Cerebral autosomal dominant arteriopathy with subcortical. Clinical Sciences

Transcription

1 Clinical Sciences Predictors of Clinical Worsening in Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy Prospective Cohort Study Hugues Chabriat, MD; Dominique Hervé, MD; Marco Duering, MD; Ophelia Godin, PhD; Eric Jouvent, MD; Christian Opherk, MD; Nassira Alili, MD; Sonia Reyes, MSc; Aude Jabouley, MSc; Nikola Zieren, PhD; Jean-Pierre Guichard, MD; Chahin Pachai, PhD; Eric Vicaut, MD; Martin Dichgans, MD Background and Purpose Predictors of clinical worsening in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy remain unknown. This study aims to identify demographic, clinical, and magnetic resonance imaging predictors of incident strokes, incident dementia, clinical deterioration, and death in patients with this genetically proven disease. Methods Two hundred ninety subjects (mean age, 50.6±11.4 years) were assessed at baseline and followed up for 36 months. Incident clinical events were recorded, and clinical scores included the Mini Mental State Examination, Mattis Dementia Rating Scale, modified Rankin Scale, and Barthel index. The number of lacunes and microbleeds, the volume of white-matter hyperintensities, and brain parenchymal fraction were assessed on baseline magnetic resonance imaging. Data were analyzed by ANCOVA, multivariable logistic regression, and Cox proportional hazard models. Results Incident stroke occurred in 55 of 278 patients (19.8%). Moderate or severe disability developed in 19 of 210 (9%) nondisabled individuals, incident dementia in 49 of 231 (20%) nondemented subjects, and 4.8% of patients died. Active smoking, the number of lacunes, and brain parenchymal fraction independently predicted incident stroke during follow-up. Gait disturbance, dementia, and brain parenchymal fraction predicted progression toward moderate or severe disability. Active smoking, disability, and brain parenchymal fraction predicted incident dementia. Age was the only significant predictor of death. Conclusions Clinical assessment and brain magnetic resonance imaging aid in predicting incident clinical events and clinical deterioration in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy. There is a bidirectional relationship between dementia and moderate or severe disability in predicting each other s onset. Active smoking is a modifiable risk factor associated with clinical progression in Notch3 mutation carriers. (Stroke. 2016;47:4-11. DOI: /STROKEAHA ) Key Words: CADASIL cerebral small vessel diseases cohort studies magnetic resonance imaging risk factors Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is an inherited small vessel disease (SVD) caused by mutations in the NOTCH3 gene whose prevalence may reach 10.7/ in the general population. 1,2 CADASIL has emerged as the most frequent hereditary cause of stroke in adults and a major cause of vascular cognitive impairment. 3 The clinical course is highly variable. Some patients develop severe manifestations while still in their 20s, whereas others remain symptom free until late age. 3,4 Genotype phenotype studies failed to identify a differential effect of individual Notch3 mutations on clinical Received July 7, 2015; final revision received October 13, 2015; accepted October 16, From the Department of Neurology, GH Saint-Louis-Lariboisière, Assistance Publique des Hôpitaux de Paris (APHP), Université Paris Denis Diderot and DHU NeuroVasc Sorbonne Paris-Cité, Paris, France (H.C., D.H, O.G., E.J., N.A., S.R., A.J.); INSERM UMR 1161, Paris, France (H.C., D.H., E.J.); Institute for Stroke and Dementia Research, Klinikum der Universität München, Ludwig-Maximilians-University, Munich, Germany (M.D., C.O., N.Z., M.D.); Department of Neuroradiology, CHU Lariboisière, Assistance Publique des Hôpitaux de Paris, Paris, France (J.-P.G.); Bioclinica Inc, Lyon, France (C.P.); Unité de Recherche Clinique, GH Saint Louis-Lariboisière, Assistance Publique des Hôpitaux de Paris, Université Paris Denis Diderot, Paris, France (E.V.); and Munich Cluster for Systems Neurology (SyNergy), Munich, Germany (M.D.). The online-only Data Supplement is available with this article at /-/DC1. Correspondence to Hugues Chabriat, MD, Service de Neurologie, Hopital Lariboisiere, Université Paris 7, 2 rue A Paré, Paris, France. hugues.chabriat@lrb.aphp.fr 2015 American Heart Association, Inc. Stroke is available at DOI: /STROKEAHA

