Article Text


What causes lacunar stroke?
  1. J M Wardlaw
  1. Correspondence to:
 Dr J M Wardlaw
 Division of Clinical Neurosciences, Western General Hospital, Crewe Rd, Edinburgh EH4 2XU UK;

Statistics from

A quarter of all ischaemic strokes (a fifth of all strokes) are lacunar type.1 Lacunar infarcts are small infarcts (2–20 mm in diameter) in the deep cerebral white matter, basal ganglia, or pons, presumed to result from the occlusion of a single small perforating artery supplying the subcortical areas of the brain.2 Although a recognised stroke subtype for over 50 years, the cause of lacunar ischaemic stroke,2 and whether it is different to cortical ischaemic stroke, remains under debate.3,4 Furthermore, lacunar stroke is not benign; 30% of patients are left dependent,5 and scant long term data suggest that up to 25% of patients have a second stroke within 5 years.6 Therefore, the prevention and treatment of this common stroke subtype may be less than ideal.


Several factors have hampered the study of lacunar stroke. Firstly, few patients die from lacunar stroke; if they do, death may occur long after the stroke, making autopsy material scant and difficult to interpret, and the small cerebral vessels require meticulous dissection. Studies of risk factors and causation have predominantly used a clinical diagnosis of stroke type, probably resulting in some misclassification. Although lacunar infarcts are associated with specific neurological syndromes, and most patients with a clinical lacunar syndrome have a small deep subcortical infarct on brain imaging (if visible), 10–20% actually have a recent small cortical infarct in a location that explains their stroke presentation.7 Similarly, 10–20% of patients with a clinical mild cortical stroke actually have a recent relevant lacunar infarct on imaging.7 Epidemiologically, these patients behave more like the lesion type on imaging than the clinical syndrome of the lesion they actually have.7 Many studies have used an inappropriate control group; age matched normal controls can only indicate whether or not a particular risk factor is associated with stroke, whereas identification of associations specific to lacunar stroke requires a control group with a different type of ischaemic stroke. Finally, some classifications, such as the Trial of Org 10172 in Acute Stroke Treatment (TOAST) method8 actually use risk factors (such as embolic sources or hypertension) to determine the stroke type, thus potentially biasing studies of differences in risk factors. Hence, inadvertent misdiagnosis of lacunar as cortical stroke, and vice versa, the paucity of pathological material, and bias in some clinical classification systems may have confounded previous pathology, prognosis, and risk factor studies.


The lacunar hypothesis supports the concept that lacunar ischaemic stroke is due to an intrinsic cerebral small arteriolar abnormality,2 in contrast to cortical ischaemic stroke, which is commonly due to embolism from the heart or large arteries. Although some studies suggest that many lacunar strokes are caused by emboli, and while it is perfectly possible for a small embolus to enter and occlude a lenticulostriate artery,9 a systematic review of risk factors including only studies using subtype definitions for ischaemic stroke free of risk factors found that atrial fibrillation and carotid stenosis were associated more with non-lacunar than lacunar infarction (relative risk (RR) of lacunar versus non-lacunar infarction: atrial fibrillation 0.51, 95% confidence interval (CI) 0.42 to 0.62; ipsilateral carotid stenosis 0.35, 95% CI 0.28 to 0.44).10 Indeed, common causes of large artery (cortical) infarction, such as emboli from the large arteries or heart,11–13 or intracranial large artery atheromatous stenoses,12 appear unlikely to cause more than 10–15% of lacunar strokes.14–17 Perhaps lacunar infarcts due to emboli or middle cerebral artery stenoses are recognisable by being larger than non-embolic/stenotic lacunes,12,18 possibly because the embolus/stenosis occluded the origin of several lenticulostriate arterioles simultaneously. There is a suggestion that lacunar stroke secondary to middle cerebral artery stenosis may be more common in south Asian populations than in western white populations, but this remains to be clarified. Only 6% of all particles (even the smallest) injected into the carotid arteries in an experimental model ended up in the lenticulostriate arteries, the rest going to the cortical arteries or their cortical branches.9 Studies that suggested stronger associations between lacunar stroke and emboli cited carotid stenoses as mild as 25%16 or cardiac abnormalities not clearly associated with embolism (for example, left ventricular hypertrophy),13 or they had no, or an inappropriate control group. It would certainly be useful to be able to infer the likely underlying mechanism of lacunar ischaemic stroke (that is, to identify the 10–15% of embolic/stenotic versus other intrinsic small vessel strokes) from the appearance of the brain lesion, as that might help target effective secondary prevention regimens, but much more work is required to see whether different patterns actually exist, before determining how closely these relate to the underlying mechanism.

