Abstract
Objective: Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is a hereditary arteriopathy caused by mutations of the Notch3 gene. The disease is characterized by T2-hyperintense lesions (subcortical white matter lesions), T1-hypointense lesions (lacunar lesions), and T2*-weighted gradient-echo (GE) lesions (cerebral microhemorrhages [CMs]) visualized on clinical MRI sequences and is considered as a model of “pure” subcortical ischemic vascular dementia. Although numerous studies have investigated the impact of white matter lesions in patients with CADASIL, the clinical importance of lacunar lesions remains unknown. Our purpose was to examine the influence of the visible MRI markers in the disease, including the load of lacunar lesions on cognitive impairment and disability in CADASIL.
Methods: We collected clinical data from 147 consecutive patients enrolled in an ongoing two-center prospective cohort study. Degree of disability was assessed by modified Rankin scale and Barthel index. Degree of cognitive impairment was assessed by Mattis Dementia Rating Scale and Mini-Mental Status Examination. T1-weighted, fluid-attenuated inversion recovery, and GE images were obtained on a 1.5-T MRI. Volume and location of lacunar lesions, white matter hyperintensities (WMHs), and CMs were assessed.
Results: There was a significant independent association between age, volume of lacunar lesions, and global cognitive function scales when analyzed in a multivariable model. In contrast, WMHs and CMs had no independent influence on cognitive function. Disability in this cohort was associated with volume of lacunar lesions, CMs, systolic blood pressure, and age but not with WMHs.
Conclusions: Among the lesions observed on conventional MRI in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), the overall lacunar lesion burden seems to have the most important impact on cognitive function and disability. These findings suggest that preventive strategies to decrease the risk of lacunar lesions as observed on MRI may reduce disease-related impairment in CADASIL. These results suggest that lacunar lesions may also play a key role in disability and cognitive impairment in more common forms of small-vessel disease.
There is accumulating evidence to suggest that different cerebral MRI lesions related to small-vessel diseases may have an important role in cognitive impairment and dementia.1–3 The extent of white matter hyperintensities (WMHs) on T2-weighted images has been linked both to cognitive impairment and dementia in cross-sectional population-based studies.4–7 Small, clinically silent brain infarctions identified on baseline MRI scan appear to be a strong risk factor for subsequent dementia in elderly individuals.2 In addition, cerebral microhemorrhages (CMs) have been associated with executive dysfunction in patients presenting with stroke or TIA.8 However, the relative contribution of each type of lesion to an individual's cognitive status and neurologic impairment is not known.
Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is a nonatherosclerotic, nonamyloid arteriopathy caused by mutations of the Notch 3 gene9 and is considered a model of “pure vascular dementia.” The main clinical manifestations of the disease include attacks of migraine with aura, mood disturbances, recurrent ischemic strokes, and progressive cognitive decline.10,11
All of the established visible MRI markers of small-vessel disease can be seen in CADASIL patients. There is evidence of widespread WMHs on T2-weighted images,12,13 lacunar lesions on T1-weighted images,9,12,14 and CMs on T2*-weighted or gradient-echo (GE) images.15,16 These MRI lesions represent different consequences of the underlying angiopathy in the disease.
In the present study, we sought to determine the relationships of these various clinical MRI markers and clinical variables on global cognitive function and disability in patients enrolled in a two-center cohort study.
METHODS
Subjects.
