Pract Neurol 8:26-38 doi:10.1136/jnnp.2007.139428
  • Review


  1. Mike O’Sullivan
  1. Humboldt Research Fellow, Neurologische Klinik, Klinikum Großhadern, Marchioninstraße 15, 81377 München, Germany
    ; michael.osullivan{at}


      Leukoaraiosis describes diffuse white matter abnormalities on CT or MR brain scans, often seen in the normal elderly and in association with vascular risk factors such as hypertension, or in the context of cognitive impairment. The term was introduced to avoid confusing an imaging appearance with a specific pathology. Neurologists often come across this appearance, but its significance is sometimes uncertain. The effects on cognitive function and gait are insidious and can be difficult to detect at the bedside, but are nevertheless important. However, gradually it is becoming clearer how leukoaraiosis relates to cerebrovascular disease, Alzheimer’s and other diseases, and how this appearance should influence treatment decisions.

      The term “leukoaraiosis” was introduced almost exactly 20 years ago by Vladimir Hachinski and his colleagues.1 A new term was needed because, with the rapid rise of CT brain imaging, it was becoming clear that diffuse low density changes in cerebral white matter were often seen in the context of dementia and vascular risk factors. Initially, this newly revealed imaging appearance was thought of as the radiological correlate of Binswanger’s disease, considered rare in the pre-imaging era (see boxes opposite). However, it became apparent that the relation between these imaged white matter lesions and pathology was more complex. Leukoaraiosis was, therefore, proposed as a purely descriptive term to avoid making assumptions about the underlying pathology, with the intention, in time, “to have labels replaced by understanding”.1 The term was derived from the Greek stem “leuko-” meaning white and referring to the white matter, and the adjective “araios” meaning rarefied.

      Leukoaraiosis is defined as diffuse, confluent white matter abnormality (low density on CT, hyperintensity on T2-weighted or FLAIR MRI; fig 1), often with irregular margins. As a purely radiological concept, the term could equally be applied to the leukodystrophies, as well as some metabolic and inflammatory disorders. However, the need for a neutral term that avoids assumptions about pathology is less pressing in patients likely on clinical grounds to have these conditions. Therefore I will focus on the vascular and degenerative forms of leukoaraiosis (table 1), which are the commonest and generate the most controversy.

      Figure 1 An MR FLAIR (fluid attenuated inversion recovery) image showing diffuse leukoaraiosis and coexistent lacunes, which appear as cerebrospinal fluid intensity “holes” (arrows).
      Table 1 Vascular causes of leukoaraiosis in older adults


      Leukoaraiosis is seen in a range of different clinical contexts and by different specialists—not just by neurologists, but also by stroke physicians and physicians who care for older adults. In large unselected stroke series like the Lausanne Stroke Registry about 7% of all patients with ischaemic stroke2 had leukoaraiosis, linked particularly with lacunar stroke which accounts for 20% of all strokes. Leukoaraiosis is also seen in patients with dementia; in early unselected series, leukoaraiosis was found in 30–40% of cases, and in recent clinical trials in vascular dementia as many as two thirds had leukoaraiosis.3 However, the link with dementia is not confined to cerebrovascular disease and up to 30% of patients with “pure” Alzheimer’s disease also have leukoaraosis;4 not surprisingly, therefore, leukoaraosis is seen in memory and “early dementia” clinics.

      In addition, leukoaraiosis may be seen in the context of normal ageing. Non-specific white matter lesions become increasingly common with age, although frank leukoaraiosis, as opposed to scattered focal lesions, is seen less commonly. Early confluent, or confluent lesions, corresponding to leukoaraiosis are seen in just over 10% of asymptomatic people aged 50–75 years and the prevalence increases with age.