2 Chabriat et al Predictors in CADASIL 5 progression, and the only established risk factor for clinical deterioration till now is age. Hence, counseling mutation carriers and selecting patients for clinical trials 5 have been difficult. Given the paucity of prognostic data, 6 10 we initiated a prospective longitudinal study to identify the predictors of clinical progression. The clinical and imaging characteristics of CADASIL much resemble those of the most severe forms of sporadic SVD. We, thus, reasoned that information derived from a longitudinal study in CADASIL might offer insights relevant to SVD in general. Here, we report the final results of this study obtained in a large cohort of patients followed up for 3 years. Methods Study Cohort Subjects were prospectively recruited between September 2003 and April 2011 through 2 major referral centers for CADASIL (University Hospital Lariboisière, Paris [n=178] and Ludwig Maximilians Universität, Munich [n=112]). They were included regardless of whether they had clinical manifestations of the disease if they were at least 18 years of age, had a documented mutation in the NOTCH3 gene, and were willing to be followed up. Details of the study protocol have been reported elsewhere. 11 A follow-up interval of 3 years was chosen based on previous longitudinal data showing that significant changes in clinical scores can be detected within an interval of 2 years 8 when following up patients seen at major referral centers and to obtain a sufficient number of clinical events. In brief, clinical and demographic data were collected at study entry and included age, sex, years of education, and vascular risk factors (hypertension, diabetes mellitus, hypercholesteremia, smoking, and alcohol use). Blood pressure was measured at baseline and at follow-up. A history of transient ischemic attacks (focal symptoms lasting <24 hours presumably of vascular origin) or stroke (rapidly evolving focal symptoms lasting 24 hours with no apparent cause other than of vascular origin), seizures, gait disturbance (any permanent difficulty to walk), psychiatric disorders (any psychological syndrome leading to treatment or hospitalization), and dementia defined according to Diagnostic and Statistical Manual of Mental Disorders, 4th Edition criteria was obtained at baseline and at follow-up. Subjects were interviewed and asked for cognitive complaints and any permanent difficulties in gait and balance and underwent a detailed neurological and neuropsychological examination at baseline and at 3 years. Global cognitive function was assessed using the Mini Mental State Examination and the Mattis Dementia Rating Scale (MDRS). The initiation/perseveration subscale of the MDRS, highly sensitive to subcortical lesions in cerebral SVD, 12 was analyzed separately. Disability was assessed using the modified Rankin Scale (mrs). Functional independence was determined using the Barthel index. Written informed consent was obtained from the study participant or a close relative if the patient was too severely disabled. This study was approved by the ethics committees of both participating centers. We used standardized procedures as recommended in therapeutic trials to continuously check both the validity and accuracy of data with regular monitoring (online-only Data Supplement). 13 Database verification, cleaning, and validation (double data entry, written queries for all inconsistencies, and corrections) were performed by Orgametrie Inc (Roubaix, France). Magnetic Resonance Imaging Data Details on scanner characteristics and magnetic resonance imaging (MRI) sequences have been reported elsewhere 11 (online-only Data Supplement). The protocol included millimetric 3-dimensional (3D) T1-weighted images, axial slices of 5 mm thickness of fluid-attenuated inversion recovery images, and T2*-weighted gradient-echo planar images. Native and preprocessed data were displayed on Unix workstations running image analysis software from Bioclinica Inc. Validated methods were used by board-certified neurologists and neuroradiologists with a long experience in SVD imaging research for evaluating white-matter hyperintensities (WMH), lacunes, and microbleeds as previously described. 11,14 The total volume of WMH was normalized to the intracranial cavity (ICC; nwmh=[volume of WMH/volume ICC] 100). Trained raters segmented lacunes manually using appropriate 2D and 3D imaging tools. In difficult cases, a consensus was obtained. Special care was taken to exclude enlarged perivascular spaces. Microbleeds were defined as rounded foci 5 mm in diameter hypointense on gradient-echo sequences. Brain volumes were determined using SIENAX and T1-weighted images. 15 Normalized brain volumes corrected for skull size (brain parenchymal fraction [BPF]=brain volume/icc) were then used for statistical analyses. Imaging variables were stratified according to their median value at baseline (Table 1) to simplify the clinical interpretation as follows: the number of lacunes was stratified into 3 and >3; microbleeds were stratified into present and not present; and BPF was stratified into and < Statistical Analysis Changes of clinical scores (Mini Mental State Examination, Barthel index, mrs, and MDRS) were determined by subtracting baseline scores from scores obtained at 3-year follow-up. To account for missing data, we proceeded as follows: for patients alive at 3 years but who did not return for follow-up because they were too severely demented and dependent, global assessment of disability was performed