Acta Neurol Scand DOI: 10.1111/ane.12266
© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd ACTA NEUROLOGICA SCANDINAVICA
Changing clinical patterns and increasing prevalence in CADASIL Moreton FC, Razvi SSM, Davidson R, Muir KW. Changing clinical patterns and increasing prevalence in CADASIL. Acta Neurol Scand: DOI: 10.1111/ane.12266. © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd. Objectives – CADASIL is a monogenic small vessel vasculopathy causing recurrent stroke. Early descriptions suggested dementia and disability were common from the 5th decade but there is evidence of marked phenotypic variability. We investigated the prevalence and clinical features of CADASIL in the west of Scotland. Methods – We undertook a retrospective review of clinical records of patients with confirmed CADASIL identified through a specialist clinic. Patients were divided to examine the effect of date of diagnosis on clinical outcomes and the characteristics at different ages. The location of pedigree members was used to estimate prevalence. Results – Twentyone different CADASIL-causing NOTCH3 mutations were identified in 49 pedigrees (61% in exon 4). Disease prevalence in Glasgow was 4.6/100,000 adults. Mutation prevalence was estimated at 10.7/100,000 population. Median age at first stroke in women (57 years) was higher than previous estimates, and stroke age in men was higher in patients diagnosed more recently (pre 2006 46 years, post 2006 56 years, P = 0.034). In patients over 58 years of age, 13/34 (38%) were living independently and 17/28 (61%) were mobile without aids when last seen. Conclusions – CADASIL prevalence is at least 4.6 per 100,000 adults. Median age of first stroke may be older than previously thought. Clinicians should consider CADASIL in the differential diagnosis even in older patients with stroke.
Introduction
From the 1950s to 1980s, a number of families with an apparently hereditary vascular dementia accompanied by a Binswanger’s-like arteriopathy were described (1–3). In 1993, two families with stroke-like episodes, dementia and leukoencephalopathy were demonstrated to have an inherited mutation on chromosome 19, and the acronym CADASIL was introduced: cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (4). Subsequent genetic studies demonstrated the disorder was due to mutations in the NOTCH3 gene on chromosome 19p13 which encodes a receptor expressed by vascular smooth muscle cells in adults (5). CADASIL is characterized by loss of vascular smooth muscle cells and fibrosis of small penetrating vessels, resulting in chronic white matter ischaemia (6).
F. C. Moreton1, S. S. M. Razvi2, R. Davidson3, K. W. Muir1 1 Institute of Neuroscience and Psychology, University of Glasgow, Southern General Hospital, Glasgow, UK; 2 Department of Neurology, Institute of Neurological Sciences, Southern General Hospital, Glasgow, UK; 3 West of Scotland Genetic Services, Southern General Hospital, Glasgow, UK
Key words: Neuroepidemiology; Neurogenetics F. C. Moreton, Institute of Neuroscience and Psychology, University of Glasgow, Southern General Hospital, 1345 Govan Road, Glasgow, G51 3TF, Scotland Tel.: +44 141 201 2144 Fax: +44 141 201 2136 e-mail: [email protected] Accepted for publication April 25, 2014
The clinical features and natural history of CADASIL have been well documented (7, 8). Migraine is reported as a common presenting symptom occurring in young adulthood, but the most frequent manifestation is subcortical ischaemic events, occurring from the fourth to sixth decades, usually in the absence of, or disproportionately severe in relation to, conventional vascular risk factors (9). Recurrent ischaemic events may lead to vascular parkinsonism and pseudobulbar palsy. Concurrently cognitive impairment, initially characterized by executive dysfunction, but extending to a more global dementia, occurs (10). Severe disability with dementia is typically described by age 50–60 (7). However, cases have been described where patients are asymptomatic even in late adulthood (11). A specialist neurovascular clinic for CADASIL in Scotland was set up in 2002 in association with a genetics clinic. In 2005, we calculated the 1
Moreton et al. disease prevalence of CADASIL in the west of Scotland as 1.98 per 100,000 adults (12), considered likely to be an underestimate consequent to under-recognition. Recognition of CADASIL was likely biased towards larger pedigrees with typical features. The first two pedigrees diagnosed with hereditary vascular dementia (subsequently confirmed as CADASIL) in our unit in 1986 had over 100 members at the time of formal diagnosis (13). Availability of genetic screening of exons 2–23 of the NOTCH3 gene, the development of prediagnostic screening tools (14), improved awareness of the disease and its manifestations, wider use of brain MRI, and availability of a local specialist clinical service all suggested that increasing numbers of cases would be diagnosed with CADASIL over time. We reviewed the presentation of CADASIL to determine whether this had changed since its initial recognition. Our aims were (i) identify any changes in patient presentation since CADASIL was recognized; (ii) examine the characteristics of older CADASIL patients; and (iii) estimate mutation prevalence. Methods Identification of cases
A neurovascular genetics clinic was established in 2002 in association with local clinical genetics services, to facilitate the diagnosis and management of patients with CADASIL. Referrals are accepted from throughout Scotland. Individuals were defined for the purposes of this study as having confirmed CADASIL if they had a pathogenic mutation on analysis of exons 2–23 of the NOTCH3 gene, demonstrated granular osmiophilic material (GOM) on biopsy or post-mortem material, or had both a personal history consistent with CADASIL and were by analysis of the family tree, an obligate gene carrier. First, 2nd and 3rd degree relatives (symptomatic or asymptomatic) of confirmed CADASIL cases in whom genetic testing had not been undertaken were defined as having a 50%, 25% or 12.5% risk of CADASIL. The genetic diagnosis of CADASIL was established by positive NOTCH3 mutation analysis covering exons 2–23. Exons 2–6, 10, 18 and 22 of the NOTCH3 gene were amplified and bidirectionally sequenced. We have previously detected mutations in these exons in our patient cohort. If these exons were negative, subsequent screening of exons 1, 7– 9, 11–17, 19–21 and 23 was undertaken. Screening of NOTCH3 is centralized to one laboratory in 2
Scotland. Those diagnosed prior to 1993 had initially been diagnosed with a hereditary vascular dementia of the Sourander type (3) subsequently confirmed as CADASIL on genetic testing. Data collection
Clinical records from the neurovascular clinic and genetic records from the West of Scotland Clinical Genetics Service were reviewed. Pedigrees were constructed and drawn using Progeny CLINICAL Version 8 (Progeny Software LLC, Delray Beach, FL, USA; www.progenygenetics.com). Patients were included in analysis of clinical information if they had attended the neurovascular genetics clinic or good quality clinical information was available. Patient age when clinical information was last available was used for analysis of clinical features. The following information was collected: presence and age of onset of migraine, stroke, psychiatric illness, cognitive disturbance and seizures; cardiovascular risk factors; age at diagnosis and presentation to secondary health services; and functional dependence measured by mobility and independence in the home. Stroke was defined as a neurological deficit of sudden onset with focal dysfunction, and symptoms lasting more than 24 h presumed to be vascular in origin. TIA was defined as above but lasting T c.239 A>G† c.328 C>T c.397 C>T c.421 C>T c.457 C>T c.505 C>T c.544 C>T c.547 T>C c.580 T>A c.619 C>T c.664 C>T c.683 T>G c.733 T>A c.778 T>C c.779 G>A c.994 C>T Not stated c.1336T>G c.2956 T>C c.3664 T>G
Arg54Cys Asp80Gly Arg110Cys Arg133Cys Arg141Cys Arg153Cys Arg169Cys Arg182Cys Cys183Arg Cys194Ser Arg207Cys Cys222Arg Phe228Cys Cys245Ser Cys260Arg Cys260Tyr Arg332Cys Cys311Ser Cys446Gly Cys986Arg Cys1222Gly
3 1 2 4 9 3 5 2 2 2 2 1 1 2 1 1 3 1 1 1 2
*Mutation description at the DNA level in coding sequence of NOTCH3. Nucleotide number starting from the A of the AGT translation initiating methionine. † Atypical mutation.
