This Article Free upon publication Abstract Full Text (PDF) Correction (v38,p1118) All Versions of this Article: 37/9/2220    most recent 01.STR.0000237236.88823.47v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Hachinski, V. Articles by Leblanc, G. G. Search for Related Content PubMed PubMed Citation Articles by Hachinski, V. Articles by Leblanc, G. G. Related Collections Cerebrovascular disease/stroke CT and MRI Animal models of human disease Behavioral Changes and Stroke Cerebral Lacunes Genetics of Stroke Pathology of Stroke Related Article

(Stroke. 2006;37:2220.)
© 2006 American Heart Association, Inc.

Original Contributions

National Institute of Neurological Disorders and Stroke–Canadian Stroke Network Vascular Cognitive Impairment Harmonization Standards

Vladimir Hachinski, MD, DSc; Costantino Iadecola, MD; Ron C. Petersen, MD, PhD; Monique M. Breteler, MD, PhD; David L. Nyenhuis, PhD; Sandra E. Black, MD; William J. Powers, MD; Charles DeCarli, MD; Jose G. Merino, MD; Raj N. Kalaria, PhD, FRCP; Harry V. Vinters, MD; David M. Holtzman, MD; Gary A. Rosenberg, MD; Anders Wallin; Martin Dichgans, MD; John R. Marler, MD Gabrielle G. Leblanc, PhD

From the London Health Sciences Centre (V.H.), University Campus, London, Ontario, Canada; the Department of Neurology and Neuroscience (C.I.), Division of Neurobiology, Weill Cornell Medical College, New York, NY; the Department of Neurology (R.C.P.), Mayo Clinic, Rochester, MN; the Department of Epidemiology and Biostatistics (M.M.B.), Erasmus Medical Center, Rotterdam, The Netherlands; the Department of Neurology and Rehabilitation (D.L.N.), University of Illinois at Chicago, Chicago, IL; the Division of Neurology Sunnybrook and Women’s College (S.E.B.), Health Sciences Centre, Toronto, Ontario, Canada; the Department of Neurology (W.J.P., D.M.H.), Washington University, School of Medicine, St. Louis, MO; the Department of Neurology (C.D.), University of California at Davis, Sacramento, CA; the Section on Stroke Diagnostics and Therapeutics (J.G.M.), National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD; the Wolfson Centre (Neuropathology), Institute for Ageing and Health (R.N.K.), Newcastle General Hospital, Newcastle upon Tyne, UK; the Department of Pathology and Laboratory Medicine (H.V.V.), Section of Neuropathology, University of California, Los Angeles, UCLA Medical Center, Los Angeles, CA; the Department of Neurology (G.A.R.), University of New Mexico, Albuquerque, NM; the Neurologische Klinik (M.D.), Ludwig-Maximilians-Universität München, Klinikum Grosshadern, München, Germany; and the National Institute of Neurological Disorders and Stroke (J.R.M., G.G.L.), National Institutes of Health, Bethesda, MD.

Correspondence to Gabrielle G. Leblanc, PhD, National Institute of Neurological Disorders and Stroke, National Institutes of Health, 6001 Executive Boulevard, Room 2136, Bethesda, MD 20892. E-mail leblancg{at}ninds.nih.gov

	Abstract

Top Abstract Introduction Clinical/Epidemiology Section Neuropsychology Section Neuroimaging Neuropathology Experimental Models Biomarkers Genetics Clinical Trials Conclusions Appendix References

 
Background and Purpose— One in 3 individuals will experience a stroke, dementia or both. Moreover, twice as many individuals will have cognitive impairment short of dementia as either stroke or dementia. The commonly used stroke scales do not measure cognition, while dementia criteria focus on the late stages of cognitive impairment, and are heavily biased toward the diagnosis of Alzheimer disease. No commonly agreed standards exist for identifying and describing individuals with cognitive impairment, particularly in the early stages, and especially with cognitive impairment related to vascular factors, or vascular cognitive impairment.

Methods— The National Institute for Neurological Disorders and Stroke (NINDS) and the Canadian Stroke Network (CSN) convened researchers in clinical diagnosis, epidemiology, neuropsychology, brain imaging, neuropathology, experimental models, biomarkers, genetics, and clinical trials to recommend minimum, common, clinical and research standards for the description and study of vascular cognitive impairment.

Results— The results of these discussions are reported herein.

Conclusions— The development of common standards represents a first step in a process of use, validation and refinement. Using the same standards will help identify individuals in the early stages of cognitive impairment, will make studies comparable, and by integrating knowledge, will accelerate the pace of progress.

Key Words: Binswangers disease • CADASIL • syndrome • cerebral infarction • cerebrovascular disorders • dementia • genetics • ischemia • lacunar infarction • leukoaraiosis • neuropsychology • stroke • vascular dementia

	Introduction

Top Abstract Introduction Clinical/Epidemiology Section Neuropsychology Section Neuroimaging Neuropathology Experimental Models Biomarkers Genetics Clinical Trials Conclusions Appendix References

 
One in 3 of us will experience a stroke, dementia or both, unless we do something about it.¹ Up to 64% of persons who have experienced a stroke have some degree of cognitive impairment,² with up to a third developing frank dementia.^3–5 Conversely, postmortem pathological studies^6–11 indicate that up to 34% of dementia cases show significant vascular pathology. Moreover, the same risk factors that make individuals prone to cerebrovascular disease also put them at risk for cognitive impairment.^12,13

Cognitive impairment that is caused by or associated with vascular factors has been termed "vascular cognitive impairment" (VCI).^14–16 VCI can occur either alone or in association with Alzheimer disease (AD). Indeed, there appears to be a strong interaction between cerebrovascular and AD pathol-ogies, such that individuals having both frequently show greater cognitive impairment than those having either pathology alone.^17–19

Because vascular risk factors are treatable, it should be possible to prevent, postpone, or mitigate VCI, as well as the vascular exacerbation of AD. However, progress in VCI research has been hindered by lack of satisfactory diagnostic criteria for the condition. None of the current stroke scales measure cognition to any extent except for the Toronto Stroke Scale, which has never been used in clinical trials. Moreover, past diagnostic criteria for cerebrovascular disease–associated cognitive impairment have focused on the most extreme form of the condition, dementia. However, only about half the population of patients with VCI exhibit dementia,^20,21 and those who do not are better candidates for clinical trials because they are at earlier stages of their illness. Another problem with the focus on dementia is that the widely accepted definition of dementia requires memory impairment as an essential feature. This definition works well for identifying patients with AD, but often misses the executive dysfunction typical of cognitive disorders with vascular bases.²²

As a first step toward developing diagnostic criteria for VCI, a workshop was convened by the National Institute for Neurological Disorders and Stroke (NINDS) and Canadian Stroke Network (CSN). It was recognized at the outset that our knowledge of VCI is currently still insufficient to develop a definitive checklist of diagnostic criteria for the condition. Rather, the goal of the workshop was to define a set of data elements to be collected in future studies aimed at more fully defining VCI, understanding its etiology, and identifying targets for treatment. Specifically, the participants were charged to do the following:

1. Develop screening questions that could be used to identify subjects with possible cognitive and behavioral impairment.

