epilepsy (failure of two appropriate drugs), more frequently with temporal lesions [2].

3.5 Headache

Headache is reported by up to 52% of patients, but primary headache is very common in the general population [2]. There is consensus that headache may be caused by haemorrhagic lesions near the pial surface or by hydrocephalus and is less likely with deep intraparenchymal lesions [2]. International Classification of Headache Disorders (ICHD-3) criteria for headache attributed to cavernous angioma are in section 7.

3.6 Focal neurological deficits and functional outcome

FND may be acute, subacute or chronic and generally result from haemorrhage or lesion growth [2]. To be considered symptomatic, they must be referable to lesion location [1]. In a series of 331 brainstem cavernous malformations (1985–2012), complete recovery (mRS 0) at 5 years was 37% without prospective haemorrhage, 18% with one and 11% with more than one; 30.3% recovered completely at 2 years, mainly within 18 months of the last haemorrhage [9]. In Scotland (1999–2003, 109 conservatively managed patients), 37% (28–46%) had a poor outcome (OHS 2–6 on at least two assessments) within 5 years [9, 57]. Functional deficit from cavernous malformation haemorrhage tends to be milder than with other vascular malformations [9].

The mRS is the most widely used measure but has interobserver variability, ceiling and floor effects and a focus on physical function; studies using SF-36, HADS (Hospital Anxiety and Depression Scale) and PROMIS (Patient-Reported Outcomes Measurement Information System) show worse outcomes than the reference population and frequent anxiety [2]. In 198 BVMC adults with familial disease, PROMIS-29 showed more anxiety (T 52.71), pain (52.50) and worse physical function (52.96), with better social participation (46.95) than the reference population; prior ICH was associated with worse anxiety, fatigue, physical function and social participation, and headaches with worse depression, fatigue, pain and sleep [58]. At a Rio de Janeiro reference centre (83 adults, 36 with familial disease), quality of life (PROMIS-29, EQ-5D-5L) did not differ between familial and sporadic forms, but in familial disease symptomatic patients had worse usual activities and physical function [59]. In the Treat_CCM trial, 31.7% of participants with questionnaire data had depression at baseline (BDI-II) and the SF-36 physical component was below the Italian population [60].

3.7 Spinal cavernous malformations

Spinal lesions manifest as radiculopathy or myelopathy with sudden, stepwise or slowly progressive onset (weakness, sensory changes, pain, sphincter dysfunction), according to Ogilvy's four presentation types [2]. In a study of KRIT1 carriers with standard imaging and GRE of the cervical and thoracic spine, 72.4% had at least one spinal lesion, most asymptomatic [3, 43]. In a small prospective series (75 patients with familial disease), 7 (9.3%) presented with symptomatic spinal haemorrhage and only 1 had prospective spinal haemorrhage [1, 61]. In a meta-analysis of 40 studies (632 intramedullary lesions of all causes), annual haemorrhage rate was 2.1% (95% CI 1.3–3.3), with an associated brain lesion in 17% and family history in 12% [9]. Spinal natural history data are scarce and subject to selection bias, and spontaneous recovery after acute haemorrhagic events is common [2].

3.8 Extraneurological manifestations

Cutaneous vascular malformations occur in 9–20% of patients [1]. In a consecutive series of 417 patients with familial disease, 38 had skin malformations: 13 capillary malformations, 15 hyperkeratotic cutaneous capillary-venous malformations (HCCVM), 8 venous malformations and 2 unclassified [1, 62]. Deep blue nodules and other subcutaneous nodules are venous malformation subtypes; some lesions were removed for bleeding, pain, protrusion, cosmetic reasons or suspected malignancy [1]. They are more frequent in KRIT1 variant carriers [1].

Retinal cavernomas and, rarely, choroidal haemangiomas occur in approximately 5% and do not usually bleed [1]. Hepatic haemangiomas were described in two Italian families with a CCM2 variant, without evidence of a higher frequency than the general population, and a renal angioma in one patient of Italian ancestry [1]. Atypical vertebral haemangiomas are bony capillary-venous malformations of vertebral bodies, T2 hyperintense but T1 hypointense (unlike common vertebral haemangioma), rarely causing pathological fractures [1]. Scoliosis, benign brain tumours and adrenal calcifications are described in CCM3, the latter presumed small vascular lesions with limited histological confirmation [1, 11].

3.9 Age-specific features

Approximately 25% of sporadic and familial cavernous malformations occur in childhood; in a series of 105 consecutive probands, up to 20% of familial index cases were under 10 years and 33% under 18 [2]. In children, discovery is often incidental during investigation of headache, head injury, developmental delay or a first seizure [2]. Cumulative seizure incidence is approximately 20% by the end of adolescence [54]. CCM3 is the form most frequently manifesting in childhood [2, 34]. In older adults, comorbidities and quality of life weigh more heavily in decisions, and surgical risks and outcomes are worse, particularly over 60 years and with brainstem lesions [2].

3.10 Natural history and treated disease

Unless otherwise stated, the figures above refer to untreated patients or those censored at intervention. Data on operated or irradiated patients describe treated disease, appear in section 7 and should not be compared directly with natural history data because surgical patients are selected according to severity and location [9, 63].

4. Clinical Recognition and Warning Signs

4.1 When to suspect

Suspicion of familial disease arises through one of three routes: imaging showing multiple lesions, family history and extraneurological manifestations. On imaging, suspicion should arise with several cerebral or spinal cavernous malformations not clustered around a DVA, or multiple punctate hypointense foci on susceptibility sequences in a young patient without vascular risk factors and with lesions in the brainstem, cerebellum, deep regions and cortex [1, 2]. Relevant family history includes seizures, FND, haemorrhagic strokes and abnormal MRI in relatives, and the 2025 guideline recommends recording a three-generation family history at diagnosis, including headaches, strokes, seizures or epilepsy, abnormal MRI, neurodevelopmental abnormalities and other neurological complications (class 1, C-EO) [2]. Absence of family history does not exclude disease, nor does a single lesion without family history [1].

Extraneurological manifestations warranting investigation are cutaneous vascular malformations (HCCVM, punctate capillary malformations and deep blue nodules), retinal cavernomas and choroidal haemangiomas, atypical vertebral haemangiomas and, in CCM3, scoliosis and benign CNS tumours (meningioma, acoustic neuroma, cerebellar astrocytoma) [1].

4.2 By age and setting

In children, suspicion arises with focal epileptic seizure, intracranial haemorrhage, headache or an incidental finding on MRI requested for head injury or developmental delay [2]. Approximately 75% of childhood ICH have an underlying vascular lesion, of which cavernous malformation accounts for a minority, according to the International Pediatric Stroke Organization and the AHA/ASA scientific statement on paediatric stroke, cited by the guideline [2]. Haemorrhage before 15 years, particularly with many lesions, points towards CCM3 [1, 34]. In a child with an apparently solitary lesion, distinguishing sporadic from familial disease requires a detailed history and GRE or SWI MRI to seek small punctate lesions and DVA [2].

In adults, the typical presentation is a first seizure, FND or haemorrhage between the second and fifth decades, or an incidental finding [2]. In Hispanic ancestry from the south-western USA or northern Mexico, the probability of familial disease with any cavernous malformation is high [2, 20]. In childhood cancer survivors with cranial irradiation, acquired cavernous malformations typically appear 2 to 20 years after radiation, and a new severe headache, seizure or FND warrants brain MRI (class 1, C-LD) [1, 2].

4.3 Warning signs and emergencies

Warning signs requiring urgent imaging are severe, prolonged or progressive headache; new or worsened FND; altered consciousness; worsening of previously controlled epilepsy; signs of hydrocephalus; and, in spinal lesions, spinal pain, weakness, sensory changes or sphincter dysfunction of sudden or progressive onset [1, 2]. Imaging should be performed as soon as possible after symptoms suggesting haemorrhage; CT is acceptable in emergencies, but MRI is definitive, ideally within two weeks of symptom onset [2]. During pregnancy, FND, acute severe headache or worsening seizures warrant non-contrast MRI, and ischaemic stroke and cerebral venous thrombosis should be excluded [1, 2].

4.4 Newborn screening

There is no newborn screening for FCCM or validated circulating biomarker [5]. Presymptomatic identification is through cascade testing in families with a known variant (section 8).

5. Diagnosis: Criteria, Tests and Algorithms

5.1 Diagnostic criteria

GeneReviews (2023 update) indicates that no published consensus clinical criteria existed and adopts an operational definition: a proband with multiple cavernous malformations, or one lesion and at least one relative with at least one lesion [1]. The 2025 guideline proposes explicit diagnostic criteria (class 2a, C-LD) [2]:

5.2 Imaging

Brain MRI is the only strongly recommended modality for diagnosis and follow-up (class 1, B-NR) and should include susceptibility-weighted sequences to determine whether there is one lesion or many (class 1, B-NR) [2]. CT shows hyperattenuation, sometimes calcification, but is non-specific and insensitive to small lesions; it is useful in emergencies [2]. Catheter angiography is generally not recommended except in the differential with arteriovenous malformation (AVM) (class 3, C-EO), and CT angiography and MR angiography have limited roles [1, 2].

The typical appearance is “popcorn” or mulberry-like on T2/FLAIR, with a hypointense haemosiderin ring; acute haemorrhage is better seen on T1 and may distort this appearance, so MRI should be repeated after resorption [2]. Very small lesions are seen only as susceptibility foci on SWI, overlapping with other-cause microbleeds and calcifications [2]. Enhancement of cavernous malformations and capillary telangiectasias does not imply tumour, but prominent enhancement suggests a haemorrhagic tumour such as melanoma or pleomorphic xanthoastrocytoma [2]. The 2017 synopsis recommended avoiding unnecessary gadolinium because of retention in the globi pallidi and dentate nuclei, of unknown clinical significance, and cautioned against ionising radiation from repeated CT [9].

The Zabramski classification describes four lesion types [1, 16]:

This division helps predict haemorrhagic risk, which is low in type 4 lesions (1.3% annually in 18 children) [1, 53]. A lesion with acute or subacute blood (T1 hyperintense), with or without oedema, represents a haemorrhagic lesion [1].

The MRI report should state field strength and sequences used; whether the lesion is solitary or multifocal; presence or absence of DVA; counts of larger lesions visible on T1/T2 and small lesions visible only on susceptibility imaging; and differential diagnosis [2, 9]. Lesions should be measured on spin-echo sequences (GRE exaggerates apparent size), with one diameter for round lesions and orthogonal measurements for asymmetrical lesions, detailing involvement of the brainstem, spinal cord, cranial nerves, cavernous sinus and intraventricular extension, as well as acute or subacute haemorrhage, extralesional haemorrhage and oedema [9]. Expressions such as “innumerable” or “too many to count” should be replaced by ranges (for example, “more than 100” or “approximately 20–30 small lesions”) or burden grading (mild, moderate, severe), because of psychological impact [2, 9]. Higher fields and thinner slices show more lesions, so differences between examinations should be interpreted cautiously, and multiple small lesions visible only on GRE or SWI are not necessarily associated with worse progression [9]. Automated counting and advanced techniques (functional MRI, tractography, quantitative susceptibility mapping, QSM, and dynamic contrast-enhanced MRI permeability, DCE-MRI) are under development [2, 9].

In the spinal cord, a lesion should be considered with neck or back pain, limb pain, paraesthesia or weakness unexplained by cerebral lesions, using a susceptibility-sensitive sequence in addition to T1 and T2 [2]. Spinal imaging should include the cervical and thoracic spine, preferably with gradient sequences [1]. The 2017 synopsis saw no justification for routine spinal MRI in patients with brain lesions without pain or myelopathic symptoms, and the 2025 guideline records that it is unclear whether detecting occult spinal lesions in asymptomatic individuals alters management [2, 9].

5.3 Genetic testing: indications and order

The 2025 guideline recommendations on genetic testing are as follows [2]:

The practical sequence is as follows. In individuals meeting clinical criteria, without prior focal radiotherapy or clustering around DVA, perform complete sequencing of the three genes with deletion/duplication analysis, sequentially or simultaneously, or a multigene panel [1, 2]. In people of Hispanic ancestry of Mexican origin or descended from Spanish settlers in the American Southwest, test first for KRIT1 c.1363C>T p.(Gln455Ter), and if negative, the remaining genes [2]. Targeted tests exist for the Ashkenazi CCM2 variant and CCM2 exon 2–10 deletion; GeneReviews allows targeted analysis first in Ashkenazi and Hispanic ancestry and considers that, in the US population, deletion/duplication analysis or chromosomal microarray (CMA) for CCM2 deletion may be first-line, whereas the 2025 guideline considers there is no strong reason to perform them first [1, 2]. Exome or genome sequencing is reserved for atypical phenotypes; most variants lie in coding regions [1]. In probands without a molecular diagnosis, DNA banking allows future reanalysis [1].

The result may guide management, particularly through the greater severity associated with CCM3 [2]. In asymptomatic relatives of a proband with a known variant, genetic testing is the best screening method, cost-effective and non-invasive; a negative familial variant test excludes the diagnosis without neuroimaging, and a negative MRI does not replace genetic testing, particularly at young ages [2].

5.4 Test sensitivity and limitations

With strict inclusion criteria (multiple lesions and/or compatible family history), a variant in one of the three genes is detected in at least 75% of families, and some studies report approximately 97% [1, 9]. These proportions describe yield in selected families and do not equal clinical sensitivity in a patient with one lesion. More than 5% of individuals with multiple lesions and/or family history have no identifiable variant in the three genes [1].