Subjects were drawn from an ongoing two-center prospective cohort study of patients with CADASIL. Subjects were recruited among consecutive CADASIL patients aged at least 18 years, evaluated at Lariboisière (Paris, France) or Ludwig-Maximilians-Universität (Munich, Germany) hospitals between October 2003 and July 2005. In all cases, diagnosis was confirmed by identification of a typical mutation in the Notch 3 gene.17–19 Patients who were pregnant or had other contraindications to MRI were excluded (two patients). Clinical and demographic data were collected by study investigators at the time of inclusion and included age, sex, history of hypertension (defined as previous diagnosis of hypertension [>140/90]) or use of antihypertensive treatment for control of blood pressure,20 dementia (defined by Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition [DSM-IV] criteria), diabetes (defined according to the World Health Organization criteria21 or as current use of hypoglycemic agent22), hypercholesteremia (defined by previous diagnosis), current medication use, and smoking history. Blood pressure measurement and laboratory evaluation (which included complete blood count, glucose, hemoglobin A1c, homocysteine, and cholesterol panel) were performed in all patients. All enrolled subjects underwent detailed baseline neurologic examination during the 2 hours preceding MRI examination, including evaluation of cognitive deficits with the Mini-Mental State Examination (MMSE)23,24 and Mattis Dementia Rating Scale (MDRS).25,26 Nine patients who did not undergo cognitive testing were excluded from the analysis. The degree of disability was based on the modified Rankin scale (mRS) and Barthel index. As in previous studies, we defined poor outcome or functional dependence as mRS score ≥ 3 or Barthel index < 95.16,27,28
Informed consent was obtained from each subject or from a close relative if the subject was too severely disabled to give written consent. This study was approved by an independent ethics committee in both participating centers.
MRI.
MRI scans were obtained by the use of a 1.5-T system (Vision, Siemens, Munich, Germany, or Signa General Electric Medical Systems, Paris, France). Three-dimensional T1-weighted sequences (Munich: repetition time [TR]/echo time [TE] 11.4/4.4 msec, slice thickness 1.2 mm, no interslice gap, 197 × 256; Paris: TR/TE 9.1/2 msec, slice thickness 0.8 mm, no interslice gap, 256 × 256), fluid-attenuated inversion recovery (FLAIR; Munich: TR/TE/inversion time [TI] 4,284/110/1,428 msec, slice thickness 5 mm, no interslice gap, 176 × 256; Paris: TR/TE/TI 8,402/161/2,100 msec, slice thickness 5.5 mm, no interslice gap, 160 × 256), and T2*-weighted GE planar imaging (Munich: TR/TE 1,056/22 msec, slice thickness 5 mm, no interslice gap, 192 × 256; Paris: TR/TE 500/15 msec, slice thickness 5.5 mm, no interslice gap, 192 × 256) were performed. MRIs from both centers were processed and analyzed together (Theralys Inc., Lyon, France).
Image processing.
All MRI sequences were spatially registered by realigning two MRI volumes. For each patient, the T1-weighted images were used as reference images. The registration algorithm used mutual information (MI) to assess the quality of match between the two volumes. A rigid three-dimensional transformation was used. According to MI and derivative values, a gradient descent technique was used to modify the transformation parameters (rotations and translations) iteratively to maximize the MI. For this purpose, a multiresolution, coarse-to-fine strategy was adopted with onset at low resolution and iterations on the basis of the results of the previous step. This strategy was chosen to increase the robustness of the registration algorithm by preventing the maximization technique from being trapped in local maximal.29
Image analysis.
Image analysis was carried out in two steps. Native and preprocessed data were displayed on identical Unix workstations running image evaluation software developed by Theralys Inc.29 An automatic computed image processing and preparation phase preceded centralized image review sessions involving study neurologists who were blinded to the clinical data, during which manual corrections and validation were carried out. Using drawing tools, a trained neurologist or neuroradiologist then validated each discrete lesion (WMHs, lacunar lesions) and corrected its delimitation if necessary.
WMH quantification.
White matter lesions were analyzed on FLAIR images. All FLAIR axial slices from the base of the cerebellum to the vertex were analyzed. A mask of lesions was generated from FLAIR images after application of the intracranial cavity (ICC) mask by applying a threshold on signal intensity derived from the signal intensities histogram. Two trained neurologists using drawing tools then validated each discrete lesion and corrected its delimitation if necessary. The total volume of WMHs was normalized to the ICC in each patient (normalized volume of WMHs or nWMHs = (volume of WMHs/volume ICC) * 100). We have previously demonstrated a high interrater reliability between the two readers (intraclass correlation coefficient = 0.998).16
Lacunar volume.