seizures 8%. Fourteen percentage of the population had migraine as their only symptom, and a further 4% of the population were asymptomatic. Prevalence of conventional risk factors for stroke was 13/98 (13%) for hypertension, 40/68 (58%) for hypercholesterolaemia and 43/90 (47%) for history of smoking; 9/41 (22%) patients had a raised homocysteine (>20 lm). Eighty-five percentage had been on an antiplatelet, 70% on a statin and 18% on an antihypertensive (although not all had been diagnosed with hypertension). Time to event analysis for median age at death, first stroke, loss of independence and gait impairment is shown in Table 2 (previously reported estimates are included for comparison (7). The difference between men and women for age at first stroke was not significant although there was a trend towards men being younger at first stroke (52 vs 57 years; P = 0.054; Fig. 1A). We postulated median age at first stroke might change with time of diagnosis. Data were divided by date of clinical diagnosis (1986–2005, n = 53 and 2006–2012, n = 52) and time to event analysis was repeated. Median age at first stroke was significantly greater in male patients diagnosed more recently (56 vs 46 years; log rank P = 0.034; Fig. 1B). Median age at first stroke was not significantly different in women (57 vs 56 years; log rank P = 0.774). There were insufficient events for time to gait impairment or loss of independence to allow valid statistical comparison. 3
Moreton et al. Table 2 Comparison of time to event analysis for death, stroke, dependence and gait impairment with current population and previous reports (in years) Variable (median, 95% confidence intervals) Death Male Female Stroke Male Female Unable to walk without assistance Male Female Requirement for carers Male Female Bedridden Male Female
Scottish population 2013 69 69 70 55 52 57 65
(66–72) (62–76) (65–75) (52.8–57.1) (45.3–58.7) (54.1–59.8) (58.0–72.0)
A
Opherk 2004 (7)
64.6 (61.7–67.6) 70.7 (67.6–73.9) 50.7 (48.2–53.1) 52.5 (50.0–54.9)
67 (56.1–77.9) 62 (55.2–68.8) 71
58.9 (56.6–61.3) 62.1 (59.7–64.4) –
71 (48.4–93.5) 64 (55.5–72.5) – – –
– 62.1 (59.7–64.4) 66.5 (63.9–69.1)
B
Smoking status, hypercholesterolaemia, hypertension and mutation site by exon were not associated with latency to stroke, loss of independence or gait impairment. However, there was a trend towards earlier loss of independence among those with hypertension (hypertension median 55 years; no hypertension median 71 years, P = 0.096). MRI scans were suitable for assessment in 73 patients. Total Scheltens score was related to age at scan (linear regression, R2 = 0.388; P < 0.01). It was not related to sex, hypertension, hypercholesterolaemia, smoking or period of clinical diagnosis on multivariate modelling. Clinical status by age
Patients were divided into tertiles on basis of age when last seen and their mobility and independence examined. In the oldest tertile (58–75 years), 13/34 (38%) of patients were living independently in their own home and a further 9/34 (26%) lived in their own home with assistance of family members (Fig. 2A). Where information was known, 17/ 28 (61%) were mobile independently; however, information on gait was missing for four nursing home residents (Fig. 2B). If these were classed in the worst gait category (wheelchair or bedbound), 17/32 (53%) of patients were independently mobile in the oldest tertile.
Figure 1. Median age of stroke onset in CADASIL. (A) Kaplan–Meier curves for median age at stroke onset in men and women. (B) Median age at stroke onset in men stratified by period of clinical diagnosis.
living adults with confirmed CADASIL were 45 (GGC), 15 (Lanarkshire) and 8 (Forth Valley), giving a prevalence of 4.6 (95% confidence interval 3.4–6.2), 3.4 (1.9–5.6) and 3.4/100,000 adults (1.4–6.7) in each health board. Average prevalence across all three health boards was 4.1/ 100,000 adults (3.2–5.2).