      White matter

      In regions of leukoaraiosis the major pathological findings are myelin pallor, enlargement of perivascular spaces, gliosis and axonal loss (fig 2).5, 6 Originally it was suggested that selective demyelination might occur as a result of “incomplete” ischaemia, but a single electron microscopy study showed that pallor was largely due to loss of nerve fibres in their entirety.7 A very similar pathology picture is observed in regions of true leukoaraiosis in asymptomatic older adults, although the pathology underlying scattered focal hyperintensities is much more diverse. Fazekas et al demonstrated that pathological heterogeneity diminishes as the lesion severity increases from small punctate hyperintensities to early confluent and then confluent lesions, which almost always have an ischaemic appearance with myelin loss, gliosis and microinfarction.8 In contrast, periventricular “caps” and a smooth halo around the lateral ventricles (or “band”) have a non-ischaemic appearance with subependymal gliosis and discontinuity of the ependymal lining.

      Figure 2 Pathology of leukoaraiosis. Two small vessels (arrows) are shown with concentric hyaline thickening, loss of smooth muscle cells and with luminal narrowing. The perivascular space is widened, and the surrounding white matter appears gliotic. (Haematoxylin and eosin. Courtesy of Dr Alistair Lammie, Cardiff, UK.)

      Although these are by far the commonest substrates of leukoaraiosis, it needs to be remembered that other quite different processes can mimic this imaging appearance. For example, disseminated white matter metastases, lymphoma and obstructive hydrocephalus have appeared as unexpected causes in postmortem series.9, 10

      Blood vessel abnormalities

      The white matter of the cerebral hemispheres is largely supplied by long, narrow penetrating arteries and arterioles that arise from branches of the major cerebral arteries on the pial surface of the brain. In regions of leukoaraiosis alterations in the structure of these vessels are an invariable feature. In the classical early series changes in the vessels seemed to span a hierarchy of severity from: hyaline thickening and arteriosclerosis (which some refer to as simple small vessel disease) (fig 2) to “lipohyalinosis”—a term which refers to a more disorganised vessel wall with foamy macrophages (complex small vessel disease), to fibrinoid necrosis.11 There are few more recent pathology data, but these suggest that the vascular pathology may have changed, with fibrinoid necrosis a much rarer finding than previously, possibly reflecting better treatment of hypertension.12

      In asymptomatic older adults, blood vessel changes are still a constant feature, but rarely progress beyond arteriosclerosis with concentric hyaline thickening. In the most comprehensive pathology survey of the brains of asymptomatic older adults9 parenchymal changes were invariably linked with a reduction in the ratio of arterial lumen to external diameter.

      Alzheimer’s disease and cerebral amyloid angiopathy

      Given its association with Alzheimer’s disease and predilection for small arteries, cerebral amyloid angiopathy might seem the most obvious cause of leukoaraiosis in Alzheimer’s disease. However, the early pathology studies of Brun and Englund showed that leukoaraiosis was associated with non-amyloid “fibrohyaline” thickening of arterial walls, as it is in ageing and cerebrovascular disease.13 This initial study also found no correlation between white matter pathology and degenerative pathology in the overlying neocortex, suggesting that leukoaraiosis is not merely secondary to the degenerative process and dying back of axons through Wallerian degeneration. Furthermore, the extent of white matter pathology in Alzheimer’s disease correlates with non-amyloid arteriosclerosis, not with the severity of amyloid angiopathy, even when both forms of microvascular pathology are present.14 In summary, in Alzheimer’s disease, leukoaraiosis co-localises with, and correlates with the severity of non-amyloid arteriosclerosis. However, this does not necessarily rule out a role for β-amyloid in the pathogenesis of these changes, as will be discussed later.

      Leukoaraiosis is also increasingly recognised as a feature of sporadic cerebral amyloid angiopathy, without Alzheimer’s disease.15 Amyloid deposits are found in more proximal portions of the small penetrating arteries than arteriosclerotic changes and it is not known whether these deposits alone, without downstream arteriosclerosis, are sufficient to cause leukoaraiosis. Notably, leukoaraiosis is found in some of the genetic amyloid angiopathies (for example, familial British dementia) suggesting that amyloid angiopathy alone is sufficient to cause this imaging appearance.