by informant telephone interviews; for other missing data, we used multiple imputation methods. Multiple imputed datasets were generated using the PROC-MI procedure of SAS with the Markov Chain Monte Carlo method. The imputation model included all variables in the analysis. We chose to generate 5 imputed datasets based Table 1. Main Clinical and MRI Features at Baseline in the 290 Patients Included in the Cohort Study Demographic Characteristics and Vascular Risk Factors Age (y), mean±sd 50.6±11.4 Sex (female/male), n (%) 160 (55.2)/130 (44.8) Hypertension, n (%) 55 (19) Hypercholesterolemia, n (%) 110 (38) Active smoking, n (%) 59 (20.3) Alcohol consumption (>2 drinks/d), n (%) 23 (7.8) Diabetes mellitus, n (%) 6 (2.1) Main clinical manifestations, n (%) Asymptomatic 13 (4.4) Migraine with aura 113 (39) TIA or stroke 190 (65.5) Gait disturbance 87 (30) Balance problems 86 (29.7) Moderate or severe disability (mrs score of 3) 51 (18) Dementia 39 (13.5) Main MRI parameters, mean±sd, range, median No. of lacunes 4.9±6.1, 0 29, 3 nwmh 0.07±0.05, , No. of MB 3.2±13.2, 0 141, 0 Brain parenchymal fraction 0.853±0.006, , MB indicates mircobleeds; MRI, magnetic resonance imaging; mrs, modified Rankin scale; nwmh, total volume of white-matter hyperintensities; and TIA, transient ischemic attack.

3 6 Stroke January 2016 on simulation studies demonstrating little gain in statistical power for higher numbers of imputations. Incident strokes and combined events were counted irrespective of whether the subject had already experienced a stroke. Age and sex were always forced into the multivariable models. Imaging variables were dichotomized according to their median values. In addition, we included parameters with a P value of <0.20 in univariate analyses. Three models were tested: a model that included all demographic and clinical factors (model 1), a model that included only MRI parameters (model 2), and a model that included all demographic and clinical factors, as well as MRI parameters (model 3). The same analyses were performed to identify factors associated with a composite end point of incident stroke, moderate or severe disability, dementia, or death. Predictors of death were analyzed using a Cox proportional hazard model. To examine factors (age, sex, history of hypertension, presence of any previous vascular risk factor, history of stroke, mrs score of 3, presence of gait disturbance, and presence of dementia) and imaging variables potentially associated with a change of clinical scores during follow-up, ANCOVA was performed with adjustment for age and sex. Additional adjustments for education level were made for cognitive scores. Tests were 2-sided, and the significance level was fixed at 5%. All analyses were conducted using SAS (release 9.3; SAS Statistical Institute, Cary, NC). Results Two hundred ninety patients were included in the study. They came from >100 families, 52 families had 2 members represented in the study (range, 2 8), and only 7 of them had >4 members. Two hundred seventy two patients (93.8%) had clinical manifestations of the disease, and 13 were asymptomatic. Forty six patients previously had transient ischemic attacks, 141 patients had at least 1 ischemic stroke (range, 1 8), only 1 individual had an intracerebral hemorrhage, and in 2 cases, the type of previous stroke was undetermined. Twenty-three patients had a positive history of both transient ischemic attacks and ischemic stroke. The other main clinical characteristics are detailed in Table 1. The mean (±SD) time to follow-up was 37 (±1.8) months. Fifty-four patients did not return for the followup visit for different reasons as communicated by the patient, the spouses, close relatives, or treating physician and using all medical records available: death (n=14), severe disability with Deceased before 3 years (n=14) Patients enrolled in the cohort study over 3 years (n=290) Severe disability with dementia (n=23) or acute stroke (n=1) at 3 years prohibiting long travel to the study center dementia (n=23), and an acute stroke at the time of follow-up (n=1; Figure, a full account of the study profile). No follow-up data were obtained in 12 patients. Baseline Predictors of Clinical Events During 3 Years Incident stroke (all ischemic) occurred in 55 of 278 patients (19.8%); 49 (80.9%) of whom had a history of stroke. In multivariable analyses considering demographic and clinical parameters (model 1), the following baseline factors were found to be associated with incident stroke: gait disturbance, a history of stroke, and active smoking. When considering MRI parameters alone (model 2), 2 baseline factors were associated with incident stroke: >3 lacunes and the presence of microbleeds. When all parameters were considered together (model 3), the following baseline factors were associated with incident strokes: >3 lacunes, active smoking, and a BPF of <0.863 (Table 2). Incident dementia developed in 49 of 231 patients (21.2%) who were not demented