Multi-Infarct Dementia

David S. Geldmacher and Peter J. Whitehouse

Cerebrovascular disease is a leading contributor to dementia worldwide. In most populations which have been studied, only Alzheimer's disease (AD) is a more common cause of dementia (8). In 1974, Hachinski et al. (24) popularized the phrase "multi-infarct dementia" (MID) to represent the syndrome of dementia accompanied by focal neurologic signs or symptoms, characterized by stepwise deterioration, and frequently associated with hypertension. In some populations with a high prevalence of hypertension (such as African American men and the Japanese), MID is more common than AD (26, 56). The nomenclature of MID is complicated by several overlapping terms. Though criteria for the diagnosis of MID were published in DSM-III-R in 1987 (2) and have been widely adopted, their reliability has been questioned and nonstandard alternatives have arisen (14). Furthermore, "vascular dementia" (VaD) has emerged as a diagnostic category that includes not only the multiple discrete infarcts of MID, but other dementing syndromes attributed to cerebrovascular origins. Among these is a dementia associated with diffuse subcortical white-matter disease putatively attributed to chronic subcortical ischemia. This state is commonly, but controversially, known as "Binswanger's disease" or "subcortical arteriosclerotic encephalopathy." In contrast, "Leuko-araiosis" was proposed by Hachinski et al. (25) as a description of radiologic and pathologic subcortical white-matter abnormalities such as those encountered in Binswanger's disease, but these changes are not obligately associated with dementia. Other less common causes of dementia, such as vasculitides, are also considered under the rubric of VaD.

MID has been considered a "subcortical dementia" (10). The term "subcortical dementia" provides a clinical shorthand for dementia with prominent motor effects and relative rarity of the "cortical syndromes" of aphasia, agnosia, and apraxia. Erkinjuntti (13) reported, however, that 65 of 79 MID patients in his series had sustained a cortical stroke and that 56% of the subjects had evidence of cortical strokes alone. Mahler and Cummings (41) have subsequently considered large-vessel and small-vessel behavioral subtypes of vascular dementia. This distinction further clouds the concept of MID as a subcortical syndrome because the behavioral neurology of large-vessel infarctions typically involves "cortical" signs. The theoretical problems inherent in a cortical-subcortical dichotomy for the description of dementia have also been previously addressed (61). The interpretation of what constitutes MID is further complicated by a lack of specificity and uniform application of proposed criteria for diagnosis. Given the high prevalence of cerebrovascular disease, strokes frequently contribute to the cognitive morbidity of individuals with dementia of all types, including AD. Although antemortem clinical evaluations and imaging may confirm the presence of multiple strokes, those techniques cannot exclude the presence of AD pathology contributing to the overall condition. For instance, the presence of cerebral infarctions may allow the clinical expression of Alzheimer-type dementia even though the pathologic criteria for AD are not met. Consequently, the frequency of pure MID in autopsy studies is 10-23%, comparable to that of "mixed dementia" with changes of both MID and AD (35) (see Towards an Understanding of the Genetics of Alzheimer's Disease and Amyloidogenisis in Alzheimer's Disease and Animal Models, for related issues).

Recurrent cerebral infarctions are, by definition, the pathophysiologic basis of MID. The risk factors for MID are, not surprisingly, those for cerebrovascular disease, especially age and hypertension. There appear to be no risks specific for the development of MID within the context of cerebrovascular disease. In about 90% of pathologically verified cases of MID there is a history of acute unilateral motor or sensory dysfunction consistent with stroke (14). There may also be a history of acute impairment of "cortical" functions manifest as aphasia, apraxia, or agnosia. Urinary dysfunction and gait disturbance have been suggested as early markers for the development of MID (38). With accumulation of ischemic brain lesions there is typically incremental impairment of memory and behavioral initiation, along with extrapyramidal features such as facial masking and rigidity.

An "ischemic score" (IS) was proposed by Hachinski et al. (23) as a means of distinguishing MID from primary degenerative dementia. A number of variants have been employed since the introduction of the original IS; a typical example is shown in Table 1. These scales share the common weaknesses that they are sensitive but not specific indicators of MID and do not address the presence or absence of AD pathology (8). In the clinical setting, an IS is most useful as an instrument for suggesting the presence of cerebrovascular contributors to a dementia syndrome.