      Leukoaraiosis and small vessel disease

      The pathological studies suggest that leukoaraiosis is one manifestation of cerebral small vessel disease. This is supported by strong pathological and clinical associations with the other major manifestation of small vessel disease—lacunar stroke.2 However, although leukoaraiosis and lacunar infarcts are often found together, in individual patients one type of imaging appearance may predominate, leading to the notion of subtypes of diffuse small vessel disease; either what has been labelled as ischaemic leukoaraiosis (defined by the combination of leukoaraiosis with a history of a clinical lacunar syndrome), or isolated lacunar infarction (in which a similar clinical presentation is accompanied by multiple lacunar lesions but no leukoaraiosis on imaging). Essentially this is the modern day imaging analogue of the distinction between diffuse white matter pathology and a state of multiple discrete lacunar infarcts, described by Pierre Marie in 1901 and christened état lacunaire. These two imaging types have recently been shown to differ in their risk factor profile; age and hypertension are most strongly associated with ischaemic leukoaraiosis while hypercholesterolaemia, diabetes mellitus and myocardial infarction are more associated with isolated lacunar infarction.16 These findings suggest some differences in pathogenesis, with leukoaraiosis, perhaps reflecting a non-atheromatous pathology of smaller calibre vessels than those implicated in lacunar infarcts. A better appreciation of distinct subtypes may explain conflicting results about the association of certain novel risk factors with stroke; for example, although homocysteine is a risk factor for both types of manifestation, it has a much stronger association with ischaemic leukoaraiosis than with isolated lacunar infarction.17


      Many patients with leukoaraiosis have a history of, or develop, lacunar stroke. But are there clinical features that can be linked specifically with leukoaraiosis? The evidence is most convincing for associations with cognitive impairment, gait disturbance and falls.

      Cognitive function

      For some time leukoaraiosis was considered by many to be a benign imaging appearance of no great consequence. Early cognitive studies gave conflicting results, at least in part because of inconsistency in the neuropsychological tests that were applied. For example, executive deficits were only patchily assessed in the early studies. However, it is now clear that leukoaraiosis is associated with deficits of information processing speed and executive function—by which we adapt to complex situations where more than one cognitive process is at play18, 19 (fig 3).

      Figure 3 The cognitive profile of ischaemic leukoaraiosis. Memory tests: Wechsler Memory Scale logical memory (WMS), paired associate learning (PAL). Executive tests: Digit Symbol (DSym), letter fluency (FAS), Trail Making part B (Trail), total errors (TE) and percentage perseverative errors (PE) on the Wisconsin Card Sorting Test. (Reproduced with permission from the BMJ Publishing Group.)

      The presence of leukoaraiosis also modifies the cognitive profile of Alzheimer’s disease. In mixed samples of patients (including criteria-defined Alzheimer’s disease, vascular dementia and mixed cases) leukoaraiosis is linked to performance on tests of executive function and processing speed across disease groups.

      Walking and falling

      As many as 80% of patients with leukoaraiosis have some degree of gait disorder.20 This association is independent of age, gender, previous stroke and hypertension.21 Furthermore, deterioration of gait is associated with progression of leukoaraiosis.22 The mechanisms underlying this gait disturbance remain unclear, although a correlation with frontal lobe atrophy,23 as well as with the white matter lesions, could imply an interruption of circuits involving the medial frontal lobe that are important for gait control.23, 24 What is clear is that gait disturbance is important. A number of studies support the link between gait disturbance and falls,22, 25 and leukoaraiosis is associated with a major increase in the risk of serious falls with fractures and hospitalisation25 (table 2). Therefore recognition of gait disturbance has implications for preventing falls and harm through falls, in part through initiating physiotherapy which has been shown to improve gait.26 A small pilot study with amantadine showed a trend for an improvement in gait steadiness, but there was no clear advantage over placebo,27 so pharmacological options remain limited but relatively unexplored.