2. Establish a minimum dataset that would be useful in common clinical practice or large-scale research studies of VCI (eg, epidemiological studies, genetic studies or clinical trials), so that different studies could pool data for comparison and cross-validation.

3. Develop an "ideal" dataset for studies focused on particular research issues.

In addition, the participants were encouraged to suggest methods or scales for quantifying specific data elements, to identify which measures need to be validated, and to point out areas of promising research.

What follows represents a compilation of the recommendations of working groups in each of the following areas: Clinical/Epidemiology, Neuropsychology, Imaging, Neuropathology, Experimental Models, Biomarkers, Genetics, and Clinical Trials.

	Clinical/Epidemiology Section

Top Abstract Introduction Clinical/Epidemiology Section Neuropsychology Section Neuroimaging Neuropathology Experimental Models Biomarkers Genetics Clinical Trials Conclusions Appendix References

 
The Clinical and Epidemiology Group approached the issue of VCI from 2 perspectives. Initially, the working group outlined the items that should be considered in a population-based study of VCI. In particular, emphasis was placed on the variables that should be gathered in an ideal study. The second approach undertaken by the group was to develop an abbreviated, more clinically oriented evaluation. This set of recommendations includes a subset of variables from the previous exercise and is meant to be practically oriented for the clinician.

Population-Based, Epidemiological Approach
1. Demographics
The group felt that the minimum set of basic data to be gathered in a VCI study should include the following variables: sex; birthdate; race/ethnicity, including birth place of origin and parents’ places of origin; years in the current country of residence; primary language; number of years of education; occupation; literacy; living situation and level of independence, including type of residence; marital status; handedness; and a contact person.

2. Informant (if available and determined relevant)
In taking information concerning the subject of interest, an informant should be sought if possible. The source of the information should be clarified as well as the amount and type of contact the person has with the subject. Additional data to be obtained should include the informant’s birthdate, gender, ethnicity/race, relationship and length of time of relationship with the subject, education, and living status with respect to the subject.

3. Family History
A history of diseases in first degree relatives should be obtained. This history should include at a minimum information about past strokes, vascular disease including myocardial infarction, history of dementia, and other neurological diseases. For all of these items the age at death should be recorded, and the age of the event (eg, stroke) should be recorded.

4. Health History
Each subject should have a thorough health history obtained including information on cardiovascular disease such as myocardial infarction, arrhythmia/atrial fibrillation, angioplasty, stent, coronary artery bypass graft or valvular surgery, pacemaker, congestive heart failure, angina, and peripheral artery disease. For each of these, the question should be phrased to include, "Have you been diagnosed with _________?" "Do you have symptoms of __________?" Care should be taken to obtain this information on all of these medical conditions.

In addition, a history should be obtained on cerebrovascular disease including stroke (hemorrhagic or ischemic), transient ischemic attack, and endarterectomy. One should also inquire about other surgeries and, in particular, whether or not any cognitive difficulties arose after surgery. Finally, a list of all medications including over-the-counter preparations should be compiled.

Other factors to be included in the history include migraine, hypertension, hyperlipidemia, diabetes mellitus, sleep disorders, sickle-cell anemia, hypercoagulable states or related conditions such as deep-vein thrombosis, pulmonary embolus or spontaneous abortion, chronic infections such as periodontal diseases and bronchitis, autoimmune diseases, depression, substance abuse including tobacco and alcohol, diet, lifestyle, renal disease, menopause and the use of contraceptives, and environmental exposures such as second-hand smoke, pesticides, and medications (which should be classified by type).

5. Evaluation
Subjective symptoms and their onset should be recorded, including cognitive and behavioral symptoms, gait problems, tremor, balance, swallowing, incontinence, and pseudobulbar affect. In addition, vital signs should be collected, including height, weight, blood pressure (orthostatic), waist circumference, ankle-brachial index, heart rate, vision, and hearing. On the neurological examination, the NIH Stroke Scale (NIHSS) should be completed, as well as timed gait, motor movements, reflexes, and Babinski signs.

A discussion was entertained as to whether in this type of an exercise the physician’s impression of cognitive impairment or vascular disease should be obtained, but this was not unanimously endorsed.

In addition, a mental status examination including components aimed at capturing vascular contributions to cognitive impairment should be completed (see the Neuropsychology Section below for recommended protocols), as well as a behavioral assessment such as the Neuropsychiatric Inventory-Q, a depression scale such as the Center for Epidemiological Studies-Depression (CES-D) or the Geriatric Depression Scale, and a functional scale such as the Pfeffer Functional Assessment Questionnaire or the Barthel Index.

Although this is not an exhaustive set of suggestions, these items were believed to be pertinent in an epidemiological study that might identify individuals with cognitive impairment.

Abbreviated Clinical Evaluation
It was also requested that the Clinical and Epidemiology Work Group make recommendations for an abbreviated clinical examination that could be performed in the context of a busy primary care physician’s practice. As such, the following subset of the above recommendations was suggested.

1. Demographics
The minimum data set should include sex, birthdate, race/ethnicity and education.

2. Informant
If available and deemed to be necessary, basic information regarding the informant’s demographics as mentioned above and the time and quality of the contact with the subject should be obtained.

3. Family History
History information concerning first degree relatives for a history of stroke, vascular disease or dementia should be obtained.

4. Health History
Historical questions concerning cardiovascular or cerebrovascular conditions, hypertension, hyperlipidemia, diabetes mellitus, alcohol use, tobacco use, physical inactivity, and medications should be obtained.

5. Evaluation
The subjective impression of the individual being evaluated should be sought with regard to the person’s general health, including whether, during the past year, the person has experienced changes in memory, speed of thinking and acting, or mood. Information should be obtained regarding functional abilities that include instrumental activities of daily living.

