Criteria for Vascular Dementia: Title and subTitle BreakReplacing Dogma With Data

INCONSISTENCY IN the diagnosis of vascular dementia has been recognized now for some years, but not of the magnitude reported by Erkinjuntti et al¹ and now confirmed by Chui et al.² The degree of inconsistency is not trivial: differences in the rate of diagnosis of vascular dementia among criteria are routinely several-fold in magnitude. Not only have Chui et al² re-emphasized this point, but they have also identified the probable reason. Despite a degree of overall similarity, the criteria themselves differ in particular respects. When tested by using computerized algorithms with yes/no branch points, the differences among the criteria exert a substantial effect on the final diagnosis, perhaps rather more than would have been the case if the criteria had been applied in standard clinical practice.

At first sight, a solution seems simple: the various sets of criteria should be disassembled into their component parts, and these parts should be tested against a gold standard. Those components proven valid would form the basis of new criteria.

However, this solution presupposes that the components are appropriate, which presents a fundamental problem. Not one of the current sets of criteria is based on evidence; all have been developed from experts' opinions. Furthermore, these criteria were developed at a time when Alzheimer disease was in the ascendancy as the principal cause of dementia. This has led to a broad diagnostic paradigm that identifies vascular dementia by the presence of dementia, based on an Alzheimer-like neuropsychological profile, with the superimposition of vascular risk factors and events. Since these criteria were developed, there has been a resurgence of interest in vascular dementia, and there are now sufficient data to review critically the bases of the criteria rather than their application.

The first fundamental problem relates to the level of severity of cognitive impairment. Dementia requires sufficient cognitive impairment to affect normal activities of daily living. Although it is quite clear that one cannot have Alzheimer disease without being demented, this is only so because the criteria for Alzheimer disease make it so. It is self-evident that the process of Alzheimer disease begins well before the development of any cognitive loss with the loss of the first synapse or the appearance of the first tangle or plaque. As there is no disease-modifying treatment available for Alzheimer disease, this is not yet a major issue. Even so, there is a move toward the identification of "cognitively-impaired, not demented" in Alzheimer disease in anticipation of disease-modifying treatments. However, disease-modifying treatments do exist for cerebrovascular disease, and it therefore makes sense to identify cases at the very earliest stages of cognitive loss prior to dementia, a stage termed vascular cognitive impairment.^{3 - 4} There is a need to move away from the concept of dementia altogether in the context of cognitive impairment due to cerebrovascular disease.

The second fundamental problem is the pattern of cognitive change on which any criteria for vascular dementia or vascular cognitive impairment is to be based. All the current criteria use a model of dementia based on Alzheimer disease. These emphasize memory loss, as is clear from Chui et al.² Cerebrovascular disease does not preferentially affect the mesial temporal lobe and would not be expected to have a disproportionate effect on memory. The evidence now available confirms this view: while memory is commonly impaired in vascular dementia, frontal and subcortical dysfunction is the most characteristic finding in the vascular dementias in general. New criteria will need to drop the emphasis on memory. While frontal and subcortical dysfunction is characteristic of many cases of vascular dementia, it is to be remembered that vascular dementia is not a single condition. Giving emphasis to frontal and subcortical dysfunction in any new criteria would probably also be a mistake because it might exclude cases due, for example, to large artery or embolic disease where cortical domains are primarily affected. Involvement of 2 or more domains without selection seems to be most appropriate.

Revision of these 2 points alone, however, is far from sufficient to resolve the many problems with the current criteria. A third fundamental problem is that of mixed dementia, (cerebrovascular disease and Alzheimer disease), a topic dogged by historical burdens. In the modern original work differentiating vascular dementia and Alzheimer diseases by Tomlinson et al,⁵ it was found that large infarct volumes were needed for vascular dementia. More recently it has become clear that there can be significant cognitive loss with very small infarct volumes and that small amounts of cerebrovascular disease can affect profoundly the age of presentation and speed of progression of what otherwise appears to be Alzheimer disease.⁶ It has also become clear that there are at least theoretical overlaps between the risk factors and pathogenesis of Alzheimer disease and vascular dementia.⁷ These facts conspire to make mixed dementia far more important than was realized at the time of the preparation of the current criteria. Mixed dementia will have to be specifically recognized as an entity because patients with this combination will need treatment for both conditions, even if the vascular component is so mild that it would have previously been dismissed as trivial. However, no good method for separating mixed dementia yet exists, and this subject will need to be developed before sound criteria can be proposed.

Only once these fundamental points have been addressed can the detailed work of developing criteria begin. This must be done on the basis of knowledge and not on the basis of supposition. Accumulation of this necessary knowledge will require the assessment of cognition and its etiological and radiological correlates in large numbers of patients with cerebrovascular disease to identify exactly how these parameters correlate with cognitive decline. This work needs to be done without the use of criteria in the first instance. If criteria are imposed at this early stage, it is very likely that they will tautologically confirm themselves. For example, if memory loss is made a criteria for the diagnosis of dementia, and if dementia is required for vascular dementia, then lo and behold! all patients with vascular dementia will have memory loss. This is self-evident nonsense.

A more subtle trap concerns the study of vascular dementia. Vascular dementia is the second most common cause of dementia, but is relatively uncommon in Alzheimer disease clinics or memory clinics. The probable explanation for this discrepancy is selection bias. It seems likely that cases of vascular dementia that more closely resemble Alzheimer disease are referred to Alzheimer disease clinics and that patients who have clearly had strokes, large or small, and cognitive loss related to these strokes, are not referred to Alzheimer clinics because the cause of their cognitive loss is already apparent. Other patients with subcortical dysfunction may simply appear slow, rather than demented, and may not be referred at all.

Thus the development and validation of diagnostic criteria for vascular dementia cannot be properly done in the setting of a memory clinic. Such data would refer only to a subgroup of all vascular dementia cases. While population-based work would be preferable, a vascular clinic would otherwise be a more appropriate forum than a memory clinic.

The failure of current criteria to produce the same diagnosis in the same patient is not the main problem in vascular dementia. Adjusting the current criteria so that they match is merely fiddling at the edges. Wholesale replacement of these criteria is needed, using fact-based criteria derived from unselected populations focussed on early disease.