There are several known limitations. Sequencing does not detect deletions and duplications of one or more exons, such as the 77.6 kb CCM2 founder deletion or multiexon KRIT1 deletions, common in Italy [1]. Targeted deletion/duplication analysis (quantitative PCR, long-range PCR, MLPA, targeted microarray) detects from one exon to the entire gene but may not define breakpoints of large deletions or adjacent gene deletions; exome and genome sequencing may detect CNV by read depth, with lower sensitivity [1]. Deep intronic variants and small CNV escape conventional panels and accounted for additional diagnoses in an Australian series [30]. Alternatively spliced CCM2 exon 1B and the fact that the first coding PDCD10 exon is exon 4 should be covered by test design [1]. Parental somatic mosaicism may not be detected in blood, and exclusively germline mosaicism is undetectable [1]. Lesional somatic variants (second hit, PIK3CA, MAP3K3) are absent from blood and cannot serve for germline diagnosis [1]. A negative result therefore does not exclude familial disease when the phenotype is suggestive.

5.5 Report content and interpretation

The report should state the gene, reference transcript with version (preferably MANE Select: NM_194454.3, NM_031443.4 and NM_007217.4), HGVS c. and p. nomenclature, zygosity, classification according to American College of Medical Genetics and Genomics and Association for Molecular Pathology (ACMG/AMP) criteria, methods used (sequencing, deletions and duplications) and uncovered regions [1, 6]. A variant of uncertain significance (VUS) neither establishes nor excludes diagnosis; familial segregation, with testing and MRI of affected and unaffected relatives, may contribute to reclassification [1]. When comparing older reports, the change in KRIT1 transcript (NM_194456.1 in GeneReviews) should be considered, although c. numbering of cited variants is unchanged [6].

5.6 Initial evaluation after diagnosis

5.7 Prenatal and preimplantation diagnosis

When the familial variant is known, prenatal diagnosis and preimplantation genetic testing (PGT) are possible [1, 2]. Perspectives on using these tests in a disorder with incomplete penetrance and variable expressivity differ, and the decision is personal [1].

6. Differential Diagnosis

6.1 Comparative table

6.2 Familial versus sporadic disease

The most frequent and important differential diagnosis is the distinction between familial and sporadic disease, because it determines counselling, genetic testing, surveillance of relatives and radiosurgery decisions. Cavernous malformations account for 5–15% of all cerebral vascular malformations, and approximately 80% of all cases are sporadic and 20% familial, estimates affected by incomplete penetrance and variable expressivity [1, 2]. A sporadic lesion is typically solitary and associated with a DVA in up to 30% of cases in some series; on 7 T MRI, all sporadic lesions studied had anomalous venous drainage, and multiple sporadic lesions are always near a DVA, which may act as an anatomical and genetic lesion “initiator” [1]. DVA is rare in familial disease [1]. Several lesions clustered around a single DVA represent sporadic disease, whereas multifocal lesions without DVA suggest familial disease [2, 9].

6.3 Large lesions and type 4 lesions

With a large lesion and acute haemorrhage, typical features may be absent and the haematoma may conceal a cavernous malformation, haemorrhagic mass or small AVM; repeat imaging after resorption clarifies most cases [1, 2]. The main mimics are haemorrhagic metastases, with greater enhancement and oedema, frequently calcified primary tumours, AVM and atrial myxoma emboli [1].

With multiple susceptibility foci without a “large” lesion, the differential includes microbleeds from amyloid angiopathy, hypertensive angiopathy, trauma, vasculitis and radiotherapy, calcifications and neurocysticercosis [1, 2]. Familial disease is favoured by a family history of seizures, FND or haemorrhages, distribution across the brainstem, cerebellum, deep and superficial regions (rather than superficial regions alone) and age 45 years or younger; genetic testing helps resolve uncertain cases [1]. Without typical lesions or family history, in a patient with ageing and vascular risk factors, tiny haemorrhagic lesions are more likely to represent microbleeds [2].

6.4 Radiation-induced cavernous malformations

CNS irradiation is associated with new lesions both in the general population and in familial disease, and radiation-induced lesions differ histologically from those present before radiation [1]. Radiotherapy history is the discriminating feature, and the 2025 guideline excludes from genetic testing patients with multiple lesions attributable to prior focal radiotherapy [2]. In children, lesions may appear within the first decade after cranial radiotherapy; the 2017 synopsis gives a dose containing an obvious typographical error, which is therefore not reproduced [9].

6.5 Hereditary haemorrhagic telangiectasia

Hereditary haemorrhagic telangiectasia (ENG, ACVRL1) is another autosomal dominant vascular disorder with CNS involvement, with cerebral vascular abnormalities in approximately 10% and spinal abnormalities, and cavernous angiomas listed as an occasional abnormality; it is distinguished by recurrent epistaxis, mucocutaneous telangiectasias and visceral AVM [64].

7. Treatments and Risk Management Strategies

7.1 General principles

There is no authorised disease-modifying pharmacological treatment, and management rests on non-randomised epidemiological, surgical and radiosurgical studies and expert opinion [1–3]. The 2025 Alliance to Cure Cavernous Malformation guideline is the current reference; there is no dedicated European Stroke Organisation (ESO) guideline, and the 2025 ESO/EANS (European Association of Neurosurgical Societies) guideline on spontaneous intracerebral haemorrhage contains no specific cavernous malformation recommendations in the text obtained [2, 65]. In Europe, an international Delphi consensus published in the EANS journal (19 experts, three rounds, agreement of at least 70%) broadly agrees with the 2025 guideline [66]. Decisions should be made by experienced multidisciplinary teams (neurology, neurosurgery, neuroradiology, genetics, epileptology) and shared with the patient.

7.2 Acute haemorrhage treatment

ICH or intraventricular haemorrhage from a cavernous malformation is managed according to general ICH guidelines: early blood pressure control, reversal of coagulopathy, intracranial pressure control and evacuation of haemorrhage with impending herniation or posterior fossa mass effect [9]. Hydrocephalus from unresectable midbrain lesions compressing the aqueduct, or from ventricular compression, may require shunting or endoscopic third ventriculostomy (ETV) [1]. MRI should be performed early, ideally within two weeks, and repeated after haematoma resorption to characterise the lesion [2].

7.3 Microsurgery

Complete resection may cure a single lesion and eliminates that lesion's haemorrhagic risk, but in familial disease does not prevent new lesions [1, 2]. Usual indications are a first haemorrhage in a surgically accessible lesion, recurrent haemorrhage with increasing disability, high deterioration risk in less accessible lesions, intractable epilepsy and mass effect with neurological decline [1, 9]. There is no evidence of different surgical outcomes between solitary and multifocal or familial lesions, although resection is more likely to be curative for a solitary lesion [2].

The balance depends on location. In a lobar lesion in a non-eloquent area, surgery after the first symptomatic haemorrhage has a relatively low permanent complication risk, equivalent to living with the lesion for 1–2 years [2, 9]. In a symptomatic lesion in an eloquent area (primary sensorimotor cortex, dominant hemisphere language areas, primary visual cortex), resection may be offered after the first haemorrhage if the lesion is accessible and risk low [2]. For deep lesions (insula, basal ganglia, thalamus), postoperative morbidity is 5–18% and mortality approximately 2%, with risk equivalent to living with the lesion for 5–10 years [9]. In the brainstem, transition from independence to dependence generally occurs after the second haemorrhage, and early morbidity affects almost half of operated patients, although most recover; conservative management after the first haemorrhage and surgery after the second constitute a reasonable alternative [2, 9]. A systematic review estimated death or non-fatal stroke risk after resection at 6%, versus 2.4% over 5 years of natural history for a lesion that had never bled and 29.5% over 5 years of recurrence after a first haemorrhage [9].

Comparative data for surgery and conservative management are observational. In the Scottish population-based study, excision was associated with greater risk of short-term disability, symptomatic ICH and new FND over 5 years, without adjustment for location, preoperative decline or technical advances [1, 57]. In prolonged follow-up of the same cohort (306 patients, 190 symptomatic, median 16 years), 37 (19%) were treated (35 by microsurgery and 2 by radiosurgery), were younger and had fewer brainstem lesions; intervention was not associated with persistent OHS ≥2 disability but was associated with OHS ≥3 (adjusted HR 2.66; 95% CI 1.19–5.94) and lesion- or intervention-related ICH or FND (HR 3.59; 1.05–12.20), without association with seizures [63]. The authors conclude that a randomised trial is needed.

The 22 surgical recommendations of the 2025 guideline appear in the table [2].

Incidental lesions have minimal haemorrhagic risk and prophylactic surgery is generally unnecessary [2]. Many surgeons avoid dissecting the associated DVA because of oedema, haemorrhage and venous infarction risks, and there are no controlled recurrence data with or without DVA resection [2]. The European–international Delphi consensus agreed not to resect DVA, to resect the haemosiderin ring only with epilepsy, to observe asymptomatic eloquent or deep lesions, to operate on deep lesions when symptomatic or after a second haemorrhage and not to consider imaging-only asymptomatic haemorrhages a surgical indication [66].

In children, longer life expectancy and plasticity make surgery a reasonable first-line option for symptomatic, growing or haemorrhagic lesions in non-eloquent areas, with better seizure control and high neurological recovery; higher-risk locations (basal ganglia, thalamus, cerebellum, brainstem, spinal cord) require individual assessment [1, 2]. In older adults, operative risks and outcomes are worse, particularly above 60 years and in the brainstem [2]. In the spinal cord, surgical risk depends on axial location, baseline function and team experience; operated patients in published series are selected (more deficit, shorter course, dorsal or dorsolateral location) [2].

Immediate postoperative MRI (within 72 hours) has a positive predictive value of 67% and negative predictive value of 97% for complete resection [2]. Long-term residual or recurrent lesions occur in 11–23% (more in children); delayed postoperative haemorrhage occurred in 5.6% of operated patients and 29% of residual or recurrent lesions, with mean follow-up of 5.6–6.7 years, and 58% of recurrences or remnants were reoperated [2]. Routine follow-up for approximately 5 years is reasonable, and paediatric guidance suggests annual MRI [2].

7.4 Stereotactic radiosurgery

Radiosurgery is not recommended in familial disease because of concern about radiation-induced new lesion genesis (class 3, C-LD in 2025; class III, level C in 2017) [1, 2, 9]. Its utility is restricted to solitary lesions with prior symptomatic haemorrhage in eloquent areas with unacceptable surgical risk [2]. A comparative meta-analysis estimated perioperative mortality of 1–2%, postoperative haemorrhage 3%, long-term morbidity 11% and efficacy 97% for microsurgery, versus periprocedural death 1%, haemorrhage 14%, morbidity 10% and efficacy 86% for radiosurgery, placing microsurgery first-line [2]. Radiosurgery is more effective when applied early after haemorrhage, with a marginal dose above 16 Gy, in non-brainstem lesions smaller than 3 cm³ and without DVA; in a small paediatric series, a marginal dose of 12 Gy may have sufficed [2]. The effect on rehaemorrhage emerges mainly after up to 2 years, a period when natural history also shows declining risk, fuelling debate about true efficacy; a systematic review estimated a haemorrhage rate of 1–2% annually after treatment and complications in approximately 2–4% [1, 46, 67]. In the brainstem, four of five studies showed reduced annual rates 2 years after radiosurgery, with mortality of 5.61% and new FND in 11.8% [9].

7.5 Laser interstitial thermal therapy

LITT has been used in drug-resistant epilepsy associated with cavernous malformation, with a haemorrhage rate of 3.1% and freedom from disabling seizures (Engel 1) in 78–83%, comparable to microsurgery [2]. In a large individual patient data meta-analysis, all LITT-treated lesions with recurrent seizures subsequently underwent microsurgery; series are single-centre and subject to selection bias, and LITT prevents intraoperative electrocorticography [1, 2].

7.6 CARE randomised surgical trial

The comparison between intervention and conservative management was the James Lind Alliance's first cavernoma priority [68]. The open-label CARE pilot trial (Cavernomas: A Randomised Effectiveness; ISRCTN41647111; NIHR128694), with blinded assessment, compared medical management plus intervention (neurosurgery or radiosurgery) versus medical management alone in children or adults with unresected symptomatic cavernoma in the United Kingdom [68, 69]. Of 511 patients screened at 28 of 40 centres, 322 (63%) were eligible, 202 were approached, 96 (48%) expressed uncertainty and 72 were randomised (median 51 years; 57% women; 92% white; 78% with prior ICH; 39% with prior seizures): 36 to intervention (12 neurosurgery, 24 radiosurgery) and 36 to medical management [69]. Retention was 93% and adherence 91%; the primary outcome (symptomatic ICH or persistent or progressive FND) occurred in 2 of 33 and 2 of 34 patients, without deaths or serious adverse events [69]. A definitive trial would require 590 to 1900 participants [69]. The pilot was not designed for efficacy and does not allow comparative conclusions.

7.7 Epilepsy management

A first seizure attributed to a cavernous malformation warrants antiseizure medication (ASM) (class 1, B-NR), because in definite CRE recurrence risk after the first unprovoked seizure exceeds 90% within 5 years; drug choice is individualised [2]. Electroencephalography (EEG) is indicated when there is doubt about the epileptic nature of episodes, their relationship to the lesion or the responsible lesion in multifocal disease, sometimes with prolonged ambulatory recording or monitoring unit admission [2]. Before surgery in a patient with seizures and a non-haemorrhagic lesion, particularly in familial disease, neurologist review and EEG are recommended to confirm the relationship and choose the lesion to resect or ablate (class 1, C-EO) [2].