On the examination, a physical examination should be done to collect vital signs and other data including height, weight, blood pressure, waist circumference, and timed gait. A mental status examination which focuses on vascular impairment should be done. This evaluation should include an assessment based on the clinical judgment of the examiner with respect to cognitive impairment and vascular contribution. However, as noted previously, the utility of such subjective, qualitative evaluations remains to be determined.

Finally, in the setting of an investigational study, certain laboratory studies should be done. These would include: collecting serum, plasma, and DNA for possible cell lines, ECG, cardiac echo, carotid ultrasound, urine studies and a MRI of the head. With respect to serum or plasma markers, the following measures could be considered: C-reactive protein, lipids, homocysteine, glucose, hemoglobin A1C, insulin, clotting factors, and fibrinogen.

	Neuropsychology Section

Top Abstract Introduction Clinical/Epidemiology Section Neuropsychology Section Neuroimaging Neuropathology Experimental Models Biomarkers Genetics Clinical Trials Conclusions Appendix References

 
Preamble and Criteria for Test Selection
VCI encompasses a large range of cognitive deficits, from relatively mild VCI no dementia to more severe vascular dementia, or combined cerebrovascular disease with other dementing conditions, such as AD.²³ The pattern of VCI cognitive deficits may include all cognitive domains, but there is likely to be a preponderance of so-called "executive" dysfunction, such as slowed information processing, impairments in the ability to shift from one task to another, and deficits in the ability to hold and manipulate information (ie, working memory).^24–26 Neuropsychological protocols must therefore be both sensitive to a wide range of abilities and especially attuned to the assessment of executive function. Timed executive function tests may be especially sensitive to VCI-related impairment because of the slowed information processing noted in this patient sample.

The Neuropsychological Working Group was charged with recommending test protocols that could be used in multicenter investigations of potential patients with VCI. Because different protocols serve different purposes, the working group was requested to produce 3 separate protocols, one that required &60 minutes, a second that required 30 minutes, and a third that required 5 minutes. It was envisioned that the 60-minute protocol be developed for use in studies that require a breakdown of cognitive abilities by domain, so the protocol contains recommended tests in 4 domains: executive/activation, language, visuospatial, and memory. In addition, tests were selected to examine neurobehavioral change and mood. Tests for the 30-minute protocol were selected from within the 60-minute protocol to be used as a clinical screening instrument for patients with suspected VCI. Finally, a 5-minute protocol was devised for potential use by primary care physicians, nurses and other allied health professionals, who need a quick screening in their office or at the bedside. The 5-minute protocol was also designed for very large epidemiological studies or clinical trials in which sensitivity and ease of administration are especially important. In addition, the 5-minute protocol was designed so that, once validated, it would be possible to be administered by telephone.

The working group was set up as an expert panel that considered the benefits and limitations of potential protocol instruments. In making its decision, the working group referred to prespecified test criteria (Table 1). It was recognized that there were no perfect tests. Instead, tests selected for inclusion in the protocols met a preponderance of the criteria. In addition, high priority was given to executive control, activation state and processing speed, word retrieval and episodic memory, to help differentiate VCI from AD and to target the executive domain.

View this table: [in this window] [in a new window]   TABLE 1. Neuropsychological Test Criteria

Proposed Neuropsychological Test Protocols
In order to maximize information obtained from a relatively brief number of tests, conventional well-validated tasks were selected. A number of novel analyses were proposed whereby multiple measures could be derived from a single simple short test. For example, word list generation cued by category can provide information relevant to language, activation, and speed of processing, set shifting, working memory and executive control.^27,28 Hence, one brief test can provide useful insight into different domains. In other words, economizing on time for administration can still yield, through neuropsychological "post-processing", multiple cognitive measures probing different anatomical regions and brain networks. A listing of the proposed 60-minute protocol tests, along with their primary citations and suggested normative sources is found in Table 2. The proposed 30-minute and 5-minute protocols are listed in Table 3.

View this table: [in this window] [in a new window]   TABLE 2. Sixty-Minute Protocol: Test List, Citations and Suggested Potential Standardization Sources

View this table: [in this window] [in a new window]   TABLE 3. Proposed 30-Minute and 5-Minute Neuropsychological Protocols

A. 60-Minute Protocol
1.
Executive/Activation

   Theoretical Framework for Fractionating Frontal Lobe Function.
It has been posited that functional domains within the frontal lobes can be fractionated, based on a theory of the evolution of 2 cortical architectonic trends, a dorsal (executive-cognitive) and a ventral (emotional-self regulatory) trend.^29,30 The "archicortical" trend evolves from the hippocampus and involves the dorsolateral prefrontal cortex, which mediates spatial and conceptual reasoning processes. The "paleocortical" trend, which mediates emotional processing, such as stimulus-reward associations, emerges from the orbitofrontal (olfactory) cortex and is linked to limbic nuclei such as the amygdala. Two additional specialized functions within the prefrontal cortex have developed later in human evolution, including action-regulation by the superior medial frontal region³¹ and metacognitive functions that appear to be subtended by the frontal polar regions, particularly on the right side.^32–34

   Test Selection.
Test procedures in the Executive/Activation domain were selected to sample these various functions. First, the working group adopted both a category (semantic) and a letter (phonemic) fluency test. The category fluency test chosen was "Animal Naming"^34–37 because of its common use, especially in elderly clinical populations, the availability of multiple normative sets, and the relative ease of potential cross-cultural applications. The test is included as part of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) neuropsychological assessment battery.³⁸ The letter fluency test chosen was the Controlled Oral Word Association Test.^39,40 This test includes 2 sets of 3 letters (CFL and PRW) that are of similar difficulty and can be used interchangeably. Multiple normative samples are available.

Both category and letter fluency tasks were chosen because of research suggesting differential performance in clinical populations across the 2 tasks and the potential of identifying separate executive as well as domain-specific (eg, language) processes.^{27,28,35,37,41–43} The working group recognized that there is a cultural generalizability issue with phonemic fluency tasks, as they would not be applicable to nonphonetic languages. Cueing by initial sounds of words rather than by letter might be appropriate in these languages, but would need to be developed and validated, if this were feasible. Similarly, letters differ in the frequency of their use across languages and we recommend that non-English investigators who wish to use the phonemic fluency task choose letters with roughly the same frequency of use in the alternate language as CFL and PRW are used in English.