Disease prevalence
Eighty-seven people with CADASIL were confirmed as alive and resident in Scotland. For the three health board areas sufficiently well-represented to allow reliable estimates, the numbers of 4
Mutation prevalence
Previously, we estimated prevalence of NOTCH3 gene mutation by establishing the location of ‘at risk’ relatives (12). Due to the large number of
Patterns and prevalence in CADASIL A
B
Figure 2. Clinical status of patients when last seen. (A) Independence, (B) Mobility. Dependence and impaired mobility increased with age but a significant proportion remain independent. Information for gait was not available for 4 patients known to be resident in nursing homes.
living relatives establishing the exact location of all individuals was not possible. Twenty-seven pedigrees were mainly resident in GGC and Lanarkshire Health Boards, with 637 known living family members. Fifty-one were definitely not resident in the area. For remaining family members, there were 141 at 50% risk, 191 at 25% risk and 93 at 12.5% risk, with the remainder at less than 12.5% risk. Estimated gene prevalence was 190 (130 estimated + 60 confirmed) NOTCH3 mutation carriers, with a predicted prevalence (symptomatic and asymptomatic) of 10.7 per 100,000 population (9.2–12.2). Discussion
Full appreciation of the clinical spectrum of any newly recognized diagnostic entity inevitably evolves over time. This analysis suggests that several features of CADASIL differ from previous descriptions, most notably an older median age at first stroke. Also, rather than being isolated cases (18), a high proportion of older patients remain independent and mobile (Fig. 2). Unlike previous studies, we did not find that an earlier onset of stroke was associated with cardiovascular risk factors (19) although there was a trend towards hypertension being associated with earlier loss of independence. Time to event analysis allows us to include patients who have not had a stroke in an analysis, rather than only calculating the mean
age at stroke. Median age of 57 years in women was outwith previous confidence intervals, but did not vary over diagnostic periods. Male median age of stroke onset (52 years) was within previous confidence intervals but in our population, those patients diagnosed more recently had a later onset of stroke than those diagnosed prior to 2006. Delayed time to first stroke in more recently diagnosed cases likely reflects increasing recognition of milder forms of the disorder, wider use of brain imaging (particularly MRI), easier availability of genetic testing, and consideration of the diagnosis even in older patients or those without a clear family history (20). It may reflect improved risk factor modification although our data was not able to assess this. We observed a trend towards higher use of antiplatelet medication in patients diagnosed earlier (40/42 vs 37/49; P = 0.09) but the length of time they were on this medication was not recorded. CADASIL can be catastrophic for some patients but severity is known to be variable. In 1998, Dichgans and colleagues reported a high prevalence of severe disability among CADASIL patients over 65 years old: 63% were categorized as modified Rankin scale 4 or 5 [unable to attend to own bodily needs or requiring constant nursing care (21)], and only 14% of patients over 60 years had no disability (22). In our patients over 58 years old, at least 53% were mobile without assistance (equivalent to mRS 3 or under) and 38% were living independently in their own home 5
Moreton et al. (equivalent to mRS 0–2). Whilst survivor bias will skew these data, it is notable that in the 26 patients still living in the 58–75 year category, 12 (46%) of them were entirely independent, and 17 of 25 (68%) mobile without aids (Fig. 2). Our cardiovascular risk factor distribution was similar to those seen in a recent study (19), but we could not confirm a relationship between hypertension and time to dependence, which may be a reflection of smaller sample size. Analysing patients who had been on antihypertensive therapy (rather than those diagnosed with hypertension) did not change these results. The minimum disease prevalence of CADASIL in Glasgow is 4.6/100,000 adult residents, a figure more than twice as high as previous estimates in Scotland and elsewhere in the United Kingdom (23). Estimated mutation prevalence is approximately 1/10,000 people. Levy and Feingold discussed a number of requirements for study design when estimating the prevalence of a genetic disease including a well-defined geographical area, homogenous diagnostic criteria, accurate denominator details based on official statistics and careful evaluation of family members (24). In fully penetrant autosomal dominant diseases where homozygotes are negligible in frequency, the disease frequency and heterozygote frequency should be the same. MRI changes in NOTCH3 mutation carriers appear fully penetrant by age 35 (25) but many heterozygotes will be symptomless for most of their life, and thus not all are recognized as affected (26). Systematic screening of asymptomatic family members in order to calculate mutation prevalence is neither ethical in routine practice nor cost-effective, and uptake of presymptomatic screening in CADASIL has been shown to be low (27). Our method for estimating mutation prevalence is based upon the rules of autosomal dominant inheritance and uses crude risk percentages. This method is believed to overestimate prevalence and whilst methods have been suggested to correct for this, they require a defined age-related risk (26), which is difficult to establish in CADASIL. A further potential inaccuracy in the data is that exact locations were not known for all relatives, and location information was taken from the index patient when first seen and only updated periodically. Thus, we cannot be sure that all relatives included in mutation estimation remain in the defined geographical areas. Whilst overestimation is possible, we must also account for under-ascertainment of CADASIL cases which is likely to include misdiagnosis and not being referred to neurovascular genetics clinic or genetics service. Testing for the gene at 6
only one centre in Scotland reduces the latter risk. If the local prevalence of 4.1/100,000 adults is extrapolated nationally (4,213,391 adults) (16) we would expect 173 (135–219) cases of CADASIL in Scotland: twice as many as are currently known to our clinic. However, there may be cases of CADASIL being managed by local services throughout Scotland. Identification of further cases of CADASIL may be enhanced by the use of prediagnostic screening scales (14, 28) which may be particularly useful in areas which do not have access to full exon screening. We were unable to apply such a scale to our data retrospectively. Whilst CADASIL remains, by the European Union definition of less than 50/100,000 population, a rare disease (29), the higher prevalence figure that we now estimate is important in considering the allocation of health resources and research funding. As this was a retrospective review of routinely acquired data, the main limitation of this study is incompleteness of clinical information, particularly for older cases, leading to potential bias. Differential attrition during follow-up may mean those who continue to attend for clinical review are more independent whereas disabled patients may have difficulties attending. Precise definitions of migraine and cognitive impairment could not used as these were not documented systematically in the clinical notes. Categorization of patients on the basis of tertiles and year of clinical diagnosis is arbitrary but avoids bias. The inclusion of a family with an atypical mutation (Asp80Gly) may be criticized but there are multiple reports of non-cysteine mutations being related to a CADASIL-like phenotype (30–32). One member of the family had the presence of granular osmiophilic material, and full exon screening performed in known affected family members revealed no other mutation. Conclusions
The prevalence of CADASIL may be as high as 1 per 10,000 people, and some affected individuals may retain independence and remain stroke free until an older age than previous estimates. Stroke physicians and neurologists should consider CADASIL as a differential even in older patients presenting with stroke. Acknowledgments Professor Hansj€ org B€ azner for kindly providing some information.