For a first seizure without haemorrhage, the initial approach is conservative with ASM; surgery may be considered after the first seizure if the patient wishes to discontinue medication or cannot tolerate it [2]. Lesionectomy, with excision of haemosiderin-stained tissue whenever possible, was associated with 88% sustained seizure freedom (at least 2 consecutive years, International League Against Epilepsy, ILAE, class 1) versus 32% with initial medical treatment alone, in class 3 evidence [2]. The 2017 synopsis reported 70–90% control after lesionectomy when seizures were sporadic or of less than 1 year's duration, with worse control after longer preoperative duration, and some authors advocate early surgery after one drug fails [9]. For persistent seizures despite ASM, epilepsy centre assessment is recommended and surgery may be considered even without strict drug-resistance criteria, because of additional haemorrhage protection [2]. Controlled epilepsy is followed annually in neurology; worsening of previously controlled epilepsy warrants MRI, EEG and possible neurosurgical assessment [1].

Patients should avoid drugs and activities increasing seizure or injury risk, comply with local driving legislation and consider ASM interactions with other drugs, contraception, pregnancy and vitamin D metabolism [2].

7.8 Headache management

Headache without a secondary cause should be classified and treated according to usual practice (class 1, C-EO) [2]. The ICHD-3 criteria for headache attributed to cavernous angioma, reproduced in the guideline, are as follows [2]:

Migraine coexisting with non-haemorrhagic lesions is treated with standard therapy [2]. In a study of 329 patients with cerebral or spinal lesions, non-aspirin non-steroidal anti-inflammatory drugs (NSAID), triptans and onabotulinumtoxinA (always at doses below 200 U) did not increase haemorrhage risk, without assessment of duration of use; one haemorrhage after onabotulinumtoxinA at doses exceeding standard migraine doses has been reported, and there are no safety data on calcitonin gene-related peptide (CGRP) inhibitors [2]. Opioids are discouraged for chronic pain because of addiction and medication-overuse headache risks [1]. Severe, prolonged or progressive headaches, or those accompanied by new or worsening deficits, require urgent imaging [1].

7.9 Rehabilitation and persistent deficits

Rehabilitation of FND rests on extrapolation from haemorrhagic stroke data [2]. Stable deficits benefit from physiotherapy and rehabilitation; after recovery from brainstem haemorrhage or surgery has stabilised, persistent peripheral facial palsy and diplopia may be treated with facial reanimation and strabismus surgery in selected cases [1, 2].

7.10 Drugs with a role in management: dosing and evidence

No drug has an authorised FCCM indication. Propranolol and atorvastatin, when used to reduce haemorrhagic risk, are off-label uses in any jurisdiction, and the doses below are trial doses, not approved doses for this disease [2, 70, 71].

Propranolol

Propranolol is used in infantile haemangiomas, with similar histopathology, and cavernous malformation shrinkage during treatment has been reported [2]. In animal models, it inhibits malformations through β1 receptor antagonism and reduces lesion development, restoring barrier function [74, 75]; another preclinical study with discordant results recommended caution [76]. Cohort results are mixed: Goldberg and colleagues found no association between β-blockers (any, β1-selective or non-selective) and haemorrhagic risk [2], whereas the Scottish audit (300 adults, 3634 person-years) found lower risk of new ICH or persistent or progressive FND in 63 β-blocker users (27 receiving propranolol) (adjusted HR 0.09; 95% CI 0.01–0.66) [77].

Treat_CCM (EudraCT 2017-003595-30; NCT03589014) was a phase 2 randomised open-label pilot study with blinded outcome assessment (PROBE design) at six Italian national rare disease reference centres [13, 72]. Between 11 April 2018 and 5 December 2019, 95 patients were assessed and 83 adults with symptomatic familial disease were randomised 2:1 to propranolol 20–320 mg/day plus standard care (57) or standard care alone (26) for 24 months, without placebo; mean age was 46 years (standard deviation 15) and 48 (58%) were women [13]. The primary outcome (new symptomatic ICH or lesion-attributable FND within 24 months) occurred in 2 of 57 (4%) patients in the propranolol group, 1.7 per 100 person-years (95% CI 1.4–2.0), and 2 of 26 (8%) in the control group, 3.9 per 100 person-years (3.1–4.7), with univariable HR 0.43 (one-sided 80% confidence interval, 0.18–0.98), considered an efficacy signal according to prespecified pilot trial criteria [13]. Hospitalisations were equal (8.2 per 100 person-years), there was one sepsis death in the control group and three propranolol discontinuations for adverse effects [13]. There was no difference in seizures, new lesions or serious events, and absolute numbers were small [2]. In patient-reported outcome analysis (71 participants with questionnaires), the proportion with depression at 2 years was lower with propranolol (28.6% vs 55.5%; p=0.047), without effect on anxiety or SF-36 [60].

The 2025 guideline describes Treat_CCM as a placebo comparison, with 95 patients and a dose from 10 mg twice daily to 160 mg twice daily [2]. The primary article shows that the trial was open-label and controlled with standard care, without placebo, with 83 patients randomised among 95 assessed, and expresses the dose as 20–320 mg/day [13]. The data presented here follow the primary article. The authors considered a phase 3 trial justified [13], but no phase 3 registration of propranolol in FCCM was located in ClinicalTrials.gov or the European trials register in October 2026, which does not demonstrate that none exists [78]. Another study, on propranolol and messenger RNA expression (NCT03474614, phase 2), ended with 4 participants [78]. In 2017, the Angioma Alliance synopsis considered that propranolol could not be recommended, because of the lack of controlled studies; in 2025 it became a class 2b, B-R recommendation [2, 9].

Statins

Statins indirectly inhibit Rho-kinase, a regulator of endothelial permeability, and reduced lesion burden in mice [1, 2]. Cohort results are mixed, with protection when combined with aspirin or a neutral effect, and limited by heterogeneity of dose, type and duration [2]. In the Scottish audit, statins (73 users) were associated with HR 0.37 (95% CI 0.01–1.07), without statistical significance [77]. A seven-study observational meta-analysis (2524 patients, 315 receiving statins; search through 27 August 2025) found lower prospective (re)haemorrhage risk in adjusted analyses (HR 0.419; 95% CI 0.186–0.942), without difference in haemorrhagic presentation [79]. In a pilot with 10 patients randomised to simvastatin 20–40 mg/day for 3 months or no treatment, DCE-MRI permeability was greater in lesions than white matter, without between-group difference [2]; a simvastatin and permeability study in New Mexico CCM1 (NCT01764451) ended with 12 participants [78].

AT CASH EPOC (NCT02603328), whose protocol was published in 2019, was a phase 1/2a single-centre (University of Chicago) study, randomised 1:1, double-blind and placebo-controlled, stratified by sex [12, 80]. It included adults aged 18–80 years with untreated solitary or familial cavernous malformations and adjudicated symptomatic haemorrhage within the previous year, excluding prior cranial irradiation and statin use in the previous year; intervention was atorvastatin 80 mg/day for 2 years [12]. The primary outcome was mean annual percentage change in lesional QSM, an iron deposition biomarker whose letter of intent for qualification as a drug development tool was accepted by the FDA (Drug Development Tool, letter of intent 000127) [12]. Between 25 July 2018 and 22 July 2022, 80 patients were randomised (41 atorvastatin, 39 placebo; median age 41 years; 51 women, 64%; 44, 55%, with a brainstem haemorrhagic lesion; 29, 36%, with familial or multifocal disease), with median follow-up of 723 days [12]. The article also states “29 (64%) women”, an internal inconsistency resolved by calculating 51/80.

In modified intention-to-treat analysis (64 patients), annual QSM change was 10.88% (standard error 7.29) with atorvastatin and 12.09% (7.54) with placebo, with an effect of −1.22 (95% CI −22.25 to 19.81; p=0.91), without difference by year, in prespecified subgroups or with imputation [12]. Symptomatic haemorrhage occurred in 6 versus 7 patients (relative risk 0.81; 95% CI 0.31–2.13), always in the index lesion, without difference in subclinical haemorrhages; no serious adverse event required discontinuation and there were no deaths [12]. The authors concluded that atorvastatin is safe after recent haemorrhage but cannot be recommended to prevent rehaemorrhage; most participants took vitamin D and avoided sex hormones, and lower-than-expected haemorrhage rates may have diluted an effect [12]. Non-significant QSM reduction in men, brainstem lesions and sporadic forms is hypothesis-generating only [12]. Subsequent exploratory analyses assessed functional and patient-reported outcomes and lesion volume as an outcome [81, 82].

7.11 Drugs and exposures modifying haemorrhagic risk

Antithrombotics

There is theoretical concern about antithrombotics, thrombolytics and other drugs affecting coagulation (fish oil, vitamin E, selective serotonin reuptake inhibitors, SSRI, NSAID), because of endothelial permeability [2]. Observational data, however, point in the opposite direction. In the Scottish audit (300 adults), 61 patients used antithrombotics (10, 16%, anticoagulation) for a mean of 7.4 years, and ICH or FND occurred in 1 of 61 (2%) versus 29 of 239 (12%) (adjusted HR 0.12; 95% CI 0.02–0.88) [83]. In the six-cohort meta-analysis (1342 patients), ICH occurred in 8 of 253 (3%) with antithrombotics and 152 of 1089 (14%) without them (incidence rate ratio 0.25; 0.13–0.51; I² 0%) [83]. The 2025 guideline attributes the mean 7.4-year period to the meta-analysis; in the article it corresponds to mean use duration in the Scottish cohort [2, 83]. Most patients in these cohorts took aspirin rather than anticoagulants, and it is unknown whether the effect is antithrombotic or anti-inflammatory; direct oral anticoagulant data are limited [2]. A meta-analysis and network meta-analysis of nine single-centre cohorts (2709 patients, 388 with antithrombotics) found lower probability of ICH presentation (odds ratio, OR, 0.56; 0.45–0.70), with hypothesis-generating conclusions only [84].

In a divergent finding, a retrospective multicentre propensity score-matched cohort from an administrative database (TriNetX; adults without prior ICH; 1540 per group) found more ICH with antithrombotics (6.3% vs 3.9%; HR 1.77; 95% CI 1.28–2.44), mainly attributed to anticoagulants, as well as more seizures, readmissions and emergency visits [85]. The available abstract does not distinguish sporadic from familial forms, and administrative data limit haemorrhage attribution [85]. The 2025 guideline considers prospective haemorrhage risk low in patients needing antithrombotics and recommends that cavernous malformation should not prevent their use when indicated (class 2a, B-NR) [2]. In 2017, the recommendation was class III, level C because of scarce data, and the synopsis allowed resection of an accessible solitary asymptomatic lesion in patients potentially needing anticoagulation, a reason absent in 2025 [2, 9]. GeneReviews, predating the guideline, advised avoiding antithrombotics except for life-threatening thrombosis, allowing consideration after individual assessment [1].

Thrombolysis in ischaemic stroke

Data are limited. In a study of 967 patients thrombolysed for presumed ischaemia, 20 had at least one cavernous malformation, and only 1 of 11 patients had evidence of lesional haemorrhage after thrombolysis, probably the cause of symptoms [1, 86]. In the 2025 guideline account, comparison of 9 patients with a lesion and 341 without showed no difference in symptomatic ICH or parenchymal haemorrhage (one ICH within the infarct territory, distant from the lesion, and one subacute haemorrhage expansion in a lesion whose aphasia was attributed to a seizure) [2]. Schwarzbach and colleagues found no significant increase in 13 patients [2]. The 2025 CCM guideline considers that, in acute ischaemic stroke with an unruptured untreated cavernous malformation, thrombolysis risk is unestablished and intravenous alteplase may be considered when indicated (class 2b, C-LD), in line with the 2019 AHA/ASA recommendation on unruptured intracranial vascular malformations: intravenous thrombolysis permissible for severe deficits with high morbidity and mortality probability outweighing anticipated haemorrhage risk (class IIb, level C-LD) [2, 87]. The 2019 AHA/ASA guideline was superseded by the 2026 early ischaemic stroke management guideline, which changed the approach to thrombolysis contraindications and addresses special situations in a table (Table 8) that could not be obtained in a readable format; therefore, the 2026 AHA/ASA position on unruptured vascular malformations is not described here [88].

Sex hormones and fertility drugs

Oestrogen is prothrombotic, and intracavernous thrombosis with poor venous outflow may promote extralesional haemorrhage; progesterone may also influence the CCM complex [2]. In two multicentre prospective cohorts (Mayo Clinic and Essen; 722 women aged 10 or older at diagnosis, sporadic and familial forms; 137 receiving hormone therapy), ICH occurred in 46 of 137 (33.6%) with hormones and 91 of 585 (15.6%) without, over up to 5 years (adjusted HR 1.56; 95% CI 1.09–2.24); in women aged 10–44 years taking oral contraceptives, adjusted HR was 2.00 (1.26–3.17) [89]. Dose, duration and route (oral, vaginal, intrauterine device) were not recorded, and progestogen-only intrauterine devices were not included; risk was higher with oral contraception combined with smoking, and the contribution of third factors such as factor V Leiden is being investigated [2]. The 2025 guideline recommends caution with exogenous oral oestrogen and progesterone (class 3, C-LD), and fertility drug data are limited [2].

Anti-inflammatory drugs, SSRI and others

GeneReviews generally advised avoiding NSAID, without an absolute contraindication in asymptomatic individuals [1]; the 2025 guideline allows non-aspirin NSAID with caution in non-haemorrhagic lesions (class 2b, C-LD) [2]. Previch and colleagues associated fish oil and SSRI with less haemorrhage, and in one cohort SSRI use was associated with less haemorrhage after adjustment, with uncertain mechanism (anxiolytic or antithrombotic effect) [2].