      Table 2 Prognosis

      Other features

      Other features include bladder instability28 and mood disturbance; interestingly early mood disturbance is a cardinal feature of one of the genetic causes of leukoaraiosis, CADASIL. The recent LADIS study has emphasised the link between leukoaraiosis and both depression and disability in the elderly, and “hidden” functional impairment in the non-disabled.29 Some of the classical features of the early descriptions of “Binswanger’s disease”, such as pseudobulbar palsy, are rarely seen now—perhaps because better treatment of hypertension has modified the severity of the vascular pathology.12


      Assessment and investigation is guided to a large extent by the clinical context in which leukoaraiosis is found:

      • Stroke and transient ischaemic attack (TIA). The approach to clinical evaluation and investigation of acute stroke/TIA should be no different in patients found to have leukoaraiosis. The clinical relevance of emerging risk factors, such as serum homocysteine which is closely linked to this appearance, remains to be established. Carotid Doppler studies should be performed in recently symptomatic patients in this as in other subtypes of stroke, because patients with leukoaraiosis still benefit from endarterectomy if they have severe carotid stenosis.30

      • Cognitive symptoms. In the initial assessment of patients presenting with cognitive symptoms, and in cognitive screening following stroke, an appreciation of the cognitive profile and the limitations of the screening tools is important. Many standard screening tools such as the Mini-Mental State Examination (MMSE) are insensitive to executive deficits and are of limited value in these patients. Detection of executive deficits can be markedly improved by adding a few additional brief tests such as Digit Symbol, letter fluency, and Trail Making.31 An extended screening instrument like the Addenbrooke’s cognitive examination, which includes letter fluency, is also likely to perform much better than the MMSE alone.32 Clock-drawing tasks can be helpful in some patients (fig 4), although they are less sensitive than fluency or Trail Making tasks. Neuropsychological assessment can be very useful and should be carried out in younger patients, those with prominent cognitive difficulties and those who present with cognitive symptoms in the absence of previous stroke, where the diagnostic possibilities are wider. In particular, physicians should be alert to patterns that do not fit well with the usual profile. For example, early loss of memory for day-to-day events (autobiographical memory), difficulties with spatial navigation and a prominent episodic memory deficit on neuropsychological testing should prompt consideration of Alzheimer’s disease even if leukoaraiosis is the predominant imaging finding. Rapid progression of cognitive and gait difficulties, over weeks to a few months, should also prompt consideration of other causes such as normal pressure hydrocephalus, malignancy or inflammatory brain disorders.

      • Walking difficulties. In patients with prominent gait difficulties and leukoaraiosis the main aim of assessment is to determine whether these features are causally linked or merely coincidental. This partly depends on characterisation of the gait disturbance and careful exclusion of features that suggest parkinsonism or another diagnosis.24 Walking difficulties can be the main presenting feature, particularly when exacerbated by intercurrent illness, and may be a common manifestation of leukoaraiosis in patients who present “off legs” to acute medical services. Most leukoaraiosis patients with gait difficulties will have coexistent cognitive problems, which should be sought.

      • Incidental. The key question in this setting is the extent to which leukoaraiosis is disproportionate to age and risk factor profile. It is difficult to approach this question objectively, partly because of the difficulties in grading leukoaraiosis severity (see below). In older people, investigation of vascular risk will often be adequate, but in younger people a much wider differential diagnosis needs to be contemplated and investigations for metabolic and inflammatory causes considered.

      Figure 4 An example of bedside cognitive testing in a patient with ischaemic leukoaraiosis. This patient had a Mini-Mental State Examination of 29/30, but the addition of a clock-drawing task revealed a cognitive deficit. Top: Patient’s attempt to draw a clock following the instruction, “Draw a clock that shows the time as a quarter to two. Draw the hands and numbers on the face so that a child could read them.” The drawing shows poor strategy, sectoring, perseveration of the digit 4 and an inability to complete the task without help. Bottom: The patient was able to copy a clock face without difficulty, showing that the problem was not because of constructional, perceptual or motor deficits.


      When the term leukoaraiosis was introduced, only CT imaging was widely available. Similar appearances are conspicuous, and more florid on T2-weighted MRI, particularly on FLAIR images, and these also distinguish leukoaraiosis from lacunar infarcts, which appear as CSF intensity “holes” (fig 1).