at baseline. In multivariable analyses, the following baseline factors were found to be associated with incident dementia: model 1, age, an mrs score of 3, active smoking, and male sex; model 2, a BPF of <0.863 and the presence of microbleeds; model 3, an mrs score of 3, a BPF of <0.863, and active smoking. Incident dementia was not found to be associated with incident stroke during followup (8/46 [17.4%] versus 38/181 [20.9%]; P=0.42). Moderate or severe disability (mrs score, 3) developed in 19 of 210 subjects (9%) with no or mild disability at baseline. In multivariable analysis, the following baseline factors were found to be associated with the development of moderate or severe disability: dementia and gait disturbance in model 1; a BPF of <0.863 in model 2 and all 3 factors in model 3 (Table 3). No significant association between incident stroke and the development of moderate or severe disability during follow-up was observed (14/44 [31.8%] versus 37/191 [19.3%]; P=0.07). Death occurred in 14 subjects (4.8%). In multivariable analysis, age was the only baseline parameter associated with Follow-up evaluation performed more than 3 months after planned end date (n=3) - Severe clinical status due to cancer (n=1) - No willingness to attend additional evaluation (n=9) - Loss to follow-up (n=3) Figure. Flowchart. Patients having completed their visit at 3 years (n=236)

4 Chabriat et al Predictors in CADASIL 7 Table 2. Summary of Demographic, Clinical, and MRI Predictors of Major Clinical Events During 3 Years Incident Events During 3 y Model 1 Model 2 Model 3 Clinical, Demographic, and Vascular Risk Factors MRI Markers 1+2 Predictors OR (95% CI) P Value Predictors OR (95% CI) P Value Predictors OR (95% CI) P Value Stroke Gait disturbance 4.6 ( ) No of lacunes, >3 5.2 ( ) < No. of lacunes, > ( ) 0.03 History of stroke 4.1 ( ) Presence of MB 2.5 ( ) < Smoking 2.6 ( ) Active smoking 2.2 ( ) 0.03 BPF* 1.1 ( ) 0.02 mrs score, 3 Dementia 222 ( ) < BPF* 28.3 ( ) Dementia Out of prediction Gait disturbance 30.5 ( ) < BPF* 18.2 ( ) 0.02 Gait disturbance 5.2 ( ) Dementia mrs score, ( ) < BPF* 11.5 ( ) < mrs score, ( ) Smoking 3.3 ( ) Presence of MB 2.4 ( ) 0.04 BPF* 6.7 ( ) Male sex 3.2 ( ) 0.03 Active smoking 3.7 ( ) 0.03 Age 1.1 ( ) 0.01 Death Age 1.1 ( ) None Not done Combined Gait disturbance 16.6 ( ) < No. of lacunes, >3 5.4 ( ) < No. of lacunes, >3 4.9 ( ) < Smoking 2.7 ( ) 0.01 BPF* 3.3 ( ) Gait disturbance 2.8 ( ) 0.02 History of stroke 2.0 ( ) 0.04 Presence of MB 2.2 ( ) 0.03 BPF* 2.4 ( ) 0.02 BPF indicates brain parenchymal fraction; CI, confidence interval; MB, microbleeds; MRI, magnetic resonance imaging; mrs, modified Rankin Scale; and OR, odds ratio. *BPF corresponds to BPF of <0.863, the median value obtained in the sample. Among 39 demented patient at entry, only 8 had an mrs score of <3 at baseline and 5 had an mrs score of 3 at 3 yr (62.5%). OR could not be estimated in the model because of the small number of subjects (in the absence of detailed dating at occurrence of each event, OR values were calculated). death (model 1). No MRI parameter was found to be independently associated with death (model 2), but all patients who died had >3 lacunes and a BPF <0.862, the median value of BPF in the whole sample. The composite end point of incident stroke, incident dementia, moderate or severe disability, or death was reached in 124 (47%) of 265 individuals with complete information for all end points. In multivariable analysis, the following baseline factors were found to be associated with the composite outcome at 3 years: model 1, gait disturbance, active smoking, and a history of stroke; model 2, >3 lacunes, a BPF of <0.863, and the presence of microbleeds; and model 3, >3 lacunes, gait disturbance, and a BPF of < The results did not change when the analysis was restricted to individuals with an mrs of <3 and without dementia at baseline (n=208). Baseline Predictors of Change in Clinical Scores During 3 Years Mean values of the total MDRS, the MDRS initiation/perseveration subscore, the Barthel index, and the mrs score all significantly deteriorated in the study cohort during the 3-year interval (Table I in the online-only Data Supplement). Focusing on demographic and clinical parameters, the following baseline factors were independently associated with a decline of the total MDRS score: age, mrs score of 3, balance problems, gait disturbance, and dementia (Table 3). The same parameters were associated with a decline in the initiation/perseveration subscore of the MDRS. Adjustment for educational level did not change the results. An mrs score of 3, gait disturbance, and dementia were further associated with a decline of the Barthel index. Incident stroke was not found to be associated with cognitive score changes (Table II in the online-only Data Supplement). Focusing on MRI parameters, the following factors were associated with a decline of the total MDRS score: the number of lacunes, presence, and number of