Developmental venous anomalies and cerebral cavernous malformations: Partners in crime

Abstract

Developmental venous anomalies (DVAs) are benign congenital veins that collect normal brain drainage into a single outlet. Cerebral cavernous malformations (CMs) are clusters of thin-walled capillary cavities prone to bleeding. When both lesions coexist, the DVA’s altered venous pressure and flow can promote CM formation or rupture. Detecting a DVA abutting an otherwise unexplained intracerebral hemorrhage can therefore raise suspicion of an occult CM as a likely cause, a clue which may be invaluable for daily clinical practice. The main focus of this review is to acknowledge the hallmark imaging appearances of DVAs and CMs, as well as their coexistence, explore the clinical consequences of mixed lesions, and emphasize that recognizing their partnership is vital for an accurate, timely diagnosis and appropriately targeted management.

Key Words: Vascular malformations; Developmental venous anomaly; Cavernous malformation; Brain magnetic resonance imaging; Central nervous system; Neuroimaging; Cerebral venous circulation; Neuroradiology; Radiology review

Core Tip: When developmental venous anomalies (DVAs) and cerebral cavernous malformations (CMs) coexist, the DVA’s altered venous pressure and flow can promote CM formation or rupture. Detecting a DVA next to an otherwise unexplained intracerebral hemorrhage can therefore raise suspicion of an occult CM as the underlying etiology. This finding can represent a clue that may be invaluable for daily clinical practice. This review aims to educate readers on the hallmark imaging appearances of DVAs and CMs, as well as their coexistence, and emphasize how recognizing their partnership is vital for an accurate, timely diagnosis, which will allow for appropriately targeted management and the avoidance of unnecessary examinations or interventions.

INTRODUCTION

Vascular malformations (VMs) of the brain are generally rare and are commonly classified according to their hemodynamics (typically as low- or high-flow) or according to the kind of blood vessel that can be found histologically[1]. Developmental venous anomalies (DVAs), cavernous malformations (CMs), and capillary telangiectasias all involve the venous circulation.

DVAS

Definition

The term DVA, previously venous angioma, cerebral venous malformation, or cerebral venous medullary malformation, was introduced by Lasjaunias et al[2] to emphasize its role as a variation of the brain venous microarchitecture rather than a true VM[3]. DVAs form when radially-oriented veins converge into a collecting vein draining into the deep or superficial venous system[4]. Some have multiple collecting veins and drain into both systems, with one predominant. Due to this morphology, on imaging studies, they resemble the caput medusae (head of medusa)[5].

Epidemiology

DVAs are the most common VMs, accounting for approximately 50% of all malformations observed in magnetic resonance imaging (MRI) studies[6]. Their incidence is estimated to be as high as 2.6% in older autopsy studies, while more recent, imaging-based studies have reported an even higher incidence of up to 6.4%[4,5,7]. They are often discovered incidentally during brain imaging, such as computed tomography (CT) and MRI.

Physiology - location

DVAs collect blood from the surrounding normal parenchyma and may act as a compensatory venous channel[8]. They represent an alternative venous outflow in case of the absence of normal subependymal or pial veins[3]. There are reported fatal complications after the surgical removal of a DVA, probably due to their important role as compensatory channels[5,9]. DVAs are more often located in the supratentorial brain parenchyma (70%) rather than the infratentorial brain (14%-29%). Supratentorially, they are predominantly found in the frontal lobes (36%-56%), followed by the parietal (12%-24%), temporal (2%-19%), and occipital (4%) lobes. Infratentorially, they are located in the cerebellum (14%-29%) and brainstem (< 5%)[10-12]. They have a wide range in size, ranging from just a small volume of the brain parenchyma to a whole brain hemisphere[5].

Histologically, they lack malformed or neoplastic elements, consisting of mature venous walls without arterial or capillary components. They may show degeneration, such as hyalinized and thickened walls. Unlike “true” VMs, their walls contain normal structural proteins, whereas VMs express angiogenic growth factors and exhibit histological fragility[6]. “Arterialized” or “fistulized” DVAs, also known as mixed-type malformations or intracerebral venous angiomas with arterial supply, are VMs that combine a DVA with an arterial component. They retain the characteristic structure of a DVA while exhibiting arteriovenous shunting, which becomes apparent during the arterial phase of angiographic imaging[5,6].

Pathogenesis - etiology

Their etiology remains unclear. The prevailing hypothesis suggests that DVAs arise from errors in venous development during embryogenesis, involving aberrant formation of normal veins and their tributaries[13,14]. As a compensatory response, the brain develops alternative venous drainage pathways during gestation and early infancy. This adaptive process implies the occurrence of an underlying pathological event in utero. Reports on DVAs in neonates are much fewer than in adults[15,16]. Despite the widespread use of MRI (in utero) and cranial ultrasound (cUS), there is limited information on their prevalence in the pediatric population[17]. Their incidence increases with age, with Brinjikji et al[15] reporting a prevalence of 1.5%, 7.1%, 9.6%, and 9.6% in neonates, toddlers, children, and adults, respectively. This discrepancy may be due to underreporting, as they are considered benign. It also suggests that DVAs may either develop postnatally or remain dormant until required.

Clinical presentation, complications, and associations

Before the widespread utilization of CT and MRI, DVAs were thought to cause major hemorrhagic events and seizures. Today, they are considered benign and asymptomatic. However, in a very small number of cases, they can lead to symptoms due to mechanical or flow-related complications[18]. Pereira et al[19], examining both adults and children, excluding cases with co-existing CMs, identified several mechanisms responsible for symptoms. These included thrombosis leading to infarction or hemorrhage, increased inflow, decreased outflow, mass effect, and parenchymal injury due to elevated venous pressure[19]. Most of the current evidence on symptomatic DVAs comes from case reports and series, making it difficult to determine their incidence and prevalence. When symptomatic, they may present with headache, seizures, focal deficits, dizziness, and ataxia, depending on their location and mechanism[19,20]. It is hypothesized that the brain parenchyma, which is drained by a DVA, shows less physiological reserves and is more vulnerable to injury. Since they constitute a less efficient drainage route, with time, they result in venous hypertension locally[11,14]. The venous congestion may result from stenosis or thrombosis of the collector veins or increased resistance in the vessel walls[21]. Demyelination, gliosis, ischemic changes, and neuronal degeneration have been observed in the brain parenchyma surrounding DVAs[11]. Thrombosis of the collecting vein is rare and can present in various ways, such as venous infarction, parenchymal hemorrhage, edema, or subarachnoid and intraventricular hemorrhage[20,21]. Current belief is that most hemorrhagic events result either from associated regional CMs or as a consequence of thrombosis of the DVA[5,22]. The annual hemorrhage risk of DVA-only lesions is 0%-1.28%, whereas most clinically significant hemorrhages arise when a CM coexists with a DVA[6,23,24].

While VMs are a known factor for epilepsy and seizures, there is limited evidence supporting DVAs as epileptogenic foci[23,25]. They may be identified during an epilepsy workup, but usually no direct correlation can be found. Some reports have mentioned seizures caused by DVA complications, such as thrombosis or hemorrhage, but direct links are rare[25]. In most cases, the seizures can be attributed to an accompanying cryptic VM or another epileptogenic lesion. DVAs can also be present in regions of malformed cortical development, such as focal cortical dysplasia, pachygyria, or polymicrogyria, which could point to a common developmental event[26]. For patients with seizures and an incidental DVA, a thorough evaluation is needed to identify any other potential causes[6].

