Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL) as a model of small vessel disease: update on clinical, diagnostic, and management aspects

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is the most common and best known monogenic small vessel disease. Here, we review the clinical, neuroimaging, neuropathological, genetic, and therapeutic aspects based on the most relevant articles published between 1994 and 2016 and on the personal experience of the authors, all directly involved in CADASIL research and care. We conclude with some suggestions that may help in the clinical practice and management of these patients.


Background
Cerebral small vessel disease (SVD) is an important cause of stroke, cognitive impairment, and mood disorders in the elderly. Besides the common sporadic forms, mostly related to age and hypertension, a minority of SVD has a monogenic cause, among which the most common and best known is cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). CADASIL provides a unique model for the study of the most prevalent forms of sporadic SVD. CADASIL is caused by mutations in the NOTCH3 gene, which maps to the short arm of chromosome 19 and encodes the NOTCH3 receptor protein, predominantly expressed in adults by vascular smooth muscle cells and pericytes [1]. Thousands of families with CADASIL have now been diagnosed worldwide in many different ethnic groups. The disorder is often overlooked and misdiagnosed. Its minimum prevalence has been estimated between 2 and 5 in 100,000 but may vary between populations [2][3][4][5].
In this article, we discuss the main clinical, genetic, neuroimaging, neuropathological, and management aspects of the disease.

Methods
The project for this paper was developed following the 9th International Congress on Vascular Dementia, 2015, held in Ljubljana, Slovenia. Articles published between 1994 to 2016 with the keyword 'CADASIL' were retrieved from the PubMed database. Figure 1 reports the number of PubMed articles on CADASIL during the past years. Only English language papers with high scientific relevance were reviewed. In total, 163 articles related to CADASIL were reviewed and included in the paper.

Clinical aspects
The main clinical features include migraine usually with aura presenting in early adulthood [6], recurrent subcortical ischemic events, mood disturbances [7], progressive cognitive impairment mostly affecting executive function, and acute encephalopathy [8,9]. Pescini et al. [10] developed a score which may be helpful in confirming the clinical suspicion of CADASIL and identifying patients in whom mutation screening has a reasonable yield.

Migraine
Migraine is often the earliest feature of the disease, being reported in about 55-75% of Caucasian cases, although it is less frequent in Asian populations [11]. The age at onset is highly variable, generally around 30 years [12]. In a recent article on the prevalence and characteristics of migraine in 378 CADASIL patients [6], a total of 54.5% of individuals had a history of migraine and 84% of these had migraine with aura (MA). Women with MA accounted for 62.4% of the total, with an earlier MA onset compared to men. Atypical auras, such as prolonged visual auras, gastrointestinal manifestations, dysarthria, confusion, focal neurologic deficits, and other uncommon manifestations, were experienced by 59.3% of individuals with MA, and in 19.7% of patients the aura was sine migraine. MA was the first clinical manifestation in 41% of symptomatic patients and an isolated symptom in 12.1%. The pathophysiological reasons leading to increased aura prevalence in CADASIL are unknown; a possible mechanism is an increased susceptibility to cortical spreading depression [13]. It is also possible that migraine in CADASIL involves the brainstem region. Alternatively, the NOTCH3 mutation itself may act as a MA susceptibility gene [14].

Subcortical ischemic events
Transient ischemic attacks and stroke are reported in approximately 85% of symptomatic individuals [15] and are related to cerebral small vessel pathology. Mean age at onset of ischemic episodes is approximately 45-50 years [16], but the range at onset is broad (3rd to 8th decade).
Ischemic episodes typically present as a classical lacunar syndrome (pure motor stroke, ataxic hemiparesis/ dysarthria-clumsy hand syndrome, pure sensory stroke, sensorimotor stroke), but other lacunar syndromes (brain stem or hemispheric) are also observed. The total lacunar lesion load, symptomatic and asymptomatic, is strongly associated with the development of severe disability with gait disturbance, urinary incontinence, pseudobulbar palsy and cognitive impairment. Strokes involving the territory of a large artery have occasionally been reported but whether these are co-incidental or related to the CADASIL pathology itself is uncertain [17][18][19]. There have been a few cases of subcortical hemorrhage, mostly in patients taking anticoagulants.