      Leukoaraiosis severity has traditionally been graded by visual rating scales. Simple scales like that of van Swieten divide the appearances into only two grades of severity. Progressively more complex scales attempt to provide more comprehensive quantitation and localisation of lesions. For example, the Fazekas scale discriminates “punctate”, “early confluent” and “confluent” white matter lesions, while the Scheltens scale adopts a 0–6 scale in multiple anatomical regions. Common problems include comparability of CT and MRI ratings (although a “unified” CT/MRI scale has been proposed33), observer variability which increases with scale complexity, and ceiling effects compared with quantitative measurements. However, even fully quantitative volumetric measurements of leukoaraiosis correlate weakly with cognitive and physical function, suggesting that T2-weighted MRI provides only a rough impression of the severity of the underlying pathology. In general, rating of leukoaraiosis severity has yet to find a clear place in the clinical evaluation of these patients.

      Newer MRI sequences can provide useful additional information. Diffusion-weighted imaging, for example, allows the distinction of new lacunar infarcts from background leukoaraiosis. In patients who present with intracerebral haemorrhage, especially lobar haemorrhage, gradient echo (T2*-weighted) images should be performed to look for evidence of previous haemorrhages or microbleeds (fig 5). The presence of multiple cortical or juxtacortical haemorrhages in patients over 55 years accurately predicts amyloid angiopathy neuropathologically,34 which may be the main substrate of leukoaraiosis or be found alongside the small vessel arteriosclerotic changes. Microbleeds may be important in other groups and have prognostic implications in predicting haemorrhage risk, although a role in guiding treatment decisions is yet to be fully worked out (see below).

      Figure 5 MR images from a 71-year-old man with probable cerebral amyloid angiopathy. The FLAIR image shows leukoaraiosis (top). Multiple juxtacortical microhaemorrhages are visible only with gradient echo MRI (bottom and arrow heads). (Reproduced with permission from Massachusetts Medical Society from Greenberg SM, N Engl J Med 2006;354:1452.)

      In terms of quantifying white matter damage, several techniques currently used in research are proving superior to T2-weighted imaging and are likely to enter the clinical arena in the near future. Diffusion tensor MRI provides a much better index of white matter damage, and simple whole-brain measurements, such as diffusion histograms, can help track disease progression.35 Diffusion tensor MRI also demonstrates the variability in the extent of white matter disruption both within lesions and in normal-appearing white matter (fig 6). Measurement of brain atrophy from volumetric T1-weighted images is also potentially valuable in these patients, as it is in patients with cortical dementias.

      Figure 6 Diffusion tensor imaging in leukoaraiosis. Right: colour-coded mean diffusivity maps. Hot colours (yellow to red to black) indicate increasing mean diffusivity (water diffusion averaged in all directions, which increases as tissue is damaged, removing the myelin and axonal membranes that restrict diffusion). The variability of diffusion within areas of leukoaraiosis is evident. Whole-brain histograms of mean diffusivity provide a much better index of disease severity and progression than the degree of leukoaraiosis, seen on MR on the left. (Images courtesy of Bioimaging Inc, Lyon, France.)


      The precise mechanisms linking the pathology of the vessel walls with tissue injury have been the source of considerable controversy. The pathology changes in white matter have been most often interpreted as “incomplete infarction”—tissue injury consistent with ischaemia not severe enough to produce frank infarction—but this remains controversial. Attention has tended to focus on ischaemic mechanisms, but proving causality has been difficult.


      Reduced white matter cerebral blood flow in the context of leukoaraiosis has been demonstrated in a range of studies and with a wide variety of techniques. The main difficulty is the distinction between true hypoperfusion and the secondary reduction in cerebral blood flow (CBF) that results from reduced function (through intact vasoneuronal coupling); in other words, it is difficult to know whether reduced blood flow is a cause or consequence of tissue damage. One observation that hints at a causal role is that CBF, measured by MR perfusion imaging, is also reduced in white matter that appears normal on T2-weighted scans.36 However, the situation is more complex than it appears at first sight for two reasons. First, newer structural imaging techniques have shown that areas that appear normal to the naked eye on T2-weighted scans often harbour important alterations in white matter integrity.37 Second, even if white matter is entirely intact structurally, this does not necessarily mean that it is functioning normally, so the possibility of a secondary reduction in CBF, though smaller, still remains.