In case of a headache, the presence of a DVA is usually an incidental finding during imaging, as most headaches resolve over time with no intervention[3,24]. However, a complicated DVA could lead to a headache[6]. Mass effect on surrounding structures can lead to various complications depending on the location of the DVA, such as cranial nerve compression or hydrocephalus[19]. There are reports of trigeminal neuralgia, tinnitus, and hemifacial spasm attributed to a DVA located in the cerebellopontine angle affecting the trigeminal and facial nerves, respectively[27-30].

DVAs have also been associated with atrophy of the surrounding parenchyma, likely due to altered venous outflow and chronic venous congestion[18]. Uchino et al[31] described a case with severe brain hemiatrophy in the context of a large DVA with enlarged, varicose drainage veins. Another reported finding is the presence of basal ganglia calcifications[18,32]. These calcifications may result from prior hemorrhage or venous congestion[5].

Understandably, any DVA that presents with neurological symptoms should prompt further investigations to exclude a cryptic accompanying VM or other lesion[6]. DVAs commonly coexist with other VMs, particularly CMs (as will be discussed below in detail). They can also coexist with sinus pericranii, a VM comprised of a communication between intracranial and extracranial veins via dilated venous channels[33]. Up to 50% of these lesions have been associated with DVAs and act as their draining channel[5,18]. Some authors support that sinus pericranii could represent an extraaxial DVA, suggesting they represent a variant anatomy[33,34]. DVAs are encountered in approximately 20% of patients with superficial venous malformations of the head and neck, which is higher than the general population[35,36]. When both a maxillofacial VM and a DVA are present, the criteria for cerebral venous metameric syndrome are met. This syndrome is a complex craniofacial disorder involving VMs in soft tissues, neural structures, and bones[37].

DVAs are also associated with lymphatic malformations of the orbital and periorbital regions[38]. DVAs can be observed in patients with blue rubber bleb nevus syndrome, a rare neurocutaneous disorder characterized by multifocal venous VMs of the skin (hands and soles) and gastrointestinal tract[39]. Other rarer genetic syndromes linked to DVAs are constitutional mismatch repair deficiency syndrome, Cowden syndrome, and diffuse intrinsic pontine gliomas[18,40]. For example, a study found that 59.1% of patients with Cowden syndrome had at least one DVA, and 36.4% had multiple DVAs[41]. In addition, CMs (discussed below in detail) can also sometimes be found in individuals with Cowden syndrome[41,42]. While neither DVAs nor CMs are core diagnostic features of Cowden syndrome, their identification on brain MRI, along with other findings (i.e., dysplastic cerebellar gangliocytoma), can prompt genetic testing and a more comprehensive evaluation for the condition. Nonetheless, a diagnosis of Cowden syndrome will typically be based on a combination of clinical findings, including mucocutaneous lesions, macrocephaly, and cancer risks, along with confirmation through phosphatase and tensin homolog gene mutation analysis[41].

There is growing interest in exploring a potential association between DVAs and demyelinating diseases, particularly multiple sclerosis. Jung et al[43] have published a case report demonstrating focal demyelination around a cerebellar DVA. Additionally, Rogers et al[44] published a small case series with patients with multiple sclerosis and accompanying DVAs. However, current evidence is poor, and further research is required.

IMAGING FINDINGS (DVAS)

DVAs are usually encountered incidentally on post-contrast brain CT and MRI and on conventional and digital subtraction angiography (DSA)[45]. They can also be assessed using advanced techniques, typically performed for other reasons, such as cUS in neonates, single photon emission CT, magnetic resonance spectroscopy, and perfusion studies, including arterial spin labeling (ASL) perfusion and bolus perfusion-weighted imaging (PWI)[46,47].

CT

On non-contrast CT, smaller DVAs may not be visualized, while larger ones with a patent collecting vein may appear isodense or slightly hyperdense compared to the brain cortex. A thrombosed collecting vein may appear markedly hyperdense[45]. Non-contrast scans can also assess locoregional parenchymal abnormalities (i.e., calcifications) in the drainage territory[48]. Post-contrast brain CT, typically performed 2-3 minutes after contrast administration, provides clearer visualization of the characteristic “caput medusae” morphology and the collecting vein, both of which enhance[10]. A “caput medusae” morphology is defined as multiple radially oriented veins that converge centrally to the collecting channel[49]. Dedicated CT angiography, usually with bolus tracking, and/or venography with 50 second-delay can provide more information on the morphology of the DVA but are usually not necessary[3,5]. Some DVAs may show ampullary dilatation of their proximal portion at the convergence of the medullary veins[10]. Newer CT technologies with dynamic subtracted angiography provide more detailed morphological visualization[5].

Parenchymal changes surrounding a DVA are not a rare occurrence and can be seen in up to 65%, as reported by San Millán Ruíz et al[10]. According to their study, the most common finding was atrophy (29.7%), followed by white matter lesions (WML) (19.3%) and dystrophic calcifications (9.6%)[10]. Brain atrophy varies, ranging from focal enlargement of a sulcus or ventricle to atrophy of the entire drainage area or even a brain hemisphere[13,31,50]. WML can occasionally be assessed with CT, but is more frequently observed in MRI due to its higher sensitivity. They resemble leukoaraiosis from other causes, such as microvascular disease, and appear as well-circumscribed non-enhancing hypodensities around the DVA[10]. Dystrophic calcifications are not unique to DVAs and are also described in other VMs, such as dural arteriovenous fistulas and in Sturge-Weber syndrome[48]. They are often overlooked due to the benign nature of DVAs, but can be seen in the surrounding white matter or basal ganglia[10]. They are better appreciated in CT rather than MRI studies and are typically unilateral[48].

MRI

DVAs are often detectable across various MRI sequences but are most clearly visualized on post-contrast T1-weighted (T1w) and susceptibility-weighted imaging (SWI) sequences[7]. Smaller DVAs can be overlooked, even in post-contrast series. In a study by Gökçe et al[7], almost half of the DVAs were not visible on non-contrast sequences. The collecting vein is more easily visualized than the “caput”[12]. DVAs typically appear as flow voids on both T1w and T2w sequences and may demonstrate phase-shift artifact[3]. The radicles may sometimes have a high signal at fluid-attenuated inversion recovery (FLAIR) images, which should not be confused with parenchymal changes[51]. They show homogeneous enhancement after gadolinium contrast administration. Their visualization is enhanced by three-dimensional (3D) gradient-echo (GE) T1w sequences, which provide high-resolution imaging and allow for multiplanar reconstructions[45]. Application of 3D sequences can also allow better morphological assessment, such as stenosis of the collecting vein[5]. MR venography and time-of-flight angiography are usually not necessary, and their role is to evaluate whether there are signs of arterial components[4]. Nonetheless, non-arterialized but large DVAs may also be visualized at time-of-flight images.