Encephalopathy
An acute encephalopathy [20] has been described in 10% of CADASIL patients, and in the majority of these it was the first major symptom, with a mean age of onset of 42 years [9]. It is frequently misdiagnosed as encephalitis, particularly if it is the initial presentation in a patient without known CADASIL. It usually evolves from a migraine attack, including confusion, reduced consciousness, seizures, and cortical signs, with spontaneous resolution.
Cognitive impairment in CADASIL first involves information processing speed and executive functions, with relative preservation of episodic memory, and is associated with apathy and depression [21][22][23][24][25] [27]. The cognitive deficits were initially attributed to subcortical origin [28]. More recently, the cerebral cortex was shown also to be affected in CADASIL [29] by direct mechanisms (cortical microinfarcts [30]), or through secondary degeneration [31]. Memory impairment in CADASIL is probably related to different causes, due to both white matter infarcts with disruption of either cortico-cortical or subcorticalcortical networks mainly in the frontal lobe and primary damage of the cortex [32,33]. In a recent longitudinal study, processing speed slowing was related to a decrease of sulcal depth, but not to global brain atrophy or cortical thinning, suggesting that early cognitive changes may be more specifically related to sulcal morphology than to other anatomical changes [34]. Moreover, studies in mouse models of CADASIL have detected dysregulation of hippocampal neurogenesis, a process essential for the integration of new spatial memory records [35].

Psychiatric disturbances
Although the clinical expression of the disease is mainly neurological, CADASIL is also characterized by psychiatric disturbances (20-41%) [36,37]. Apathy and major depression are commonly observed in CADASIL. Also bipolar disorder and emotional incontinence are present in a consistent percentage of patients [38]. The direct consequence is a negative effect on patient's quality of life and caregiver burden, with different degrees. Other psychiatric manifestations, such as psychotic disorders, adjustment disorders, personality disorders, drug addiction, and abuse of substances, are less frequent. The pathogenesis of psychiatric disturbances in CADASIL is incompletely understood, but, similarly to other cerebrovascular diseases, may depend on the damage of the cortical-subcortical circuits, leading to the consideration of mood disorders in CADASIL under the concept of 'vascular depression'.
In conclusion, CADASIL symptoms may be highly variable, usually starting in adulthood, but also reported in older age [39]. The wide clinical spectrum includes complex migraine attacks with prominent aura, acute confusional states or coma, lacunar strokes, and pure psychiatric or cognitive presentations. Ischemic events, in different brain areas, may involve also the optic nerve [40]; retinal involvement is also described [41,42]. Few reports also suggest the presence of myopathy or peripheral neuropathy, the first possibly related to involvement of mitochondrial dysfunctions, but these findings may be coincidental. Typical and atypical phenotypes are reported in Table 1.

Modulating factors
CADASIL is characterized by a large phenotypic variability both across and within families. No clear genotype-phenotype correlation has been found to date [43] and both modulating genetic and risk/environmental factors probably play a role.
Among the conventional risk factors, hypertension and smoking are associated with early age of onset of stroke [9,43,44] and with clinical worsening towards dementia [45,46]. In a recent longitudinal study, smoking was associated with a three-fold increase in stroke risk [46]. Homocysteine levels have been found to be higher in patients with CADASIL who have migraine, but this finding needs replicating [43].
Opherk et al. [47,48] found that white matter lesion volumes are influenced by genetic factors distinct from the causative NOTCH3 mutation, but the nature of the responsible modulating genetic factors is currently unknown. The role of Apolipoprotein E (APOE) is debated -Gesierich et al. [49] suggested a modifying influence of APOE ɛ2 on the volume of white matter hyperintensities, while Singhal et al. [43] detected no relationship between phenotype and APOE ɛ4 genotype.
Gunda et al. [50] hypothesized on a possible role of ovarian hormones in sex-related differences in CADA-SIL, data that need further confirmation.
Another point of interest is the finding of a nondipper pattern of nocturnal blood pressure [56]. A lower nocturnal blood pressure fall may be associated with incidence and/or worsening of deep white matter lesions in CADASIL. The pathogenesis of abnormal blood pressure profile in CADASIL remains to be clarified. Central and peripheral mechanisms controlling blood pressure variations may be involved.
The association between CADASIL and right-to-left cardiac shunt has been reported but remains under debate [57,58].