Patterns and prevalence in CADASIL Conflict of interest and sources of funding The authors declare no conflicting interests. This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
16. 17.
References 1. VAN BOGAERT L. Encephalopathie sous-corticale progressive (Binswanger) a evolution rapide chez deux soeurs. Med Hellen 1955;24:961–72. 2. STEVENS DL, HEWLETT RH, BROWNELL B. Chronic familial vascular encephalopathy. Lancet 1977;1:1364–5. 3. SOURANDER P, WALINDER J. Hereditary multi-infarct dementia. Morphological and clinical studies of a new disease. Acta Neuropathol 1977;39:247–54. 4. TOURNIER-LASSERVE E, JOUTEL A, MELKI J et al. Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy maps to chromosome 19q12. Nat Genet 1993;3:256–9. 5. JOUTEL A, CORPECHOT C, DUCROS A et al. Notch3 mutations in CADASIL, a hereditary adult-onset condition causing stroke and dementia. Nature 1996;383:707–10. 6. RUCHOUX MM, MAURAGE CA. CADASIL: cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy [Article]. J Neuropathol Exp Neurol 1997;56:947–64. 7. OPHERK C, PETERS N, HERZOG J, LUEDTKE R, DICHGANS M. Long-term prognosis and causes of death in CADASIL: a retrospective study in 411 patients. Brain 2004;127:2533–9. 8. PETERS N, HERZOG J, OPHERK C, DICHGANS M. A twoyear clinical follow-up study in 80 CADASIL subjects. Stroke 2004;35:1603–8. 9. DESMOND DW, MORONEY JT, LYNCH T, CHAN S, CHIN SS, MOHR JP. The natural history of CADASIL: a pooled analysis of previously published cases. Stroke 1999;30:1230–3. 10. PETERS N, OPHERK C, DANEK A, BALLARD C, HERZOG J, DICHGANS M. The pattern of cognitive performance in CADASIL: a monogenic condition leading to subcortical ischemic vascular dementia. Am J Psychiatry 2005;162:2078–85. 11. PESCINI F, BIANCHI S, SALVADORI E et al. A pathogenic mutation on exon 21 of the NOTCH3 gene causing CADASIL in an octogenarian paucisymptomatic patient. J Neurol Sci 2008;267:170–3. 12. RAZVI SSM, DAVIDSON R, BONE I, MUIR KW. The prevalence of cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy (CADASIL) in the west of Scotland. J Neurol Neurosurg Psychiatry 2005;76:739–41. 13. ST CLAIR D, BOLT J, MORRIS S, DOYLE D. Hereditary multi-infarct dementia unlinked to chromosome 19q12 in a large Scottish pedigree: evidence of probable locus heterogeneity. J Med Genet 1995;32:57–60. 14. PESCINI F, NANNUCCI S, BERTACCINI B et al. The cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) scale. Stroke 2012;43:2871–6. 15. SCHELTENS P, BARKHOF F, LEYS D et al. A semiquantative rating scale for the assessment of signal hyperinten-
18.
19.
20.
21.
22.
23.
24.
25. 26.
27.
28.
29.
30.
31.
32.
sities on magnetic resonance imaging. J Neurol Sci 1993;114:7–12. ANON. Mid-2011 population estimates Scotland. Scotland: National Records of Scotland, 2012. MORETON FC, BAEZNER H, DAVIDSON R, MUIR KW. CADASIL phenotype associated with a cysteine-sparing NOTCH 3 mutation. Cerebrovasc Dis 2012;33(Suppl. 2):930–1. LEE YC, YANG AH, SOONG BW. The remarkably variable expressivity of CADASIL: report of a minimally symptomatic man at an advanced age. J Neurol 2009;256:1026–7. ADIB-SAMII P, BRICE G, MARTIN RJ, MARKUS HS. Clinical spectrum of CADASIL and the effect of cardiovascular risk factors on phenotype. Stroke 2010;41: 630–4. RAZVI SSM, DAVIDSON R, BONE I, MUIR KW. Is inadequate family history a barrier to diagnosis in CADASIL? Acta Neurol Scand 2005;112:323–6. RANKIN J. Cerebral vascular accidents in patients over the age of 60. II. Prognosis. Scott Med J 1957;2: 200–15. DICHGANS M, MAYER M, UTTNER I et al. The phenotypic spectrum of CADASIL: clinical findings in 102 cases. Ann Neurol 1998;44:731–9. NARAYAN SK, GORMAN G, KALARIA RN, FORD GA, CHINNERY PF. The minimum prevalence of CADASIL in northeast England. Neurology 2012;78:1025–7. LEVY M, FEINGOLD J. Estimating prevalence in singlegene kidney diseases progressing to renal failure. Kidney Int 2000;58:925–43. CHABRIAT H, LEVY C, TAILLIA H et al. Patterns of MRI lesions in CADASIL. Neurology 1998;51:452–7. HARPER P. The epidemiology of Huntington’s disease. In: G BATES, HARPER PS, JONES L, eds. Huntington’s disease, 3rd edn. Oxford: Oxford University Press, 2002;159–97. REYES S, KURTZ A, HERVE D, TOURNIER-LASSERVE E, CHABRIAT H. Presymptomatic genetic testing in CADASIL. J Neurol 2012;259:2131–6. PANTONI L, PESCINI F, NANNUCCI S et al. Comparison of clinical, familial, and MRI features of CADASIL and NOTCH3-negative patients. Neurology 2010;74:57– 63. VASSILIOU A. Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions on Rare Diseases – Europe’s challenges. Brussels: Commission of the European Communities, EU, 2008. SCHEID R, HEINRITZ W, LEYHE T et al. Cysteine-sparing Notch3 mutations: CADASIL or CADASIL variants? Neurology 2008;71:774–6. MIZUNO T, MURANISHI M, TORUGUN T et al. Two Japanese CADASIL families exhibiting Notch3 mutation R75P not involving cysteine residue. Intern Med 2008;47:2067–72. KIM Y, CHOI EJ, CHOI CG et al. Characteristics of CADASIL in Korea: a novel cysteine-sparing Notch3 mutation. Neurology 2006;66:1511–6.