The diagnosis of MID depends on the establishment of dementia — that is, a sustained decrement from previously attained levels of cognitive ability, sufficient to interfere with everyday activities, without an associated impairment of consciousness. Dementia may be stable or progressive. If strokes are the cause of a dementia, it is conceivable that there might be an improvement in cognitive status as the deficits from an acute stroke resolve without returning to baseline. When dementia is accompanied by a history of strokes temporally linked to stepwise deterioration in intellectual abilities, the clinical diagnosis of MID is obvious, though mixed dementia is also a possibility. A more difficult diagnostic situation is the patient with a history of strokes not temporally associated with onset of worsening of cognitive impairment. Recently, Chui et al. (9) proposed criteria for the diagnosis of "ischemic vascular dementia," based on the model for diagnosis of AD (44). These criteria are summarized in Table 2. An even more broadly defined set of international diagnostic criteria for research studies of VaD has been proposed (52), but these have been criticized for being overly inclusive and failing to address the importance of temporal association of vascular events with onset of intellectual impairment (12). Of particular note is the inability of any criteria, short of autopsy examination, to differentiate mixed dementia from MID. These factors have led to considerable controversy over the clinical usefulness of the "vascular dementia" concept (7, 49). Hachinski (22) has further argued that diagnostic criteria for vascular dementia fail to account for the fact that it is a syndromic diagnosis of multiple origins and outcomes.

Because they are sensitive to site of dysfunction as opposed to the mechanism causing it, neuropsychological tests have been incapable of consistently distinguishing between MID, AD, and mixed dementias (41). Gainotti et al. (19) reported that AD patients were more likely than those with MID to make "globalistic" or "odd" type errors on Raven's Colored Progressive Matrices task, and on a design copy task were more likely to demonstrate the "closing-in" phenomenon — that is, copying figures such that they overlap the model. Mendez and Ashla-Mendez (45) suggested that unstructured neuropsychological tasks, such as the Tinker Toy test, may be able to distinguish between AD and MID, because of prominent aspontaneity in the latter. As with other neuropsychological measures, the ranges of performance of AD and MID patients overlap, which limits the diagnostic specificity in any individual patient. Furthermore, how well these results generalize to a populations not selected for the "classic" clinical courses of the syndromes is unknown. Rothlind and Brandt (53) have proposed the use of a Frontal/Subcortical Assessment Battery as a supplement to common bedside cognitive examinations for differentiating dementia types characterized by prominent subcortical pathology from AD.

As with most central nervous system diseases, imaging studies have an important role in the diagnosis of MID. In contrast to the diagnosis of AD, in which cerebral images are used to "rule out" structural changes contributing to the dementia, the images in MID can clearly identify significant pathology. In the neuropathologically verified series of Erkinjuntti's group (14), 74% of MID patients had cortical infarcts and 13% had deep infarcts on x-ray computed tomography (CT). Magnetic resonance imaging (MRI) is more sensitive to lesions in the brain than CT, but this is not necessarily an advantage in the diagnosis of MID. Cavities present on T1-weighted images are consistent with cerebral infarction, but many of the changes observed on MRI may represent the effects of healthy aging, such as dilated perivascular spaces. The typical changes include small, focal areas of increased signal as well as patchy or confluent periventricular white-matter hyperintensity on T2-weighted images. These nonspecific changes are the basis of the term "leukoaraiosis" (LA). It is important to recognize that a large volume of diffuse signal change may be present on CT or MRI without meaningful impairment of cognition. Nonetheless, LA is a frequent correlate of MID. In Erkinjuntti et al.'s (15) clinical series, 72% of MID patients had LA, as opposed to 19% of AD patients.

For many years, "cerebral arteriosclerosis" was considered an important component of most senile dementia — hence the popular use of the phrase "hardening of the arteries" as a synonym for dementia. This perception understandably led to extensive study of cerebral blood flow and metabolism, but with little concern over clinical differentiation of dementia types. The earliest studies employed inert gas measures of global cerebral metabolic rate for O2 (CMRO2). Such studies demonstrated diminished cerebral metabolism in demented subjects, both with and without known cerebrovascular disease (39).