Both of the fluency tasks will be administered and scored in the standard manner. However, additional scoring options may lead to a richer understanding of frontal executive functions. For example, the number of words generated in the first 15 seconds of a 1-minute word-list generation task reflects more automatic processing and provides an index of speed and activation.^43,44 The number of words generated in the subsequent 45 seconds yields information on working memory, set shifting and executive control, obtainable through careful analysis of the strategies used to cluster words. Category cueing, such as animals or fruits, may rely on posterior left parietal-temporal functioning,²⁷ whereas phonemically-cued word-list generation has demonstrated relationship with left dorsolateral frontal integrity.⁴³

In addition to the fluency tasks, the working group chose to include the Digit Symbol-Coding subtest from the Wechsler Adult Intelligence Scale, Third Edition.⁴⁵ This task provides a direct measure of processing speed and activation. Additional supplemental tasks of incidental free and cued recall are available for this task.⁴⁶ The Trailmaking Test⁴⁷was also chosen to provide an additional measure of information processing speed and set shifting.³² Finally, the working group recommends that additional scoring options available from the revised Hopkins Verbal Learning Test (HVLT-R) can provide measures of strategic learning reflecting dorsolateral frontal function, in addition to episodic memory indices.^48–50 By examining the strategies used to learn a super-span list, a measure of executive organization can be derived.

Test procedures to probe self-regulatory functions and metacognitive processes are still in development, but these behaviors can be inferred through the use of questionnaires, administered to caregivers. Likewise, the working group recommends that both simple and choice reaction time tasks should be considered as additions to the protocol in the future, because such activation tasks have been shown to be sensitive to frontal executive function, but as of yet may be cost prohibitive and require special equipment. As part of the validation process for the test protocols, we recommend the completion of exploratory and confirmatory factor analyses. These analyses may point toward a bifurcation of Executive/Activation tasks into timed and untimed tests.

2. Visuospatial
The working group selected the Rey-Osterreith Complex Figure⁵¹ copy condition as the primary visuospatial test. The memory condition of the test was selected as a supplemental measure. This well-known, untimed spatial task requires both organizational and visuoperceptual skill. Multiple scoring systems are available, including a standard 36-point method of determining accuracy of the final product^52,53and more qualitatively based systems that include a study of subject’s organizational ability.⁵⁴ Standardization samples are available for the major scoring systems.

3.
Language/Lexical Retrieval

The working group chose the Short Form (15 item) of the Boston Naming Test (BNT), Second Edition, as a measure of visual confrontation naming. The BNT, which is associated with the Boston Diagnostic Aphasia Examination,⁵⁵ has been well studied. Shorter versions, using either 15 or 30 words, have been found mostly equivalent and are relatively reliable and valid as screening tests, with age and education effects varying in different studies.^56–58 The correlation between the Short Form and the 60-item BNT is 0.97.⁵⁵ Normative information is exclusively based on the 60-item version, and it is suggested by the authors to extrapolate this information for the 15-item version.⁵⁵ As mentioned above, "animal" fluency can serve as a less structured lexical retrieval task as well as that of a test of executive function. Verbal fluency tasks have been widely used for many decades with some discriminative value in differentiating cognitive impairment and dementia from normal aging as well as VCI from AD.^59,60

More detailed tests of semantics and syntax including the Pyramids and Palm Trees Test and the Token Test were considered but not included in the basic battery. These could be supplemental tests to probe comprehension and semantic understanding more thoroughly. Similarly, a test of apraxia was considered but not included at this time because of the expertise required in scoring and lack of available validated batteries. Apraxia testing could be considered for future use if an appropriate brief tool were available.

4.
Memory/Learning

The working group favored a list-learning test over a paragraph-recall test or a paired association–learning test because list-learning tests can generate acquisition scores initially and with repeated administration, as well as a short- and long-delayed recall. In addition, list-learning tests are easier to develop in other languages and cultures than paragraph-recall tests.

After much debate, the HVLT-R⁶¹ was chosen as the preferred list-learning test. Strengths of the HVLT-R include its multiple alternate forms, its use in clinical trials and its relatively brief administration time. The working group recognized that the HVLT-R does not include either an interference list or a cued recall condition, both of which have been found to be sensitive to VCI-related cognitive impairment in clinical samples.^60,62,63 For this reason, the working group includes the California Verbal Learning Test, Second Edition (CVLT-2⁶⁴), as an alternative to the HVLT-R, which may be used by investigators who require the additional information gained from this test (eg, cued recall), who are studying subjects capable of completing a 16-word list test, and who have the increased time necessary to complete the CVLT-2.

In addition to either the HVLT-R or the CVLT-2, the working group recommended the development of a brief test of recognition memory created by repeating the BNT items with foils in a forced choice picture recognition paradigm. Cued recall of the pairing of the symbols and numbers in the Digit Symbol-Coding test can also be completed.⁴⁶

5.
Neuropsychiatric/Depressive Symptoms

The working group recommended the Neuropsychiatric Inventory, Questionnaire Version.⁶⁵ This test is derived from the original NPI, but can be completed by a caregiver without the need for an interviewer. The test probes most behavioral domains that are affected in VCI as well as other disease conditions. In addition, in order to more thoroughly probe depressive symptoms, the working group recommends the Centre for Epidemiologic Studies-Depression Scale (CES-D) developed at the National Institute of Mental Health.⁶⁶ It has 20 items, takes about 10 minutes and can be self-reported or administered as a questionnaire by an examiner. The CES-D has been used for over twenty years and has been previously validated in NINDS Stroke Data Bank patients using a structured psychiatric interview and established diagnostic criteria: a score of 16+ had been found to be highly predictive of clinical depression (sensitivity 86%, specificity 90%, positive predictive value 80%).⁶⁷ It has been shown to have high concurrent validity with other depression measures in geriatric stroke patients (both observer-reported and self-reported)⁶⁸ and has been used to assess poststroke depressive symptomatology in several studies.^68,69 It has also been used in the Cardiovascular Health Study.⁷⁰ In addition to the examination of depression and if time permits, investigators may consider completing a measure of apathy, such as Starkstein’s Apathy Scale⁷¹ in light of studies demonstrating apathy in patients with suspected VCI.⁷²

6.
Premorbid Status

To obtain premorbid history of cognitive status, the 16-item Informant Questionnaire for Cognitive Decline in the Elderly^73,74 should be completed by a person knowledgeable of the patient.

The Mini-Mental State Examination (MMSE⁷⁵) is widely used in all dementia studies and would be a sensible supplement to the above protocol.