7
© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd ACTA NEUROLOGICA SCANDINAVICA
Changing clinical patterns and increasing prevalence in CADASIL Moreton FC, Razvi SSM, Davidson R, Muir KW. Changing clinical patterns and increasing prevalence in CADASIL. Acta Neurol Scand: DOI: 10.1111/ane.12266. © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd. Objectives – CADASIL is a monogenic small vessel vasculopathy causing recurrent stroke. Early descriptions suggested dementia and disability were common from the 5th decade but there is evidence of marked phenotypic variability. We investigated the prevalence and clinical features of CADASIL in the west of Scotland. Methods – We undertook a retrospective review of clinical records of patients with confirmed CADASIL identified through a specialist clinic. Patients were divided to examine the effect of date of diagnosis on clinical outcomes and the characteristics at different ages. The location of pedigree members was used to estimate prevalence. Results – Twentyone different CADASIL-causing NOTCH3 mutations were identified in 49 pedigrees (61% in exon 4). Disease prevalence in Glasgow was 4.6/100,000 adults. Mutation prevalence was estimated at 10.7/100,000 population. Median age at first stroke in women (57 years) was higher than previous estimates, and stroke age in men was higher in patients diagnosed more recently (pre 2006 46 years, post 2006 56 years, P = 0.034). In patients over 58 years of age, 13/34 (38%) were living independently and 17/28 (61%) were mobile without aids when last seen. Conclusions – CADASIL prevalence is at least 4.6 per 100,000 adults. Median age of first stroke may be older than previously thought. Clinicians should consider CADASIL in the differential diagnosis even in older patients with stroke.
Introduction
From the 1950s to 1980s, a number of families with an apparently hereditary vascular dementia accompanied by a Binswanger’s-like arteriopathy were described (1–3). In 1993, two families with stroke-like episodes, dementia and leukoencephalopathy were demonstrated to have an inherited mutation on chromosome 19, and the acronym CADASIL was introduced: cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (4). Subsequent genetic studies demonstrated the disorder was due to mutations in the NOTCH3 gene on chromosome 19p13 which encodes a receptor expressed by vascular smooth muscle cells in adults (5). CADASIL is characterized by loss of vascular smooth muscle cells and fibrosis of small penetrating vessels, resulting in chronic white matter ischaemia (6).
F. C. Moreton1, S. S. M. Razvi2, R. Davidson3, K. W. Muir1 1 Institute of Neuroscience and Psychology, University of Glasgow, Southern General Hospital, Glasgow, UK; 2 Department of Neurology, Institute of Neurological Sciences, Southern General Hospital, Glasgow, UK; 3 West of Scotland Genetic Services, Southern General Hospital, Glasgow, UK
Key words: Neuroepidemiology; Neurogenetics F. C. Moreton, Institute of Neuroscience and Psychology, University of Glasgow, Southern General Hospital, 1345 Govan Road, Glasgow, G51 3TF, Scotland Tel.: +44 141 201 2144 Fax: +44 141 201 2136 e-mail: [email protected] Accepted for publication April 25, 2014
The clinical features and natural history of CADASIL have been well documented (7, 8). Migraine is reported as a common presenting symptom occurring in young adulthood, but the most frequent manifestation is subcortical ischaemic events, occurring from the fourth to sixth decades, usually in the absence of, or disproportionately severe in relation to, conventional vascular risk factors (9). Recurrent ischaemic events may lead to vascular parkinsonism and pseudobulbar palsy. Concurrently cognitive impairment, initially characterized by executive dysfunction, but extending to a more global dementia, occurs (10). Severe disability with dementia is typically described by age 50–60 (7). However, cases have been described where patients are asymptomatic even in late adulthood (11). A specialist neurovascular clinic for CADASIL in Scotland was set up in 2002 in association with a genetics clinic. In 2005, we calculated the 1
Moreton et al. disease prevalence of CADASIL in the west of Scotland as 1.98 per 100,000 adults (12), considered likely to be an underestimate consequent to under-recognition. Recognition of CADASIL was likely biased towards larger pedigrees with typical features. The first two pedigrees diagnosed with hereditary vascular dementia (subsequently confirmed as CADASIL) in our unit in 1986 had over 100 members at the time of formal diagnosis (13). Availability of genetic screening of exons 2–23 of the NOTCH3 gene, the development of prediagnostic screening tools (14), improved awareness of the disease and its manifestations, wider use of brain MRI, and availability of a local specialist clinical service all suggested that increasing numbers of cases would be diagnosed with CADASIL over time. We reviewed the presentation of CADASIL to determine whether this had changed since its initial recognition. Our aims were (i) identify any changes in patient presentation since CADASIL was recognized; (ii) examine the characteristics of older CADASIL patients; and (iii) estimate mutation prevalence. Methods Identification of cases
A neurovascular genetics clinic was established in 2002 in association with local clinical genetics services, to facilitate the diagnosis and management of patients with CADASIL. Referrals are accepted from throughout Scotland. Individuals were defined for the purposes of this study as having confirmed CADASIL if they had a pathogenic mutation on analysis of exons 2–23 of the NOTCH3 gene, demonstrated granular osmiophilic material (GOM) on biopsy or post-mortem material, or had both a personal history consistent with CADASIL and were by analysis of the family tree, an obligate gene carrier. First, 2nd and 3rd degree relatives (symptomatic or asymptomatic) of confirmed CADASIL cases in whom genetic testing had not been undertaken were defined as having a 50%, 25% or 12.5% risk of CADASIL. The genetic diagnosis of CADASIL was established by positive NOTCH3 mutation analysis covering exons 2–23. Exons 2–6, 10, 18 and 22 of the NOTCH3 gene were amplified and bidirectionally sequenced. We have previously detected mutations in these exons in our patient cohort. If these exons were negative, subsequent screening of exons 1, 7– 9, 11–17, 19–21 and 23 was undertaken. Screening of NOTCH3 is centralized to one laboratory in 2
Scotland. Those diagnosed prior to 1993 had initially been diagnosed with a hereditary vascular dementia of the Sourander type (3) subsequently confirmed as CADASIL on genetic testing. Data collection
Clinical records from the neurovascular clinic and genetic records from the West of Scotland Clinical Genetics Service were reviewed. Pedigrees were constructed and drawn using Progeny CLINICAL Version 8 (Progeny Software LLC, Delray Beach, FL, USA; www.progenygenetics.com). Patients were included in analysis of clinical information if they had attended the neurovascular genetics clinic or good quality clinical information was available. Patient age when clinical information was last available was used for analysis of clinical features. The following information was collected: presence and age of onset of migraine, stroke, psychiatric illness, cognitive disturbance and seizures; cardiovascular risk factors; age at diagnosis and presentation to secondary health services; and functional dependence measured by mobility and independence in the home. Stroke was defined as a neurological deficit of sudden onset with focal dysfunction, and symptoms lasting more than 24 h presumed to be vascular in origin. TIA was defined as above but lasting T c.239 A>G† c.328 C>T c.397 C>T c.421 C>T c.457 C>T c.505 C>T c.544 C>T c.547 T>C c.580 T>A c.619 C>T c.664 C>T c.683 T>G c.733 T>A c.778 T>C c.779 G>A c.994 C>T Not stated c.1336T>G c.2956 T>C c.3664 T>G
Arg54Cys Asp80Gly Arg110Cys Arg133Cys Arg141Cys Arg153Cys Arg169Cys Arg182Cys Cys183Arg Cys194Ser Arg207Cys Cys222Arg Phe228Cys Cys245Ser Cys260Arg Cys260Tyr Arg332Cys Cys311Ser Cys446Gly Cys986Arg Cys1222Gly
3 1 2 4 9 3 5 2 2 2 2 1 1 2 1 1 3 1 1 1 2
*Mutation description at the DNA level in coding sequence of NOTCH3. Nucleotide number starting from the A of the AGT translation initiating methionine. † Atypical mutation.