Developing technology subsequently allowed regional cerebral blood flow (CBF) measurements using the gamma-emitter 133Xe and multiple extracranial radiation detectors for planar or tomographic imaging. Simultaneously, a greater understanding of dementia subtypes improved the discriminative abilities of the techniques. Patients with vascular dementia, including MID, demonstrate patchy, irregular areas of decreased CBF consistent with areas of infarction or ischemia, whereas AD patients have more uniform frontal, parietal, and temporal decreases in CBF (36, 62). There is no general agreement that diminished CBF by 133Xe methods correlates with dementia severity. Some studies have found good correlation in MID only (23), and others have reported it in AD only (62); however most studies have found it in both (6).

Positron emission tomography (PET) using 15O allows detailed mapping of O2 metabolism. Neither AD nor MID patients typically demonstrate chronic ischemia by this method (18). Despite early enthusiasm for [18F]fluorodeoxyglucose (FDG) PET as a useful technique for the differentiation of MID and AD (4), subsequent investigations have not been as conclusive (6).

Single photon emission computed tomography (SPECT) is more widely available than PET and has been used clinically to differentiate MID from AD, though the validity of SPECT for this purpose is not known. Neither of the two isotopes in general use, 123I-labeled amphetamine (IMP) and 99mTc-labeled hexamethylpropylene amine oxime (HMPAO), has been shown to be superior in the differential diagnosis of dementia (20). As with other imaging modalities, MID patients tend to show patchy or multifocal hypoperfusion whereas AD patients show more diffuse changes, but there is sufficient overlap to prevent diagnostic surety in any individual patient (55).

The reported frequency of MID in demented populations ranges from 4.5% to 39% (34). Karasawa and Homma (33) have suggested that the prevalence of MID, at least in Japan, has decreased since 1980 as the result of fewer strokes affecting the elderly.

Jorm et al.'s (29) extensive review of previous studies provides the basis for much of the current understanding of the demographics of MID. They calculated the prevalence of MID as doubling with every 5.3 years of age, which is in contrast to a popular perception that the prevalence of MID declines after age 75 because of mortality associated with recurrent strokes (43). Men are affected with MID more frequently — as opposed to AD, which is more common among women (29). In Europe, there is also a trend toward higher rates of MID in rural populations than in urban ones (34).

Meta-studies of the epidemiology of MID have been complicated by the lack of clear-cut and uniform diagnostic criteria. Another problem in the interpretation of MID epidemiology is that the illness is often defined on the basis of its risk factors regardless of temporal course. As pointed out by Kase (34), in the presence of dementia, the IS items of (a) history of hypertension, (b) history of stroke, (c) evidence of associated atherosclerosis, and (d) focal findings on neurologic exam are considered sufficient to diagnose MID. Prospective studies, using uniform diagnostic criteria and paying careful attention to the timing and character of stroke and dementia, will be required to more fully understand the epidemiology and natural history of MID.

Tomlinson, Blessed, and Roth's landmark article (59) on the neuropathology of demented older individuals clarified the importance of AD pathology in senile dementia. It also reported a 20% frequency of multiple, discrete infarcts. These findings, along with Hachinski et al.'s (24) popularization of the term MID, defined the role of focal infarctions as a cause of dementia. Lacunar infarctions, also known as lacunes, are commonly implicated as a major contributor to MID because of the "subcortical" features often prominent in the clinical presentation of the illness. Lacunes are small cavitary lesions attributed to the occlusion of deep penetrating arteries. There is no uniform definition based on size, but most lacunes are less than 2 cm in diameter.

Lacunar infarctions are strongly associated with a history of hypertension. In Fisher's (16) report, 97% of 114 autopsy cases of lacunar infarction had a diagnosis of hypertension, though more recent studies with stricter criteria for hypertension suggest rates ranging from 60% to 75% (47). The importance of lacunes per se as contributors to the dementia has been questioned. Both Tomlinson et al. (59) and Fisher (17) minimized the role of these lesions in cognitive deficits. Cases of MID with lacunes also typically show myelin-stain evidence for extensive white-matter degeneration (27, 48). Whether an accumulation of lacunes themselves is able to produce dementia in the absence of associated noncavitary white-matter damage is unknown. Though frequently referred to as demyelination, electron microscopy (EM) indicates that axons within the myelin-stain lesions are lost as well (63). Because the diffuse white-matter changes and the cavitary lesions almost always co-occur and share a common pathophysiology, it is unlikely that their differential effects will be elucidated from human clinical material. The problem in differentiating "pure" MID pathologically is one factor contributing to the evolution of the more inclusive concept of ischemic vascular dementia.