      Blood-brain barrier dysfunction?

      The leading non-ischaemic theory is that blood-brain barrier disruption leads to white matter damage, presumably through toxic effects of serum proteins.38 In patients with leukoaraiosis, extravasation of proteins that are usually confined to the plasma, such as IgG, complement and fibrinogen has been identified in white matter, in parallel with the pattern of leukoaraiosis.39 Also, there is a raised CSF:serum albumin ratio in patients with “vascular dementia” which correlates with the degree of leukoaraiosis, suggesting that the damaged white matter may be the site of blood-brain barrier leakage.40 More recently, careful MRI studies of tissue contrast uptake have demonstrated reduced blood-brain barrier integrity in vivo and that this relates to the severity of visible white matter lesions.41

      However, the development of new regions of leukoaraiosis in areas of abnormal perfusion or altered blood-brain permeability has not been shown in longitudinal studies. The core problem is that both mechanisms may occur in parallel, and without the ability to experimentally manipulate one or other, it will be difficult to disentangle their effects, or demonstrate causality.

      The endothelium?

      Ischaemic and blood-brain barrier theories are not necessarily mutually exclusive. Emerging evidence of endothelial dysfunction in leukoaraiosis provides a framework to unify these theories; alterations in blood flow and haemodynamics, and blood-brain barrier function, may both be part of a broader failure of endothelial function.42 Studies of circulating endothelial markers are beginning to hint at a specific pattern of endothelial function that is associated with leukoaraiosis. A pro-coagulant pattern has been found in ischaemic leukoaraiosis, which contrasts with the pattern linked with lacunar infarction.

      Leukoaraiosis and β amyloid?

      The association with a non-amyloid angiopathy at the pathological level might be seen as evidence that leukoaraiosis is simply coincidental in Alzheimer’s disease. However, recent evidence suggests a far more complex interaction. In Alzheimer’s disease, and in patients with mild cognitive impairment, some of whom will have prodromal Alzheimer’s, white matter changes correlate with serum levels of the Aβ1-40 peptide, which is the predominant peptide found in vessel deposits in cerebral amyloid angiopathy.43 But intriguingly, plasma Aβ also correlates with the severity of white matter lesions in the population-based Rotterdam study.44 Given that cerebral amyloid angiopathy is a rare cause of leukoaraiosis at the population level, this suggests that Aβ has an impact on white matter damage even in patients with arteriosclerotic small vessel disease and no pathological evidence of cerebral amyloid angiopathy. The underlying mechanisms of this association are not yet clear. Conversely, pathological studies in Alzheimer’s disease suggest that arteriosclerotic small vessel disease can drive more extensive amyloid deposition and neurofibrillary tangle formation,45 which is consistent with epidemiological evidence that vascular risk factors are important, and suggests that the interaction may work in both directions.


      In several prospective studies, leukoaraiosis has been associated with an increased risk of ischaemic stroke, cerebral haemorrhage, vascular death and all-cause mortality.27 Increased mortality occurs not just through stroke and vascular death but also via an increased risk of pneumonia and falls (table 2).

      Stroke and TIA

      In acute stroke or TIA, there is no evidence to suggest that a different approach to treatment should be taken in the presence of leukoaraiosis. Very recently, the implications of finding leukoaraiosis in patients eligible for thrombolysis have been evaluated in the Canadian Alteplase for Stroke Effectiveness Study (CASES). Although the risk of rt-PA related haemorrhage was higher the chances of a good outcome were not reduced.46 Patients with leukoaraiosis also benefit from carotid endarterectomy, despite higher perioperative risks.30 So in general, based on current evidence, the same approach should be taken except, perhaps, if anticoagulants are to be used (see below).