seizures 8%. Fourteen percentage of the population had migraine as their only symptom, and a further 4% of the population were asymptomatic. Prevalence of conventional risk factors for stroke was 13/98 (13%) for hypertension, 40/68 (58%) for hypercholesterolaemia and 43/90 (47%) for history of smoking; 9/41 (22%) patients had a raised homocysteine (>20 lm). Eighty-five percentage had been on an antiplatelet, 70% on a statin and 18% on an antihypertensive (although not all had been diagnosed with hypertension). Time to event analysis for median age at death, first stroke, loss of independence and gait impairment is shown in Table 2 (previously reported estimates are included for comparison (7). The difference between men and women for age at first stroke was not significant although there was a trend towards men being younger at first stroke (52 vs 57 years; P = 0.054; Fig. 1A). We postulated median age at first stroke might change with time of diagnosis. Data were divided by date of clinical diagnosis (1986–2005, n = 53 and 2006–2012, n = 52) and time to event analysis was repeated. Median age at first stroke was significantly greater in male patients diagnosed more recently (56 vs 46 years; log rank P = 0.034; Fig. 1B). Median age at first stroke was not significantly different in women (57 vs 56 years; log rank P = 0.774). There were insufficient events for time to gait impairment or loss of independence to allow valid statistical comparison. 3
Moreton et al. Table 2 Comparison of time to event analysis for death, stroke, dependence and gait impairment with current population and previous reports (in years) Variable (median, 95% confidence intervals) Death Male Female Stroke Male Female Unable to walk without assistance Male Female Requirement for carers Male Female Bedridden Male Female
Scottish population 2013 69 69 70 55 52 57 65
(66–72) (62–76) (65–75) (52.8–57.1) (45.3–58.7) (54.1–59.8) (58.0–72.0)
A
Opherk 2004 (7)
64.6 (61.7–67.6) 70.7 (67.6–73.9) 50.7 (48.2–53.1) 52.5 (50.0–54.9)
67 (56.1–77.9) 62 (55.2–68.8) 71
58.9 (56.6–61.3) 62.1 (59.7–64.4) –
71 (48.4–93.5) 64 (55.5–72.5) – – –
– 62.1 (59.7–64.4) 66.5 (63.9–69.1)
B
Smoking status, hypercholesterolaemia, hypertension and mutation site by exon were not associated with latency to stroke, loss of independence or gait impairment. However, there was a trend towards earlier loss of independence among those with hypertension (hypertension median 55 years; no hypertension median 71 years, P = 0.096). MRI scans were suitable for assessment in 73 patients. Total Scheltens score was related to age at scan (linear regression, R2 = 0.388; P < 0.01). It was not related to sex, hypertension, hypercholesterolaemia, smoking or period of clinical diagnosis on multivariate modelling. Clinical status by age
Patients were divided into tertiles on basis of age when last seen and their mobility and independence examined. In the oldest tertile (58–75 years), 13/34 (38%) of patients were living independently in their own home and a further 9/34 (26%) lived in their own home with assistance of family members (Fig. 2A). Where information was known, 17/ 28 (61%) were mobile independently; however, information on gait was missing for four nursing home residents (Fig. 2B). If these were classed in the worst gait category (wheelchair or bedbound), 17/32 (53%) of patients were independently mobile in the oldest tertile.
Figure 1. Median age of stroke onset in CADASIL. (A) Kaplan–Meier curves for median age at stroke onset in men and women. (B) Median age at stroke onset in men stratified by period of clinical diagnosis.
living adults with confirmed CADASIL were 45 (GGC), 15 (Lanarkshire) and 8 (Forth Valley), giving a prevalence of 4.6 (95% confidence interval 3.4–6.2), 3.4 (1.9–5.6) and 3.4/100,000 adults (1.4–6.7) in each health board. Average prevalence across all three health boards was 4.1/ 100,000 adults (3.2–5.2).