Hypertension and diabetes are said to be strongly associated with lacunar ischaemic stroke. However, in studies using risk factor free ischaemic stroke subtype definitions, there was only a marginal excess of hypertension with lacunar versus non-lacunar infarction (RR 1.11; 95% CI 1.04 to 1.19) and no difference for diabetes (RR 0.95; 95% CI 0.83 to 1.09).10 Nor was there clear evidence of any association between smoking, prior transient ischaemic attack, excess alcohol consumption, or raised cholesterol in lacunar compared with non-lacunar infarction.10


After lacunar ischaemic stroke, a recurrence is more likely to be lacunar then cortical; 47% of recurrences after a lacunar stroke were lacunar compared with 15% of recurrences after a non-lacunar stroke).19 If lacunar and cortical ischaemic strokes were due to the same mechanisms, then recurrent strokes would be equally likely to be cortical after a lacunar stroke as lacunar, which appears not to be the case.

Lacunar ischaemic stroke also appears to be more closely associated with white matter lesions (WMLs) than does cortical ischaemic stroke. WMLs are abnormal areas of hypodensity (on computed tomography scans) or hyperintensity (on T2 weighted magnetic resonance imaging (MRI)) in the deep hemispheric and periventricular white matter and brain stem.20 They are in turn associated with cognitive decline,21 and increased risk of future stroke, particularly lacunar type.22 WMLs also progress rapidly after lacunar stroke.23 After lacunar stroke, 15–20% develop dementia (more than might reasonably be attributed to the amount of brain damage caused by a lacunar infarct).6,21,24 After lacunar ischaemic stroke, new “silent lacunar infarcts” occur on follow up imaging.25 Asymptomatic small deep white matter infarcts, in addition to the symptomatic lesion, have been seen on MR diffusion imaging at presentation with lacunar ischaemic stroke.26 The imaging appearance of these asymptomatic lacunar infarcts suggests that they are slightly older than the presenting lesion. Cerebral microhaemorrhages, which are tiny, apparently asymptomatic bleeds (or “leaks”) in the brain are also associated with lacunar stroke and WMLs.27 These observations suggest that most lacunar strokes are the clinically focal manifestation of what is actually a diffuse abnormality of the small cerebral arterioles, which, if extensive enough, can also manifest clinically as cognitive decline and dementia.28


Detailed pathology studies in the 1950s identified abnormal small deep perforating (lenticulostriate) arteries resulting in lacunar infarcts,29 termed segmental arterial wall disorganisation (since called lipohyalinosis or fibrinoid necrosis). In lipohyalinosis, the vessel wall appears thickened, with focal dilation, and eventually leads to disintegration of the wall with an infarct around it. This arteriolar abnormality remains the most commonly described defect to date.30,31 However, there is debate about the pathology and its relationship to symptoms; many studies did not ascertain that the lesion seen at autopsy was symptomatic, some studied multiple small “holes” in the brain regardless of symptoms, and many lesions came from few patients.32 Lipohyalinosis is found in areas of the brain corresponding with WMLs on imaging.33

The nature of the intrinsic arteriolar abnormality remains unresolved. It may be microatheroma, poor cerebral blood flow, or vasospasm. It is difficult to see why atheroma would affect the small arterioles when there is a lack of association with the better understood large artery atheroma in lacunar infarction,10,14 Vasospasm, inducible in animal models with extreme hypertension, causes fibrinoid necrosis, but the association of lacunar infarction with any excess of hypertension (more than other types of ischaemic stroke) in patients is weak (RR 1.11; 95% CI 1.04 to 1.19).10 Thickened vessels may narrow or occlude the lumen, leading to ischaemia, but relatively few truly occluded vessels have been seen pathologically.29,30 Thickened vessel walls may be stiff and impair autoregulation, and indeed patients with WMLs may have impaired autoregulation.34 Some studies found reduced cerebral blood flow (CBF) in patients with WMLs,35 but others did not.36 CBF is difficult to quantify: “reduced” CBF may be artefactual,37 or simply the consequence of a reduction in normal tissue to supply. None of this explains what starts the vessel abnormality, only the features or behaviour of the vessels and the possible mechanisms for damaging the brain, once the abnormality is established.


The underlying abnormal lenticulostriate artery can be seen in about 10% of lacunar infarcts on detailed MRI scans.38 Although this appearance could be an intra-luminal thrombus (a small arteriolar “hyperdense middle cerebral artery sign” equivalent), other features suggest that blood products had passed into the vessel wall and perivascular space. The infarcts appear to be around, rather than at the end of the abnormal segment. Lesions that resemble an “incomplete” lacunar infarct (perivascular oedema related lesions) at autopsy39 suggest that the “infarct” was actually due to oedema fluid leaking from the arteriole and damaging adjacent tissue.