      Leukoaraiosis and anticoagulants

      The association of leukoaraiosis with treatment-associated haemorrhage is especially important as it may be partly preventable by carefully weighed treatment decisions. In the SPIRIT trial of oral anticoagulation following non-disabling cerebral ischaemia, patients in sinus rhythm received either anticoagulation with a target INR of 3–4.5, or aspirin. The trial was stopped early because of an unacceptable increase in cerebral haemorrhage, and leukoaraiosis was found to be a strong independent risk factor for this complication, with an odds ratio of 9.2.47 Furthermore, in patients with a previous ischaemic stroke, leukoaraiosis is a strong risk factor for subsequent symptomatic haemorrhage,48 either of lobar or deep subcortical type.

      It is not yet clear whether the risk of bleeding can be predicted by the pattern or severity of leukoaraiosis, or any associated imaging features. Severity is probably important. In SPIRIT, patients with “severe” leukoaraiosis had a haemorrhage risk 2.5 times that of “moderate” leukoaraiosis and a dose-response relation was also seen in the case-control study of intracerebral haemorrhage. However, all of the studies of bleeding risk are based on collapsing leukoaraiosis severity into only 1 or 2 grades (corresponding to the van Swieten scale) on CT assessment. The potential role of more detailed severity rating, based on MRI, or of additional MRI techniques such as the detection of microbleeds is unclear. There is at least one further source of uncertainty; the relation between leukoaraiosis and bleeding risk seems to vary in different clinical contexts. For example, there was a striking difference between intracranial haemorrhage risk in patients with “ischaemia of presumed arterial origin” (SPIRIT), and stroke in the context of atrial fibrillation. This almost certainly reflects pathological heterogeneity.47

      Given these major uncertainties, what recommendations can be made? Certainly, warfarin should not be used for recurrent stroke/TIA in patients with leukoaraiosis who are in sinus rhythm and have no other clear indication; high intensity anticoagulation (INR >3) is hazardous in these patients with an annual risk of major bleeding of 7% (about half of which will be intracranial and approximately a third fatal). Warfarin probably should be given to patients with previous stroke and atrial fibrillation, but with caution. The influence of leukoaraiosis on bleeding risk is less clear in this group47 but, nevertheless, cognitive aspects and the risk of falls should be carefully evaluated. Better methods of risk stratification are certainly needed. New imaging approaches that provide a deeper insight into the underlying pathology may help, but other aspects such as better evaluation of the risk of falls will also be important.

      Risk factors

      Currently, treatment of vascular risk factors is often all that physicians have at their disposal in patients who are found to have leukoaraiosis. But even here, there has long been a concern that blood pressure lowering could exacerbate white matter ischaemia because of impaired autoregulation. But so far clinical trials of anti-hypertensive therapy have not identified a subgroup of patients who do deteriorate in practice. In addition, in the PROGRESS study of blood pressure lowering with perindopril, the effect of treatment to slow the progression of white matter disease was greatest in the subgroup with leukoaraiosis at baseline.49 However, patients with severe leukoaraiosis are probably not well-represented in previous trials, and assessment of severity has been limited, so the existence of a subgroup vulnerable to hypotensive treatment remains possible.

      There is less evidence about how leukoaraiosis should influence treatment in other contexts such as cognitive impairment, or in asymptomatic older people. Many physicians would treat risk factors in these individuals, and observational studies lend some support to this strategy, but more data are needed.


      The concept of leukoaraiosis has now been with us for two decades and has generated controversy, debate and fruitful investigation. In the last decade, a number of the controversies have faded: for example, the association with cognitive function has been convincingly demonstrated, the deficits defined more clearly and the prognostic importance of leukoaraiosis has become clearer. Far from being an incidental and harmless consequence of ageing, leukoaraiosis is a clear predictor of risk of stroke, falls and, most importantly, treatment-related cerebral haemorrhage. In an ageing society, and with increasing use of imaging and potentially hazardous treatments, an understanding of the meaning and significance of this imaging appearance will be valuable to neurologists and physicians in general.