SWI is very sensitive in detecting pathologies that cause inhomogeneity of the magnetic field[45]. A DVA appears as a low-signal structure on SWI and T2* sequences. In phase imaging, its signal remains consistent with other veins but is inverted relative to that of calcifications[45]. SWI provides a more detailed assessment of the entire structure and has been shown to be more sensitive than conventional T2*-weighted imaging. SWI is not significantly affected by low velocities, making it very effective to assess low-flow lesions such as DVAs[52,53]. SWI in high and ultra-high field strengths (3T and 7T) can better visualize a DVA[54].

MRI is superior to CT in evaluating parenchymal abnormalities[3]. WML appears as high-signal areas on T2/FLAIR and low-signal areas in T1w sequences without diffusion restriction, mass effect, or contrast enhancement[5]. 3D FLAIR is more sensitive in detecting these lesions[51]. They can range from small foci to a large confluent lesion in the drainage territory[10]. Their incidence ranges from 11.6%-30.7% and tends to occur more frequently in older patients[10,11,55,56]. They likely represent edema, gliosis, ischemia, demyelination, or glial metaplasia[55,56]. A case series by Jung et al[56] evaluated the apparent diffusion coefficient values of the WML and showed increased values compared to normal parenchyma, possibly due to congestion. Takasugi et al[57] found hypointense foci on SWI within the drainage territory in 59% of cases, which were more common with WML, likely representing microhemorrhages or tiny CMs.

Functional MRI (fMRI) depends on blood oxygen level-dependent contrast, and changes are attributed to both extravascular tissue and local capillaries and veins. Theoretically, DVAs can contribute to the fMRI signal and can potentially lead to pseudoactivations during presurgical mapping[58]. Various examples of DVAs on CTs and MRIs are shown in Figures 1 and 2.

Figure 1 Developmental venous anomalies. A and B: Sagittal (A) and axial (B) maximum intensity projection reconstruction of a contrast-enhanced computed tomography venography demonstrate a very large developmental venous anomaly in the right frontal lobe, evident by the “caput medusae” morphology (dashed oval) converging in collector veins (thin arrows) and draining in the right internal cerebral vein (thick arrow). Developmental venous anomalies are rarely this large and are usually not easily discernible in computed tomography scans; nonetheless, this example is provided for educational purposes due to its size and characteristic “caput medusae” morphology; C-E: Maximum intensity projection, of T1-weighted sagittal (C), coronal (D) and axial reconstruction images (E) following intravenous gadolinium administration demonstrate a developmental venous anomaly in the left parietal lobe, apparent by the characteristic “caput medusae” (dashed circle) and draining veins (arrows).

Figure 2 Developmental venous anomalies. A and B: T1-weighted (T1w) axial (A) image following intravenous gadolinium administration, and (B) susceptibility-weighted image demonstrate ectopic veins in the right cerebellar hemisphere (dashed oval); C: Maximum intensity projection of a T1w axial image following intravenous gadolinium administration clearly displays the convergence of the ectopic veins into a singular draining vein (arrow), findings consistent with a developmental venous anomaly; D: Axial non-contrast enhanced computed tomography image performed due to headaches demonstrates hyperdense linear abnormalities in the left frontal lobe (dashed circle) initially perceived as suspicious for subarachnoid hemorrhage. The possibility of the finding representing a vascular structure was also entertained due to the slightly hyperdense appearance of the blood pool in additional vascular structures such as the superior sagittal sinus (thick arrow) and straight sinus (curved arrow); E and F: Susceptibility-weighted axial images from a subsequent magnetic resonance imaging reveal the finding to represent ectopic veins (dashed oval) with a “caput medusae” appearance, converging into a draining vein (thin arrow), findings consistent with a developmental venous anomaly; G-I: Maximum intensity projection of a T1w (G) coronal and axial reconstruction image (H) following intravenous gadolinium administration demonstrate a large developmental venous anomaly in the right cerebellar hemisphere, apparent by the characteristic “caput medusae” (dashed circle) and the draining vein (arrow); I: Maximum intensity projection coronal reconstruction of a phase contrast venography display the collector vein (arrow) draining in the convergence of the right transverse with the superior sagittal dural venous sinus (thick arrow).

Perfusion

Perfusion studies such as PWI, ASL, and CT have shown that the normal-looking parenchyma surrounding a DVA may show hemodynamic abnormalities, even in uncomplicated DVAs[47,59]. Hanson et al[47] performed CT perfusion studies and found increased cerebral blood volume (CBV), cerebral blood flow (CBF), and mean transit time (MTT) in the hemisphere affected by DVAs, indicating altered perfusion. Sharma et al[60] published a case series on PWI analysis of DVAs. The analysis revealed increased relative CBV, relative CBF (rCBF), and relative MTT, indicating hemodynamic changes. DVAs with associated CMs showed greater MTT prolongation and a lesser rCBF increase, suggesting stronger venous outflow restrictions[60]. Research has shown that GE sequences are more sensitive to blood flow in larger vessels compared to spin-echo[61]. While GE sequences offer higher contrast-to-noise ratios, their increased susceptibility effects around large vessels can result in an overestimation of CBF and CBV in these areas[59]. Although perfusion changes with DVAs are mostly common on PWI, they can also be present on ASL studies as well; however, they are more unusual (Figure 3)[46]. Larvie et al[62] reported that 76% of cases showed metabolic abnormalities on PET scans in the surrounding brain parenchyma, most commonly presenting as glucose hypometabolism, with greater severity in older patients. These hemodynamic differences may contribute to CM formation and potential hemorrhagic risk. Nonetheless, existing data are limited, and additional studies are needed to clarify hemodynamic patterns and their clinical implications.

Figure 3 Developmental venous anomaly image. A and B: T1-weighted axial (A) and coronal (B) images following intravenous gadolinium administration demonstrate a developmental venous anomaly in the left temporal lobe, evident by the characteristic “caput medusae” appearance and the draining vein (dashed circle); C and D: When compared to the contralateral side, dynamic susceptibility contrast magnetic resonance perfusion demonstrates increased relative cerebral blood volume (C), and arterial spin-labeling perfusion displays increased relative cerebral blood flow (D) (dashed oval) in the corresponding territory of the developmental venous anomaly.

DSA

CT and/or MRI findings alone are adequate to characterize a VM as a DVA. Angiography has a role in symptomatic DVAs and those with atypical presentations, such as hemorrhage, in order to determine the true cause of bleeding[3,45]. A DVA opacifies at the same time as normal veins, although those located in the frontal lobes may opacify earlier, showing a capillary blush[59]. In case of hemorrhage, DSA clearly demonstrates the “caput medusae” in the area affected by hemorrhage with signs of venous obstruction, such as venous stasis, a missing collector (in case of thrombosis), and delayed washout[7]. DVAs with accompanying VMs, arteriovenous shunting, and “arterialized” DVAs demonstrate capillary blush during the arterial phase and early venous filling[5,7,45].

cUS

Horsch et al[16] described the imaging findings of DVAs in a case series using cUS. The vast majority demonstrated hyperechoic brain parenchyma surrounding a DVA, a finding that did not correlate with MRI findings in most cases[16,51]. Most of the DVAs exhibited venous waveforms on Doppler, while a few showed arterial-like form[16]. It is unknown whether those lesions represented “arterialized” DVAs.

Treatment options

In the vast majority of cases, DVAs do not require follow-up or treatment. In case of thrombosis, there are reports of successful anticoagulation therapy[20,63-65]. Current belief is to treat a thrombosed DVA similarly to cortical or dural venous sinus thrombosis and to screen for prothrombotic conditions[3,5]. However, there is not enough evidence to support it[3,14]. The decision to initiate anticoagulation should be carefully assessed, considering the potential risk of hemorrhage. Resection of a DVA is contraindicated, as they act as compensatory venous channels and previous resections have led to fatal complications due to venous infarction[1,5,9]. A neurosurgeon should be informed in advance to avoid accidental resection. If surgery is necessary due to a complication, such as hemorrhage or edema, careful preservation of the DVA’s collecting vein is crucial to prevent complications[6,20]. Hemorrhagic “arterialized” DVAs are believed to follow a more aggressive clinical course with higher hemorrhage risk[5,6]. They have been managed using surgery, proton beam radiosurgery, embolization, and gamma knife treatment[14,66-69]. Unintentional surgical removal of the DVA component led to acute deterioration, while gamma knife treatment targeting the DVA resulted in radiation necrosis, likely due to venous congestion[68,70]. Treating asymptomatic “arterialized” DVAs should be carefully considered on a case-by-case basis, targeting the arterial component.

CEREBRAL CMS

Definition and epidemiology

CMs, also known as cavernous angiomas, cavernous hemangiomas, cryptic VMs, or cavernomas, are low-flow vascular lesions, mainly encountered in the supratentorial region (63%-81%) but could also be found in infratentorial white matter, mainly the pons and cerebellar hemispheres, as well as the spinal cord. Although relatively rare in the general population (0.4% to 0.8%), they represent the second most common cerebral VM, accounting for approximately 5%-15% of all vascular anomalies[71]. The majority of cases (85%) arise sporadically, while the remaining 15% are associated with familial inheritance or radiation exposure[71]. No gender differences have been reported[72]. CMs are occasionally found alongside a DVA, a condition known as a mixed VM. Recognizing the presence of an associated DVA on imaging is crucial, as it can assist in correct diagnosis and guide surgical decisions, since DVAs serve as essential venous drainage routes, and any disruption can lead to venous infarction or increased complications. The presence of a DVA is more commonly linked to the nongenetic familial variant of the disease[73]. Moreover, occasionally, CMs coexist with capillary telangiectasia, indicating that these lesions may be part of the same pathological entity[74,75].

Clinical presentation

In most cases, they are asymptomatic and discovered incidentally. Clinical manifestations include hemorrhage, epileptic seizures, and neurological deficits. Hemorrhagic risk is estimated between 2% to 6% per year and increases with a history of previous hemorrhage, brainstem location, as well as familial predilection[76,77]. Furthermore, factors such as age, female sex, deep white matter location, size, multiplicity, and the presence of associated DVAs have not been associated with increased hemorrhagic risk[77]. The annual incidence of seizures in individuals with CMs ranges between 1.5% and 2.4%[76].

Pathology

They are composed of abnormally dilated, thin-walled capillary networks devoid of normal intervening brain parenchyma. Their pathophysiology is multifactorial, encompassing genetic mutations, endothelial dysfunction, pro-inflammatory signaling pathways, and aberrant angiogenic processes[78]. In familial CMs, three primary gene mutations (CCM1, CCM2, and CCM3) are implicated in the disease pathogenesis. These mutations disrupt cell adhesion, migration, angiogenesis, and endothelial barrier integrity, resulting in the formation of fragile, dilated capillaries that are highly susceptible to leakage and hemorrhage[76].

IMAGING FINDINGS (CMs)

CT

The sensitivity of CT in detecting CMs is limited, with a detection rate ranging between 30% and 50%[79]. Sensitivity decreases significantly in smaller lesions (< 1 cm)[74,75]. CMs do not exhibit distinct imaging characteristics; they typically appear as well-defined lesions that may be either hypodense or hyperdense, depending on the presence of calcifications and hemosiderin deposits. Following contrast administration, mild enhancement may be observed. Edema or mass effect is uncommon unless a recent hemorrhage has occurred[80]. CT is typically the initial and preferred imaging modality in acute settings to rule out hemorrhage[81].

MRI

MRI remains the gold standard for detecting and further characterizing CMs[81]. Conventional MRI sequences, particularly T1w and T2w imaging, exhibit high sensitivity and specificity for identifying clinically symptomatic CMs[75]. Their signal characteristics on MRI vary depending on lesion size and hemorrhagic status. Small, punctate lesions typically appear hypointense on T2w imaging. In contrast, larger CMs exhibit a characteristic reticulated pattern of mixed hypo- and hyperintense signals, resembling the so-called “mulberry-like” or “popcorn” appearance on T2 and FLAIR sequences. Increased signal in T1 sequences reflects the presence of methemoglobin, indicative of subacute hemorrhage. A characteristic hypointense rim surrounding the lesion could also be noted. Usually, CMs lack prominent enhancement, while they could exhibit low to moderate enhancement, particularly if there is associated blood-brain barrier disruption, recent hemorrhage, or adjacent gliosis. Additionally, the presence of deoxyhemoglobin and hemosiderin results in susceptibility effects, leading to a decrease in signal intensity on T2w sequences. Thus, GE sequences and particularly T2*-weighted have proven valuable in the detection of CCMS. Multiple studies have demonstrated that T2w GE (GRE) imaging offers increased sensitivity compared to other MRI sequences for the detection of sporadic and familial cases of CM lesions[80].

SWI represents an advanced imaging technique particularly well-suited for the detection of VMs, owing to its high sensitivity to deoxyhemoglobin and iron deposition. SWI has demonstrated higher sensitivity compared to GRE imaging in the detection of CMs, especially in familial cases[82,83]. Moreover, SWI also provides enhanced sensitivity for the detection of associated venous anomalies and potential telangiectasias, especially in cases where contrast media is contraindicated[84]. Nevertheless, the hemosiderin-induced “blooming effect” could artifactually exaggerate the apparent size of the lesions; thus, both GRE and SWI sequences should be interpreted carefully[84]. Additionally, small hemorrhages or cortical vessels could be difficult to distinguish in SWI sequences and to differentiate from CMs. As a result, correlation with spin echo images is essential[80]. Advanced imaging techniques, including DTI and fMRI, have emerged as useful imaging techniques for preoperatively planning these lesions[84]. Multiple studies have investigated the role of high-field MRI in both experimental and clinical contexts in the detection of CM, demonstrating higher diagnostic accuracy and enhanced preoperative planning[85,86]. Overall, high-field MRI combined with SWI imaging improves the diagnostic accuracy for detecting CMs.

In 1994, Zabramski et al[87] proposed an MRI-based classification system for CMs, based on their imaging characteristics, particularly on T1w, T2w, and susceptibility-sensitive sequences, which remains one of the most widely used methods for characterizing these lesions. The classification system is outlined in Table 1.

Table 1 Demonstration of the Zabramski classification system for cavernous malformations.

	T1	T2	GRE/SWI	Clinical manifestation
I (subacute hemorrhagic lesions)	Hyperintense center	Hyper or hypointense center, surrounded by a hypointense (hemosiderin) rim	-	Usually symptomatic
II (classic “popcorn-like” lesions)	Heterogeneous signal	Heterogeneous and hypointense (hemosiderin) rim	Hypointense (hemosiderin) rim with blooming	The most prevalent type of cavernous malformation; recurrent asymptomatic or symptomatic episodes
III (chronic lesions)	Hypointense or isointense	Hypointense or isointense	Hypointense rim with blooming (marked hemosiderin deposition)	Clinically silent
IV (punctate lesions)	Poorly seen	Poorly seen	Small, punctate hypointense foci	Asymptomatic

GRE: T2-weighted gradient-echo; SWI: Susceptibility-weighted imaging.

Type I lesions exhibit a hyperintense core on T1w imaging, indicative of subacute hemorrhage due to methemoglobin accumulation. The lesion’s size and intensity may fluctuate over time, reflecting the dynamic nature of the hemorrhagic process[88]; Type II lesions, considered the most prevalent type, exhibit the classic “popcorn-like” imaging, displaying a heterogeneous core with mixed signal intensities on both T1w and T2w sequences and a well-defined hypointense hemosiderin ring on T2-GRE or SWI images; Type III lesions represent chronic hemosiderotic lesions, appearing hypointense or isointense on T1w and T2w MRI, demonstrating ‘’blooming effect’’ on T2-GRE or SWI, indicating prior hemorrhagic activity; and Type IV lesions include small punctate lesions, mainly visualized on susceptibility-sensitive sequences (T2-GRE or SWI). Classifying lesions as type IV could be challenging, as their exact pathological nature has not been definitively determined[85].

The clinical applicability of the Zabramski classification is still being investigated. A retrospective cohort study evaluating the imaging evolution and clinical trajectory of 255 untreated patients with sporadic CMs over a follow-up period of approximately five years provided evidence that the Zabramski classification may facilitate risk stratification and contribute to treatment planning, particularly in determining the necessity for surgical intervention[89]. Furthermore, a recent study by Saari et al[88] established an association between the radiological characteristics of the Zabramski classification and their clinical relevance, emphasizing that type I lesions have a higher likelihood of becoming symptomatic. Nikoubashman et al[90] suggested an additional category (type V lesions) accounting for cavernomas presenting with gross extralesional hemorrhage. Various examples of CMs on CTs and MRIs, including familial cerebral CM cases, are shown in Figures 4, 5 and 6.

Figure 4 Cerebral cavernous malformation. A: Computed tomography scan performed in a 65-year-old woman due to memory deficits displayed a mildly hyperdense lesion in the right parietal lobe (dashed circle), which raised suspicion for a hemorrhage; B and C: On the subsequent magnetic resonance imaging examination performed, the lesion (dashed circle) was hypointense on T1-weighted images (B), without significant or discernible contrast enhancement following intravenous gadolinium administration (C); D: On T2-weighted images, the lesion (dashed circle) displays a heterogeneously hyperintense center with a low signal periphery (arrows) compatible with a hemosiderin rim; E: Susceptibility-weighted imaging shows intense blooming of the lesion (dashed circle). The findings were consistent with a cerebral cavernous malformation diagnosis.

Figure 5 Familial cerebral cavernous malformation syndrome. A and B: Computed tomography scan performed due to an acute onset headache in a 25-year-old man displayed a heterogeneously hyperdense hemorrhagic lesion in the pons (dashed circle) and an additional mildly hyperdense lesion in the left cerebellar hemisphere (curved arrow); C and D: On the subsequent magnetic resonance imaging examination performed, the pontine lesion (dashed circle) displays a hyperintense signal on T1-weighted (C) and fluid-attenuated inversion recovery images (D) with a low-signal hemosiderin rim (thick arrow); D: Also note the hyperintense signal intensity surrounding the pontine lesion due to vasogenic edema; E-H: Susceptibility-weighted imaging verify the hemosiderin presence and rim (thick arrow) in the pontine lesion (dashed circle) and also display hemosiderin presence with blooming in the left cerebellar hemisphere lesion (curved arrow). Several additional bilateral hemispheric microbleeds are also noted on the susceptibility-weighted images (thin arrows). The findings are suggestive of a Zabramski type I cerebral cavernous malformation (CM) in the pons as indicated by the recent bleed (T1 central hyperintensity) and the surrounding vasogenic edema. The left cerebellar hemispheric lesion represented a Zabramski type II cerebral CM (hemosiderin rim; “popcorn” appearance; blooming; no evidence of recent bleed on the remaining examination). In this setting, the additional bilateral cerebral hemispheric lesions most likely represent Zabramski type IV cerebral CM. The constellation of all the above findings is highly suspicious for familial cerebral CM syndrome.

Figure 6 Familial cerebral cavernous malformation syndrome. Magnetic resonance imaging examination performed due to dizziness in a 50-year-old woman. A-C: A lesion (dashed circle) centered between the cerebellar hemispheres and the scolex is demonstrated. The lesion displays a heterogeneous signal with a few areas of increased signal intensity centrally on T1-weighted images (A), while on T2-weighted (B) and fluid-attenuated inversion recovery (C) images it demonstrates a characteristic “popcorn” appearance with a heterogeneous signal centrally and a low signal hemosiderin rim peripherally; D: T1-weighted axial image following intravenous gadolinium administration demonstrates no discernible enhancement and an ectopic draining vein (curved arrow); E-H: Susceptibility-weighted images verify the hemosiderin presence of the lesion. The findings are suggestive of a Zabramski type II cerebral cavernous malformation with a concomitant developmental venous anomaly (ectopic draining vein) (curved arrow). In addition, susceptibility-weighted images display hemosiderin deposition in the cortical surfaces of the cerebellar and cerebral hemisphere, the pons and midbrain (thick arrows), suggesting superficial siderosis from a prior hemorrhage with subarachnoid extension. Multiple bilateral microbleeds (thin arrows) are also noted, suggesting a diagnosis of familial cerebral cavernous malformation syndrome.

DSA

DSA is not a routine tool for CMs, as they are angiographically occult. DSA has been proven particularly useful in cases with atypical imaging or clinical features, to exclude alternative VMs[81].

Differential diagnosis

The imaging characteristics of CMs are relatively distinctive, with a narrow differential diagnosis. Nevertheless, acute or subacute hemorrhage could obscure their characteristic imaging features, thereby complicating accurate diagnosis. In the setting of recent hemorrhage, affected lesions often appear as well-circumscribed, space-occupying lesions, frequently associated with perilesional edema and mass effect. The differential diagnosis includes both neoplastic and non-neoplastic entities, such as gliomas, metastases, hemorrhagic stroke, and arteriovenous malformations. The diagnostic challenge increases when multiple punctate lesions are detected on SWI/GRE imaging, as their appearance overlaps with various other pathological entities, including hemorrhagic metastatic disease, hypertensive microangiopathy, cerebral amyloid angiopathy, and traumatic diffuse axonal injury[80]. Recurrent hemorrhagic episodes, the presence of a hypointense hemosiderin rim on T2-GRE or SWI, and well-defined lesion encapsulation are key imaging features that strongly support CM diagnosis. On the other hand, significant mass effect, homogeneous single-aged blood products, and expansile hemorrhage are quite uncommon in CMs and more indicative of other pathologies, including arteriovenous malformations[91].