To assess the number of lacunar lesions and lacunar lesion volume, T1-weighted images were segmented in four tissue classes consisting of white matter, cortical and deep gray matter, CSF intensity lesions, and nonbrain partitions. After segmentation, the class “CSF intensity lesions” was automatically isolated from other tissue classes. Subsequently, two raters (F.B., R.C.) isolated lacunes from CSF voxels using appropriate two- and three-dimensional imaging tools. All of the hypointense lesions with both a signal identical to that of CSF and a diameter > 2 mm (this diameter criterion was used to exclude most of Virchow–Robin spaces30,31) were selected. Large-vessel infarctions were excluded.
The total volume of lacunes in each patient was normalized to the ICC (normalized lacunar volume or nLV = (volume of lacunes/volume ICC) * 1,000). Good interrater reliability has been demonstrated for both the volume and number of lacunes (intraclass correlation coefficient = 0.830 and 0.824).16
Microhemorrhages.
Microhemorrhages were defined as rounded foci ≤ 5 mm in diameter hypointense on GE sequences and distinct from vascular flow voids, leptomeningeal hemosiderosis, or nonhemorrhagic subcortical mineralization. The location and number of microhemorrhages were recorded. Twenty pairs of images were analyzed separately by two raters (A.V., M.D.). Raters had access to T1- and T2-weighted spin-echo sequences for comparison and validation. It was agreed before the study that only lesions considered to be independent hemorrhagic foci would be counted. Each lesion was marked electronically by the reader on the screen, and all lesions were recorded in the corresponding three-dimensional space for each examination. All disagreements were resolved by consensus. High interrater reliability for the number of CMs has been previously shown (intraclass correlation coefficient = 0.964).16
Statistical methods.
Continuous variables that were normally distributed were compared using Pearson correlations. Dichotomous clinical variables were compared using the χ2 or Fisher exact test and to continuous variables using Student's t test considering equal or unequal variances. The Wilcoxon rank sum test was used to compare dichotomous variables to continuous variables that were not normally distributed. The number of microhemorrhages was analyzed in three classes (0, 1 to 3, or ≥4) to divide the subjects into tertiles. Because of their nonnormal distributions and concern for nonlinear effects, both nWMHs and nLVs were analyzed in three classes to divide the subjects into tertiles. All p values were two-tailed, and the criterion for significance was p < 0.05.
Multivariable logistic regression models were used to find predictors of disability (mRS, Barthel indices) and cognitive impairment (Mattis global rating scale, MMSE); candidate covariates were those associated with clinical scales in univariate analysis (p < 0.10). To control for any center effects, an analysis of the interaction between the two centers and all predictor variables was performed. The final model variables were selected by the stepwise regression analysis (automated by JMP statistical software, SAS Institute, Cary, NC) with entry and removal values set to 0.05.
RESULTS
Baseline characteristics, according to the presence or absence of dementia, are presented in table 1. Compared with individuals without dementia, those subjects who met DSM-IV criteria for dementia were older, had more CMs, and had greater WMHs and lacunar lesion volumes in χ2 comparisons or using the Wilcoxon rank sum test (table 1 and figure). Lacunar lesions most commonly occurred in the basal ganglia, thalamus, brainstem, and frontal and parietal white matter regions. The mean number of lacunar lesions detected was 13.6 (95% CI 11.1 to 16.1). The mean MDRS and MMSE scores were 91.4 and 17.5, respectively, in the dementia group. All patients with dementia were classified in the most severe category of MDRS. The degree of agreement between these two scales in categorizing a patient with dementia was good (κ = 0.64).