Ionising radiation

CNS irradiation is associated with new lesions in familial disease [1]. Repeated CT and dental radiographs in carriers, particularly children, are also concerning [2, 9].

7.12 Non-pharmacological interventions and lifestyle

The 2025 guideline recommendations on lifestyle and comorbidities are: aerobic activity (class 1, C-LD); avoid binge drinking (class 1, C-LD); consider a diet low in processed foods and emulsifiers because of intestinal permeability's role (class 2a, C-EO); assess and treat hypertension according to standards, with limited data on its role as a risk factor (class 1, A, derived from studies outside CCM); and assess anxiety and refer to appropriate resources (class 2a, C-EO) [2].

Smoking, analysed as a binary variable, was not associated with haemorrhage in several cohorts, but in one familial cohort pack-years showed a trend towards more haemorrhage, and combination with hormones increased risk compared with hormones alone, so smoking is discouraged [2]. No studies link alcohol to haemorrhage, but recent excessive consumption before haemorrhages has been reported [2]. Hypertension, analysed as a binary variable, did not increase haemorrhagic risk in most cohorts, consistent with the non-arteriolar nature of haemorrhages; in one study, lower systolic pressure was associated with lower lesion counts and a trend towards less haemorrhage in familial disease [1, 2]. Physical activity was not related to haemorrhage timing in one study, particularly aerobic exercise and light weights [2].

The 2017 synopsis listed theoretical risk situations, without disease-specific evidence, that continue to inform counselling [9]:

In animal models, an emulsifier-rich diet increased lesions; there are no human trials, and a diet low in processed foods appears reasonable, without data on probiotics or prebiotics [1, 2]. Anxiety is more frequent than in the general population and should be assessed [2].

7.13 Neurological management recommendations in the 2025 guideline

The table compiles the 21 neurological and medical management recommendations of the 2025 guideline [2].

Regarding vitamin D, the inverse relationship between 25-hydroxyvitamin D and aggressive behaviour (early onset, two or more symptomatic haemorrhages or high burden) and the association between low levels and haemorrhage at presentation derive from cohorts [1, 2]. In a cohort of 364 patients (2018 patient-years), supplementation was associated with 70% less prospective haemorrhage, with variable doses and formulations and no adjustment for obesity, sun exposure or diet [2]. The guideline considers a 25-hydroxyvitamin D level below 20 suboptimal for bone health, without stating the unit in the text, and identifies some ASM, northern latitudes, body mass index above 30 and little sun exposure as deficiency risk factors [2]. In 2017, the synopsis considered that there was no evidence vitamin D prevented haemorrhage and statins should be limited to approved indications [9].

7.14 Regulatory status

No drug is FDA- or EMA-authorised for FCCM, and no marketing authorisation application for this indication was located through October 2026 [2, 78]. The EMA granted four orphan designations, recognising rare disease status and product interest but not constituting authorisation: at designation, FCCM affected no more than 3 in 10,000 people in the European Union (EU), a regulatory threshold rather than a precise epidemiological estimate [90, 91]. FDA orphan designations were not consulted.

Generic atorvastatin and propranolol have national EU authorisations without a centralised European public assessment report; national summaries of product characteristics were not consulted [2]. Since no medicine is authorised for the indication, funding assessments by INFARMED or Brazil's National Commission for the Incorporation of Technologies into SUS (CONITEC) do not apply.

7.15 Clinical trials

These trials do not allow indirect comparisons: they differ in population (symptomatic familial disease in Treat_CCM, recent haemorrhage with predominantly sporadic forms in AT CASH EPOC), design (open-label versus double-blind), outcome (clinical versus imaging biomarker) and duration. Treat_CCM's HR of 0.43 rests on 2 events in each arm (2 of 57 versus 2 of 26), with a one-sided 80% confidence interval appropriate to a pilot, and does not demonstrate efficacy. The negative atorvastatin trial contrasts with favourable statin associations in cohorts and the observational meta-analysis, illustrating observational studies' limitations for inferring causality [12, 79]. Absolute haemorrhage rates were low in all trials, requiring large samples to detect modest relative reductions.

7.16 Developments in the last 6 to 12 months and investigational therapies

The 12-month follow-up after AT CASH EPOC completion, published in 2026, recorded symptomatic haemorrhage in 4 patients at 3, 49, 84 and 225 days after atorvastatin withdrawal, versus 1 patient at 395 days after placebo withdrawal, all in the index lesion and 3 in patients discontinuing more than 2 years after their last haemorrhage [73]. Haemorrhage-free survival was lower after atorvastatin withdrawal (log-rank χ² 4.136; p=0.042), but the cohort was small, blinding was partially broken in 5 patients, there were more resections in the former placebo group and no systematic subclinical haemorrhage screening [73]. The authors propose considering the risk of stopping high-dose atorvastatin in this context; it is unknown whether the phenomenon applies to other statins or doses or whether gradual withdrawal attenuates it, making this a hypothesis-generating observation (level: trial post hoc analysis) [73].

REC-994 (tempol) is a redox-cycling nitroxide, a superoxide dismutase mimetic, identified through an untargeted discovery platform; in phase 1 studies in healthy volunteers it was well tolerated up to 800 mg, with dose-proportional exposure [97]. SYCAMORE (NCT05085561; Recursion Pharmaceuticals) was a phase 2 randomised triple-blind placebo-controlled trial with oral REC-994,200 mg or 400 mg once daily; it included 62 adults with symptomatic cavernous malformation between March 2022 and June 2025, and the primary outcome was adverse event incidence and severity [95]. In September 2024, the company announced that the safety outcome was met, with encouraging trends in exploratory MRI measures (announcement, not peer reviewed) [2]. According to Alliance to Cure, which followed 62 participants at 16 centres, initial trends towards lower lesion volume and functional improvement were not sustained in the long-term extension, Recursion discontinued the programme in May 2025 and, in May 2026, the organisation signed a licence agreement transferring to it results, remaining drug, biosamples and regulatory permissions to use REC-994 in its current form in symptomatic cavernous malformation [96]. These data derive from an announcement and patient organisation; no peer-reviewed publication was located in October 2026 [96]. GeneReviews calls this compound “Rec-944”, a typographical error corrected here [1].

Selective ROCK2 inhibitors are at an early stage. NRL-1049, which reduced lesion burden and haemorrhage in mice, was tested in a phase 1 single ascending dose study (25 to 250 mg) in healthy volunteers: 9 of 24 (37.5%) had at least one dose-dependent adverse event (dizziness 16.7%, headache 8.3%, syncope 8.3%), without serious events, with a maximum tolerated dose of 150 mg [28, 94]. No NRL-1049 trial in patients was located in ClinicalTrials.gov [78]. For OV888 (GV101), another ROCK2 inhibitor, no ClinicalTrials.gov registration was located, and its cavernous malformation clinical development status could not be confirmed in the sources consulted [78].

Daratumumab (anti-CD38 antibody), at 16 mg/kg intravenously weekly for 8 weeks (fixed protocol regimen, with no maximum dose defined in the source), is registered for a single-arm trial in 10 adults with refractory symptomatic familial disease who are not surgical candidates (Beijing Tiantan Hospital), not yet recruiting in the registry [98]. The CARE PREP study designed an international precision platform to test several drugs, with stratification between sporadic and familial forms, involving 14 countries; targeted pharmacological therapies rank fourth among the ten James Lind Alliance priorities [99]. Prospective cohorts that will underpin future trials are ongoing: CCM_Italia (NCT06983132, paediatric and adult familial disease, 3 T MRI and circulating biomarkers), a French cohort of rare diseases of the brain and eye vessels (NCT07650110), the Tiantan cohort of untreated malformations (NCT06214767), the trial-readiness and biomarker studies in symptomatic haemorrhage (NCT03652181, NCT04467489) and the University of California, San Francisco severity modifier study (NCT01764529, completed in 2025) [78, 100].

7.17 Pregnancy and breastfeeding

Prepregnancy counselling includes genetics consultation in familial or multifocal disease (class 1, C-EO), selecting the ASM with least teratogenicity, starting folic acid before conception (class 1, A) and monitoring ASM levels [1, 2]. Epilepsy in pregnancy is associated with higher maternal mortality; ASM reduces this risk and, although fetal exposure may increase adverse outcomes depending on drug, dose and gestational stage, that risk is generally lower than from untreated maternal seizures [1]. Resection of asymptomatic lesions to make pregnancy safer is not recommended (class 3, C-LD) [2].

Studies from the 1980s and 1990s suggested an aggressive course during pregnancy, but recent series showed no increase in symptoms or haemorrhage versus the non-pregnant state [2]. Witiw and colleagues recorded 349 pregnancies and 49 haemorrhages during reproductive age, only 3 during pregnancy: 1.15% per person-year in pregnancy versus 1.01% outside pregnancy (15–44 years), assuming lesion presence from 15 years [2]. Kalani and Zabramski (64 patients, 28 sporadic and 36 familial, 168 pregnancies) estimated symptomatic haemorrhage risk at approximately 3% per pregnancy, 1.8% in sporadic and 3.6% in familial disease, with only 5 haemorrhages total, 2 in the same woman, making the difference uncertain [1, 101]. In the prospective Mayo Clinic registry (160 women; 90 diagnosed before 46 years, 25.6% familial), haemorrhage rate outside pregnancy was 10.4% annually (95% CI 7.5–14.0) and no haemorrhages occurred in 32 pregnancies after diagnosis (26.9 pregnancy-years; 0%, 95% CI 0–13.6; p=0.09) or during delivery [102]. A systematic review of reports and series (94 patients) found pregnancy haemorrhage rates of 0.9–3% in comparative studies; among symptomatic women, haemorrhage occurred in 70.9% and seizures in 34.1%, with publication bias; caesarean delivery was frequent, more from caution than evidence [103]. Seizures are the most common haemorrhage manifestation in pregnancy [1].

A Chinese multicentre prospective registry, published in July 2026, studied women with brainstem cavernous malformation, most with recent haemorrhage (63 women, 72 pregnancies, 371.7 patient-years, 34 symptomatic haemorrhages) [104]. Annual haemorrhage rate during pregnancy and up to 6 weeks after delivery or miscarriage was 20.4%, versus 7.2% outside pregnancy (adjusted incidence rate ratio 2.56; 95% CI 1.11–5.93), with values of 5.9%, 16.8% and 34.6% across the three trimesters and 38.3% postpartum; the increase was significant only in the third trimester and postpartum and confined to periods with haemorrhage in the preceding year (46.7% vs 9.4%) [104]. These results diverge from the 2025 recommendation (class 2a, B-NR) that pregnancy probably does not alter risk, but apply to a selected subgroup (brainstem with recent haemorrhage), not asymptomatic or incidental lesions [2, 104]. In that subgroup, data warrant close clinical surveillance in the third trimester and postpartum, consistent with close postpartum surveillance proposed by the 2026 systematic review [103, 104].

For FND, acute severe headache or worsening seizures during pregnancy or breastfeeding, perform non-contrast MRI (class 2a, C-EO), also excluding ischaemic stroke and cerebral venous thrombosis [1, 2]. If haemorrhage occurs during pregnancy, weigh severity and recurrence risk against surgical risk at that time; surgery is reserved for acute deterioration or persistent infratentorial haemorrhage [2, 103]. Vaginal delivery is appropriate for most women unless prevented by neurological deficit or recent haemorrhage [2, 102]. Clinical events during labour are extremely rare in the literature [1]. Regarding drugs, atorvastatin should be discontinued when pregnancy is recognised and breastfeeding is not recommended; propranolol should be used during pregnancy only if benefit justifies risk, with reports of intrauterine growth restriction and neonatal bradycardia, hypoglycaemia and respiratory depression [70, 71]. Hormone therapy should be used cautiously (section 7.11) [2].

7.18 Perioperative care, anaesthesia and transition

The guidelines consulted contain no FCCM-specific anaesthesia or perioperative recommendations beyond those already described: ICH management according to general standards, review of anticoagulation and antithrombotics according to the 2025 recommendation, early postoperative MRI and post-resection surveillance [2, 9]. Young children or those with intellectual disability may require sedation or anaesthesia for MRI, influencing decisions about routine imaging [2, 9]. Transition from paediatric to adult care also lacks specific guidance in the sources consulted; disclosure of diagnosis to the child and progressive decision-making independence should be prepared with parents and a genetic counsellor (class 1, C-EO) [2].

7.19 Source divergences and evolving recommendations

The main changes between the 2017 synopsis and 2025 guideline are as follows: genetic testing changed from class I, level B to class 2a, B-NR, and the recommendation to offer testing to at-risk adult relatives gave way to recommendations centred on prior counselling and psychological consequences; propranolol changed from not recommendable to class 2b, B-R; antithrombotics changed from a class III, level C recommendation to the indication that the lesion should not prevent their use (2a, B-NR); future anticoagulation ceased to be a reason to resect an asymptomatic lesion; the recommendation against asymptomatic lesion resection changed from class III, level B to 3, C-LD; and recommendations were added on pregnancy, spinal cord, LITT, children and older adults [2, 9]. The 2017 synopsis recommended CT in the first week and MRI ideally within two weeks of symptom onset; the 2025 guideline retains MRI within two weeks as the definitive investigation [2, 9].

Among contemporary sources, the most relevant divergences concern antithrombotics (favourable prospective cohorts and guideline versus an unfavourable TriNetX administrative cohort, mainly for anticoagulants) [2, 83, 85], pregnancy (no increased risk in the general patient population versus an increase in brainstem disease with recent haemorrhage) [2, 102, 104], statins (favourable cohort association versus neutral trial) [12, 79] and targeted CCM2 founder variant analysis as a first step (allowed by GeneReviews for the US population and not prioritised by the 2025 guideline) [1, 2]. The European–international Delphi consensus agrees with the 2025 guideline on observation of asymptomatic lesions and surgery for deep lesions after a second haemorrhage [2, 66].