B. 30-Minute Protocol
Already available batteries such as the modified MMSE⁷⁶ and Cognitive Abilities Screening Instrument⁷⁷ were considered because they have been used in population studies of the elderly and dementia. However, the concern was their sensitivity in a VCI context. Hence, a subset of the 60-minute protocol for screening purposes was suggested to include semantic and phonemic fluency, Digit Symbol-Coding and the revised Hopkins Verbal Learning Test, in addition to the CES-D and Neuropsychiatric Inventory. The Trail Making Test A and B could be used as a supplemental measure and MMSE would also be prudent if not already administered.

C. 5-Minute Protocol
There was considerable debate as to whether or not MMSE would be sufficient for a brief, minimal data set. However, this was rejected because it insufficiently probes executive function and because its 3-word recall test may be insensitive to the more subtle memory impairment often encountered in VCI. It was also felt that it would be advantageous if this brief protocol could be administered by telephone. The recommended protocol consists of selected subtests from the Montreal Cognitive Assessment available with instructions and sample means in English and French at www.mocatest.org (MoCA⁷⁸), including a 5-word immediate and delayed memory test, a 6-item orientation task and a 1-letter phonemic fluency test (the letter F). The MoCA may be used without permission, free of charge, for clinical or educational noncommercial purposes (Copyright Ziad Nasreddine, MD).

Supplemental tests, not all of which would be amenable to telephone administration, could comprise all or part of the remainder of the MoCA, which includes a cube and a clock drawing task with a simple scoring routine, a 3-item picture naming task, a short "Trails B" paradigm and other brief attention, language and abstraction tasks, and would take an additional 5 minutes. If there is more time available, some investigators may also wish to complete the original Trailmaking Test,⁴⁷ a semantic fluency test or the MMSE, provided that the MMSE is completed on a different day or an hour or more after the 5-minute protocol if done on the same day.

Summary
It is important to note that that these protocols are offered as a basic assessment, appropriate for different purposes. Supplemental tasks can be used on a project basis and if further emphasis is needed on particular functions. Validation of the protocols needs to be completed to see how well they detect documented cognitive impairment and how well they do so in relation to cerebrovascular disease. In particular, the 5-minute protocol, and the supplementary items from the MoCA, will require further age and education standardization in English and French as well as other languages. Further standardization would also be needed for the phonemic fluency in nonphonetic languages such as Chinese, and for the brief test of recognition memory using items from the short BNT with suitable foils. The goal is to encourage all comers—clinicians, epidemiologists, vascular medicine and dementia researchers alike—to consider the use of the recommended tests to obtain minimal cognitive data sets, according to the particular application, in order to improve communication, comparability and dialogue across cultures, patient groups, and studies and achieve better understanding and treatment of VCI around the world.

	Neuroimaging

Top Abstract Introduction Clinical/Epidemiology Section Neuropsychology Section Neuroimaging Neuropathology Experimental Models Biomarkers Genetics Clinical Trials Conclusions Appendix References

 
Neuroimaging in Studies of VCI
The main role of neuroimaging in the study of VCI so far has been to describe the brain, not diagnose it. Thus, neuroimaging plays a fundamentally different role in the study of VCI than it does in other conditions.^79,80 This focus on description rather than diagnosis results from the facts that (1) vascular and degenerative pathology frequently coexist,⁹ and (2) there are no pathognomonic radiological features of VCI. Different researchers have used a variety of terms and definitions to describe the changes in the brains of people with VCI, making comparison between studies difficult; this in turn has limited the understanding of the neuroimaging features of this condition. The use of a common minimal research data set with standardized terminology in all clinical studies can help overcome this obstacle. A minimal data set does not preclude researchers from adding their own neuroimaging descriptors. Rather, it simply stipulates those variables that must be measured and shared by all investigators. Implementation of this standard will allow comparisons between studies (population enrolled, outcome) and facilitate the pooling of data across research groups.

Prospective studies of VCI must include measures of ischemic brain injury as well as AD-type pathology, a prevalent confounder of brain-behavior relationships in VCI.^17,81 Whereas quantitative measurements are ideal, reliable qualitative scales may provide important data for large cohort studies. Before these scales can be accepted, however, intra- and interobserver reliability must be established, and the new scale must be validated against quantitative measurements. In addition, the means of conversion to existing scales need to be determined.⁸²

MRI is the ideal imaging tool for cognitive disorders because it is the most sensitive modality, and it offers the greatest amount of reliable data. The minimally acceptable field strength is 1.0 T, but 1.5 T or greater is preferred. The following sequences are required: 3D T1-weighted, T2-weighted, fluid-attenuated inversion recovery, and gradient-echo. The first 3 sequences provide information on the anatomy and presence of infarction and other pathology, whereas the latter detects large and small, acute and chronic hemorrhages.⁸³ In addition, diffusion-weighted images, and quantification of the apparent diffusion coefficient is encouraged because it gives information about acute strokes and integrity of the white matter fibers.⁸⁴ Images must be acquired parallel to the AC-PC line, which goes from the superior surface of the anterior commissure to the center of the posterior commissure.⁸⁵ CT has limited use in VCI research because it measures only severe disease, the findings are difficult to quantify, and there is significant radiation exposure (especially if 2 or more scans are obtained). Two general CT techniques can be used: the axial acquisition at +15° to the orbital-meatal line, and hippocampal acquisition at –30° to the orbital-meatal line, using 2-mm slice thickness.

Table 4 lists the recommended and acceptable MRI measures for prospective studies of VCI, and Table 5 lists the signal characteristics that differentiate perivascular space and infarcts, as used in the Cardiovascular Health Study (CHS).⁸⁶ The use of these standardized definitions is encouraged. Volume measurements, normalized for head size to take into account gender effects, are recommended to quantify atrophy and white matter hyperintensities, but validated qualitative scales are also acceptable.^87,88 The size and location of infarcts and hemorrhages must be specified,⁸⁵ preferably using standardized atlases, as described in Table 4. Table 6 lists the information that can be obtained from CT scans. Qualitative and quantitative scales of ventricular size have to be developed and validated. Hippocampal atrophy and white matter hyperintensities can be described using qualitative scales.^89,90 In the chronic stage it is difficult to differentiate hemorrhages from infarcts, as both appear hypodense; for this reason, the number, volume, and location of discrete hypodensities, rather than infarcts and hemorrhages, should be described. Acute and subacute hemorrhages are clearly different from infarcts and should be described as such.