Table 1 Baseline characteristics of subjects in cohort by DSM-IV–based diagnosis of dementia
Figure Box-and-whisker plots of visible MRI lesions in patients with CADASIL with and without dementia
Box borders represent the 25th and 75th percentiles, the middle bar is at the median, and the whiskers extend to 1.5 interquartile ranges. Statistical testing is by Wilcoxon rank sum test. * p = 0.01; ** p = 0.001; *** p = 0.0002. (A) Normalized white matter hyperintensity volume (nWMH). (B) Normalized lacunar lesion volume (nLV). (C) Number of cerebral microhemorrhages (CMs). CADASIL = cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy.
Univariate analysis of clinical and demographic variables demonstrated a significant correlation between age and MDRS rating scale (table 2). Similar results were obtained using the MMSE (data not shown). A positive association was also noted between MDRS and MMSE and history of hypercholesteremia (χ2; p = 0.04). In contrast, no significant association was found between systolic blood pressure (SBP), diastolic blood pressure (DBP), history of hypertension, and scales of global cognitive function.
Table 2 Analysis of factors associated with cognitive impairment as measured by MDRS in CADASIL cohort
All MRI visible markers of the disease significantly correlated with measures of global cognitive function as assessed by the MDRS or the MMSE in univariate analyses (table 2, only results obtained with MDRS are shown). nLVs and nWMHs showed a significant correlation with both MDRS and MMSE scores, whereas the number of CMs correlated only with the MDRS score.
Stepwise multivariable logistic regression using significant MRI and clinical variables was performed to determine independent predictors of cognitive impairment in CADASIL patients. Increased nLV (grade 2 vs grade 0 or 1; p = 0.0004) and older age (p < 0.0001) were independently associated with worse MDRS scores. Of note, the effect of nWMHs and number of CMs was not significant on MDRS scores. Similarly, multivariable modeling showed that increased nLV (grade 2 vs grade 0 or 1; p = 0.006) and older age (p < 0.0001) were independently associated with lower MMSE scores. Again, nWMHs and number of CMs were not found to be significant.
The effect of MRI markers and clinical variables on disability was also assessed. Univariate analysis revealed that all MRI markers, SBP, and age correlated with poor functional outcome as measured by the mRS (table 3). History of hypertension and past or current smoking were significant in addition to the above-described variables when the Barthel index was used as a measure of disability (data not shown).
Table 3 Analysis of factors of poor outcome in CADASIL patients
Stepwise multivariable logistic regression was performed using MRI markers and significant clinical variables to determine independent predictors of functional disability in CADASIL. The final model variables were selected by stepwise analysis. In the final maximally adjusted model, increased nLV (grade 2 vs grade 0 or 1; p = 0.0001), number of CMs (p = 0.03), SBP (p = 0.005), and age (p = 0.0007) were found to be independently associated with poor functional outcome (mRS score ≥ 3). A similar analysis using the Barthel index demonstrated that increased nLV (grade 2 vs grade 0 or 1; p = 0.0001), presence of CMs (p = 0.0004), SBP (p = 0.01), and age (p = 0.007) were independently associated with poor outcome.
In all of above-described analyses, we found no significant center effect that modified the interpretation of our results.
DISCUSSION
The major findings from this large two-center cohort study of CADASIL patients are that among the visible MRI markers in the disease, lacunar lesions and age are independent predictors of disability and cognitive impairment. In addition, we found that SBP and CMs are independently associated with disability in our patients.