Neuroimaging
Cerebral Magnetic Resonance Imaging (MRI) in CADA-SIL shows age-and disease stage-dependent extensive white matter abnormalities, seen as large symmetrical hyperintense signals on T2-weighted and FLAIR images [59], and lacunes, i.e., fluid-filled cavities with cerebrospinal fluid-isointense signal. Diffusion-weighted imaging may show acute and subacute infarctions, and can also be used to quantify chronic white matter changes through diffusion tensor metrics. White matter changes often involve the anterior temporal lobe, the external capsule and the superior frontal gyrus [60]. Anterior temporal pole changes have been shown to have a high sensitivity and specificity for the disease (approximatively 90% for each) and are useful in diagnosis [61]. In Asian populations, anterior temporal lobe involvement is less common [62]. External capsule changes also have a high sensitivity (approximatively 90%) but a lower specificity (approximately 50%). In fact, a recent systematic analysis showed a similar involvement of the external capsule in CADASIL and sporadic SVD [63]. Corpus callosal signal abnormalities, rarely occurring in sporadic SVD, are described in CADASIL [59] (Fig. 2); such abnormalities are also a feature of multiple sclerosis, which is one reason for the misdiagnosis of CADASIL as multiple sclerosis. Cerebral microbleeds, which are shown by Gradient-echo images with dot-like hypointense lesions, occur in a variable proportion of cases (30-70%) and usually increase with age and risk factors such as high blood pressure, and intracerebral hemorrhage have also been described in a number of patients [64][65][66], especially those of Asian origin [67]. Nevertheless, the clinical significance of the cerebral microbleeds observed in CADASIL has not been clearly elucidated.
The MRI findings that have been most strongly related to clinical deficit are number and volume of lacunes, brain atrophy, and white matter damage as assessed by diffusion tensor imaging [69,70]. CADASIL has primarily been defined as a subcortical disease, but recent studies with high-field 7-Tesla MRI, which allows higher resolution imaging, have detected primary involvement of the cortex, including the demonstration of cortical microinfarcts and early diffuse cortical alterations [23] in frontal and parietal regions most often following a symmetrical pattern in both hemispheres. These changes were not related to cortical thinning or to subcortical lesions and have been hypothesized to be secondary to venous vascular density or intramyelin edema [71]. Similar data have also been reported in experimental models [72]. Moreover, subcortical changes might induce secondary cortical changes [69,[72][73][74]. In a longitudinal study, incident lacunes were followed by cortical thinning specifically in connected brain regions [31]. Another recent development is the use of diffusion tensor imaging to characterize tissue damage [75,76]. The typical change in diffusion metrics observed in CADASIL patients is a reduction in fractional anisotropy (a measure for the directionality of diffusion) and an increase in the apparent diffusion coefficient or mean diffusivity (a measure for the extent of diffusion) [69,77]. Histogram analysis has been proposed as a sensitive tool to measure CADASIL-related changes cross-sectionally as well as longitudinally [78][79][80].