microbleeds, as well as BPF at baseline. The same parameters were associated with a worsening of the initiation/perseveration subscore of the MDRS (Table 4). The presence and number of MB, as well as the BPF, were further independently associated with a worsening of the mrs and a decline of the Barthel index. When all potential clinical and MRI predictors were considered together in multivariable analysis, an mrs score of >3, the number of lacunes, the presence of microbleeds, and BPF all were independently associated with worsening of the total MDRS score during follow-up (Table 5). The only significant predictor of worsening in the initiation/perseveration subscore was an mrs score of 3. Sensitivity analyses showed that these results remained essentially unchanged when statistical analysis was restricted to the 236 patients who received their clinical evaluation at 3 years (Tables III, IV, V, VI, and VII in the online-only Data Supplement). Discussion The main findings from this largest ever prospective study on Notch3 mutation carriers were as follows: first, a substantial proportion of adult patients with CADASIL experience major neurological end points or deteriorate clinically within a period of 3 years. Second, clinical status is a major independent predictor of subsequent clinical worsening. Specifically, gait disturbance represents a strong and independent predictor of cognitive decline, and there is a bidirectional

5 8 Stroke January 2016 Table 3. Baseline Demographic and Clinical Predictors of Change in the Main Clinical Scores Between Baseline and Follow-Up (36 Months) ΔMDRS (Total) ΔInitiation/Perseveration MDRS ΔBarthel Index ΔModified Rankin Scale Mean (SE) P Value Mean (SE) P Value Mean (SE) P Value Mean (SE) P Value Sex (%)* Male 6.0 (0.5) (0.32) (0.6) (0.02) 0.25 Female 4.0 (0.6) 2.18 (0.29) 5.77 (0.5) 0.53 (0.02) Age tertile 1st (<47 y) 0.78 (0.4) (0.12) (0.3) (0.03) nd (<56 y) 5.02 (0.4) 2.29 (0.30) 4.76 (0.8) 0.59 (0.02) 3rd 9.01 (0.9) 4.88 (0.60) 9.74 (1.0) 0.83 (0.04) Migraine with aura (%) No 5.4 (0.5) (0.34) (0.6) (0.01) 0.79 Yes 4.3 (0.5) 2.23 (0.24) 3.66 (0.6) 0.64 (0.02) Stroke (%) No 4.4 (0.5) (0.27) (1.0) (0.03) 0.35 Yes 5.3 (0.4) 2.92 (0.30) 4.79 (0.6) 0.56 (0.01) Modified Rankin Scale score, 3 (%) No 2.8 (0.3) < (0.16) (0.5) 0.05 nd nd Yes 14.9 (1.4) 7.62 (1.06) 10.5 (1.1) nd Balance problems (%) No 3.9 (0.3) (0.23) (0.3) (0.02) 0.45 Yes 7.5 (0.9) 3.61 (0.42) 9.02 (1.2) 0.70 (0.02) Gait disturbance (%) No 2.60 (0.3) < (0.15) (0.3) (0.02) 0.12 Yes 10.5 (1.1) 5.33 (0.61) 10.9 (1.1) 0.79 (0.01) Dementia (%) No 2.88 (0.3) < (0.17) (0.3) (0.01) 0.77 Yes 18.3 (1.8) 6.73 (1.06) 21.4 (2.5) 0.56 (0.02) MDRS indicates Mattis Dementia Rating Scale; and nd, not done. Where possible, variables were adjusted for age and sex: *after adjustment for age only and after adjustment for sex only. relationship between dementia and disability in predicting each other s onset. Third, the number of lacunes and brain volume are major independent predictors of clinical decline. Finally, active smoking is a modifiable risk factor for clinical worsening. Almost half of our patients experienced the composite end point of incident stroke, incident dementia, moderate or severe disability, or death within 3 years after study inclusion. The most common event was incident stroke followed by incident dementia and development of moderate or severe disability. These findings refine observations from previous retrospective studies 4 and demonstrate that adult Notch3 mutation carriers are at high risk of clinical worsening. We found a history of stroke to be associated with a 4-fold increased risk of incident stroke. However, there was no significant relationship between a history of stroke and worsening of clinical scores during the 3-year interval. Also, incident strokes were not associated with incident dementia or a decline of cognitive scores although there was a trend for an association between incident stroke and the development of moderate or severe disability during follow-up. Together, these findings indicate that although clinical strokes represent a risk factor for future strokes, they are not the only factor contributing to clinical worsening. Additional studies are warranted to determine whether the accumulation of so-called silent infarcts or the development of cerebral atrophy are the main source of clinical worsening during 3 years. 16,17 Also, the lack of association with age and hypertension further supports that the progression of the disease itself is the main contributor of clinical worsening in such a time frame although investigations are still needed to evaluate the potential effect of various blood pressure levels or fluctuations on the progression of the disease. Among the most consistently found predictors of clinical worsening was gait disturbance. We found the latter to predict a decline of global cognitive function, executive function, and the Barthel score during follow-up. Also, gait disturbance independently predicted the composite end point of incident stroke, severe disability, dementia, or death in a model that included clinical and demographic variables and in the full model. This is in keeping with previous studies in elderly subjects that found gait and balance to be associated with