8. Reproductive Options and Family Counselling

8.1 Recurrence risks

FCCM is transmitted in an autosomal dominant manner, and each child of a heterozygote has a 50% probability of inheriting the variant, regardless of parental clinical status [1]. Inheriting the variant does not mean developing symptoms, given incomplete age-dependent penetrance: up to half of carriers remain asymptomatic throughout life, although most have MRI lesions [1, 33]. Severity cannot be predicted within a family, and CCM3 tends to be earlier and more severe [2, 9].

8.2 Parental evaluation, de novo variants and mosaicism

Many probands have a symptomatic parent, but de novo variants occur, more frequently in PDCD10, in an unknown proportion because subtle parental signs are rarely assessed [1, 24]. In an apparently isolated proband, parental molecular testing is recommended when a molecular diagnosis exists, and/or brain MRI with GRE or SWI; a history of seizures, deficits or haemorrhages guides which parent to assess [1]. If the variant is absent from both parents, possibilities are a de novo variant or parental germline mosaicism; leucocyte testing may not detect somatic mosaicism and does not detect exclusively germline mosaicism [1]. An apparently negative family history is confirmed only after parental assessment [1]. Clinical diagnosis through multiple lesions may arise in an isolated case, and molecular diagnosis may exist without family history [1].

8.3 Cascade testing and screening relatives

With a pathogenic or likely pathogenic variant identified in the proband, counsel on dominant inheritance and identify at-risk relatives through the pedigree (class 1, C-EO) [2]. In asymptomatic relatives of any age, familial variant testing is the preferred screening approach because it is cost-effective and non-invasive; a negative result excludes diagnosis without MRI [1, 2]. If the familial variant is unknown, offer brain and/or spinal MRI with SWI or GRE [1]. Symptomatic relatives undergo MRI with SWI [1]. In carriers, MRI with recommended sequences investigates non-specific symptoms and establishes baseline lesion burden [2].

Testing asymptomatic individuals raises ethical issues, particularly in children: there is no preventive or curative therapy, but confirmation lowers the neuroimaging threshold for symptoms and allows avoidance of exposures carrying theoretical risk [1, 2]. Genetics consultation before screening asymptomatic relatives is recommended (class 2a, C-EO), and at-risk relatives should receive information about psychological consequences from a genetics specialist before deciding (class 1, C-EO) [2]. The 2017 synopsis recommended offering testing to at-risk adult relatives [9].

8.4 Counselling minors

Disclosing diagnosis to an at-risk child is ethically complex because symptoms usually emerge in adulthood, except in CCM3 [2]. The 2025 guideline recommends shared decision-making between parents and genetic counsellor before disclosing diagnosis to the child, respecting increasing autonomy (class 1, C-EO) [2]. The ideal MRI age in asymptomatic carrier children is uncertain, but MRI is mandatory for neurological symptoms, and sedation needs in young children should be considered [2].

8.5 Reproductive planning, prenatal and preimplantation diagnosis

The best time to assess risk and discuss prenatal testing or PGT is before pregnancy, and counselling should be offered to affected or at-risk young adults [1]. In familial or multifocal disease, genetic counselling before pregnancy is recommended (class 1, C-EO) [2]. With a known familial variant, prenatal diagnosis and PGT are possible [1, 2]. Opinions on these tests differ, particularly in a disorder with incomplete penetrance, variable expressivity and the possibility of an asymptomatic life, and the decision is personal [1]. In probands without molecular diagnosis, DNA banking allows these tests to be offered in future if reanalysis identifies the variant [1]. Counselling should be provided by genetics professionals (clinical geneticist, genetic counsellor or specialist genetics nurse) [1].

9. Special Considerations: Population Diversity and Epidemiology

9.1 Prevalence and incidence

Frequency estimates measure different things and should not be confused. Prevalence of cavernous malformations of any cause is 0.16% in incidental MRI findings and up to 0.5% in autopsy series, increasing with age [2]; GeneReviews indicates 0.4–0.9% in the general population [1]. In the Mayo Clinic Study of Aging (Olmsted County, Minnesota), non-clinical MRI in 2715 participants aged 50–89 years showed cavernous malformations in 12 (0.44%; weighted 0.46%, 95% CI 0.05–0.86), with presumed familial disease in 0.07% [3, 10]. This older cohort from one US region does not represent national or paediatric prevalence and probably underestimates asymptomatic lesions [1].

Among all cavernous malformation cases, approximately 80% are sporadic and 20% familial [2]. Orphanet assigns familial disease a worldwide point prevalence in the class 1–5 per 10,000, with a mean value of 15 per 100,000, based on expert estimation [7]. The difference from Flemming's 0.07% (70 per 100,000) reflects different methods: expert estimate versus an MRI cohort of older people with presumptive classification [7, 10]. The 3 in 10,000 threshold used in EMA orphan designations is regulatory, not an epidemiological measure [90].

Detection incidence in Scotland in adults over 16 years during the 2000s was initially estimated at 0.56 per 100,000 annually [2, 105]. The latest analysis of the same audit (300 cases diagnosed in 1999–2003 and 2006–2010) estimated crude incidence at 0.72 (95% CI 0.64–0.80) per 100,000 adults annually, similar by sex, with 0.40 (0.34–0.47) for symptomatic and 0.31 (0.26–0.37) for incidental lesions, increasing with age (incidence rate ratio 1.09 per decade) [50]. Current incidence is probably higher because of wider MRI availability [2].

9.2 Founder populations

Familial disease is more frequent in populations with founder effects. In Hispanics from the south-western USA and northern Mexico, with approximately 400 years of regional history, KRIT1 c.1363C>T explains most familial cases (70% according to GeneReviews; 16 of 21 families in the original description), and the cutaneous profile is distinctive [1, 2, 20, 44]. Familial clustering in Mexican Americans was the first observation in Rigamonti's 1988 series [15]. In US Ashkenazi Jews, CCM2 c.30+5_30+6delinsTT is recurrent [2, 41]. The CCM2 exon 2–10 deletion clusters in white families with eighteenth-century ancestry in the south-eastern USA (at least 50 families) [2, 42]. In Europe, KRIT1 c.987C>A p.(Cys329Ter) is a possible Sardinian founder, and in Italy multiexon KRIT1 deletions represented 50% of sequencing-negative cases, versus 5% in the USA and 4% in France [1, 40]. In the French series from the time CCM2 was identified, genetic linkage distributed families across 7q (40%), 7p (20%) and 3q (40%) [23].

9.3 Regional data

In the USA, the main cohorts are BVMC (familial disease), Mayo Clinic and University of Chicago [12, 51, 58, 102]. In Europe, population data come from the Scottish audit, a long-term prospective population-based cohort [47, 48, 50], supplemented by French (national reference centre), Italian (Treat_CCM, CCM_Italia), German (Essen) and Dutch (Amsterdam) cohorts [13, 49, 52, 89, 100]. The international cohort of 1592 patients brought together 16 tertiary centres [38]. Outside these regions, there are Chinese data on brainstem lesions and pregnancy [104], an Australian genetic series [30] and the Turkish case described in section 10 [35].

In Brazil, the Rio de Janeiro Familial Cavernous Malformation Project studied nine families; four were genetically mapped and all had KRIT1 variants; imaging changed from T2* GRE to SWI, only two index cases underwent surgery and no relative underwent surgery during 7 years of follow-up [106]. The same reference centre assessed quality of life in 83 adults, 36 (43.4%) with familial disease [59], and published a pathophysiology and biomarker review [29]. The sources consulted provide no Portuguese epidemiological data, which does not demonstrate their non-existence.

9.4 Special populations

Children represent approximately 25% of cases, with a greater proportion of CCM3 and need for MRI sedation [2]. Childhood cancer survivors with cranial irradiation are at risk of acquired cavernous malformations, large- and small-vessel arterial vasculopathy, chronic encephalopathy and early cognitive decline, exacerbated by radiation from repeated CT and dental radiographs; in a Delphi panel of 45 survivorship experts, there was 70–89% agreement on repeat imaging in 1–2 years and stroke risk factor screening (lipids, glycated haemoglobin, glucose), high agreement on avoiding antiplatelets and less than 70% agreement on referring a new lesion to neurology or neurosurgery; long-term proton therapy effects are unknown [2]. In older adults, comorbidities and quality of life influence decisions and surgical risks are higher [2]. Women of reproductive age require attention to hormonal contraception and pregnancy planning (sections 7.11 and 7.17).

9.5 European networks and patient organisations

The closest European reference network is VASCERN (European Reference Network on Rare Multisystemic Vascular Diseases), with working groups for hereditary haemorrhagic telangiectasia, hereditary aortic disease, medium-sized arterial disease, neurovascular disease (NEUROVASC, including CADASIL and moyamoya), primary lymphoedema and vascular anomalies (VASCA); its main page does not mention cerebral cavernous malformation, and the VASCA document on “CM” refers to capillary rather than cavernous malformation [14]. No specific European FCCM care pathway document was identified.

Patient organisations include the US Alliance to Cure Cavernous Malformation (formerly Angioma Alliance), sponsoring the 2017 and 2025 guidelines, maintaining the international patient registry and a DNA and tissue bank, and licensing REC-994 [1, 96, 107]; Cavernoma Alliance UK, which funded the Scottish audit [83, 108]; and CCM Italia, the Italian community associated with the University of Turin and CCM_Italia [100, 109]. General epilepsy organisations (American Epilepsy Society, Epilepsy Foundation) are resources cited by GeneReviews [1].

10. Illustrative Clinical Case

10.1 Presentation

Yayıcı Köken and colleagues described a Turkish family with familial disease associated with a KRIT1 initiation-region variant in 2026 [35]. The proband, a 14-year-old boy, was referred to paediatric neurology after a focal seizure with left upper limb numbness. Five months earlier, he had undergone surgery for a right parietal scalp mass, painful and palpable for approximately 2 years, growing faster over the preceding year, firm, approximately 5 cm, with a broad base adherent to the dura; the excised bone specimen measured 8 × 7.4 × 3 cm and represented an intraosseous calvarial haemangioma, with a “spoke-wheel” radiographic pattern, diploic expansion on CT and heterogeneous T2 signal with intense MRI enhancement, interpreted by the authors as unrelated to the cavernous malformations [35].

10.2 Findings

SWI showed multiple supratentorial punctate hypointense foci and bilateral cerebellar lesions compatible with cavernous malformations, and spinal MRI showed an intramedullary cervical lesion with a hyperintense core and hypointense haemosiderin ring [35]. EEG revealed right temporoparietal epileptiform activity. The proband had pulsatile headaches two to three times weekly, with weekly episodes lasting 3–4 hours responding to paracetamol; he had no skin or retinal lesions, and neurological examination was normal [35]. His father had died at 41 years from intracerebral haemorrhage, without genetic testing [35].

10.3 Genetic investigation

Exome sequencing (GRCh38; mean depth 125.83× and 97.74% of the target region with at least 20× coverage in the proband) identified heterozygous KRIT1 NM_004912.4:c.2dup, absent from gnomAD, without CCM2 or PDCD10 variants [35]. Normalised to the MANE Select transcript, it corresponds to NM_194454.3:c.2dup (NC_000007.14:g.92242134dup), with predicted protein effect p.(Met1?) because it affects the initiation codon; the authors describe it as p.(Met1IlefsTer31), a designation not confirmed by VariantValidator [6, 35]. The variant was classified as likely pathogenic using ACMG/AMP criteria PVS1, PM2, PP4 and PP1 [35]. His asymptomatic 16-year-old brother was a carrier and had multiple cavernous malformations on SWI without spinal lesions and with normal ophthalmological examination; the clinically unaffected mother was not a carrier on exome sequencing, and the authors did not perform Sanger confirmation, based on sequencing quality and concordant segregation [35].

10.4 Management and course

The proband started levetiracetam and remained seizure-free during 6 months of follow-up; asymptomatic intracranial lesions remained under radiological surveillance [35]. Figure 1 of the article (proband panels a–f, with radiography, CT and MRI of the calvarial haemangioma and SWI showing cerebral and cerebellar lesions, and brother panels g–h, with SWI) can be viewed at https://pmc.ncbi.nlm.nih.gov/articles/PMC13586551/figure/mgg370304-fig-0001/; Figure 2 (sagittal cervical T2 MRI with the intramedullary lesion) and Figure 3 (pedigree and KRIT1 domains) are at https://pmc.ncbi.nlm.nih.gov/articles/PMC13586551/figure/mgg370304-fig-0002/ and https://pmc.ncbi.nlm.nih.gov/articles/PMC13586551/figure/mgg370304-fig-0003/ [35]. The Figure 3 legend calls KRIT1 “malcavernin”, whereas this name belongs to CCM2 [4, 35].

10.5 Commentary

The case brings together three expressions of the same variant in one family: asymptomatic radiological disease in the brother, focal epilepsy in the proband and fatal intracerebral haemorrhage in the father, whose genotype could not be determined [35]. Management agrees with the 2025 guideline: ASM after the first lesion-attributed seizure (class 1, B-NR), MRI with a susceptibility sequence and spinal imaging given multiplicity, three-gene genetic testing and cascade testing of relatives [2]. Identifying the brother as a carrier illustrates cascade testing's utility, and absence of Sanger confirmation and functional variant study are report limitations. Coexisting intraosseous calvarial haemangioma cannot establish a disease relationship from a single case; atypical vertebral haemangiomas described in FCCM are the recognised skeletal manifestation [1].