View this table: [in this window] [in a new window]   TABLE 4. Imaging: MRI Measures

View this table: [in this window] [in a new window]   TABLE 5. Imaging: CHS Lesion Scoring

View this table: [in this window] [in a new window]   TABLE 6. Imaging: CT Measures

Retrospective studies based on chart abstraction of radiological reports are of limited value because radiological reports seldom have sufficient detail. Whereas certain imaging features are frequently mentioned in clinical reports (eg, "age appropriate atrophy", "white matter changes of vascular origin"), these are not sufficiently reliable for clinical research studies and, thus, should not be recorded, because they are only confounding information. Data to record are the presence of large infarctions, small subcortical infarctions, acute hemorrhage, ventricular enlargement, and other pathology.

Other neuroimaging techniques and applications require further research before they can be used routinely in prospective studies of VCI. The areas for further research are detailed in Table 7.

View this table: [in this window] [in a new window]   TABLE 7. Imaging: Areas for Further Research

	Neuropathology

Top Abstract Introduction Clinical/Epidemiology Section Neuropsychology Section Neuroimaging Neuropathology Experimental Models Biomarkers Genetics Clinical Trials Conclusions Appendix References

 
Optimal Brain Handling and Processing at Autopsy
The pathological diagnosis of VCI requires systematic evaluation of potentially relevant features. To achieve this, there must be some uniformity of autopsy protocols across centers, although individual approaches can be based on the strengths of a particular program. A large number of centers may follow the Alzheimer’s Disease Research Centers, USA (ADRC) guidelines, in which 1 cerebral hemisphere (usually left) is fixed and the other hemisphere frozen. If this approach is used, it is important to realize that lesions in the frozen hemisphere may be missed. Therefore, it is recommended that centers fix and freeze alternate coronal blocks from each cerebral hemisphere. Centers with major interests in neuroimaging correlative studies may prefer most of the brain to be fixed. Nonetheless, it is recommended that all centers should snap-freeze at least some material to obtain good quality mRNA, even with long postmortem intervals between death and freezing of tissue. The quality of the mRNA and viable yields aquired can be assessed case by case.⁹¹ Sampling of fixed brains should follow the CERAD protocol and include anterior (centrum semiovale) and posterior white matter blocks.

Assessing the neuropathological substrates of VCI obviously involves assessment of (1) parenchymal lesions, including infarcts and hemorrhages, and (2) the vascular abnormalities that may have caused them. The vascular lesions that contribute to full-blown dementia syndromes (during life) often show much more severe abnormalities than those contributing to milder VCI.^92–94 In addition, systemic factors (eg, hypotension, hypoglycemia) may cause brain lesions in the absence of severe vascular disease. Finally, parenchymal abnormalities may be present that are not obviously associated with either vascular disease or systemic factors; these include Alzheimer or hippocampal lesions.

The National Alzheimer’s Coordinating Center (NACC; www.alz.washington.edu) vascular dataset is a reasonable starting point for guiding one in the assessment of cerebrovascular disease, but could be refined to be more informative. The guidelines below will differentiate between a "minimal" informative dataset, and an "ideal" dataset for each type of abnormality being considered.

Data to Be Collected
1. Atherosclerosis of the Basal, Peripheral and Meningeal Vessels
Minimal Dataset
Comment should be made on the severity of basal atherosclerosis. Optimally, one should photograph this, especially noting similarities and differences between the anterior versus posterior portions of the circle of Willis, and left versus right sides. Representative histologic sections of the major arteries could be provided. The presence and severity of dolichoectasia and fusiform aneurysm(s) should be noted. Severity of stenosis of major arteries should be assessed, and can easily be estimated in ‘quartiles’, 0 to 25%, 26% to 50%, 51% to 75%, etc. The presence of atherosclerosis in distal (meningeal) arteries should be assessed.

Ideal Data Set
Further information from antemortem data on atherosclerosis of the cervical arteries may be acquired from angiographic studies. Severity of atherosclerotic lesions may be verified by dissecting the carotid and vertebral arteries in the course of a complete necropsy. This is more practical (though time-consuming) for the carotid than vertebral arteries; in practice it is rarely done. The status of the vessels in the heart, kidneys and other vascular beds should be assessed in the course of a complete autopsy. In addition, histological sections should be obtained to assess atherosclerosis, including plaque calcification, hemorrhage (remote or recent), ulceration, mural or complete thrombi, etc, as well as changes in distal arteries (eg, over the cerebral convexities).

2. Microvascular Disease: Arteriosclerosis
In general, arterial disease is considered more significant than venous disease. Venous adventitial fibrosis should be evaluated, however, because it has been linked to cognitive abnormalities and neuroradiologic lesions in some studies.^95,96 Areas to sample are those specified in the CERAD protocol,⁹⁷ together with anterior and posterior white matter blocks from the frontal and parieto-occipital regions, respectively. The middle cerebral artery-anterior cerebral artery watershed zones bilaterally are appropriate regions to assess for watershed ischemic change, and also to screen for cerebral amyloid angiopathy (CAA).

Minimal Data Set
Assessment of the severity of arteriosclerosis can be quite subjective. To minimize this subjectivity, templates are best used and findings in a tissue section matched to these (Ann McKee, personal communication, 2005). Any degree of inflammation should be documented, including presence of lymphocytes or macrophages centered on blood vessels (and not necessarily a function of brain ischemia). Perivascular hemosiderin should be noted as evidence of remote hemorrhage, even if minimal in amount. The presence of fibroid necrosis and Charcot-Bouchard microaneurysms should be described (realizing, however, that these are nonspecific complications of microvascular disease).

Ideal Data Set
Special stains such as Masson trichrome, elastica van Gieson, and Movat pentachrome should be used in conjunction with immunohistochemistry (eg, for smooth-muscle -actin, collagen types, etc) to provide a more illuminating picture of cerebral microvascular (arteriosclerotic) disease. The sclerotic index (SI=1–[internal diameter/outer diameter]),^98,99 an approximate measure of arterial/arteriolar stenosis, may be calculated for at least a subset of microvessels.

3. Microvascular Disease: CAA
Minimal Data Set
CAA can be evaluated quite accurately on hematoxylin/eosin–stained sections. Optimally, however, Congo red or thioflavin stains and amyloid-ß immunocytochemistry would be used to assess the extent and severity of this angiopathy. Key assessments should include focal versus widespread, meningeal versus cortical, and arteriolar versus capillary (or both) involvement of the meningocortical microvasculature. Quantification of CAA severity can be roughly approximated by evaluating the degree of involvement of individual arterial walls (Vonsattel grading^99,100), and multiplying by numbers of affected arteries per tissue section. Evidence of perivascular hemorrhage around affected arteries should be recorded, whether old (hemosiderin, hematoidin) or recent, as should the presence of CAA-associated inflammation (usually granulomatous) and other CAA-associated microangiopathies,¹⁰⁰ including microaneurysm formation, fibroid necrosis, etc.