The degree and spectrum of cognitive impairment in CADASIL has been widely described in numerous studies.32–35 Of all the candidate variables examined in our study, nLV, age, CMs, nWMHs, and history of hypercholesterolemia showed the strongest univariate association with global cognitive scales (table 2). However, only nLV and age remained significant in multivariable modeling. These findings are in line with previous reports showing that lacunar lesions are associated with cognitive impairment in various population-based studies.2,36 Importantly, the association of the other clinical MRI markers in the disease (namely, WMHs and CMs) did not remain significant in the final adjusted model. These results may suggest that among the most easily detectable MRI lesions used in clinical practice, lacunar lesions, but not WMHs, have the strongest clinical impact on cognitive status in CADASIL. These data are consistent with recent findings demonstrating the importance of lacunar lesion volumes but not WMH volumes in elderly patients with subcortical vascular disease and cognitive impairment.37
Disability has been described in CADASIL38 and has been associated with various clinical variables and MRI markers in different studies.22,39,40 Although there were some differences between disability measures and clinical variables in univariate analyses, these differences did not remain in the final multivariate model. Our results suggest that when all these variables are taken into account in a large cohort of patients, nLV and CMs in association with blood pressure and age are independently related to poor functional outcome, as measured by mRS or Barthel index. These results suggest that MRI markers distinct from WMHs (namely, nLV and CMs) may be more reflective of the degree of disability in patients with CADASIL. The possible impact of SBP on clinical disability is in line with previous results obtained in longitudinal studies.41,42 This independent effect of SBP on disability may be mediated through its influence on mean diffusivity and brain atrophy in CADASIL patients.39,42 It is possible that we were unable to detect a similar association with hypertension because of the small numbers of individuals in our cohort with this diagnosis. Alternatively, these data may suggest that in the setting of an existing cerebral microangiopathy, small elevations in blood pressure, even in the absence of hypertension per se, play the most important role in disease-related disability. Our results raise the possibility that antihypertensive therapy may be of benefit to reduce disease-associated disability in CADASIL patients.
All of the current MRI markers of the disease were highly correlated to cognitive status and functional outcome and were also highly correlated to each other (data not shown). Therefore, we cannot exclude some bias caused by these intercorrelations in the multivariable regression models. However, despite these possible biases, the strengths of association with the various MRI markers seen in this study are largely consistent with what has been previously reported in CADASIL.15,16,39,43
The present results emphasize that the degree but not the extent of tissue lesions (as visualized by conventional MRI sequences) may play the key role in the development of disability in CADASIL. These results are in line with the data obtained using diffusion tensor imaging (DTI) showing that the degree of white matter damage but not the extent of WMHs correlated strongly with cognitive function scales.44 Whether the influence of white matter damage as assessed by DTI outside the lacunar lesions has an independent impact cognitive function (apart from other MRI markers) will require further studies. Additionally, studies that measure degree of brain atrophy may help to further refine the precise relationships between these MRI markers.42
Our results show that among the lesions observed on conventional MRI, the load of lacunar lesions and the number of CMs are most strongly associated with cognitive decline and disability in CADASIL. In contrast, the extent of WMHs distinct from these lesions and assessed by FLAIR may have only limited clinical consequences. More generally, these results suggest that studies of MRI correlates of disability and cognitive impairment in sporadic cerebral small-vessel diseases should consider these two key markers in association with WMHs in the future. These relationships remain to be definitively confirmed in future prospective studies. Finally, although the MRI markers were precisely quantified in the present analyses, the cerebral localization of lacunar lesions and CMs was not performed. Because the localization of such lesions may have a strong influence on cognitive status in vascular dementia,45 future studies analyzing both lesion burden with anatomic location may serve to further account for the differences seen in the cognitive function of CADASIL patients.
ACKNOWLEDGMENT
The authors thank Prof. Eric Vicaut and Ms. Carole Boutron for their invaluable assistance in the organization and management of the cohort database.
Footnotes
-
Editorial, see page 131
Supported by PHRC grant AOR 02-001 (DRC/APHP) and performed with the help of ARNEVA (Association de Recherche en Neurologie Vasculaire), Hopital Lariboisière, France; the Deutsche Forschungsgemeinschaft (SFB596/TPA4); and a grant from EISAI Medical Res. Inc. (Germany). M.O'S. is an Alexander von Humboldt Fellow and is also supported by the European Neurological Society and the Peel Medical Research Trust.
Disclosure: The authors report no conflicts of interest.
Received October 12, 2006. Accepted in final form March 1, 2007.
REFERENCES
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan