Vascular dementia may be easier to treat than diagnose

Vascular dementia (VaD) is the second most common dementia after Alzheimer’s disease (AD). Post-stroke dementia increases the risk of long-term disability and mortality [1–3]. Healthcare utilization and resource costs for VaD exceed those for AD [4]. Population-based studies estimate the age-adjusted prevalence of dementia to be approximately 30% among people with a history of stroke [5,6], and individuals with incident stroke have a twofold increased risk of dementia compared with stroke-free individuals [7,8]. This increased risk appears to be independent of age, sex, education, hemispheric location of the cerebrovascular lesion and type of stroke. Hospital-based studies reveal post-stroke dementia prevalence rates of 5.9–32%. The highest prevalence is within 3 months of stroke and the lowest prevalence is beyond 1 year of stroke [9]. Approximately 20% of incident stroke cases develop dementia [8,10–12].

Unlike AD, the concept of VaD and vascular cognitive impairment remains controversial, and diagnostic criteria have been difficult to validate [13]. Several groups have issued standardized criteria for diagnosing VaD. The two most widely known are the National Institute of Neurological Disorders and Stroke and the Association Internationale pour le Recherche et l’Enseignement en Neurosciences (NINDS–AIREN) criteria [14], and the Alzheimer’s Disease Diagnostic and Treatment Centers (ADDTC) criteria [15]. The Canadian Study of Health and Aging demonstrated that, depending on the diagnostic criteria used, prevalence rates of post-stroke dementia ranged from 3.1% (International Classification of Diseases, 10th edition [ICD-10]) to 29.1% (Diagnostic and Statistical Manual of Mental Disorders, 3rd edition [DSM–III]) [16]. In the Cardiovascular Health Study, the NINDS–AIREN and ADDTC criteria failed to identify the same group of subjects [17].

The NINDS–AIREN criteria enable the diagnosis of VaD with varying degrees of confidence [14]. Probable VaD is defined by:

•	Cognitive decline from a previously higher level of functioning in memory and other cognitive domains, with cognitive impairment severe enough to interfere with activities of daily living
•	Cerebrovascular disease on both neurological examination and radiological studies
•	A temporal relationship between stroke and cognitive decline

Radiological criteria include the following, either alone or in some combination: multiple large vessel infarcts; single, strategically placed infarcts (e.g., angular gyrus, thalamus, basal forebrain and posterior or anterior communicating artery territories); multiple basal ganglia and white matter lacunes; and extensive periventricular white matter lesions (involving at least 25% of white matter). Definite VaD is diagnosed:

•	If criteria for probable VaD are met
•	If there is histopathological evidence of cerebrovascular disease
•	If there is no histopathological evidence of other possible causes of the cognitive loss

Possible VaD is diagnosed:

•	If there are no radiological data, but there is clinical evidence of cerebrovascular disease
•	If there is absence of a temporal relationship between dementia and stroke
•	In individuals with a subtle onset of dementia and variable course

Although the NINDS–AIREN criteria are the most commonly used criteria in clinical trials, they have important limitations [14]. The clinical criteria for documenting cognitive impairment have been criticized for being modeled on the pattern of cognitive impairments of AD [18]. Memory impairment is required for diagnosing VaD, although individuals with cognitive impairment following stroke may not demonstrate memory impairment. The NINDS–AIREN criteria may identify mixed dementia better than dementia solely caused by cerebrovascular disease. By requiring marked limitation in functional status for diagnosing VaD, the criteria tend to capture only advanced stages of VaD, leaving little opportunity to identify milder or earlier stage disease where therapeutic interventions may be most efficacious.

The radiological criteria of NINDS–AIREN also have limitations. The radiological criteria have high specificity but low sensitivity, often excluding cases for which cerebrovascular disease appears to have been the cause of cognitive and functional decline [19]. Inexperienced interpreters of magnetic resonance imaging studies apply the radiological criteria with poor inter-rater reliability (κ = 0.17) [20]. In one study, the radiological criteria did not reliably distinguish between stroke patients with and without dementia [21]. Finally, even among autopsy-confirmed cases of VaD, the criteria have demonstrated poor validity and inter-rater reliability [22].

There is little consensus on the best method for detecting and describing cognitive impairment in the context of VaD. The most frequently used screening tool for VaD is the Mini Mental State Examination (MMSE) [23]. The most common presentation of VaD is subcortical cognitive impairment [18]; however the MMSE may lack sensitivity for detecting deficits in executive function and processing speed, which reflect compromise of frontal subcortical systems [24]. Specific screening tests of executive functions such as the executive clock drawing test (CLOX) [25] and the Executive Interview (EXIT25) [26] have been developed to improve diagnostic sensitivity. Different brief cognitive screening batteries have been proposed to either supplement or supplant MMSE screening [27–29]. The National Institutes of Health (US) recently held a vascular cognitive impairment harmonization conference (April 2005) to determine if a consensus could be reached on this important issue. Despite these efforts, the diagnostic validity of specific screening batteries remains to be established. In the interim, it is important to recognize that the incidence and prevalence of VaD may be underestimated due to the lack of sufficiently sensitive screening tools [10,30].

Limitations in the accurate identification of VaD consequently limit treatment options for patients. Clinical screening tools that are sensitive to cognitive deficits resulting from common causes of cerebrovascular compromise must be developed. Ideally, such tools should capture an index of general cognitive function while additionally screening for attention, concentration, processing speed, mental flexibility, reasoning, receptive and expressive language skills, visual–spatial skills, praxis, and verbal and visual memory. Cognitive screening should include some assessment for the presence or absence of pre-existing cognitive impairment. One such measure is the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) [31]. This is a fruitful area for future research.

Despite limitations in screening and diagnosing VaD, there have been a number of promising treatment trials. The major randomized clinical trials of drug therapies in recent years have tested drugs that were approved for treatment of AD. There have been two multicenter, randomized, double-blind placebo-controlled trials of donepezil in possible or probable VaD as defined by NINDS–AIREN criteria [32,33]. Donepezil is a centrally acting cholinesterase inhibitor approved for use in mild to moderate AD. The two trials used the same doses of active drug (5 and 10 mg/day) and had the same length of follow-up of 24 weeks. Not surprisingly, the results were similar.

A total of 952 patients were screened to randomize 603 subjects in the Donepezil 307 Vascular Dementia Study [33]. Screen failures occurred in 16% of subjects because they had an MMSE score demonstrating cognitive impairment that was either too mild (MMSE > 26) or too severe (MMSE < 10) [23]. Patients were excluded if they had had a new stroke within 28 days before baseline assessment. Among randomized subjects, 71% had probable VaD and 29% had possible VaD. A total of 69% had hypertension and 18% had diabetes mellitus. Abnormalities were detected on head imaging in 98% of subjects. A total of 28% did not have a history of either stroke or transient ischemic attack (TIA).

Primary outcome measures were the Alzheimer’s disease Assessment Scale-Cognitive Subscale (ADAS-cog) [34] and the Clinicians Interview-Based Impression of Change-plus version (CIBIC-plus) [35]. The ADAS-cog principally assesses language, memory and praxis. Scores range from zero (best) to 70 (worst) points. Subjects on donepezil at either dose demonstrated significant improvement versus placebo in the ADAS-cog at every follow-up visit (6, 12, 18 and 24 weeks). Improvement in global function based on the CIBIC-plus was observed in a greater proportion of actively treated, than placebo-treated, individuals for subjects taking 5mg/day of donepezil, but not for subjects taking twice the dose. This lack of dose–response suggests that the magnitude of the benefit was marginal and subject to misclassification using CIBIC-plus.

Cholinomimetic side effects such as nausea (16%, 10 mg/day group), muscle cramps (13%) and anorexia (8.3%) were associated with donepezil use. Stroke occurred in 3% of subjects treated with placebo, 4% treated with donepezil at 5 mg/day, and 5.8% treated with donepezil at 10 mg/day. Recurrent and first-ever strokes are to be expected in this older cohort replete with vascular risk factors. The number of strokes was too small to know whether the arithmetic trend toward higher numbers of stroke with higher doses of donepezil was more than simply coincidence and investigators rated the majority of strokes as unrelated to treatment.

The Donepezil 308 Study screened 887 subjects and randomized 616 subjects [32]. Study subjects were excluded if they had suffered a stroke in the preceding 28 days. A total of 67% had a history of stroke and 76% had radiographic evidence of cerebral infarction. Significant differences in favor of donepezil were again seen at every follow-up visit on the ADAS-cog, with the placebo group demonstrating no cognitive deterioration over the 24-week follow-up. In contrast to the Donepezil 307 Study, significant differences on the CIBIC-plus were observed for both the 5 and 10 mg/day donepezil groups as compared with placebo. Reassuringly, stroke as an adverse event was not more common in the actively treated groups than in the placebo group. Cholinomimetic side effects were again seen. A combined analysis of the two Phase III trials demonstrated that donepezil caused significant improvements in cognition, global function and ability to perform instrumental activities of daily living [36].