11. Prognosis and Long-Term Follow-up

11.1 Prognosis

FCCM prognosis is heterogeneous. Up to half of carriers remain asymptomatic throughout life [1], but in a tertiary-centre cohort with follow-up from birth, lifetime first symptomatic haemorrhage risk was estimated at approximately 80% and first-seizure risk at approximately 45% [38]; the figures are not contradictory because they describe distinct populations (carriers generally versus referred patients). In the prospective cohort of 479 patients, 60% had experienced at least one seizure by 80 years [54]. In prospectively followed familial patients, symptomatic ICH rate was 2.8 per 100 patient-years, higher after previous haemorrhage (4.5) than without (2.0) and increasing with lesion burden [51]. In carriers asymptomatic at diagnosis, approximately 21% had a first event within 5 years and 31% within 10 years [52]. New lesions form at 0.2–1 lesion per person-year [1].

Worse prognostic factors are presentation with ICH or FND, brainstem location, higher lesion burden, CCM3 and, during pregnancy, a brainstem lesion with haemorrhage in the preceding year [38, 46, 51, 104]. Recurrence risk peaks in the first years after haemorrhage and declines markedly after 5 years, so decisions should use 5-year horizons rather than extrapolating annual rates throughout life [47, 48]. Haemorrhage and seizures are independently associated with greater disability [38]. In the Scottish audit, after a median 15 years without treatment, 24% of patients were dependent and 2% had died from lesion-related causes [48]. Lesion-related epilepsy has a good prognosis with ASM in most patients: at 10 years, 62% had been seizure-free for 5 years [55]. Quality of life is below the reference population, particularly after ICH and with headaches [58].

11.2 Surveillance

There is no routine MRI schedule with an established interval for asymptomatic carriers, and routine surveillance in asymptomatic individuals is debatable [1–3]. Scheduled MRI may document stability of lesions that have bled or grown, through shared decision-making; subclinical changes on surveillance MRI may herald symptomatic haemorrhage, but their clinical impact is controversial [2]. The table summarises recommended surveillance.

Children and adults with familial disease should also be informed about avoidance of or caution with sex hormones, the absence of an absolute contraindication to antithrombotics when indicated, smoking cessation and avoiding excessive alcohol [2].

12. Summary and Final Considerations

Familial cerebral cavernous malformations result from germline loss-of-function variants in KRIT1, CCM2 or PDCD10, transmitted in an autosomal dominant manner with incomplete age-dependent penetrance. Each lesion requires a second somatic hit in the remaining allele, and somatic PIK3CA variants may accelerate growth; CCM complex loss dysregulates endothelial MEKK3 and RhoA/ROCK signalling and compromises the blood–brain barrier. CCM3 is the earliest and most severe form, CCM1 the most frequent and includes the Hispanic c.1363C>T founder variant, and CCM2 includes founder variants in Ashkenazi Jews and the south-eastern USA.

Diagnosis rests on MRI with a susceptibility-sensitive sequence revealing multiple lesions without DVA, and three-gene sequencing with deletion/duplication analysis; targeted Hispanic variant testing is the first step in that population. A negative test does not exclude disease with a suggestive phenotype, and a negative MRI does not exclude carrier status in a relative, who should be tested for the familial variant.

Natural history is marked by epileptic seizures, haemorrhages and FND, with higher haemorrhagic risk after a first haemorrhage, in brainstem lesions and with greater lesion burden, and declining recurrence risk after 5 years. Management is mainly conservative: ASM after the first seizure, surgery weighed according to location and haemorrhage number, radiosurgery discouraged in familial disease, and MRI surveillance guided by symptoms and shared decision-making. Antithrombotics should not be withheld when indicated, and sex hormones require caution. Pregnancy does not appear to increase risk in most patients, but a recent registry suggests increased risk in the third trimester and postpartum in women with a brainstem lesion and recent haemorrhage.

There is no authorised pharmacological therapy. Propranolol showed safety and an efficacy signal in an open-label pilot trial, atorvastatin was neutral in a randomised trial and its withdrawal was associated with rehaemorrhages in an exploratory analysis, REC-994 was discontinued by the company and licensed to the patient organisation, and selective ROCK2 inhibitors are in phase 1. CARE demonstrated that randomised comparison of intervention and conservative management is feasible, and precision medicine platforms under preparation may test several drugs simultaneously. Genetic counselling, cascade testing and informed reproductive planning are central components of care for these families.

13. References

1. Flemming KD, Smith E, Marchuk D, Derry WB. Familial Cerebral Cavernous Malformations. 2003 Feb 24 [updated 2023 Jul 27]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993-. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1293/
2. Akers AL, Albanese J, Alcazar-Felix RJ, Al-Shahi Salman R, Awad IA, Connolly ES, et al. Guidelines for the Diagnosis and Clinical Management of Cavernous Malformations of the Brain and Spinal Cord: Consensus Recommendations Based on a Systematic Literature Review by the Alliance to Cure Cavernous Malformation Clinical Advisory Board Experts Panel. Neurosurgery. 2026;98(1):3-22. doi:10.1227/neu.0000000000003459. PMID: 40396744. PMC12680281.
3. GeneGuidance. KRIT1 — clinical gene curation (accessed 2026-09-26).
4. GeneGuidance. CCM2 — clinical gene curation (accessed 2026-09-26).
5. GeneGuidance. PDCD10 — clinical gene curation (accessed 2026-09-26).
6. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: Accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID: 28967166.
7. Orphanet. Familial cerebral cavernous malformation (ORPHA:221061) [Internet]. Paris: INSERM; [cited 2026 Oct 10]. Available from: https://www.orpha.net/en/disease/detail/221061
8. Orphanet. Cerebral cavernous malformations (ORPHA:164) [Internet]. Paris: INSERM; [cited 2026 Oct 10]. Available from: https://www.orpha.net/en/disease/detail/164
9. Akers A, Al-Shahi Salman R, A Awad I, Dahlem K, Flemming K, Hart B, et al. Synopsis of Guidelines for the Clinical Management of Cerebral Cavernous Malformations: Consensus Recommendations Based on Systematic Literature Review by the Angioma Alliance Scientific Advisory Board Clinical Experts Panel. Neurosurgery. 2017;80(5):665-680. doi:10.1093/neuros/nyx091. PMID: 28387823. PMC5808153.
10. Flemming KD, Graff-Radford J, Aakre J, Kantarci K, Lanzino G, Brown RD Jr, Mielke MM, Roberts RO, Kremers W, Knopman DS, Petersen RC, Jack CR Jr. Population-based prevalence of cerebral cavernous malformations in older adults: Mayo Clinic Study of Aging. JAMA Neurol. 2017;74:801-5. PMID: 28492932.
11. Shenkar R, Shi C, Rebeiz T, Stockton RA, McDonald DA, Mikati AG, Zhang L, Austin C, Akers AL, Gallione CJ, Rorrer A, Gunel M, Min W, Marcondes de Souza J, Lee C, Marchuk DA, Awad IA. Exceptional aggressiveness of cerebral cavernous malformation disease associated with PDCD10 mutations.Genet Med. 2015;17:188-96. PMID: 25122144.
12. Awad IA, Alcazar-Felix RJ, Stadnik A, Kinkade S, Jhaveri A, Lee J, et al. Safety and efficacy of atorvastatin for rebleeding in cerebral cavernous malformations (AT CASH EPOC): a phase 1/2a, randomised placebo-controlled trial. Lancet Neurol. 2025;24(4):295-304. doi:10.1016/s1474-4422(25)00036-5. PMID: 40120614. PMC12080613.
13. Lanfranconi S, Scola E, Meessen JMTA, Pallini R, Bertani GA, Al-Shahi Salman R, et al. Safety and efficacy of propranolol for treatment of familial cerebral cavernous malformations (Treat_CCM): a randomised, open-label, blinded-endpoint, phase 2 pilot trial. Lancet Neurol. 2023;22(1):35-44. doi:10.1016/s1474-4422(22)00409-4. PMID: 36403580.
14. VASCERN — European Reference Network on Rare Multisystemic Vascular Diseases [Internet]. Paris: VASCERN; [cited 2026 Oct 10]. Available from: https://vascern.eu
15. Rigamonti D, Hadley MN, Drayer BP, Johnson PC, Hoenig-Rigamonti K, Knight JT, et al. Cerebral cavernous malformations. Incidence and familial occurrence. N Engl J Med. 1988;319(6):343-347. doi:10.1056/nejm198808113190605. PMID: 3393196.
16. Zabramski JM, Wascher TM, Spetzler RF, Johnson B, Golfinos J, Drayer BP, et al. The natural history of familial cavernous malformations: results of an ongoing study. J Neurosurg. 1994;80(3):422-432. doi:10.3171/jns.1994.80.3.0422. PMID: 8113854.
17. Dubovsky J, Zabramski JM, Kurth J, Spetzler RF, Rich SS, Orr HT, et al. A gene responsible for cavernous malformations of the brain maps to chromosome 7q. Hum Mol Genet. 1995;4(3):453-458. doi:10.1093/hmg/4.3.453. PMID: 7795602.
18. Günel M, Awad IA, Anson J, Lifton RP. Mapping a gene causing cerebral cavernous malformation to 7q11.2-q21. Proc Natl Acad Sci U S A. 1995;92(14):6620-6624. doi:10.1073/pnas.92.14.6620. PMID: 7604043. PMC41570.
19. Johnson EW, Iyer LM, Rich SS, Orr HT, Gil-Nagel A, Kurth JH, Zabramski JM, Marchuk DA, Weissenbach J, Clericuzio CL, Davis LE, Hart BL, Gusella JF, Kosofsky BE, Louis DN, Morrison LA, Green ED, Weber JL. Refined localization of the cerebral cavernous malformation gene (CCM1) to a 4-cM interval of chromosome 7q contained in a well-defined YAC contig. Genome Res. 1995;5:368-80. PMID: 8750196.
20. Sahoo T, Johnson EW, Thomas JW, Kuehl PM, Jones TL, Dokken CG, et al. Mutations in the gene encoding KRIT1, a Krev-1/rap1a binding protein, cause cerebral cavernous malformations (CCM1). Hum Mol Genet. 1999;8(12):2325-2333. doi:10.1093/hmg/8.12.2325. PMID: 10545614.
21. Laberge-le Couteulx S, Jung HH, Labauge P, Houtteville JP, Lescoat C, Cecillon M, et al. Truncating mutations in CCM1, encoding KRIT1, cause hereditary cavernous angiomas. Nat Genet. 1999;23(2):189-193. doi:10.1038/13815. PMID: 10508515.
22. Liquori CL, Berg MJ, Siegel AM, Huang E, Zawistowski JS, Stoffer T, et al. Mutations in a gene encoding a novel protein containing a phosphotyrosine-binding domain cause type 2 cerebral cavernous malformations. Am J Hum Genet. 2003;73(6):1459-1464. doi:10.1086/380314. PMID: 14624391. PMC1180409.
23. Denier C, Goutagny S, Labauge P, Krivosic V, Arnoult M, Cousin A, et al. Mutations within the MGC4607 gene cause cerebral cavernous malformations. Am J Hum Genet. 2004;74(2):326-337. doi:10.1086/381718. PMID: 14740320. PMC1181930.
24. Bergametti F, Denier C, Labauge P, Arnoult M, Boetto S, Clanet M, et al. Mutations within the programmed cell death 10 gene cause cerebral cavernous malformations. Am J Hum Genet. 2005;76(1):42-51. doi:10.1086/426952. PMID: 15543491. PMC1196432.
25. Akers AL, Johnson E, Steinberg GK, Zabramski JM, Marchuk DA. Biallelic somatic and germline mutations in cerebral cavernous malformations (CCMs): evidence for a two-hit mechanism of CCM pathogenesis. Hum Mol Genet. 2009;18(5):919-930. doi:10.1093/hmg/ddn430. PMID: 19088123. PMC2640209.
26. Pagenstecher A, Stahl S, Sure U, Felbor U. A two-hit mechanism causes cerebral cavernous malformations: complete inactivation of CCM1, CCM2 or CCM3 in affected endothelial cells. Hum Mol Genet. 2009;18:911-8 PMID: 19088124.
27. Ren AA, Snellings DA, Su YS, Hong CC, Castro M, Tang AT, Detter MR, Hobson N, Girard R, Romanos S, Lightle R, Moore T, Shenkar R, Benavides C, Beaman MM, Müller-Fielitz H, Chen M, Mericko P, Yang J, Sung DC, Lawton MT, Ruppert JM, Schwaninger M, Körbelin J, Potente M, Awad IA, Marchuk DA, Kahn ML. PIK3CA and CCM mutations fuel cavernomas through a cancer-like mechanism. Nature. 2021;594:271-6. PMID: 33910229.
28. Morrison L, Gutierrez J, Ayata C, Lopez-Toledano M, Carrazana E, Awad I, et al. Current and Future Treatment Options for Cerebral Cavernous Malformations. Stroke Vasc Interv Neurol. 2024;4(3):e001140. doi:10.1161/svin.123.001140. PMID: 41584470. PMC12778514.
29. Fontes-Dantas FL, da Fontoura Galvão G, Cunha AM, de Sena Murteira Pinheiro P, Morandi V, de Souza JM. Familial Cerebral Cavernous Malformations: Pathophysiology, Genetics, Biomarkers, and Treatment Perspectives. J Neurochem. 2026;170(1):e70342. doi:10.1111/jnc.70342. PMID: 41486746. PMC12766151.
30. Sikta N, Gooley S, Green TE, Hoeper O, Witkowski T, Bennett C, et al. Improving genetic diagnostic yield in familial and sporadic cerebral cavernous malformations: detection of copy number and deep Intronic variants. Hum Mol Genet. 2025;34(15):1286-1293. doi:10.1093/hmg/ddaf077. PMID: 40401429. PMC12278727.
31. Alcazar-Felix RJ, Shenkar R, Benavides CR, Bindal A, Srinath A, Li Y, et al. Except for Robust Outliers, Rapamycin Increases Lesion Burden in a Murine Model of Cerebral Cavernous Malformations. Transl Stroke Res. 2025;16(3):859-867. doi:10.1007/s12975-024-01270-9. PMID: 38980519. PMC11711328.
32. Hale AT, Kundishora AJ, Kahle KT. Genetically Guided Pharmacotherapy for Structural Neurovascular Lesions. Neurol Genet. 2026;12(4):e200413. doi:10.1212/nxg.0000000000200413. PMID: 42453853. PMC13366159.
33. Denier C, Labauge P, Brunereau L, Cavé-Riant F, Marchelli F, Arnoult M, Cecillon M, Maciazek J, Joutel A, Tournier-Lasserve E. Clinical features of cerebral cavernous malformations patients with KRIT1 mutations. Ann Neurol. 2004;55:213-20. PMID: 14755725.
34. Denier C, Labauge P, Bergametti F, Marchelli F, Riant F, Arnoult M, Maciazek J, Vicaut E, Brunereau L, Tournier-Lasserve E, Société Française de Neurochirurgie. Genotype-phenotype correlations in cerebral cavernous malformations patients. Ann Neurol. 2006;60:550-6. PMID: 17041941.
35. Yayıcı Köken Ö, Yanartaş MS, Aygün H, Ceylan AC, Karaali K, Kazan MS. Pediatric Familial Cerebral Cavernous Malformation Associated With a Novel KRIT1 Initiation-Region Frameshift Variant. Mol Genet Genomic Med. 2026;14(9):e70304. doi:10.1002/mgg3.70304. PMID: 42755143. PMC13586551.
36. Riant F, Bergametti F, Fournier HD, Chapon F, Michalak-Provost S, Cecillon M, Lejeune P, Hosseini H, Choe C, Orth M, Bernreuther C, Boulday G, Denier C, Labauge P, Tournier-Lasserve E. CCM3 Mutations Are Associated with Early-Onset Cerebral Hemorrhage and Multiple Meningiomas.Mol Syndromol. 2013;4:165-72. PMID: 23801932.
37. Liquori CL, Berg MJ, Squitieri F, Leedom TP, Ptacek L, Johnson EW, Marchuk DA. Deletions in CCM2 are a common cause of cerebral cavernous malformations. Am J Hum Genet. 2007;80:69-75. PMID: 17160895.
38. Dammann P, Santos AN, Mavarani L, Guey S, Chabriat H, Herve D, et al. Lifetime Risk of First Symptomatic ICH or Seizure in Familial Cerebral Cavernous Malformations: A Multicenter Patient Data Analysis. Neurology. 2025;105(3):e213798. doi:10.1212/wnl.0000000000213798. PMID: 40638881. PMC12428507.
39. National Center for Biotechnology Information. ClinVar [Internet]. Bethesda (MD): NLM. VCV000005721, VCV000005728, VCV000193463, VCV002759819 [cited 2026 Oct 10]. Available from: https://www.ncbi.nlm.nih.gov/clinvar/
40. Cau M, Loi M, Melis M, Congiu R, Loi A, Meloni C, Serrenti M, Addis M, Melis MA. C329X in KRIT1 is a founder mutation among CCM patients in Sardinia. Eur J Med Genet. 2009;52:344-8. PMID: 19454328.
41. Gallione CJ, Solatycki A, Awad IA, Weber JL, Marchuk DA. A founder mutation in the Ashkenazi Jewish population affecting messenger RNA splicing of the CCM2 gene causes cerebral cavernous malformations. Genet Med. 2011;13:662-6. PMID: 21543988.
42. Gallione CJ, Detter MR, Sheline A, Christmas HM, Lee C, Marchuk DA. Genetic genealogy uncovers a founder deletion mutation in the cerebral cavernous malformations gene. Hum Genet. 2022;141:1761-1769. PMID: 35488064.
43. Mabray MC, Starcevich J, Hallstrom J, Robinson M, Bartlett M, Nelson J, Zafar A, Kim H, Morrison L, Hart BL. High prevalence of spinal cord cavernous malformations in the Familial Cerebral Cavernous Malformations Type 1 Cohort. AJNR Am J Neuroradiol. 2020;41:1126-30. PMID: 32467184.
44. Manole AK, Forrester VJ, Zlotoff BJ, Hart BL, Morrison LA. Cutaneous findings of familial cerebral cavernous malformation syndrome due to the common Hispanic mutation. Am J Med Genet A. 2020;182:1066-72. PMID: 32100472.
45. Geraldo AF, Alves CAPF, Luis A, Tortora D, Guimarães J, Abreu D, Reimão S, Pavanello M, de Marco P, Scala M, Capra V, Vaz R, Rossi A, Schwartz ES, Mankad K, Severino M. Natural history of familial cerebral cavernous malformation syndrome in children: a multicenter cohort study. Neuroradiology. 2023;65:401-14. PMID: 36198887.
46. Horne MA, Flemming KD, Su IC, Stapf C, Jeon JP, Li D, et al. Clinical course of untreated cerebral cavernous malformations: a meta-analysis of individual patient data. Lancet Neurol. 2016;15(2):166-173. doi:10.1016/s1474-4422(15)00303-8. PMID: 26654287. PMC4710581.
47. Al-Shahi Salman R, Hall JM, Horne MA, Moultrie F, Josephson CB, Bhattacharya JJ, et al. Untreated clinical course of cerebral cavernous malformations: a prospective, population-based cohort study. Lancet Neurol. 2012;11(3):217-224. doi:10.1016/s1474-4422(12)70004-2. PMID: 22297119. PMC3282211.
48. Sandmann ACA, Vandertop WP, White PM, Verbaan D, Coutinho JM, Al-Shahi Salman R, et al. Long-term outcomes of untreated cerebral cavernous malformations: a prospective, population-based cohort study. Lancet Reg Health Eur. 2025;57:101410. doi:10.1016/j.lanepe.2025.101410. PMID: 41132766. PMC12541635.
49. Sandmann ACA, Kempeneers MA, van den Berg R, Verbaan D, Vandertop WP, Coutinho JM. Clinical course of patients with conservatively managed cerebral cavernous malformations. Eur Stroke J. 2024;9(3):667-675. doi:10.1177/23969873241246868. PMID: 38624046. PMC11418428.
50. Sandmann ACA, Coutinho JM, Verbaan D, Vandertop WP, White PM, Al-Shahi Salman R. Incidence of cerebral cavernous malformations among adults in Scotland: a prospective, population-based study. J Neurol Neurosurg Psychiatry. 2026;97(9):790-795. doi:10.1136/jnnp-2025-338343. PMID: 41986053. PMC13479685.
51. Weinsheimer S, Nelson J, Abla AA, Ko NU, Tsang C, Okoye O, et al. Intracranial Hemorrhage Rate and Lesion Burden in Patients With Familial Cerebral Cavernous Malformation. J Am Heart Assoc. 2023;12(3):e027572. doi:10.1161/jaha.122.027572. PMID: 36695309. PMC9973654.
52. Popa CR, Hervé D, Alili N, Dubus E, Taleb A, Fernandes F, et al. Risk of Clinical Events in Presymptomatic Familial Cerebral Cavernous Malformations. Eur J Neurol. 2026;33(6):e70669. doi:10.1111/ene.70669. PMID: 42334220. PMC13288596.
53. Nikoubashman O, Di Rocco F, Davagnanam I, Mankad K, Zerah M, Wiesmann M. Prospective hemorrhage rates of cerebral cavernous malformations in children and adolescents based on MRI appearance. AJNR Am J Neuroradiol. 2015;36:2177-83. PMID: 26272978.
54. Fox CK, Nelson J, McCulloch CE, Weinsheimer S, Pawlikowska L, Hart B, Mabray M, Zafar A, Morrison L, Zabramski JM, Akers A, Kim H. Seizure incidence rates in children and adults with familial cerebral cavernous malformations. Neurology. 2021;97:e1210-6. PMID: 34389651.
55. Sandmann ACA, Vandertop WP, White PM, Verbaan D, Coutinho JM, Al-Shahi Salman R. Seizures and Epilepsy in Patients With Untreated Cerebral Cavernous Malformations: A Prospective, Population-Based Cohort Study. Neurology. 2025;105(11):e214387. doi:10.1212/wnl.0000000000214387. PMID: 41232061. PMC12615012.
56. Bergey GK. Guidance in the Medical Treatment of Recent-Onset Seizures Associated With Cerebral Cavernous Malformations. Neurology. 2025;105(11):e214444. doi:10.1212/wnl.0000000000214444. PMID: 41232059.
57. Moultrie F, Horne MA, Josephson CB, Hall JM, Counsell CE, Bhattacharya JJ, Papanastassiou V, Sellar RJ, Warlow CP, Murray GD, Al-Shahi Salman R, et al. Outcome after surgical or conservative management of cerebral cavernous malformations. Neurology. 2014;83:582-9. PMID: 24994841.
58. Tsang C, Nelson J, Vassar R, Ko NU, McCulloch C, Smith ER, et al. Association of Quality of Life Domains and Clinical Symptoms in Patients With Familial Cerebral Cavernous Malformation. J Am Heart Assoc. 2025;14(17):e035926. doi:10.1161/jaha.124.035926. PMID: 40879027. PMC12553445.
59. Cunha AM, da Fontoura Galvão G, Marcondes de Souza J. Comparative Analysis of the Health-Related Quality of Life Between Patients with Familial and Sporadic Forms of Cerebral Cavernous Malformation. World Neurosurg. 2025;198:124023. doi:10.1016/j.wneu.2025.124023. PMID: 40306412.
60. Meessen JMTA, Abete-Fornara G, Zarino B, Castori M, Tassi L, Carriero MR, et al. Patient-reported outcome measures in patients with familial cerebral cavernous malformations: results from the Treat_CCM trial. Front Neurol. 2024;15:1338941. doi:10.3389/fneur.2024.1338941. PMID: 38419711. PMC10899450.
61. Alalfi MO, Lanzino G, Flemming KD. Clinical presentation, hemorrhage risk, and outcome in patients with familial cavernous malformations: a pragmatic prospective analysis of 75 patients. J Neurosurg. 2023. Epub ahead of print. PMID: 36933254.
62. Sirvente J, Enjolras O, Wassef M, Tournier-Lasserve E, Labauge P. Frequency and phenotypes of cutaneous vascular malformations in a consecutive series of 417 patients with familial cerebral cavernous malformations. J Eur Acad Dermatol Venereol. 2009;23:1066-72. PMID: 19453802.
63. Sandmann ACA, Verbaan D, Coutinho JM, Vandertop WP, White PM, Al-Shahi Salman R. Intervention or Conservative Management for Symptomatic Cerebral Cavernous Malformations: A Prospective, Population-Based Cohort Study. Neurology. 2026;107(5):e218442. doi:10.1212/wnl.0000000000218442. PMID: 42585604. PMC13528903.
64. Jones KL, Jones MC, del Campo M. Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 744 (Hereditary Hemorrhagic Telangiectasia).
65. Steiner T, et al.; European Stroke Organisation (ESO), European Association of Neurosurgical Societies (EANS). Guideline on stroke due to spontaneous intracerebral haemorrhage. Eur Stroke J. 2025. doi:10.1177/23969873251340815. PMC12098356.
66. Tasiou A, Brotis AG, Kalogeras A, Tzerefos C, Alleyne CH Jr, Andreou A, et al. Cavernous malformations of the central nervous system: An international consensus statement. Brain Spine. 2023;3:102707. doi:10.1016/j.bas.2023.102707. PMID: 38020995. PMC10668094.
67. Poorthuis MHF, Rinkel LA, Lammy S, Al-Shahi Salman R. Stereotactic radiosurgery for cerebral cavernous malformations: a systematic review. Neurology. 2019;93:e1971-e1979. PMID: 31659093.
68. Loan JJM, Bacon A, van Beijnum J, Bhatt P, Bjornson A, Broomes N, et al. Feasibility of comparing medical management and surgery (with neurosurgery or stereotactic radiosurgery) with medical management alone in people with symptomatic brain cavernoma - protocol for the Cavernomas: A Randomised Effectiveness (CARE) pilot trial. BMJ Open. 2023;13(8):e075187. doi:10.1136/bmjopen-2023-075187. PMID: 37558454. PMC10414059.
69. Al-Shahi Salman R, Kitchen N, Forsyth L, Ganesan V, Hall PS, Harkness K, et al. Medical management and intervention (using neurosurgical resection or stereotactic radiosurgery) versus medical management alone for symptomatic brain cavernoma: the CARE pilot RCT. Health Technol Assess. 2025;29(38):1-24. doi:10.3310/gjrs5321. PMID: 40820590.
70. INDERAL LA (propranolol hydrochloride) long-acting capsules. Prescribing information. NDA 018553. Effective 2026 Aug 24. DailyMed set ID 35d28979-36b1-4630-b85e-a44e0a443734 [cited 2026 Oct 10]. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=35d28979-36b1-4630-b85e-a44e0a443734
71. Pfizer/Viatris. LIPITOR (atorvastatin calcium) tablets, for oral use. Prescribing information. NDA 020702. Effective 2024 Apr 15. DailyMed set ID a60cc18b-0631-4cf0-b021-9f52224ece65 [cited 2026 Oct 10]. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=a60cc18b-0631-4cf0-b021-9f52224ece65
72. Mario Negri Institute for Pharmacological Research. Treat_CCM: Propranolol in Familial Cerebral Cavernous Malformation. ClinicalTrials.gov identifier: NCT03589014 [updated 2022 Feb 25; cited 2026 Oct 10]. Available from: https://clinicaltrials.gov/study/NCT03589014
73. Ali B, Shenkar R, Lee J, Alcazar-Felix RJ, Thompson RE, Stadnik A, et al. Recurrent Symptomatic Hemorrhage in Cerebral Cavernous Malformations After Discontinuation of Atorvastatin or Placebo. J Am Heart Assoc. 2026;15(3):e46943. doi:10.1161/jaha.125.046943. PMID: 41608890. PMC13055477.
74. Li W, Shenkar R, Detter MR, Moore T, Benavides C, Lightle R, et al. Propranolol inhibits cavernous vascular malformations by β1 adrenergic receptor antagonism in animal models. J Clin Invest. 2021;131(3):144893. doi:10.1172/jci144893. PMID: 33301422. PMC7843213.
75. Oldenburg J, Malinverno M, Globisch MA, Maderna C, Corada M, Orsenigo F, et al. Propranolol Reduces the Development of Lesions and Rescues Barrier Function in Cerebral Cavernous Malformations: A Preclinical Study. Stroke. 2021;52(4):1418-1427. doi:10.1161/strokeaha.120.029676. PMID: 33618555.
76. Shenkar R, Moore T, Benavides C, Lightle R, Detter MR, Hobson N, et al. Propranolol as therapy for cerebral cavernous malformations: a cautionary note. J Transl Med. 2022;20(1):160. doi:10.1186/s12967-022-03360-4. PMID: 35382850. PMC8981698.
77. Zuurbier SM, Hickman CR, Rinkel LA, Berg R, Sure U, Al-Shahi Salman R, et al. Association Between Beta-Blocker or Statin Drug Use and the Risk of Hemorrhage From Cerebral Cavernous Malformations. Stroke. 2022;53(8):2521-2527. doi:10.1161/strokeaha.121.037009. PMID: 35410492. PMC9311291.
78. ClinicalTrials.gov [Internet]. Bethesda (MD): National Library of Medicine. Search for “cerebral cavernous malformation”, “cavernous angioma”, “cavernoma”, “OV888”, “NRL-1049”, “BA-1049” via API v2 [cited 2026 Oct 10]. Available from: https://clinicaltrials.gov
79. Erel BC, Uslu I, Kara B, Turkmenel MG, Celikkiran ZE, Gultekin E, et al. The effect of statins on the risk of bleeding in cerebral cavernous malformations: A systematic review and meta-analysis. Surg Neurol Int. 2026;17:182. doi:10.25259/sni_73_2026. PMID: 41952755. PMC13054363.
80. Polster SP, Stadnik A, Akers AL, Cao Y, Christoforidis GA, Fam MD, et al. Atorvastatin Treatment of Cavernous Angiomas with Symptomatic Hemorrhage Exploratory Proof of Concept (AT CASH EPOC) Trial. Neurosurgery. 2019;85(6):843-853. doi:10.1093/neuros/nyy539. PMID: 30476251. PMC7054711.
81. Sader G, Alcazar-Felix RJ, Jhaveri A, Ali B, Stadnik A, Lee J, et al. Functional and Patient-Reported Outcomes in the Atorvastatin versus Placebo Trial of Hemorrhagic Cerebral Cavernous Malformations: An Exploratory Study. Cerebrovasc Dis. 2026;:1-9. doi:10.1159/000552209. PMID: 42030203. PMC13186120.
82. Alcazar-Felix RJ, Thompson RE, Stadnik A, Kinkade S, Sader G, Jhaveri A, et al. Relevance of Lesion Volume as an Outcome in Cerebral Cavernous Malformation Drug Trial: Exploratory Analyses in Atorvastatin Treatment of Cavernous Angiomas with Symptomatic Hemorrhage Exploratory Proof of Concept. Neurosurgery. 2026;98(6):1395-1401. doi:10.1227/neu.0000000000003814. PMID: 41117517.
83. Zuurbier SM, Hickman CR, Tolias CS, Rinkel LA, Leyrer R, Flemming KD, et al. Long-term antithrombotic therapy and risk of intracranial haemorrhage from cerebral cavernous malformations: a population-based cohort study, systematic review, and meta-analysis. Lancet Neurol. 2019;18(10):935-941. doi:10.1016/s1474-4422(19)30231-5. PMID: 31401075. PMC6744367.
84. Musmar B, Salim H, Abdelgadir J, Spellicy S, Adeeb N, Zomorodi A, et al. Antithrombotic Therapy in Cerebral Cavernous Malformations: A Systematic Review, Meta-Analysis, and Network Meta-Analysis. J Am Heart Assoc. 2024;13(6):e032910. doi:10.1161/jaha.123.032910. PMID: 38471833. PMC11010038.
85. Salim HA, Lakhani DA, Hajjeh O, Essibayi MA, Chen H, Balar A, et al. Antithrombotic therapy and risk of intracranial hemorrhage in patients with cerebral cavernous malformations: a multicenter propensity-matched cohort study. J Neurol. 2026;273(6):365. doi:10.1007/s00415-026-13712-5. PMID: 42240840. PMC13236725.
86. Erdur H, Scheitz JF, Tütüncü S, Fiebach JB, Endres M, Werring DJ, Nolte CH. Safety of thrombolysis in patients with acute ischemic stroke and cerebral cavernous malformations. Stroke. 2014;45:1846-8. PMID: 24788970.
87. Powers WJ, Rabinstein AA, Ackerson T, Adeoye OM, Bambakidis NC, Becker K, Biller J, Brown M, Demaerschalk BM, Hoh B, Jauch EC, Kidwell CS, Leslie-Mazwi TM, Ovbiagele B, Scott PA, Sheth KN, Southerland AM, Summers DV, Tirschwell DL. Guidelines for the early management of patients with acute ischemic stroke: 2019 Update to the 2018 Guidelines for the Early Management of Acute Ischemic Stroke: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association. Stroke. 2019;50:e344-e418. PMID: 31662037.
88. Prabhakaran S, Gonzalez NR, Zachrison KS, Adeoye O, Alexandrov AW, Ansari SA, et al. 2026 Guideline for the Early Management of Patients With Acute Ischemic Stroke: A Guideline From the American Heart Association/American Stroke Association. Stroke. 2026;57(8):e316-e436. doi:10.1161/str.0000000000000513. PMID: 41582814.
89. Zuurbier SM, Santos AN, Flemming KD, Schmidt B, Jabbarli R, Lanzino G, et al. Female Hormone Therapy and Risk of Intracranial Hemorrhage From Cerebral Cavernous Malformations: A Multicenter Observational Cohort Study. Neurology. 2023;100(16):e1673-e1679. doi:10.1212/wnl.0000000000206888. PMID: 36754635. PMC10115495.
90. European Medicines Agency. EU/3/17/1948: Orphan designation for the treatment of familial cerebral cavernous malformation (4-hydroxy-2,2,6,6-tetramethylpiperidine-N-oxyl; tempol) [Internet]. Amsterdam: EMA; 2017 Dec 12 [updated 2022 Oct 11; cited 2026 Oct 10]. Available from: https://www.ema.europa.eu/en/medicines/human/orphan-designations/eu-3-17-1948
91. European Medicines Agency. EU/3/14/1387: Orphan designation for the treatment of familial cerebral cavernous malformations (exisulind) [Internet]. Amsterdam: EMA; 2014 Dec 16 [cited 2026 Oct 10]. Available from: https://www.ema.europa.eu/en/medicines/human/orphan-designations/eu-3-14-1387
92. European Medicines Agency. EU/3/22/2713: Orphan designation for the treatment of familial cerebral cavernous malformations (propranolol) [Internet]. Amsterdam: EMA; 2022 Nov 11 [cited 2026 Oct 10]. Available from: https://www.ema.europa.eu/en/medicines/human/orphan-designations/eu-3-22-2713
93. European Medicines Agency. EU/3/22/2714: Orphan designation for the treatment of familial cerebral cavernous malformations (atorvastatin) [Internet]. Amsterdam: EMA; 2022 Nov 11 [cited 2026 Oct 10]. Available from: https://www.ema.europa.eu/en/medicines/human/orphan-designations/eu-3-22-2714
94. Madden S, Awad I, Lopez-Toledano MA, Morrison L, Gutierrez J, Ngo LY, et al. Safety, Tolerability, and Pharmacokinetics of NRL-1049, a Rho-Associated Kinase Inhibitor, in Healthy Volunteers: A Phase 1, First-in-Human, Single-Ascending Dose, Randomized, Placebo-Controlled Trial. CNS Drugs. 2025;39(9):865-877. doi:10.1007/s40263-025-01198-0. PMID: 40627117. PMC12354486.
95. Recursion Pharmaceuticals Inc. A Two-Part Study of REC-994 in the Treatment of Adults With Symptomatic Cerebral Cavernous Malformation (CCM) (SYCAMORE). ClinicalTrials.gov identifier: NCT05085561 [updated 2026 Jun; cited 2026 Oct 10]. Available from: https://clinicaltrials.gov/study/NCT05085561
96. Alliance to Cure Cavernous Malformation. Recursion's REC-994 [Internet]. 2026 May 12 [cited 2026 Oct 10]. Available from: https://www.alliancetocure.org/home/research-clinical-trials/completed-research-projects/rec-994/
97. Alfa R, Considine T, Virani S, Pfeiffer M, Donato A, Dickerson D, et al. Clinical pharmacology and tolerability of REC-994, a redox-cycling nitroxide compound, in randomized phase 1 dose-finding studies. Pharmacol Res Perspect. 2024;12(3):e1200. doi:10.1002/prp2.1200. PMID: 38655895. PMC11040693.
98. Beijing Tiantan Hospital. A Study on the Safety and Preliminary Efficacy of the Anti-CD38 Monoclonal Antibody Daratumumab in the Treatment of Refractory Familial Cerebral Cavernous Malformations. ClinicalTrials.gov identifier: NCT07026604 [posted 2025 Jun 18; cited 2026 Oct 10]. Available from: https://clinicaltrials.gov/study/NCT07026604
99. Al-Shahi Salman R, Adamczuk K, Parker RA, Weir CJ, Latini R, Kim H, et al. Developing a Randomised Efficacy PREcision medicine Platform trial design for Cavernomas: the CARE PREP study. Southampton (UK): National Institute for Health and Care Research; 2025 (Efficacy and Mechanism Evaluation). Available from: https://europepmc.org/books/NBK613140. PMID: 40168499.
100. Lanfranconi S, Scola E, Novelli D, Poggesi A, Pescini F, Pavanello M, et al. Natural history of familial cerebral cavernous malformations: the CCM_Italia cohort study. Front Neurol. 2025;16:1668098. doi:10.3389/fneur.2025.1668098. PMID: 41561324. PMC12812676.
101. Kalani MY, Zabramski JM. Risk for symptomatic hemorrhage of cerebral cavernous malformations during pregnancy. J Neurosurg. 2013;118:50-5. PMID: 23039155.
102. Joseph NK, Kumar S, Brown RD Jr, Lanzino G, Flemming KD. Influence of Pregnancy on Hemorrhage Risk in Women With Cerebral and Spinal Cavernous Malformations. Stroke. 2021;52(2):434-441. doi:10.1161/strokeaha.120.031761. PMID: 33493052.
103. Palermo M, Albanese A, Doglietto F, Olivi A, Sturiale CL. Cerebral cavernous malformations in pregnancy: A systematic review of case reports and case series of hemorrhagic risk and outcomes. Neurosurg Rev. 2026;49(1):176. doi:10.1007/s10143-026-04136-w. PMID: 41612088. PMC12855374.
104. Zheng JJ, Weng JC, Xu XY, Guo J, Yuan WC, Zhang GJ, et al. Association Between Pregnancy and Symptomatic Hemorrhage in Women With Brainstem Cavernous Malformations and Bleeding in the Preceding Year: A Prospective Multicenter Study. J Am Heart Assoc. 2026;15(14):e049628. doi:10.1161/jaha.126.049628. PMID: 42466515. PMC13477402.
105. Al-Shahi R, Bhattacharya JJ, Currie DG, Papanastassiou V, Ritchie V, Roberts RC, Sellar RJ, Warlow CP, et al. Prospective, population-based detection of intracranial vascular malformations in adults: the Scottish Intracranial Vascular Malformation Study (SIVMS). Stroke. 2003;34:1163-9. PMID: 12702837.
106. Domingues F, Gasparetto EL, Andrade R, Noro F, Eiras A, Gault J, et al. Familial cerebral cavernous malformations: Rio de Janeiro study and review of the recommendations for management. Arq Neuropsiquiatr. 2008;66(4):795-799. doi:10.1590/s0004-282x2008000600003. PMID: 19099113.
107. Alliance to Cure Cavernous Malformation [Internet]. [cited 2026 Oct 10]. Available from: https://www.alliancetocure.org
108. Cavernoma Alliance UK (CAUK) [Internet]. [cited 2026 Oct 10]. Available from: https://www.cavernoma.org.uk
109. CCM Italia — La Community Italiana per le Malformazioni Cavernose Cerebrali [Internet]. Torino: Università di Torino; [cited 2026 Oct 10]. Available from: https://www.ccmitalia.unito.it