Neuropathology
The pathological hallmark of CADASIL is deposition of granular osmiophilic material (GOM) in close relation to vascular smooth muscle cells, seen on electron microscopy. GOM can also be detected in vessels of extracerebral tissues, including skin and muscle. Consequently, skin biopsy has been considered an important diagnostic tool [81], with a specificity of 100%, and variable sensitivity from 45% to 100% according to the different reports [82][83][84]. The reasons for this discrepancy may be related to different factors, including different stages of the disorder, the size of the biopsy and the number of arteriolar and precapillary vessels in the specimen, or the focality of lesions. It is also important to differentiate GOM from dense bodies and non-specific granular debris present in degenerated smooth muscle cells. The reason why the small-to-medium sized cerebral arteries are most severely affected and why there are no extracerebral clinical disease manifestations is still not fully understood. Further, the composition of GOM remains partly unclear. Previous immunoelectron microscopy studies reported the accumulation of the NOTCH3 ectodomain in proximity of GOM formation [85]. More recent investigation by post-embedding immunoelectron microscopy analysis using antibodies against the extraand intracellular portions of NOTCH3, showed the NOTCH3 ectodomain to be a major component of GOM [86][87][88]. Immunohistochemistry analysis using a NOTCH3 monoclonal antibody was proposed as a further reliable tool for CADASIL diagnosis [89]. Although high specificity and sensitivity, ranging from 90% to 100%, were reported, this high level of accuracy has not been reported by any other group and the technique is not currently widely used in diagnosis [90].
In conclusion, preferably, skin biopsy analysis should include both NOTCH3 immunohistochemistry and electron microscopy. For electron microscopy analysis at least 4-6 arteriolar vessels for each patient should be examined. In most CADASIL centers, skin biopsy is only reserved in those cases with suspected CADASIL, but who did not have a NOTCH3 mutation or who had an unclassified variant in NOTCH3. Genetic testing is becoming an increasingly accessible diagnostic tool and, in patients with a clinical suspicion of CADASIL, NOTCH3 Sanger sequencing of EGFr encoding exons, if available, is the first choice for confirming the clinical diagnosis. This has the added advantage that testing of family members (both symptomatic and asymptomatic) is feasible once the mutation is known in the index patient.
There are few brain neuropathological analyses published in CADASIL, and these have mainly focused on cerebral vessel changes that are reported to be consistently narrowed with intimal thickening and degeneration of vascular smooth muscle cells and GOM deposition [91]. In 11 patients, Craggs et al. [92] reported disruption of either cortico-cortical or subcortico-cortical networks in the white matter of the frontal lobe, explaining motor deficits and executive dysfunction present in CADASIL.

Genetics
NOTCH3 is the causative gene for CADASIL, which maps to the short arm of chromosome 19 (19p13.2-p13.1) [1,93]. It is a large gene containing 33 exons [94]. The NOTCH3 gene codes for a single-pass transmembrane receptor protein, with a large extracellular domain, with 34 EGFRs (tandem epidermal growth factor-like repeat domains) and 3 LNR repeats, a transmembrane portion and an intracellular domain containing ankyrin-repeat motifs and a PEST domain [95]. NOTCH3 protein is expressed in vascular smooth muscle cells, whose degeneration results in progressively impaired cerebrovascular autoregulation, hypoperfusion, and ischemia [84]. More than 200 different NOTCH3 gene CADASIL-associated mutations have been reported, the majority of them consisting of stereotyped missense substitutions of a single base leading to loss or gain of a cysteine residue within one of the EGFRs of NOTCH3 extracellular domain [96]. Initially, most mutations were found to cluster in exons 3 and 4, but there are now many reports of mutations in other exons encoding the extracellular portion of the protein [2,39,97,98]. Therefore, screening should include all EGFR encoding exons (exon 2-24). A few patients are reported with homozygous or compound heterozygous (Dotti unpublished data, [99]) NOTCH3 mutations; in these patients, the phenotype is almost always within the classical CADASIL spectrum [100][101][102][103][104][105].
Whether non-cysteine variants can cause CADASIL is debated. Small NOTCH3 deletions, duplications, splice site mutations, and deletion/insertions associated with CADASIL have been described. Most of these mutations lead to a numerical cysteine acid change [106]. In one study, a small intronic deletion leading to intron 3 retention was found to be associated with a GOM-positive CADASIL patient; the 25 additional amino acids in the EGFR likely disturb the disulphide bond formation [107]. Several studies have reported atypical, cysteinesparing, NOTCH3 mutations associated with a familial CADASIL phenotype. These variants may be interpreted as causative mutations only if the complete screening of the EGFR-encoding exons excludes other classic mutations and GOM are present in skin biopsy [106,108,109]. However, one study suggested that GOM analysis might be less sensitive in patients with non-cysteine involving mutations [110].
Mutations causing loss of function (frame-shift deletion or nonsense mutations) are rarely reported in CADASIL patients, and their role is still debated [106,111,112]; they should be carefully analyzed to determine their pathogenicity. NOTCH3 mutations located outside of the EGFR encoding exons, reported in few cases, are not associated with a CADASIL phenotype [113][114][115].
A simple flow chart for CADASIL diagnosis is shown in Fig. 3.