Disease prevalence
Eighty-seven people with CADASIL were confirmed as alive and resident in Scotland. For the three health board areas sufficiently well-represented to allow reliable estimates, the numbers of 4
Mutation prevalence
Previously, we estimated prevalence of NOTCH3 gene mutation by establishing the location of ‘at risk’ relatives (12). Due to the large number of
Patterns and prevalence in CADASIL A
B
Figure 2. Clinical status of patients when last seen. (A) Independence, (B) Mobility. Dependence and impaired mobility increased with age but a significant proportion remain independent. Information for gait was not available for 4 patients known to be resident in nursing homes.
living relatives establishing the exact location of all individuals was not possible. Twenty-seven pedigrees were mainly resident in GGC and Lanarkshire Health Boards, with 637 known living family members. Fifty-one were definitely not resident in the area. For remaining family members, there were 141 at 50% risk, 191 at 25% risk and 93 at 12.5% risk, with the remainder at less than 12.5% risk. Estimated gene prevalence was 190 (130 estimated + 60 confirmed) NOTCH3 mutation carriers, with a predicted prevalence (symptomatic and asymptomatic) of 10.7 per 100,000 population (9.2–12.2). Discussion
Full appreciation of the clinical spectrum of any newly recognized diagnostic entity inevitably evolves over time. This analysis suggests that several features of CADASIL differ from previous descriptions, most notably an older median age at first stroke. Also, rather than being isolated cases (18), a high proportion of older patients remain independent and mobile (Fig. 2). Unlike previous studies, we did not find that an earlier onset of stroke was associated with cardiovascular risk factors (19) although there was a trend towards hypertension being associated with earlier loss of independence. Time to event analysis allows us to include patients who have not had a stroke in an analysis, rather than only calculating the mean
age at stroke. Median age of 57 years in women was outwith previous confidence intervals, but did not vary over diagnostic periods. Male median age of stroke onset (52 years) was within previous confidence intervals but in our population, those patients diagnosed more recently had a later onset of stroke than those diagnosed prior to 2006. Delayed time to first stroke in more recently diagnosed cases likely reflects increasing recognition of milder forms of the disorder, wider use of brain imaging (particularly MRI), easier availability of genetic testing, and consideration of the diagnosis even in older patients or those without a clear family history (20). It may reflect improved risk factor modification although our data was not able to assess this. We observed a trend towards higher use of antiplatelet medication in patients diagnosed earlier (40/42 vs 37/49; P = 0.09) but the length of time they were on this medication was not recorded. CADASIL can be catastrophic for some patients but severity is known to be variable. In 1998, Dichgans and colleagues reported a high prevalence of severe disability among CADASIL patients over 65 years old: 63% were categorized as modified Rankin scale 4 or 5 [unable to attend to own bodily needs or requiring constant nursing care (21)], and only 14% of patients over 60 years had no disability (22). In our patients over 58 years old, at least 53% were mobile without assistance (equivalent to mRS 3 or under) and 38% were living independently in their own home 5
Moreton et al. (equivalent to mRS 0–2). Whilst survivor bias will skew these data, it is notable that in the 26 patients still living in the 58–75 year category, 12 (46%) of them were entirely independent, and 17 of 25 (68%) mobile without aids (Fig. 2). Our cardiovascular risk factor distribution was similar to those seen in a recent study (19), but we could not confirm a relationship between hypertension and time to dependence, which may be a reflection of smaller sample size. Analysing patients who had been on antihypertensive therapy (rather than those diagnosed with hypertension) did not change these results. The minimum disease prevalence of CADASIL in Glasgow is 4.6/100,000 adult residents, a figure more than twice as high as previous estimates in Scotland and elsewhere in the United Kingdom (23). Estimated mutation prevalence is approximately 1/10,000 people. Levy and Feingold discussed a number of requirements for study design when estimating the prevalence of a genetic disease including a well-defined geographical area, homogenous diagnostic criteria, accurate denominator details based on official statistics and careful evaluation of family members (24). In fully penetrant autosomal dominant diseases where homozygotes are negligible in frequency, the disease frequency and heterozygote frequency should be the same. MRI changes in NOTCH3 mutation carriers appear fully penetrant by age 35 (25) but many heterozygotes will be symptomless for most of their life, and thus not all are recognized as affected (26). Systematic screening of asymptomatic family members in order to calculate mutation prevalence is neither ethical in routine practice nor cost-effective, and uptake of presymptomatic screening in CADASIL has been shown to be low (27). Our method for estimating mutation prevalence is based upon the rules of autosomal dominant inheritance and uses crude risk percentages. This method is believed to overestimate prevalence and whilst methods have been suggested to correct for this, they require a defined age-related risk (26), which is difficult to establish in CADASIL. A further potential inaccuracy in the data is that exact locations were not known for all relatives, and location information was taken from the index patient when first seen and only updated periodically. Thus, we cannot be sure that all relatives included in mutation estimation remain in the defined geographical areas. Whilst overestimation is possible, we must also account for under-ascertainment of CADASIL cases which is likely to include misdiagnosis and not being referred to neurovascular genetics clinic or genetics service. Testing for the gene at 6
only one centre in Scotland reduces the latter risk. If the local prevalence of 4.1/100,000 adults is extrapolated nationally (4,213,391 adults) (16) we would expect 173 (135–219) cases of CADASIL in Scotland: twice as many as are currently known to our clinic. However, there may be cases of CADASIL being managed by local services throughout Scotland. Identification of further cases of CADASIL may be enhanced by the use of prediagnostic screening scales (14, 28) which may be particularly useful in areas which do not have access to full exon screening. We were unable to apply such a scale to our data retrospectively. Whilst CADASIL remains, by the European Union definition of less than 50/100,000 population, a rare disease (29), the higher prevalence figure that we now estimate is important in considering the allocation of health resources and research funding. As this was a retrospective review of routinely acquired data, the main limitation of this study is incompleteness of clinical information, particularly for older cases, leading to potential bias. Differential attrition during follow-up may mean those who continue to attend for clinical review are more independent whereas disabled patients may have difficulties attending. Precise definitions of migraine and cognitive impairment could not used as these were not documented systematically in the clinical notes. Categorization of patients on the basis of tertiles and year of clinical diagnosis is arbitrary but avoids bias. The inclusion of a family with an atypical mutation (Asp80Gly) may be criticized but there are multiple reports of non-cysteine mutations being related to a CADASIL-like phenotype (30–32). One member of the family had the presence of granular osmiophilic material, and full exon screening performed in known affected family members revealed no other mutation. Conclusions
The prevalence of CADASIL may be as high as 1 per 10,000 people, and some affected individuals may retain independence and remain stroke free until an older age than previous estimates. Stroke physicians and neurologists should consider CADASIL as a differential even in older patients presenting with stroke. Acknowledgments Professor Hansj€ org B€ azner for kindly providing some information.