Two other types of discrete infarctions contribute to many cases of MID. Large-vessel infarctions are usually identifiable by history with features of hemiparesis, hemianopia, aphasia, and so on. These are also unequivocally evident on CT or MRI. The volume of tissue loss from such lesions is an important factor in the development of dementia. Tomlinson et al. (59) reported that all their autopsy subjects with greater than 100 ml of tissue loss were demented. However, it is clear that dementia can follow much smaller losses of brain tissue if these are strategically located (11). The second type of cortical lesion contributing to MID is the micro-infarct. These have been reported as the sole basis of dementia (32, 59) and consist of 0.5-to 2-mm-diameter lesions within the cortical ribbon. They are associated with a history of transient ischemic attacks (48).

Other factors which predispose to the development of multiple cerebral infarctions are associated with MID or VaD. Conditions leading to thromboembolic showers, such as endocarditis or atrial myxoma, can lead to the rapid development of a demented state often after a period of acute encephalopathy or coma. Autoimmune vasculitides, such as in systemic lupus erythematosus or granulomatous angiitis of the central nervous system, contribute to areas of cerebral ischemia and infarction. They can be associated with long-term cognitive impairments. Tertiary Lyme disease and syphilis can also cause dementia on the basis of vasculitic thromboses. Cerebral amyloid angiopathy, though often linked to AD, may lead to multiple intracerebral hemorrhages and play a significant role in the development of VaD (28). One other lesion of vascular origin which can present as dementia is chronic subdural hematoma. These intracranial fluid collections can mimic the fluctuating, stepwise cognitive deterioration and prominent motor symptoms characteristic of MID, and they are largely reversible with surgical drainage of fluid and relief of mass effect.

To date, there remains no concise explanation for the pathogenesis of MID except for infarctions causing loss of brain volume or loss of strategic, localized, areas integral to normal cognition, or a combination of these two factors.

Although CBF is diminished in MID, this is a feature common to most dementia and probably represents a response to reduced cerebral metabolism, rather than the cause of the cognitive impairment. Some MID patients show foci of elevated regional oxygen extraction fraction (rOEF) suggestive of areas of chronic compensated ischemia (21). Rogers et al. argued (51) that a state of insufficient blood flow to the brain precedes the onset of dementia in MID patients by up to 2 years. Brown and Frackowiak (6) have cautioned, however, that such rOEF changes are not common among MID patients and therefore cannot be the major factor in the development of most MID. Two conditions associated with global diminution in CBF — cardiac disease (58) and hypertension (3) — have nonetheless been long recognized as contributors to impairment on neuropsychological testing. Meyer et al. (46), for example, reported that careful control of blood pressure improved cognition in some in MID patients, but overcontrol (with presumed diminution of CBF) worsened cognitive performance.

MID and, more inclusively, VaD are associated with changes in the blood-brain barrier (BBB). Elevated cerebrospinal fluid (CSF) concentrations of albumin and immunoglobulin G (IgG) have been reported for MID patients (40), though other studies have found no difference for albumin (1) or IgG (5). Interestingly, Blennow et al. (5) also reported increased CSF/serum ratios for albumin in AD patients with white-matter lesions or vascular risk factors. This indicates that BBB dysfunction in VaD may result from risk factors for cerebrovascular disease rather than represent a unique contributor to MID. Wallin and Blennow (60) have argued that, because myelin lipids are significantly reduced in vascular dementia, the myelin sheath is a primary lesion site. They further hypothesize that the high metabolic demands of the oligodendrocytes render them prone to ischemic damage. These views are at odds with (a) the PET data, which suggest that chronic ischemia is not a contributor to MID (6), and (b) the EM studies, which show axonal loss in areas of noncavitary demyelination (63). Although myelin loss and BBB dysfunction may contribute to some vascular dementia syndromes, their causative role in MID is questionable. One of the difficulties in assessing the pathophysiology of VaD is the considerable frequency of dementia with findings of both vascular disease and AD. Although this may simply represent the co-occurrence of two common illnesses, there is evidence that links cerebrovascular disease and AD pathology. Kalaria et al. (31), for instance, found that cerebral ischemia promotes deposition of potentially neurotoxic amyloid in the brain. Sofroniew et al. (57) reported that focal cerebral damage causes neuronal loss in the nucleus basalis of Meynert similar to that observed in AD. Furthermore, such changes in the basal forebrain, when associated with AD, have been linked to alterations of cerebral vascular regulation and diminution of CBF (54). The synthetic sites for the biogenic amines are also affected in AD (42, 50). Degeneration in these sites, the locus coeruleus and dorsal raphe nuclei, may adversely affect cerebrovascular function, because norepinephrine and serotonin also influence vascular autoregulation (53). The distinction between causes of vascular and "primary degenerative" dementias may therefore be more difficult than is commonly accepted.