Ideal Data Set
For more rigorous quantification of CAA extent/severity than that described above, grid-counting techniques in conjunction with amyloid-ß immunohistochemistry are being developed (M.P. Frosch, personal communication, 2005).

4. Miscellaneous Microangiopathies
CAA and arteriosclerosis (lipohyalinosis) dwarf all other forms of cerebral microvascular disease in terms of their clinical importance. However, neuropathologists must also be vigilant for other types of disease, while realizing that these rarely present with dementia. These disorders include vasculitides (non-CAA associated), disseminated intravascular lymphoma, thrombotic thrmbocytopenic purpura, etc. Vigilance must be maintained for new or previously poorly characterized familial ischemic vascular dementia syndromes similar to cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), cerebral autosomal recessive arteriopathy with subcortical infarcts and leukoencephalopathy (CARASIL), and hereditary endotheliopathy retinopathy nephropathy and stroke (HERNS).¹⁰¹

5. Parenchymal Abnormalities Associated With Cerebrovascular Disease
As a general rule, all infarcts must be assessed and recorded in terms of their number, size, location and estimated age. Acute lesions are less likely to be of importance in chronically evolving cognitive impairment, though they are obviously a marker of the duration and severity of cerebrovascular disease.

Minimal Data Set
Cystic lesions should be assessed as being large (arbitrarily >1 cm in maximal dimension), small (<1 cm), or microinfarcts (by definition not visible to the naked eye but detected on histologic sections). The term "lacunar infarct" should be reserved for cystic lesions that are smaller than 1 cm in size and located in the basal ganglia, brain stem or deep white matter (but not the cerebral cortex). Laminar necrosis is a distinct ischemic lesion usually associated with severe anoxic-ischemic encephalopathy or hypotension. Watershed infarcts, often symmetrical in the 2 cerebral hemispheres (and frequently in the anterior cerebral artery/middle cerebral artery watershed territory), should be noted. Assessment and significance of hippocampal injury, whether multifocal, segmental or multisegmental, or diffuse, is discussed below. Cribriform change is understood by most neuropathologists to represent (noninfarctive) enlargement of perivascular spaces. This change is questionably associated with cerebrovascular disease but should nevertheless be recorded, as should its location, whether in white matter or cerebral gray matter.

A major hemorrhagic component in an ischemic lesion suggests significant extravasation of blood into an infarct, or reabsorbed parenchymal hemorrhage. CAA may be associated with large or small bleeds (the latter sometimes visible only with microscopy) that are either preagonal or old. Such hemorrhages should be assessed in terms of number, age and estimated size, and an attempt made to link them to vascular disease present in the same brain specimen.

6. Leukoencephalopathy
It is deemed important to record minimally whether the anterior versus posterior white matter (or both), and periventricular versus deep white matter (or both) are affected. Optimally, these assessments should be made in myelin-stained sections (Luxol Fast Blue, Kluver-Barrera). A semiquantitative evaluation of "myelin loss" or, more accurately, "loss of myelinated tissue" (making no a priori judgment as to whether myelin pallor results from myelin or axon loss) can be made using an internal control such as the middle cerebellar peduncle on a semiquantitative scale of 0 to 3. Record should be made of whether the change is diffuse or multifocal.

Ideal Data Set
Quantitative assessment of large (grossly visible) infarct size, number, and location should be made. If possible, this should be done by taking digital images of brain slices and maintaining these in an easily searchable archive. Similarly, the topography of microinfarcts should be recorded, together with a quantitative assessment of their density in given locations. It may also be possible to coregister infarcts discovered at autopsy with antemortem neuroimaging data. The value of postmortem imaging (eg, of brain slices) is somewhat controversial because some MRI sequences are not interpretable in this material. Postmortem imaging may have value for assessing cortical thickness and "regularity", which can in turn be a surrogate marker for cortical scarring that results from microinfarcts. For optimal investigation of white matter lesions, immunohistochemistry could be employed using primary antibodies to myelin oligodendrocyte glycoprotein, phosphorylated neurofilament, glial markers, ubiquitin and amyloid precursor protein (APP).

7. Hippocampal Lesions
Given the importance of hippocampal structures in memory storage and retrieval, the hippocampus merits attention out of proportion to its size relative to the neocortex and subcortical white matter. It is also known to be a structure that frequently shows anoxic-ischemic change in the brains of elderly individuals.¹⁰²

Minimal Data Set
Both the anterior and posterior hippocampus and the amygdala should be assessed. The presence of focal microinfarcts/scars versus diffuse or segmental (CA1, prosubiculum) neuron loss and astrocytic gliosis should be assessed. Note that patchy neuron loss and gliosis in a heavily "Alzheimerized" brain may be difficult or impossible to distinguish from anoxic-ischemic change. Hippocampal injury resembling hippocampal sclerosis is now reported in autopsy brain specimens from individuals with fronto-temporal dementia, especially the variant associated with motor neuron disease.^103,104 Hippocampal sclerosis latter can be fairly easily distinguished from anoxic-ischemic hippocampal injury by the judicious use of ubiquitin immunohistochemistry.

Ideal Data Set
Serial blocks of the hippocampus and quantitative assessment of neuronal loss and astrogloisis should be carried out; however, this may be feasible only in dedicated research laboratories.¹⁰⁵

8. Incomplete Ischemic Injury
This entity, though suspected to be of importance clinically, does not have a widely appreciated or uniformly agreed on morphoanatomic correlate.⁹⁴ It may manifest as tissue rarefaction, which can be assessed by routine hematoxylin/eosin–stained tissue sections. Immunohistochemistry using antibodies which detect injury response (such as microgliosis, astrocytosis, or the presence of other "reactive" cells or surrogate markers of dendritic, synaptic or axonal damage) may, if assessed quantitatively, yield clues to subinfarctive brain injury. In this regard, morphological changes of reactive cells (for example, numbers of ramified processes in microglia), may be more important to assess than the number or density of these cell types. It will be important to establish correlations between these parameters and neuroimaging data. Purkinje cells of the cerebellum are important to examine in a given case of suspected hypoxic-ischemic encephalopathy, simply because of their vulnerability to this type of insult.