Patterns and prevalence in CADASIL Conflict of interest and sources of funding The authors declare no conflicting interests. This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
16. 17.
References 1. VAN BOGAERT L. Encephalopathie sous-corticale progressive (Binswanger) a evolution rapide chez deux soeurs. Med Hellen 1955;24:961–72. 2. STEVENS DL, HEWLETT RH, BROWNELL B. Chronic familial vascular encephalopathy. Lancet 1977;1:1364–5. 3. SOURANDER P, WALINDER J. Hereditary multi-infarct dementia. Morphological and clinical studies of a new disease. Acta Neuropathol 1977;39:247–54. 4. TOURNIER-LASSERVE E, JOUTEL A, MELKI J et al. Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy maps to chromosome 19q12. Nat Genet 1993;3:256–9. 5. JOUTEL A, CORPECHOT C, DUCROS A et al. Notch3 mutations in CADASIL, a hereditary adult-onset condition causing stroke and dementia. Nature 1996;383:707–10. 6. RUCHOUX MM, MAURAGE CA. CADASIL: cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy [Article]. J Neuropathol Exp Neurol 1997;56:947–64. 7. OPHERK C, PETERS N, HERZOG J, LUEDTKE R, DICHGANS M. Long-term prognosis and causes of death in CADASIL: a retrospective study in 411 patients. Brain 2004;127:2533–9. 8. PETERS N, HERZOG J, OPHERK C, DICHGANS M. A twoyear clinical follow-up study in 80 CADASIL subjects. Stroke 2004;35:1603–8. 9. DESMOND DW, MORONEY JT, LYNCH T, CHAN S, CHIN SS, MOHR JP. The natural history of CADASIL: a pooled analysis of previously published cases. Stroke 1999;30:1230–3. 10. PETERS N, OPHERK C, DANEK A, BALLARD C, HERZOG J, DICHGANS M. The pattern of cognitive performance in CADASIL: a monogenic condition leading to subcortical ischemic vascular dementia. Am J Psychiatry 2005;162:2078–85. 11. PESCINI F, BIANCHI S, SALVADORI E et al. A pathogenic mutation on exon 21 of the NOTCH3 gene causing CADASIL in an octogenarian paucisymptomatic patient. J Neurol Sci 2008;267:170–3. 12. RAZVI SSM, DAVIDSON R, BONE I, MUIR KW. The prevalence of cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy (CADASIL) in the west of Scotland. J Neurol Neurosurg Psychiatry 2005;76:739–41. 13. ST CLAIR D, BOLT J, MORRIS S, DOYLE D. Hereditary multi-infarct dementia unlinked to chromosome 19q12 in a large Scottish pedigree: evidence of probable locus heterogeneity. J Med Genet 1995;32:57–60. 14. PESCINI F, NANNUCCI S, BERTACCINI B et al. The cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) scale. Stroke 2012;43:2871–6. 15. SCHELTENS P, BARKHOF F, LEYS D et al. A semiquantative rating scale for the assessment of signal hyperinten-
18.
19.
20.
21.
22.
23.
24.
25. 26.
27.
28.
29.
30.
31.
32.
sities on magnetic resonance imaging. J Neurol Sci 1993;114:7–12. ANON. Mid-2011 population estimates Scotland. Scotland: National Records of Scotland, 2012. MORETON FC, BAEZNER H, DAVIDSON R, MUIR KW. CADASIL phenotype associated with a cysteine-sparing NOTCH 3 mutation. Cerebrovasc Dis 2012;33(Suppl. 2):930–1. LEE YC, YANG AH, SOONG BW. The remarkably variable expressivity of CADASIL: report of a minimally symptomatic man at an advanced age. J Neurol 2009;256:1026–7. ADIB-SAMII P, BRICE G, MARTIN RJ, MARKUS HS. Clinical spectrum of CADASIL and the effect of cardiovascular risk factors on phenotype. Stroke 2010;41: 630–4. RAZVI SSM, DAVIDSON R, BONE I, MUIR KW. Is inadequate family history a barrier to diagnosis in CADASIL? Acta Neurol Scand 2005;112:323–6. RANKIN J. Cerebral vascular accidents in patients over the age of 60. II. Prognosis. Scott Med J 1957;2: 200–15. DICHGANS M, MAYER M, UTTNER I et al. The phenotypic spectrum of CADASIL: clinical findings in 102 cases. Ann Neurol 1998;44:731–9. NARAYAN SK, GORMAN G, KALARIA RN, FORD GA, CHINNERY PF. The minimum prevalence of CADASIL in northeast England. Neurology 2012;78:1025–7. LEVY M, FEINGOLD J. Estimating prevalence in singlegene kidney diseases progressing to renal failure. Kidney Int 2000;58:925–43. CHABRIAT H, LEVY C, TAILLIA H et al. Patterns of MRI lesions in CADASIL. Neurology 1998;51:452–7. HARPER P. The epidemiology of Huntington’s disease. In: G BATES, HARPER PS, JONES L, eds. Huntington’s disease, 3rd edn. Oxford: Oxford University Press, 2002;159–97. REYES S, KURTZ A, HERVE D, TOURNIER-LASSERVE E, CHABRIAT H. Presymptomatic genetic testing in CADASIL. J Neurol 2012;259:2131–6. PANTONI L, PESCINI F, NANNUCCI S et al. Comparison of clinical, familial, and MRI features of CADASIL and NOTCH3-negative patients. Neurology 2010;74:57– 63. VASSILIOU A. Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions on Rare Diseases – Europe’s challenges. Brussels: Commission of the European Communities, EU, 2008. SCHEID R, HEINRITZ W, LEYHE T et al. Cysteine-sparing Notch3 mutations: CADASIL or CADASIL variants? Neurology 2008;71:774–6. MIZUNO T, MURANISHI M, TORUGUN T et al. Two Japanese CADASIL families exhibiting Notch3 mutation R75P not involving cysteine residue. Intern Med 2008;47:2067–72. KIM Y, CHOI EJ, CHOI CG et al. Characteristics of CADASIL in Korea: a novel cysteine-sparing Notch3 mutation. Neurology 2006;66:1511–6.
7