The presence of T1 hyperintensity within perilesional edema surrounding a hemorrhagic lesion has proven to be a valuable diagnostic feature, aiding in the differentiation of CMs from other hemorrhagic lesions[92]. The same authors also reported a significantly higher occurrence of the hypointense signal or rim on T2-weighted images in CMs compared to neoplastic lesions[92]. In cases where CMs occur in atypical locations, such as the cavernous sinus or extraaxial regions, the differential diagnosis becomes more complex, as these lesions may mimic other vascular, neoplastic, or inflammatory lesions, necessitating careful imaging interpretation and, in some cases, additional diagnostic workup. In addition, identifying a DVA within the vicinity of a hemorrhagic brain lesion may strongly suggest an underlying CM as the incriminating etiology, further representing a clue that can assist in the differential diagnosis approach.

Treatment and follow-up

In asymptomatic cases, surgical resection may be an option for a solitary CM in a noneloquent area and easily accessible, aiming to decrease the risk of future hemorrhage. However, surgery is generally recommended for CMs associated with epilepsy, as well as for symptomatic, easily accessible lesions when the surgical risks are comparable to those of living with the condition for approximately two years[81]. Managing infratentorial: CMs are particularly challenging due to the high risk of early postoperative mortality and morbidity. In such cases, surgical intervention is typically reserved for patients who have experienced a second symptomatic hemorrhage[81]. Radiotherapy may serve as an alternative treatment for solitary CMs that have previously led to symptomatic hemorrhage, especially when the lesion is located in an eloquent brain region where the risks of surgical intervention are prohibitively high[81].

DVAS AND CEREBRAL CMS

Epidemiology

Over the past fifty years, the advancement of CT and the advent of MRI have enabled the work-up, diagnosis, and follow-up of patients with coexistent DVAs and CMs in the central nervous system (CNS). A series of studies have estimated the prevalence of DVA-CM coexistence to be up to 33% (Tables 2 and 3)[93-102]. Despite high-field MRI imaging, many authors, however, suggest that the prevalence of DVA-CM concurrence might have been underestimated because many small DVAs are missed on imaging and observed merely during operation time[101,103]. The location of the coexistent lesions, age of patients, and clinical presentation have been correlated with the concurrence of DVA-CM mixed malformations.

Table 2 Association of developmental venous anomalies and cavernous malformations in some of the largest case series in literature.

Ref.	Publication year	Number of developmental venous anomalies	Coexistence with cavernous malformations
Ostertun et al[93], 1993	1993	21	33%
Wilms et al[94], 1994	1994	65	23%
Huber et al[50], 1996	1996	43	30%
Töpper et al[95], 1999	1999	67	18%
Meng et al[96], 2014	2014	1839	11%
Brinjikji et al[97], 2017	2017	1689	7%

Table 3 Association of cavernous malformations and developmental venous anomalies in some of the largest case series in literature.

Ref.	Publication year	Number of cavernous malformations	Coexistence with developmental venous anomalies
Porter et al[98], 1997	1997	173	13%
Abe et al[99], 1998	1998	102	23%
Abdulrauf et al[100], 1999	1999	55	24%
Wurm et al[101], 2055	2005	58	26%
Zhang et al[102], 2013	2013	41	27%

Abe et al[99] found that 83% of the DVA-CM lesions were infratentorial (25 out of 30 cases). In their retrospective analysis of 55 patients harboring CMs, Abdulrauf et al[100] observed a statistically significant difference between DVA-CM and CM-only cases; 67% of the mixed malformation cases were infratentorial, compared with 27% of the CM-only group. Interestingly, in a series of 86 patients who underwent resection of hemorrhagic brainstem CMs, all of them were associated with a DVA[104]. In a large study of 1689 individuals with DVAs, coexistence with CMs was observed in 6,9% (116/1689); DVA-CM lesions were more likely to be noticed in the brainstem than DVA alone (15.5% vs 3.2%) and basal ganglia or thalamus (6% vs 3.9%)[97]. The infratentorial location of a DVA increased the chance of coexistence with a CM, according to a large observational study by Meng et al[96], too. In 41 cases of cerebellar CMs, Zhang et al[102] observed 26.8% DVA-CM coexistence (11/41); no statistically significant difference was estimated regarding age, sex, size, and clinical presentation between patients with cerebellar CM alone and patients with DVA-CM coexistence in the cerebellum.

Brinjikji et al[97] observed a statistically significant correlation between age and DVA-CM coexistence. While the simultaneous presence of DVAs and CMs was noticed in < 1% of individuals under 10 years old and < 5% under 20 years old, concurrence was 12% in people older than 70 years old[97]. The authors concluded that DVA-CM lesions have a de novo formation basis rather than congenital, and age-related changes in the venous network could trigger the development of a CM in the vicinity of a DVA[97]. In a large MRI observational study of 1839 Chinese patients with DVAs, there was an association with CM in 205 of them (11.1%)[96]. However, there was no statistically significant difference in the DVA-CM concurrence in the four age groups studied (< 20, 20-39, 40-59, > 60 years old)[96].

In a retrospective review of 112 patients with CM, 81 were familial, and 18 were sporadic[105]. A DVA was associated with 44% of the sporadic CM cases (8/18) but only in 1/81 patients with familial-based CM[105]. Petersen et al[105] justified the lower percentage of DVA-CM prevalence in other studies by the mixed population, i.e., sporadic and familial. Moreover, the authors suggested a different pathogenetic mechanism for the sporadic CMs, linking their formation with a pre-existing DVA[105]. Zhang et al[102] observed multiple CMs in 16.7% of cases of cerebellar CMs without DVA present (5/30 cases) and in no case of cerebellar DVA-CM concurrence (0/11 cases), further supporting a different development mechanism for familial and sporadic cases of CMs, since multiplicity of CMs is related to a familial origin of them.

A symptomatic hemorrhagic event is the most common severe clinical manifestation of a DVA-CM concurrence in the CNS. Although not statistically significant due to the study’s small series, Abdulrauf et al[100] observed symptomatic hemorrhage in 62% of DVA-CM cases and 38% of cases with CM alone. Repeat symptomatic hemorrhage was also higher in the DVA-CM group than in the CM-only group: 23% vs 9.5%; 15% of DVA-CM cases were associated with focal neurological deficits, and another 15% were found during headache work-up or incidentally[100]. A higher annual hemorrhagic rate for DVA-CM cases was also observed by Porter et al[98] when compared with CM-only patients: 3.4% vs 0.4%. In their research with both supra- and infratentorial DVAs published the same year, Töpper et al[95] noticed that in all cases with symptomatic hemorrhage, a CM was in the vicinity of the DVA (5/5 cases).

Pathophysiology

Over the last decades, the clinical impact of a hemorrhagic event by DVA-CM malformations has urged many researchers to explore their pathophysiological basis. Despite the classical McCormick classification of VMs into four main types (DVA, CM, arteriovenous malformation, and capillary telangiectasias), histopathology analyses reveal a continuum among the different subtypes with mixed malformations in all possible combinations[103,106,107]. CM-DVA coexistence is considered the most common mixed VM[95].