6 Chabriat et al Predictors in CADASIL 9 Table 4. Baseline MRI Predictors of Change in the Main Clinical Scores Between Baseline and Follow-Up MRI at Baseline ΔMDRS ΔInitiation/Perseveration MDRS ΔBarthel Index ΔModified Rankin Scale β (SE) P Value β (SE) P Value β (SE) P Value β (SE) P Value nwmh (0.02) (0.009) (0.03) (0.001) 0.99 No. of lacunes 0.88 (0.17) < (0.06) < (0.19) (0.01) 0.19 Microbleeds (%) No 4.85 (0.48) (0.20) (0.33) (0.02) 0.01 Yes 11.7 (1.0) 4.22 (0.43) 8.93 (1.14) 0.88 (0.03) Number 0.31 (0.08) (0.05) (0.16) (0.005) 0.01 BPF 1.07 (0.16) < (0.07) < (0.24) (0.01) Variables were adjusted for age and sex. BPF indicates brain parenchymal fraction; MDRS, Mattis Dementia Rating Scale; MRI, magnetic resonance imaging; and nwmh, total volume of white-matter hyperintensities. cognitive decline and dementia. 18,19 Gait disturbance can be easily assessed and may thus prove useful to predict the risk of clinical progression. We further found a modified Rankin Scale score of 3 to predict clinical worsening, in particular cognitive decline. The mrs score of 3 independently predicted a decline in global cognitive function and executive function, and it was associated with incident dementia both in a model that included clinical and demographic variables and in the full model. Interestingly, there was a reciprocal relationship; in that, the presence of dementia independently predicted the progression toward moderate or severe disability in a model that included clinical and demographic variables and in the full model. Together, these observations highlight a bidirectional relationship between disability as assessed by the mrs score and dementia diagnosed by Diagnostic and Statistical Manual of Mental Disorders, 4th Edition. MRI markers also independently predicted new clinical events, cognitive decline, and disability in our patients. Among the strongest and most consistent predictors was the load of lacunes. Patients with >3 lacunes were found to have 5-fold increased risk of both incident stroke and of reaching the composite end point of incident stroke, incident dementia, moderate or severe disability, or death both in the full model and when considering imaging parameters alone. The number of lacunes further predicted cognitive decline as assessed by the total MDRS score and the executive subscore. Brain atrophy emerged as another important predictor. A lower BPF independently predicted incident stroke, incident dementia, the development of moderate or severe disability, and a decline of all clinical scores. These results are in line with previous cross-sectional data that found the number of lacunes and cerebral atrophy to be the 2 main MRI markers associated with clinical severity in CADASIL. 16,20 Our findings extend previous studies by showing that the same imaging parameters also predict future clinical worsening and can thus be considered markers of active disease. Microbleeds whose clinical value remains debated in sporadic SVD 21 showed some predictive capacity in model 2 but not in the full model, suggesting that these lesions are probably less relevant for clinical risk prediction. This study indicates that active smoking more than doubles the risk of incident stroke and more than triples the risk of incident dementia. This effect was strong and remained significant when adjusting for potential clinical and imaging confounders. Our findings agree with previous cross-sectional data in patients with CADASIL, 10 and active smoking has been reported to also increase the risk of stroke in sporadic SVD. 22 Smoking has multiple effects on the microvasculature, including altered nitric oxide signaling of calcium channels, 23 depletion of the free pool of tissue-type plasminogen activator 24 and enhanced free-radical production. 25 In this study, active smoking was also found to increase the risk of incident dementia independently of other potential risk factors, whereas a previous history of stroke did not change this risk. Smoking has also been shown to double the risk of dementia Table 5. Multivariate Linear Regression Analysis Including Demographic, Clinical, and Magnetic Resonance Imaging Predictors of Cognitive Decline Found Significant in Univariate Analysis ΔMDRS (Total) ΔInitiation/Perseveration MDRS β (SE) P Value β (SE) P Value Male sex 0.73 (1.6) (0.6) 0.63 Age 0.15 (0.08) (0.03) 0.12 Modified Rankin Scale score, (3.5) (1.4) 0.03 Balance problems 3.7 (2.2) (0.8) 0.07 Gait disturbance 5.0 (2.7) (1.0) 0.19 No. of lacunes 0.37 (0.17) (0.06) 0.06 MB 0.17 (0.07) (0.05) 0.17 BPF 0.31 (0.16) (0.06) 0.07 BPF indicates brain parenchymal fraction; MB, microbleeds; and MDRS, Mattis Dementia Rating Scale.

