Epidemiology of Ischaemic Stroke Subtypes Do Differences in Epidemiology Provide Evidence for a Distinct Lacunar Arterial Pathology

Background Lacunar ischaemic stroke accounts for around one quarter of all strokes, and is presumed to result from the occlusion of a single perforating artery supplying the deep subcortical areas of the brain. The underlying arterial pathology is poorly understood, but is thought to differ from the atherothrombotic processes that occlude larger intra- and extracranial arteries causing most other ischaemic stroke subtypes. Progress in understanding the aetiology of lacunar stroke has been limited by the lack of informative autopsy studies, and the difficulties in studying small blood vessels using brain imaging. One alternative approach is to compare the epidemiology of ischaemic stroke subtypes, since differences in the epidemiology may reflect and inform about different underlying pathologies. Methods I performed two systematic literature reviews to identify studies presenting data on (1) the risk factors for, and (2) the outcome of, different ischaemic stroke subtypes. I extracted relevant data from included studies and performed a series of meta-analyses comparing risk factor profiles, and risks of death, recurrent stroke and myocardial infarction (MI) in patients with lacunar versus non-lacunar ischaemic stroke. To address some of the unanswered questions and controversies surrounding the causes of ischaemic stroke we set up the Edinburgh Stroke Study (ESS), which I co-ordinated. We recruited patients with stroke and transient ischaemic attack seen at our hospital between 2002 and 2005, and followed them for 1-4 years for death, recurrent stroke and MI. To overcome the methodological limitations of the studies included in my reviews and of my meta-analyses, I carried out a large collaborative individual patient data analysis in which I combined data from five stroke registries - including the ESS - that had used similar robust methodology, and performed a series of analyses comparing the risk factor profiles of patients with lacunar versus nonlacunar ischaemic stroke. In an updated meta-analysis, I combined this data with existing published studies that had used an unbiased method of classifying ischaemic stroke subtypes. Using the ESS data, I compared the risks of recurrent stroke and MI, and patterns of recurrent stroke subtypes in patients with lacunar versus nonlacunar stroke. Results In my systematic review of risk factors I found evidence of classification bias in many studies, where systematic error was introduced through the use of classification methods that included risk factors in the definitions of stroke subtypes. This led to overestimation of some risk factor-stroke subtype associations and, in particular, to apparently stronger associations between hypertension and diabetes and lacunar compared with non-lacunar ischaemic stroke. When I included only unbiased studies, I found a significantly reduced prevalence of atrial fibrillation (AF) and severe carotid stenosis and a trend towards a reduced prevalence of ischaemic heart disease (IHD) in lacunar patients. I found a very slight excess of hypertension among lacunar patients, but no difference in the prevalence of diabetes, or any other risk factor studied. In my collaborative individual patient data analysis, I confirmed a significantly lower prevalence of severe carotid stenosis, AF and previous IHD in patients with lacunar ischaemic stroke, but found no difference in the prevalence of hypertension, diabetes, or any other risk factor studied, even after adjusting for confounding factors. These results were largely confirmed in my updated metaanalysis, although there was a slight excess of hypertension among lacunar compared with non-lacunar ischaemic strokes. In my systematic review of outcome after lacunar versus non-lacunar ischaemic stroke, I found a lower risk of death following lacunar compared with non-lacunar stroke which attenuated but persisted long-term; a higher recurrent stroke risk in non-lacunar patients during the first month only; and limited data on recurrent stroke subtypes suggesting that ischaemic stroke subtypes may breed true to type. Data on MI risk were extremely sparse. My analyses of data from the ESS showed no difference overall in risk of recurrent stroke between patients with lacunar versus non-lacunar ischaemic stroke, but some evidence for a lower very early recurrence risk among lacunar patients. There was evidence that recurrent stroke subtypes breed true, since patients with a lacunar stroke at baseline were much more likely to have a lacunar than a non-lacunar recurrence. We identified five times as many MI events following stroke than have been previously reported in the published literature, and found a non-significantly reduced risk of MI in patients with lacunar compared with non-lacunar ischaemic stroke. Conclusions My comparisons of the epidemiology of lacunar versus non-lacunar ischaemic stroke subtypes revealed differences in the risk factor profiles and risks of recurrent stroke and myocardial infarction which suggest that a distinct, nonatherothrombotic arteriopathy underlies many lacunar ischaemic strokes. My analyses of recurrent stroke subtype patterns suggest that recurrent ischaemic strokes subtypes tend to breed true, providing further support for a distinct lacunar arteriopathy. Contrary to widespread belief, hypertension and diabetes do not appear to be more important in the aetiology of lacunar stroke than in other types of ischaemic stroke. These findings support other lines of evidence for a distinct lacunar arteriopathy, and highlight the need for further research into the aetiology of lacunar ischaemic stroke.