Handbook of Practical Medicine Stroke

9781405127660_4_005.qxd 10/13/07 11:05 AM Page 239 5.7 Differentiating haemorrhagic transformation of an infarct from intracerebral haemorrhage 239 lesion, in the second and third weeks after the stroke. Frequency, risk factors and clinical relevance of These areas also enhance markedly with X-ray and are haemorrhagic transformation of an infarct thought to correspond with areas where the capillaries are leaky, where there is blood–brain barrier break- Our systematic review of CT and MR scanning indicated down, and where there is frank petechial haemorrhage at that some degree of petechial haemorrhage occurs in postmortem. 53 15–45% of patients and of symptomatic haematoma formation in 2.5–5%, at some point within 1–2 weeks after stroke. 374 This may be an under- or overestimate for The influence of observer variability and visual perception the following reasons: Some of the variability in the reported frequency of • the studies were generally small and few were of petechial haemorrhage must be a result of inter-observer consecutive patients or prospective; variation, but this is less likely to apply to focal hae- • not all patients were followed up, only survivors or matomas. 62,374,375 The visual perception of the density of those who remained in hospital for the study period; normal brain is influenced by the density of adjacent • the definition of HTI was not stated in all the tissue; normal brain next to the low density of an infarct publications; looks of higher density than it really is, and so can be • the influence of inter-observer variability was not mistaken for areas of haemorrhage (Fig. 5.56). To avoid taken into account; this mistake, the density of the brain can be measured on • the generation of scanner used varied and hence the the CT console to distinguish petechial haemorrhage sensitivity of the diagnosis of HTI; from normal brain. There are no data on observer vari- • the number of patients given antithrombotic drugs ability in detection of HTI on MR. was often not stated (the frequency of HTI gener- ally increased with the amount of antithrombotic or thrombolytic drug given). There was limited information suggesting that the major risk factors for HTI were large infarcts and increas- ing doses of antithrombotic or thrombolytic drugs, but we were not able to find an association with cardio- embolic stroke as has been previously suggested. We did not find any association with raised blood pressure. The Multicentre rt-PA Acute Stroke Survey (MASS, an observational study of patients receiving open label thrombolysis) found associations between increased patient age, stroke severity, plasma glucose and falling platelet count with asymptomatic and symptomatic HTI. 376 While minor degrees of HTI are regarded as asym- ptomatic, there is some evidence that any degree of bleeding inside the head may be associated with a worse outcome. 377,378 A combined analysis of all trials of Fig. 5.56 CT brain scan to illustrate the effect of altered brain plasminogen actvator showed that any asymptomatic density on visual perception. The scan was from a 40-year-old intracranial haemorrhage reduced the probability of a man at 24 h after a left middle cerebral artery occlusion causing good outcome by 30% (95% CI 60% reduction to 12% an extensive left hemispheric infarct (thin arrows). The areas of increase) – although not statistically significant, all hyperattenuation (thick arrows) within the low density of the the studies were underpowered to detect a statistically infarct were interpreted as being caused by haemorrhage by the significant effect for which a trial of more than 4000 clinician because of their apparent brightness. However, the 378 patients would be needed. actual density when measured on the CT scanning console was the same as normal grey matter, indicating that the areas of hyperattenuation were in fact islands of surviving, i.e. non- Possible mechanisms of haemorrhagic transformation of infarcted, brain and not haemorrhage at all. Area 1 (normal an infarct right insular cortex) was 51 Hounsfield units (HU: unit for measurement of density on CT) and area 2 (apparent increased Traditionally, HTI was considered to occur when an density within the left infarct) was 46 HU. Blood generally arterial occlusion, usually embolic from the heart, registers at around 70–80 HU. resulted in ischaemia of the distal capillary bed and then, .. ..

9781405127660_4_005.qxd 10/13/07 11:05 AM Page 240 240 Chapter 5 What pathological type of stroke is it, cerebral ischaemia or haemorrhage? with fragmentation of the embolus, this ischaemic area The idea that embolism is the cause of HTI has become was resubjected to arterial pressure, resulting in rupture rather entrenched. However, closer examination of the of the necrotic arterioles and capillaries. 379 This notion literature shows that the situation is more complic- arose from work with postmortem brains, but is clearly ated than the reperfusion–haemorrhagic transformation biased towards patients who die in the early stages of hypothesis would suggest. 380 Shortly before Fisher and their stroke and who are therefore more likely to have Adam’s (1951) 279 work became so widely publicized and had a large cerebral infarct. The signs of haemorrhage accepted, Globus and Epstein 381 published the results resolve with time so that, in patients dying weeks or of experimental cerebral infarction in monkeys and months after their stroke, it may no longer be possible dogs and some observations on postmortem brains to distinguish the relative contribution of infarct and from stroke patients. They observed that haemorrhage haemorrhage in the residual lesion. into infarcted tissue was often worse when the occluded (c) (a) (b) (d) (e) Fig. 5.57 Unenhanced CT brain scan at 2.5 h after onset deteriorated with worsening of the left hemiparesis and of left hemiparesis. (a) The right middle cerebral artery (MCA) drowsiness, and repeat unenhanced CT showed extensive main stem is hyperattenuated (arrow) and (b) there is early haemorrhagic transformation of the right MCA territory infarct parenchymal hypoattenuation in the right insular and with blood in the right lateral ventricle (d,e), and (d) persistent posterior frontal cortex consistent with an early ischaemic right MCA hyperattenuation (occlusion, arrow). There was also stroke. (c) CT angiography confirmed the occluded right MCA remote haemorrhage in the right cerebellum. In other words, (arrow). Intravenous thrombolysis was given immediately after haemorrhagic transformation is not simply a consequence the CT angiogram. (d,e) Eighteen hours later the patient of recanalization. .. ..