      Practice points

      • Do not be content to overlook a radiological report of “some small vessel disease [of uncertain significance]”. Ask yourself which small vessel disease, consider vascular risk factors and think about cognitive aspects.

      • Alzheimer’s disease and sporadic amyloid angiopathy are commonly associated with leukoaraiosis.

      • Cognitive deficits are difficult to detect with many screening tools (for example, the Mini-Mental State Examination).

      • Active treatments such as carotid endarterectomy and thrombolysis should not be withheld; patients with leukoaraiosis still often benefit from them in the right clinical context. But different counselling should be given, quoting higher treatment risks.

      • Leukoaraiosis is associated with an increased risk of intracerebral haemorrhage with warfarin therapy in some cohorts. Warfarin also needs careful consideration given the morbidity and mortality from falls.

      Otto Binswanger (1852–1929) was a Swiss psychiatrist and neurologist who came from a famous family of physicians. After receiving his medical degree in 1877, Binswanger spent the next few years as assistant to Theodor Meynert in Vienna and Karl Friedrich Otto Westphal at the Charité Hospital in Berlin. From 1882 until 1919 he was a professor and director of psychology at the University of Jena. One of his more famous patients was German philosopher Friedrich Nietzsche.

      Binswanger’s disease: a potted history

      In 1894, in a paper on the differential diagnosis of general paresis of the insane, Otto Binswanger linked marked atrophy of the cerebral white matter and arteriosclerosis to a syndrome of progressive decline of mental functions with depression and personality change, diminished power in the legs and tremor.i Diffuse white matter injury distinct from ischaemic infarcts had already been described by Durand-Fardel under the heading, atrophie interstitielle du cerveau, but Binswanger was probably the first to draw a causal link between such diffuse abnormalities and a progressive clinical syndrome. However, he did not provide microscopic descriptions and this lack of detail has led to perennial uncertainty over exactly what pathological entity he was describing. The autopsy description just comments on granular deposits on the dura mater at the skull base, minimal intracranial atherosclerosis, enlargement of the lateral ventricles and multiple ependymal thickenings. In 1962, while reviewing the original descriptions and noting that the patient had a history of syphilis, Olszewski suggested that the changes in the dura mater and ependyma made neurosyphilis a more likely diagnosis.ii

      For decades, Binswanger’s disease was recognised as an entity but thought to be rare. This perception changed with the advent of CT and initially leukoaraiosis on CT was assumed to reflect Binswanger’s disease; a critique of this view was one of the pillars of Hachinski’s original paper introducing the term leukoaraiosis.

      In more recent times, use of the eponymous term has declined while inconsistency has grown. One prevalent myth is that Binswanger’s disease can be diagnosed definitively at autopsy. In fact, the entity is not well-defined at pathological, clinical or radiological levels. Even if the pathology had been well defined there is evidence that the pathology may have changed with time and the full-blown syndrome of dementia, pseudobulbar palsy and gait disturbance is now rare, perhaps reflecting advances in the treatment of vascular risk factors. The lack of precise definition seriously undermines the value of the concept of Binswanger’s disease; one can have sympathy with the view that it is “neither Binswanger’s, nor a disease”.iii

      • i. Blass JP, Hoyer S, Nitsch R. A translation of Otto Binswanger’s article, ‘The delineation of the generalized progressive paralyses’ Arch Neurol 1991;48:961–72.

      • ii. Olszewski J Subcortical arteriosclerotic encephalopathy. World Neurol 1965;3:359–73.

      • iii. Hachinski V. Binswanger’s disease: neither Binswanger’s nor a disease. J Neurol Sci 1991;103:1.


      I am very grateful to Dr Tom Hughes, Dr Fergus Rugg-Gunn and Dr David Sharp for helpful comments on the manuscript, to Dr Alistair Lammie for helpful discussions of the pathology and for figure 2, and to Dr Anand Viswanathan and Dr Steven Greenberg for helpful comments on the link with cerebral amyloid angiopathy and figure 5. I am an Alexander von Humboldt fellow in Germany and supported by the Peel Medical Research Trust. This article was reviewed by Heine Mattle, Berne, Switzerland.