Galantamine, another cholinesterase inhibitor, was tested in a multicenter clinical trial that included patients with VaD [37]. Subjects in the actively treated group received 24 mg/day. A subgroup analysis of the 81 placebo and 171 galantamine subjects with probable VaD demonstrated a trend toward improvement on the ADAS-cog change scores of 1.9 points (p = 0.06).

Memantine is a moderate affinity uncompetitive N-methyl-D-aspartate (NMDA) receptor antagonist approved by the FDA (US) for the treatment of moderate to severe AD. The therapeutic rationale for testing the drug in VaD comes from the observation that ischemia can lead to glutamate-mediated excitotoxic cell death. There have been two randomized clinical trials of memantine in VaD [38,39]. The Memantine in Mild to Moderate Vascular Dementia trial (MMM) 300 involved 321 patients randomized to either placebo or memantine at 10 mg twice daily [38]. All subjects were required to have probable VaD based on NINDS–AIREN criteria. Subjects had to have at least mild to moderate cognitive impairment at baseline, with an MMSE of 12–20 points. The mean MMSE for enrolled participants was 17 points. CIBIC-plus and ADAS-cog were the primary end points. Memantine did not significantly affect the CIBIC-plus, but at 28 weeks of follow-up, memantine did significantly improve the ADAS-cog versus placebo. Efficacy was greatest for subjects with the poorest baseline ADAS-cog and MMSE scores. There was minimal decline in cognitive performance over 28 weeks based on the ADAS-cog, with a mean change from 0 to 28 weeks of 1.6 points. Despite the minimal placebo cognitive slide, investigators interpreted the improvement in ADAS-cog as evidence that memantine likely provided symptomatic rather than disease-modifying benefits.

The MMM 500 trial screened 844 subjects and randomized 579 subjects to either placebo or memantine at 10 mg twice daily [39]. All subjects were required to have probable VaD based on NINDS–AIREN and the DSM-III-revised criteria, and a Hachinski Ischemic Score of at least 4 points [40]. Subjects had to have baseline MMSE of 10–20 points. The dementia had to have started at least 1 year before enrollment. Mean MMSE for enrolled participants was nearly 18 points. The two study groups significantly differed in favor of memantine by a mean of 1.75 points in the change in ADAS-cog scores from baseline. The largest treatment effect was in subjects with a baseline MMSE under 15 points.

Several patterns have emerged from the recent series of multicenter clinical trials in VaD. In contrast to AD, little or no cognitive decline has been observed among placebo-treated patients on the ADAS-cog over a period of 24 weeks or more of follow-up. The relatively stable cognitive trajectory of subjects with NINDS-AIREN-defined VaD supports the idea that the study populations are not merely subjects with symptomatic AD and coincidental cerebrovascular disease. Experimental drug treatments have usually been tested in subjects who have not had recent symptomatic focal cerebrovascular events such as TIA or ischemic stroke. This may partly explain why placebo- treated subjects tended to have a stable cognitive trajectory.

Patients who have suffered from an ischemic stroke tend to demonstrate significant recovery from neurological deficits in the first 3–12 months following onset of symptoms. One would expect some degree of spontaneous cognitive recovery in the first few months following a symptomatic stroke. Instituting drug therapy with a cholinesterase inhibitor or the NMDA antagonist memantine in the subacute period following stroke might enhance cognitive recovery rather than merely raising cognitive performance over a stable baseline. To date, clinical trials of drug therapy for VaD have demonstrated the benefits on cognition and instrumental activities of daily living to be modest, and durability of the benefits beyond 6 months of treatment has not been established. Beneficial effects on functional status might be enhanced if drug treatment were performed in conjunction with acute rehabilitation.

VaD may be easier to treat than diagnose. Despite limitations in screening and diagnosing VaD, there have been encouraging results from recent drug trials in subjects diagnosed with probable or possible VaD based on NINDS–AIREN criteria. Several questions remain. Future studies should clarify how soon after stroke drug treatment should be initiated and whether a limited course of drug treatment could augment the long-term benefits of physical and occupational therapy. There should also be a vigorous discussion about what constitutes minimally acceptable improvement in cognition to justify long-term drug therapy. In parallel with drug development, there should be further research into practical screening tests for VaD and less severe forms of cognitive impairment due to cerebrovascular disease, and this treatment can be delivered to a greater percentage of those who could benefit.