7 10 Stroke January 2016 in the general population and to alter white-matter microstructure, 26 although direct effects on the microvasculature were not always detected. 27 Further investigations are needed to determine the effects of smoking on the vasculature and brain tissue in patients with CADASIL. Specific strengths of this study include the prospective design, the large number of most incident events available for analysis, a centralized data management, external monitoring, and the use of validated protocols for image analysis following clinical trial standards. Limitations include a relatively long period of recruitment, lack of analysis of potential effects of drugs in use, and a substantial number of patients lost to follow-up although this was accounted for in sensitivity analyses. The bidirectional relationship between disability and dementia in predicting each other onset may in part result from aspects of disability contributing to the diagnosis of dementia and vice versa. However, they do represent different aspects of disease. Also, results on lowfrequency events such as death need to be interpreted with caution. In conclusion, this study demonstrates that the combination of clinical, demographic, and MRI markers may aid in predicting the risk of clinical worsening in CADASIL and that this risk is in part modifiable. These results may have implications for the counseling and management of subjects with CADASIL and for the planning of future therapeutic trials. Acknowledgments We acknowledge all patients who participated in this research, their families, and the association CADASIL France for their active collaboration, Jocelyne Ruffié and Solange Hello for their involvement in the practical organization of the study, M.G. Bousser and A. Kurtz for their advice and constant support during the study, and the Unité de Recherche Clinique of Saint-Louis/Lariboisiere hospital, Paris France (V. Jouis and L. Guery) for their technical support. Sources of Funding This study was supported by grants from the French Ministry of Health (Regional and National PHRC AOR ), Association de Recherche en NEurologie Vasculaire, the Vascular Dementia Research Foundation, and the Fondation Leducq (Transatlantic Network of Excellence on the Pathogenesis of Small Vessel Disease of the Brain; None. Disclosures References 1. Joutel A, Corpechot C, Ducros A, Vahedi K, Chabriat H, Mouton P, et al. Notch3 mutations in CADASIL, a hereditary adult-onset condition causing stroke and dementia. Nature. 1996;383: doi: /383707a0. 2. Moreton FC, Razvi SS, Davidson R, Muir KW. Changing clinical patterns and increasing prevalence in CADASIL. Acta Neurol Scand. 2014;130: doi: /ane Chabriat H, Joutel A, Dichgans M, Tournier-Lasserve E, Bousser MG. Cadasil. Lancet Neurol. 2009;8: doi: / S (09) 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(Pt 11): doi: /brain/ awh Dichgans M, Markus HS, Salloway S, Verkkoniemi A, Moline M, Wang Q, et al. Donepezil in patients with subcortical vascular cognitive impairment: a randomised double-blind trial in CADASIL. Lancet Neurol. 2008;7: doi: /S (08) Liem MK, Lesnik Oberstein SA, Haan J, van der Neut IL, Ferrari MD, van Buchem MA, et al. MRI correlates of cognitive decline in CADASIL: a 7-year follow-up study. Neurology. 2009;72: doi: /01. wnl Holtmannspötter M, Peters N, Opherk C, Martin D, Herzog J, Brückmann H, et al. Diffusion magnetic resonance histograms as a surrogate marker and predictor of disease progression in CADASIL: a two-year follow-up study. Stroke. 2005;36: doi: /01. STR a4. 8. Peters N, Herzog J, Opherk C, Dichgans M. A two-year clinical followup study in 80 CADASIL subjects: progression patterns and implications for clinical trials. Stroke. 2004;35: doi: /01. STR f1. 