9. Mixed Vascular and Parenchymal Pathology
Cerebrovascular disease and AD pathology frequently coexist in the brains of elderly individuals.¹⁰⁶ The presence and degree of pathology related not only to AD but also to Lewy Body Disease and other types of dementing illness should be carefully documented, so that the extent of comorbidity may be accurately assessed in a given brain specimen. These assessments can be made using standard criteria enunciated by NACC, CERAD, and others.⁹⁷ It is recommended that Braak staging be provided on all cases. In an effort to further assess the extent of AD pathology, amyloid-ß (Aß) burden staging.

What Information Should the Neuropathologist Provide?
In the final autopsy report the neuropathologist has the option of providing (1) a final list of diagnoses, reflecting her/his assessment of the relative significance of different types of structural abnormality in the central nervous system (CNS) and their relative contributions to a neurodegenerative illness, or (2) an unbiased summary of the parenchymal and vascular pathology present in a given case, leaving subsequent interpretation of the findings to others. It may be useful to calculate an all-encompassing "score" to estimate the degree of vascular pathology present in these cases; this is being attempted by some groups.⁹⁴

The recommendations for obtaining essential information above are summarized in Table 8.

View this table: [in this window] [in a new window]   TABLE 8. Neuropathological Features of VCI Minimal Dataset to Be Collected

	Experimental Models

Top Abstract Introduction Clinical/Epidemiology Section Neuropsychology Section Neuroimaging Neuropathology Experimental Models Biomarkers Genetics Clinical Trials Conclusions Appendix References

 
The relationship between vascular dysfunction, brain injury, and cognitive changes (eg, VCI) in humans is only partly understood. Animals can be useful for modeling specific forms of vascular dysfunction that may lead to brain dysfunction, damage, and VCI. Animal models can potentially provide insight into how specific molecules, cells, and systemic conditions participate in vascular and parenchymal injury. Such models can also assist in defining targets for therapy. Finally, animal models can be used to perform preclinical assessment of new therapies to help guide human trials, as well as to develop biomarkers. In this section, we highlight some of the key issues surrounding the use of currently available animal models to study VCI, and provide recommendations for future research and model development.

Questions Addressed
1. What Animal Models Do We Currently Have to Understand the Pathophysiology of VCI?
A handful of experimental models are available to study VCI; however, these are not nearly as developed as those used to study acute ischemic brain injury or AD. This lack partly reflects the current lack of clarity in the definition of the clinical phenotype of this condition in humans, as well as the multitude of pathogenic pathways that can lead to the phenotype of VCI. Genetic and other risk factors can have a profound effect on the expression of vascular causes for cognitive impairment and on brain responses to vascular disease. These issues need to be further studied in virtually all the models listed below. In addition, the mechanisms by which vascular pathologies impair brain function leading to cognitive deficits in these models are poorly understood and need to be further studied.

Current models that provide insight into VCI include those in rodents and primates. Mice are attractive because of the ability to manipulate specific genes and to perform more detailed mechanistic studies. Primate models have been less fully used, but offer important advantages as well. These include the fact that nonhuman primate brains, like those of humans, possess large amounts of white matter; this feature seems particularly important given observed correlations between white matter injury and cognitive dysfunction in humans. In addition, disease states in primate models provide more direct functional and anatomic analogy to human cerebral diseases. In the aged primate, amyloid-ß (Aß) deposition in the microvasculature and in plaques analogous to similar to those seen in AD have been described. Well-described models of vascular injury, very relevant for studying the evolution and consequences of focal ischemia and chronic cerebral injury in humans, have been developed and can inform work in smaller animal systems. An additional advantage of nonhuman primates is the possibility to examine the chronic neurobehavioural consequences of cerebrovascular disease in subjects whose higher cognitive functions and behavior more closely model those of humans.

CAA
   Models Available:
(1) Transgenic mice over-expressing mutant forms of human APP that lead to AD, eg, APPsw.^107,108 These mice over-produce wild-type human Aß and develop parenchymal plaques and CAA. (2) Transgenic mice expressing mutant forms of human APP that lead to familial CAA, eg, APPDutch, APPDutch/Iowa.^109,110 These mice produce mutant Aß. (3) Squirrel, rhesus, marmosets, and others. These primate species are known to develop age-dependent CAA and parenchymal plaques.^111,112,123

   Pros.
Rodent models exhibit pathology similar to that seen in human disease, including hemorrhages, Aß, inflammation around amyloid, smooth-muscle and pericyte degeneration, basement membrane alterations, and concentration of pathology around leptomeningeal and penetrating arteries. Aged primate subjects show age-related accumulation of Aß deposits, accumulation of parenchymal pathology, vascular alterations, and cognitive impairment (but these changes have only been studied in a select genus species subgroup).

   Cons.
Concomitant over-expression of APP in some transgenic models does not occur in humans, and the presence of mutant Aß in Dutch or Dutch/Iowa mutants does not occur in most individuals with CAA. In addition, there is no strong evidence yet for the presence of infarcts that are associated with CAA in humans, although this issue has not yet been fully explored.

   Future Directions/Recommendations.
The development of more models in which there is selective vascular CAA without parenchymal plaques would be useful to the field. Also, animals that develop CAA in an anatomical distribution even more closely resembling that seen in human conditions would also be useful.

CADASIL
   Models Available.
Transgenic mice expressing Notch3 mutations at R90C¹¹³ or R133C (Martin Dichgans, unpublished data, 2006) have been produced.

   Pros.
Mice develop similar vascular pathology and vascular reactivity problems similar to that seen in human disease.^113–115

   Cons.
The mice do not exhibit ischemic lesions. The small amount of white matter in rodents limits comparison to humans, in which there is extensive white matter damage.

   Future Directions/Recommendations.
Detailed behavioral testing in these models needs to be performed. Studies in brain slices examining white matter function and susceptibility to injury may be informative.

Chronic Oligemia Models
   Model.
Bilateral carotid ligation in rats and baboons leads to sustained reduction in forebrain blood flow (oligemia) and white matter changes.^116–118

   Pros.
White matter changes and cognitive impairment are seen.

   Cons.
This paradigm is technically more difficult to implement in mice, so that it is not as easy to assess the effects of genetic manipulation.

   Future Directions.
Bilateral carotid coiling may provide another useful model in mice. We also need to develop other models, in both mice and primates, with chronic reduction in cerebral blood flow.

Hypertensive Vasculopathy
   Models.
(1) Spontaneously hypertensive, stroke-prone rats (SHR-SP). These rats develop spontaneous hemorrhages and ischemic lesions when fed a Western diet with salt.^119,120 (2) Mice over-expressing ^</