Increased “leakage” of intravenously injected MR contrast agent across the blood–brain barrier have been found on detailed MR imaging in patients with WMLs.40 There are numerous other examples in a rather scattered literature pointing to “leaky” small arterioles predisposing to WMLs and lacunar ischaemic stroke, of which the following are but a few. Extravasated plasma proteins have been found at autopsy in WMLs in patients with ischaemic cerebrovascular disease in life.41 Discrete areas of extravasated plasma proteins have been seen around small cerebral arterioles in hypertensive primates.42

These combined observations suggest that the initiating step for most lacunar ischaemic stroke and WMLs may be failure of the arteriolar endothelium (that is, the blood–brain barrier) allowing extravasation of blood components into the vessel wall, and consequently vessel wall, perivascular neuronal, and glial cell damage.28,39,41,42 This would explain the features described histologically,29,31,32 and the association with microhaemorrhages (small blood leaks).27

Is there other evidence of endothelial failure? Patients with isolated lacunar infarction, or lacunar infarction plus WMLs, have elevated systemic plasma markers of endothelial activation (plasma intercellular adhesion molecule 1, thrombomodulin, and tissue factor pathway inhibitor) compared with age matched normal controls.43 However, in the absence of non-lacunar controls, it is unclear if these patterns are specific to lacunar stroke. Several studies have identified retinal microvascular abnormalities that were associated with increased risk of stroke, cognitive impairment, and white matter lesions. Those most closely related to these cerebral abnormalities (microaneurysms, retinal haemorrhages, and soft exudates) were most commonly seen when there was breakdown of the blood–retinal barrier.45 Unfortunately most of these studies did not examine stroke subtypes, so more work is required.


The mechanism underlying, and the long term consequences of, lacunar ischaemic stroke, the cause of a quarter of all ischaemic strokes, are poorly understood, but are not benign. Evidence supports the hypothesis that, in most lacunar strokes, the vascular abnormality is pathologically diffuse, even if the clinical manifestations are focal, and result from small vessel endothelial damage, subtle increase in blood–brain barrier permeability, and leakage of substances toxic to the brain into the perivascular tissue. As originally proposed in the lacunar hypothesis, only a small proportion of lacunar stroke appears to result from artery to artery or cardiogenic emboli or intracranial large artery stenoses. These latter embolic/stenotic lesions may be recognisable by their size (being larger).

It might be questioned how this presumably gradual process could produce a sudden lacunar infarct. In fact, lacunar stroke symptoms (more than other stroke subtypes) not uncommonly progress after onset.44 Perhaps the interstitial fluid composition eventually reaches a critically abnormal point where the axons cease to transmit signals, or the vessel wall thickening narrows the lumen and reduces blood flow, or fails to allow nutrients out and waste products of metabolism back into the circulation. This diffuse endothelial failure might also account for additional asymptomatic lacunar infarcts observed at presentation with,26 or during follow up after,25 a symptomatic lacunar infarct with no underlying embolic source. These asymptomatic lesions, gradually multiplying, could be the source of the WMLs. Neurones can switch off suddenly, even when the underlying process is gradual; for example, neurones switch off electrical activity (manifest as a stroke) as cerebral blood flow falls below about 25 ml/100 g brain/min, thus the fall in blood flow may have been gradual but the symptoms are sudden.

An important target for new therapeutic approaches to prevent or treat a common form of ischaemic stroke and cognitive decline may have been overlooked. It is important to understand this process better, not simply in order to reduce the burden of lacunar stroke, but because the same mechanism may also underlie WMLs and consequent cognitive decline to dementia. Lacunar ischaemic stroke has for too long been simply lumped together with other stroke subtypes; while it is likely that many older stroke patients will share common vascular risk factors, this tendency has hindered understanding of what appears to be an importantly different stroke subtype that may require different acute treatment and prevention. For example, the association with microhaemorrhages, a possible risk factor for intracerebral haemorrhage, is worrying, and may increase the risk of haemorrhage complicating anti-thrombotic drug treatment compared with their use in cortical ischaemic stroke. Further studies, using detailed, accurate, and unbiased patient classification, are needed to examine risk factors for lacunar stroke including blood–brain barrier function (now possible with detailed MRI),40 how to identify those lacunar strokes that are due to embolism, and clarify long term outcomes. Study of small arteriolar abnormalities in other vascular beds (for example the retina, where the arterioles can be directly observed) and systemic endothelial abnormalities will help both in understanding the mechanisms of lacunar ischaemic stroke and in identifying diagnostic and prognostic markers.


I would like to acknowledge the many people who have discussed with me the possible mechanisms of lacunar ischaemic stroke or authored some of the supporting papers, in particular Professor C Warlow, Professor M Dennis, Dr C Sudlow, Dr A Lammie, Dr J Bamford, Professor P Sandercock, Dr P Armitage, and Dr G Mead.


View Abstract


  • Competing interests: none declared

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.