9. Molko N, Pappata S, Mangin JF, Poupon F, LeBihan D, Bousser MG, et al. Monitoring disease progression in CADASIL with diffusion magnetic resonance imaging: a study with whole brain histogram analysis. Stroke. 2002;33: Adib-Samii P, Brice G, Martin RJ, Markus HS. Clinical spectrum of CADASIL and the effect of cardiovascular risk factors on phenotype: study in 200 consecutively recruited individuals. Stroke. 2010;41: doi: /STROKEAHA Viswanathan A, Guichard JP, Gschwendtner A, Buffon F, Cumurcuic R, Boutron C, et al. Blood pressure and haemoglobin A1c are associated with microhaemorrhage in CADASIL: a two-centre cohort study. Brain. 2006;129(Pt 9): doi: /brain/awl Mok VC, Wong A, Lam WW, Fan YH, Tang WK, Kwok T, et al. Cognitive impairment and functional outcome after stroke associated with small vessel disease. J Neurol Neurosurg Psychiatry. 2004;75: ICH. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use ICH Harmonised Tripartite guideline. Guideline for Good Clinical Practice. E6 (r1) Wardlaw JM, Smith EE, Biessels GJ, Cordonnier C, Fazekas F, Frayne R, et al; STandards for ReportIng Vascular changes on neuroimaging (STRIVE v1). Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol. 2013;12: doi: /S (13) O Sullivan M, Jouvent E, Saemann PG, Mangin JF, Viswanathan A, Gschwendtner A, et al. Measurement of brain atrophy in subcortical vascular disease: a comparison of different approaches and the impact of ischaemic lesions. Neuroimage. 2008;43: doi: / j.neuroimage Viswanathan A, Godin O, Jouvent E, O Sullivan M, Gschwendtner A, Peters N, et al. Impact of MRI markers in subcortical vascular dementia: a multi-modal analysis in CADASIL. Neurobiol Aging. 2010;31: doi: /j.neurobiolaging Jouvent E, Viswanathan A, Mangin JF, O Sullivan M, Guichard JP, Gschwendtner A, et al. Brain atrophy is related to lacunar lesions and tissue microstructural changes in CADASIL. Stroke. 2007;38: doi: /STROKEAHA Verghese J, Wang C, Lipton RB, Holtzer R, Xue X. Quantitative gait dysfunction and risk of cognitive decline and dementia. J Neurol Neurosurg Psychiatry. 2007;78: doi: /jnnp Marquis S, Moore MM, Howieson DB, Sexton G, Payami H, Kaye JA, et al. Independent predictors of cognitive decline in healthy elderly persons. Arch Neurol. 2002;59: Viswanathan A, Gschwendtner A, Guichard JP, Buffon F, Cumurciuc R, O Sullivan M, et al. Lacunar lesions are independently associated with disability and cognitive impairment in CADASIL. Neurology. 2007;69: doi: /01.wnl Martinez-Ramirez S, Greenberg SM, Viswanathan A. Cerebral microbleeds: overview and implications in cognitive impairment. Alzheimers Res Ther. 2014;6:33. doi: /alzrt Bezerra DC, Sharrett AR, Matsushita K, Gottesman RF, Shibata D, Mosley TH Jr, et al. Risk factors for lacune subtypes in the Atherosclerosis Risk in Communities (ARIC) Study. Neurology. 2012;78: doi: /WNL.0b013e31823efc Gerzanich V, Zhang F, West GA, Simard JM. Chronic nicotine alters NO signaling of Ca(2+) channels in cerebral arterioles. Circ Res. 2001;88:

8 Chabriat et al Predictors in CADASIL Wang L, Kittaka M, Sun N, Schreiber SS, Zlokovic BV. Chronic nicotine treatment enhances focal ischemic brain injury and depletes free pool of brain microvascular tissue plasminogen activator in rats. J Cereb Blood Flow Metab. 1997;17: doi: / Hanna ST. Nicotine effect on cardiovascular system and ion channels. J Cardiovasc Pharmacol. 2006;47: doi: /01. fjc e. 26. Gons RA, van Norden AG, de Laat KF, van Oudheusden LJ, van Uden IW, Zwiers MP, et al. Cigarette smoking is associated with reduced microstructural integrity of cerebral white matter. Brain. 2011;134 (Pt 7): doi: /brain/awr Court JA, Johnson M, Religa D, Keverne J, Kalaria R, Jaros E, et al. Attenuation of Abeta deposition in the entorhinal cortex of normal elderly individuals associated with tobacco smoking. Neuropathol Appl Neurobiol. 2005;31: doi: /j x.