Since the 1990s, the mechanism of “hemorrhagic angiogenic proliferation” has been suggested[108]. Venous and capillary malformations cause microhemorrhage due to hemodynamic disturbance or vessel fragility, which in turn leads to reactive angiogenesis and the formation of CMs[108]. The expression of angiogenic factors has been verified in all types of CNS VMs[109]. In 1998, two studies on radiation therapy’s effect on CM development were published. Larson et al[110] noticed the formation of a CM in the irradiated brain field of 6 children with a CNS neoplasm. Histopathology of the resected lesions showed features of both CMs and CTs[110]. The radiation effect on the cerebral microcirculation could induce the formation of CTs and their evolution into CMs[110]. In the case of a 3-year-old boy with medulloblastoma, Maeder et al[111] observed a small hemorrhagic lesion close to a known DVA 3 years after irradiation of the neoplasm. Two years later, the lesion acquired imaging and histopathologic features of a CM[111]. Stenosis of the deep collector vein of the DVA, shown on T1w MR images, and the probable high venous pressure were held responsible for the CM formation[111]. Dillon et al[112] had already reported intraoperative high venous pressure within a DVA associated with recurrent cerebral hemorrhages. Obstruction of venous outflow was shown on CT angiography[112]. Radiation-induced CM formation supports the concept of “hemorrhagic angiogenic proliferation” proposed by Awad et al[108].

The concept of a DVA contributing to hemorrhage risk in an associated CM was further supported by Porter et al[104]. As previously noted, in their retrospective review of hemorrhagic brainstem CM resection in 86 patients, a DVA was found intraoperatively in all 86 of them[104]. Moreover, in 2 patients with relapse, post-operative MR imaging was clear for CM lesions, but DVAs were present[104]. Two case reports, in 2003 and 2005, described the de novo development of a CM adjacent to a pre-existing DVA without any apparent triggering factor, such as radiation. Cakirer reported the evolvement of a well-circumscribed cerebral hematoma close to a pre-existing DVA into a DVA-CM mixed malformation within 1 year[113]. The CM was verified histologically, but no DVA stenosis was revealed on angiography[113]. In the case report by Campeau et al[114], a cerebellar CM was formed in the vicinity of a pre-existing DVA in a patient with relapsing headache and dizziness.

Therefore, an important question arises: Could the abnormal draining vein of the DVA, which may be shared by or lie adjacent to the CM, be the main factor in inducing CM development? The frequency of DVA-CM malformations may have been underestimated in the literature. Many small venous anomalies become visible only intraoperatively and after the CM is resected[101]. Sasaki et al[115] noticed intraoperatively that small veins from a CM drained to the central, dilated medullary vein of the DVA. Wilson et al[116] proposed three possible causes of raised venous pressure in a venous malformation or its distal radicles: (1) Thrombosis; (2) Acute increase in intracranial pressure; and (3) Outflow restriction of the vein while entering the central vein or venous sinus.

Interestingly, in 3 DVA-CM relapses where the CM had been resected, and the large draining vein of the DVA had been left untouched, histology of the relapses revealed arteriovenous angiomas or CTs, not CMs[101]. These results supported the CNS VM continuum theory, suggesting that a DVA is the primary lesion and the other three VMs arise because of its dysfunctional venous drainage[101,103]. The next step was to try to comprehend the hemodynamic behavior and angioarchitectural structure of the DVA-CM lesions by zooming in. In 22/23 DVA-CM lesions, Hong et al[117] measured acute angulation (< 90°) of the distal medullary or draining vein. The CM was also found adjacent to the angulated draining vein in 22/23 of the mixed lesions[117]. The authors suggested that the turbulent flow caused mainly by these two angioarchitectural factors may lead to vessel injury, microhemorrhage, and neoangiogenesis[117].

In a multivariate analysis of 76 DVA-CM cases vs 503 DVA-only cases, six angioarchitectural factors were associated with DVA-CM concurrence: (1) Infratentorial location of the DVA; (2) Deep direction of the terminal or draining vein to which the caput medusae joins; (3) Torsion of the draining vein; (4) Presence of ≥ 5 medullary veins; (5) ≥ 54.68% stenosis rate of the medullary veins; and (6) Angle ≤ 10650 of the draining vein[118].

According to Shama et al[60], a local increase in venous pressure due to restricted venous drainage could also contribute to microhemorrhage and CM formation. In their prospective perfusion MRI study, 10 DVA-CM cases showed a greater prolongation of MTT and a lesser increase or even decrease of rCBF compared to 15 DVA-only cases[60]. Integration of clinical parameters into the DVA-CM formation theories was missing. Kumar et al[119] compared the clinical data of 145 sporadic DVA-CM cases with 300 age-matched DVA-only cases. A history of severe infection, such as Lyme disease, or chronic inflammatory disease, such as inflammatory bowel disease, increased the risk of developing a CM fivefold and twofold, respectively[119].

The authors provided three probable mechanisms: (1) DVA thrombosis due to inflammation; (2) Angiogenesis induced by inflammation-enhanced vascular permeability; and (3) Implication of the Toll-like receptor 4 (TLR4)[119]. Gram-negative bacteria activate TLR4 and increase CM formation in mice[120]. Additionally, polymorphisms that increase TLR4 expression levels have been linked with a higher number of CMs in humans[120]. Overall, the main pathophysiological mechanisms and theories of CM formation in the vicinity of a known DVA are depicted in Figure 7.

Figure 7 De novo formation of a developmental venous anomaly-associated sporadic cavernous malformation. A: Events of temporary or permanent high venous pressure in a developmental venous anomaly (DVA) may induce hemodynamic instability and vessel fragility, leading to microhemorrhage in the adjacent parenchyma. The attraction of proliferative angiogenic factors can eventually lead to the formation of a sporadic cavernous malformation sharing the DVA’s venous drainage system; B: Thrombosis and stenosis or/and the presence of favoring angioarchitectural factors raise venous pressure within the DVA (blue arrows). The resulting hemodynamic disturbance and turbulent flow may affect the rigidity of venous walls, leading to microhemorrhage. Neoangiogenesis in the vicinity of the DVA produces new vascular malformations of a broad histologic spectrum, mainly cavernous malformations (red arrows). Environmental factors such as radiation therapy, inflammatory and infectious agents, and aging may also be associated with DVA thrombosis and increased vessel fragility (brown arrows).

Imaging findings

Imaging findings in coexistent DVAs and CMs will be the combination of those described for each malformation separately. Awareness of the possibility of their coexistence may prove invaluable in clinical practice. This is because recognition of a DVA close to a brain hemorrhage may raise suspicion for an underlying CM as the most likely etiology. Additional examinations and follow-up imaging following hematoma resolution will further enhance diagnostic confidence. Figures 8, 9, 10 and 11 illustrate clinical examples in which identifying the synchronous presence of a DVA adjacent to a brain hemorrhage (or suspected brain hemorrhage) enabled the diagnosis of a probable underlying CM as the most likely etiology, thereby avoiding unnecessary examinations or interventions.