Genetic counseling
Genetic counseling should be offered to all patients diagnosed with CADASIL and should always precede predictive genetic testing of family members. For predictive genetic testing, the Huntington's disease guidelines can be used [116,117]. It is important to note that a MRI scan in an unaffected family member can have a similar impact as a genetic test if it shows the characteristic MRI hallmarks of the disease. Genetic counselling also plays an important role in family planning and questions about options for prenatal diagnosis or pre-implantation genetic diagnosis may be discussed in countries where these techniques are available [118,119].

Treatment
An effective treatment for CADASIL is not available. In the absence of curative approaches, treatments should be directed towards the search of possible diseasemodifying strategies to mitigate clinical manifestations [120]. The review of available data shows that only a few and very preliminary studies have been reported thus far. In Tables 2 and 3, we summarized studies focused on symptomatic outcomes and those focused on cerebral and peripheral blood flow changes with a possible effect on disease mechanisms.
In view of the evidence of a more severe disease course in individuals with vascular risk factors, particularly smoking and hypertension, control of vascular risk factors is an important part of CADASIL management. Concerning the use of antiplatelet drugs, such as aspirin or clopidogrel, most neurologists apply the guidelines used for sporadic stroke when treating CADASIL patients. However, the appropriateness of this approach is undetermined. In fact, the thrombotic genesis of ischemic events in this disease has not been proven so far. On the other hand, many reports have stressed the presence of microhemorrhages (microbleeds) in a considerable percentage of CADASIL patients [121]. For these reasons, the safety of antiplatelet drugs in this disease remain to be clarified [65]. Similarly, the benefit of thrombolysis is uncertain, although it has been suggested that it is of benefit in sporadic lacunar stroke [122].
Following the data reported by Keverne et al. [121] showing a cholinergic neuronal impairment in CADASIL, a multicenter trial in 168 patients with donepezil was performed; this showed no improvement in the primary outcome of the Vascular Dementia Assessment scalecognitive subscale. Improvements were noted on several measures of executive function, but the clinical relevance of these findings is not clear [123]. Complications of CADASIL, such as depression and migraine, appear to respond to similar treatments to those used in sporadic disease, although further studies are required to determine the efficacy of these approaches in CADASIL patients.

Differential diagnosis
A differential diagnosis in some cases may be hard. However, in CADASIL, in comparison to multiple sclerosis, temporo-polar MRI lesions are present in a high percentage, the optic nerve and spinal cord are generally not involved, and oligoclonal bands in the cerebrospinal fluid are absent. Corpus callosum hyperintensities are described both in CADASIL and in multiple sclerosis [124,125]. In relation to cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CARASIL), in CADASIL, extraneurological findings (early-onset spondylosis and early-onset alopecia) are absent, clinical onset is generally later, MA is a typical sign, and inheritance is autosomal dominant [126]. Other hereditary leukoencephalopathies to take into account for a correct differential diagnosis are MELAS, Krabbe disease, Fabry disease, CARASAL, hereditary diffuse leukoencephalopathy with axonal spheroids, and other more rare monogenic forms of SVD [127].

Highlights and key questions for management of CADASIL patients
-Clinical presentation: The clinical presentation is usually stereotyped with MA, small subcortical strokes, or cognitive impairment in middle-aged patients with a family history of stroke or dementia, but the spectrum of the disease is wide and atypical phenotypes have been reported. There is wide variation in phenotype both within and between families.
-Cognitive impairment: Cognitive impairment is a prominent feature in CADASIL. While its clinical pattern is typically subcortical, several recent data suggest that the cerebral cortex is also involved and contributes to the cognitive profile abnormalities. -Brain MRI: Cerebral MRI in CADASIL shows extensive white matter changes, with frequent involvement of temporal pole and external capsule. No abnormality is pathognomonic, but confluent bilateral anterior temporal pole T2hyperintensities are highly suggestive. Recent advances in neuroimaging have also revealed involvement of the cortex. -Vascular risk factors: Smoking seems to be clearly associated with an increase in the stroke risk both in cross-sectional and longitudinal studies. Moreover, smoking and hypertension are associated with an earlier onset of stroke and more rapid progression in CADASIL. Evidence is less robust for other risk factors and more study is needed. However, in the absence of effective therapeutic approaches, control of vascular risk factors is recommended. Also, genetic modulating factors other than the NOTCH3 mutation probably play a role, but their nature remains to be determined. -Genetic: More than 200 different NOTCH3 gene mutations have been reported in exons 2-24, involving the loss or gain of cysteine residues. The role of non-canonic NOTCH3 mutations (e.g., cysteine-sparing mutations, mutations outside EGFR domains, frameshift or nonsense mutations) is Migraine Acetazolamide [152][153][154] As prophylaxis, reduces the frequency of migraine attacks Sodium valproate [155] Anecdotal cases Cognitive decline Acetylcholinesterase inhibitor [121] Not efficacious on the primary end-point (Vascular Dementia Assessment scale after 18 weeks), but some improvement in relation to frontal-subcortical dysfunction Primary and secondary stroke prevention Antiplatelet drugs [66,123] Unproven and debated benefits Table 3 Therapeutic studies on blood flow in CADASIL Drug Effect Acetazolamide Related to an increase of cerebral perfusion evaluated by perfusion MRI [156], transcranial Doppler sonography [157] and Tc-99 m extracellular domain brain perfusion SPECT [158,159] Atorvastatin No effects on cerebral blood flow [160], tested with transcranial Doppler L-arginine Induced-vasoreactivity [161], tested with transcranial Doppler Sapropterin (200-400 mg bid in 24 months) Final results for the primary end-point (mean difference in reactive hyperemia index) were not significant for any improvement of peripheral vasoreactivity [162] Novel molecular therapeutic target (as stem cell factor and granulocyte-colony stimulating factor) Ongoing [163] debated; in these cases, the search for GOMs in skin is mandatory, although in some cases it may be negative. When identifying a non-canonic mutation, sequencing of all EGFR-encoding exons of the NOTCH3 gene is mandatory as well as the cosegregation analysis of the non-canonic mutation with the CADASIL MRI phenotype within the family. -Neuropathology: GOM is the neuropathological hallmark of CADASIL. Specificity is high, but sensitivity is variable. Mutation testing is the gold standard, and the use of skin biopsy should be limited to the rare situations when the diagnosis is strongly suspected despite negative genetic testing. -Genetic counseling: Genetic counseling should always be offered to patients diagnosed with CADASIL and should precede all predictive genetic testing. Predictive genetic testing should only be offered to adults (>18 years) and should be performed by trained staff with genetic counselling experience, using a standard protocol such as that used in Huntington's disease. -Treatment: In the absence of specific data for CADASIL, most neurologists use aspirin in secondary prevention after ischemic strokes in older patients, for example, those aged over 40, but there is no evidence for or against its use. Whether this strategy is appropriate in CADASIL is undetermined and will require further investigation, given the possible increased hemorrhagic risk. Patients that need to undergo anticoagulation for a clear indication such as high risk atrial fibrillation should be carefully followed given the reported risk of intracerebral hemorrhage. A multicenter randomised controlled trial using donepezil to improve cognitive dysfunction showed no benefit on the primary endpoint and only improvements of uncertain clinical significance on several measures of executive function. New, rational therapeutic interventions for CADASIL are in early phases of pre-clinical development.

Conclusions
CADASIL is the most frequent monogenic cerebral SVD, and has received a lot of attention in recent years as a model for SVD, with an increased number of cases. However, more work remains to be done in order to better explore the underlying mechanism of its pathogenesis. The studies on animal models (see elsewhere in this issue) will improve the understanding of the pathogenic mechanisms and may open new therapeutic strategies.
Funding Hugh Markus is supported by an NIHR Senior Investigator award and his work is supported by the Cambridge University Hospital Comprehensive NIHR Biomedical Research Unit.
Authors' contributions IDD: Writing the manuscript. AM: Neuropathology section. LP: Therapy. SB, SP: Genetics. MTD: Clinical aspects. HM: Clinical and genetic aspects; revision of whole draft. AF: Organization, coordination and revision. AR: Cardiologic and ophthalmologic aspects. All other authors contributed to the revision and the organization of the manuscript. All authors read and approved the final manuscript.

Competing interest
The authors declare that they have no competing interests.