Although a number of animal models for the development of MID have been employed, none have been satisfactory. Rodents tend not to have profound long-lasting behavioral effects from cerebral infarctions, and the multiple or diffuse, gradually acquired lesions characteristic of MID in humans have not been reproduced. The promising technique of inducing embolic ischemia in rats by injecting 35-mm-diameter microspheres into rat carotid arteries produced effects on memory, but these were not sustained (37).

Drugs of many classes and presumed mechanisms of action have been tried in the treatment of the cognitive symptoms in MID, but none have consistently been demonstrated to be effective. No agent has been approved for such use in the United States. There are, however, potential means of symptomatic treatment. Improvement among selected MID patients on a screening instrument for cognition, the Cognitive Capacity Screening Exam (CCSE), was reported with treatment of vascular risk factors such as hypertension and smoking. Similar treatments did not affect the cognition of AD patients in the same paradigm (46). In systemic conditions that decrease CBF, such as valvular heart disease and hypertension, neuropsychological test performance can improve with treatment of the causative factor(s) (30).

Alteration of the course of the illness may also be accomplished. Reduction of blood pressure is a primary goal of treatment in order to diminish the risk for recurrent stroke (43). Other risk factors, such as smoking and diabetes mellitus, can be addressed to reverse or slow the progression of vascular pathology. Any treatment approach that reduces the likelihood of stroke, such as carotid endarterectomy in moderate stenoses or the use of aspirin or ticlopidine in primary and secondary prevention, is likely to alter the course of MID, but no definitive analyses have been reported. It is important, however, to emphasize that many of the vascular changes contributing to strokes are the result of long-term pathologic processes which are not reversed with treatment. As Meyer et al. (46) found, overreduction in blood pressure can actually worsen cognition. That risk factor modification can affect the course of MID after diagnosis has not been conclusively demonstrated, but a reduction in VaD prevalence has been attributed to attention to risk factors (26).

Multi-infarct dementia is a syndrome which varies according to the site, size, nature, number, and timing of the lesions. Although criteria for the diagnosis of vascular dementia as a whole have been proposed, the long-term utility of such criteria has been questioned (22). No specific risk factors beyond those for cerebral ischemia have been identified, but it is likely that with control of the risk factors, progression of the illness, and perhaps current function, can be affected. The challenge lies in the early identification of those at risk for subsequent development of cognitive impairments and intervention. Prevention of vascular dementia through risk factor management may have further impact because of potential interactions between cerebral ischemia and the expression of AD.

Hachinski (22) has claimed that "Few areas in medicine are as ripe for action as the vascular dementias." The success of further efforts to understand VaD depends on several factors. Included among them are (a) a commonly accepted definition of what constitutes VaD and (b) the recognition that multiple, potentially treatable causes contribute to a final common clinical state of dementia. Early recognition of risk, and subsequent intervention, are then possible before the evolution of the dementia. The development of more useful animal models and new techniques of functional imaging to understand the pathogenesis of dementia in the face of vascular compromise will be vital in settling many of the controversies surrounding the field today. Despite those controversies, and the impediments to progress engendered by them, it is apparent that prevention and treatment of VaD is an achievable goal.

This work was supported by NIA Alzheimer's Disease Research Center grant AG08012.

published 2000