Figure 8 Cerebral cavernous malformation and developmental venous anomaly. A: Non-contrast computed tomography scan of a 40-year-old man presenting to the emergency department with an acute onset headache demonstrated a hyperdense-hemorrhagic lesion (dashed circle) in the right middle cerebellar peduncle, abutting the 4th ventricle; no contrast enhancement was shown. There was no relevant medical history or additional lesions; B and C: Magnetic resonance imaging examination performed a few days later demonstrates a central hyperintense signal intensity in T1-weighted images (B) and hyperintense signal centrally with a low signal hemosiderin rim on T2-weighted images (C), mild surrounding vasogenic edema was also noted on T2-weighted images. The findings were mostly suspicious of a Zabramski type I cerebral cavernous malformation with recent bleeding (T1 hyperintensity and mild surrounding vasogenic edema); D: On susceptibility-weighted images, intense lesion blooming due to hemosiderin was also noted. However, in addition, a concomitant developmental venous anomaly (evident by the caput medusae morphology) was detected in close vicinity to the lesion (dashed rectangle), thus adding further diagnostic confidence to a cerebral cavernous malformation as the underlying diagnosis for the hemorrhagic lesion.

Figure 9 Cerebral cavernous malformation and developmental venous anomaly. A: Non-contrast-enhanced computed tomography in a 60-year-old man presenting to the emergency department due to dizziness and a medical history of arterial hypertension. A mildly hyperdense lesion was depicted in the right thalamus (dashed circle). Considering the location of the lesion and the medical history of arterial hypertension, it was suspicious of a hypertensive bleed. The lesion remained unchanged in follow-up computed tomography examinations, and therefore, further magnetic resonance imaging was decided; B: On T2-weighted images, the lesion displayed heterogeneous signal intensity centrally with a “popcorn-like” appearance and a peripheral low-signal hemosiderin rim; C: On T1-weighted images, heterogeneous, mostly hypointense, signal was noted centrally with mildly increased T1 signal in the periphery; D: On susceptibility-weighted imaging, the lesion displayed intense blooming. The findings were suggestive of a Zabramski type II cerebral cavernous malformation rather than a hypertensive bleed; E-G: This scenario was further validated by the depiction of an adjacent ectopic draining vein (arrows) on T1-weighted sagittal (E), coronal (F) and axial (G) maximum intensity projection reconstructions following intravenous gadolinium administration, thus indicating the concomitant presence of a developmental venous anomaly.

Figure 10  Cerebral cavernous malformation and developmental venous anomaly. A: Non-contrast-enhanced computed tomography obtained in a 67-year-old man presenting to the emergency department due to impaired balance shows a hyperdense lesion in the pons (dashed circle); B and C: On T1-weighted images the lesion displays a heterogeneous but mostly hyperintense signal, suggesting recent hemorrhage (B), while on T2-weighted images (C), a peripheral low-signal hemosiderin rim is noted, and a “popcorn-like” or “mulberry-like” appearance is depicted centrally; D-F: On susceptibility-weighted imaging, the lesion displayed intense blooming. The findings were mostly suggestive of a cerebral cavernous malformation (Zabramski type I or II, due to mixed findings of a “popcorn” appearance and recent bleed) as the underlying diagnosis, which was further verified by the depiction of a large adjacent developmental venous anomaly (arrows).

Figure 11  Cerebral cavernous malformation and developmental venous anomaly. A: Non-contrast-enhanced computed tomography obtained in a 56-year-old man presenting to the emergency department with left-hand weakness shows an intracerebral hematoma in the right frontal lobe (thick arrow). The etiology remained unclear, as no helpful findings were depicted in the computed tomography angiography and venography study performed (not shown); B: On T1-weighted images, the hematoma displays a heterogeneous hyperintense signal due to the recent hemorrhage; C: On fluid-attenuated inversion recovery images, mild surrounding vasogenic edema is also noted (curved arrow); D: Hemosiderin presence is also verified on susceptibility-weighted imaging; E-G: On T1-weighted images sagittal (E), coronal (F) and axial (G) reconstructions following intravenous gadolinium administration, an ectopic vein with “caput medusae” morphology is noted immediately adjacent to the internal border of the hematoma (thin arrows), indicating the presence of a developmental venous anomaly and thus pointing towards an underlying recently bled cerebral cavernous malformation as the underlying etiology for the hematoma; H: An additional developmental venous anomaly was also noted on the same side in a more posterior location (dashed circle).

Treatment

Most of the debate on the management and treatment of mixed DVA-CM malformations has focused on two issues: Leave the DVA untouched while resecting a CM, or not? Does the presence of a DVA next to a CM increase its hemorrhagic risk or not? CM excision while preserving the associated DVA, which drains normal parenchyma, remains the most common surgical strategy[104,121]. Brain swelling, infarction, and death have been reported in cases of resecting DVA’s prominent draining veins[101,104]. Wurm et al[101] challenged this strategy, believing that the large vein of the DVA at the length of the CM was the primary abnormality. The large vein of the DVA was coagulated and dissected. Six cases were operated on for the first time, and another three were recurrences[101]. Intraoperative swelling was not observed, and all patients’ post-operative follow-up was uneventful[101]. Zhang et al[102] also reported good long-term prognosis after resecting 36 cerebellar CMs while coagulating and dissecting the distal radicles of the DVA. In only 2/36 patients was there severe cerebellar edema, and one of them needed a decompressing craniotomy[102]. The mean follow-up period was 22.3 months[102].

Association of a DVA with a CM has long been considered a risk factor for a more aggressive clinical course than a CM-only lesion[100,101,108]. In their retrospective multivariable analysis of 154 CM cases, Kashefiolasl et al[122] associated the age > 45 years old, the infratentorial location of the CM, and the presence of a DVA with a higher hemorrhagic risk. However, in a large single-center observational study with 731 non-familial, non-spinal DVA-CM cases, the presence of a DVA close to a CM was associated with a lower risk of presenting with intracerebral hemorrhage and equal risk of (re)-hemorrhage with CM-only cases[123]. The size of the hemorrhage did not relate to the presence of a DVA either[123]. Initial presentation with intracerebral hemorrhage and brainstem location of the mixed malformation were the only independent risk factors for (re)-hemorrhage[123]. These results seriously impact clinical counseling by questioning the argument that the presence of a DVA close to a CM constitutes an indication to resect the CM[123,124].

CONCLUSION

Cerebral VMs represent a heterogeneous group of entities with distinct pathophysiological mechanisms, imaging characteristics, and clinical implications. DVAs and CMs are the most common venous malformations, and when they coexist, the DVA’s altered venous pressure and flow can promote CM formation or rupture, causing a symptomatic hemorrhagic event, which is the most common severe clinical manifestation of a DVA-CM concurrence. Thus, detecting a DVA adjacent to an otherwise unexplained intracerebral hemorrhage may be the clue to an important diagnosis, as it can raise suspicion of an occult CM as the underlying etiology of the bleed. Recognizing the hallmark imaging appearances of DVAs and CMs, along with the sometimes subtle signs of their coexistence, can be invaluable for achieving an accurate and timely diagnosis, ensuring appropriately targeted management, and avoiding unnecessary interventions.