Handbook of Practical Medicine Stroke

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 645 12.1 Pathophysiology of acute ischaemic stroke 645 process, characterized by an evolving zone of bioener- photon emission tomography (SPECT). 12,13,19,34,39,58,59 3 getic upheaval. This penumbral concept provides an The parameters they map, the criteria they use to important insight into the nature of tissue survival after identify the penumbra and their main advantages and stroke and an important target for therapy. Changes limitations are listed in Table 12.2. 19 in water homeostasis, including shrinkage of the extra- cellular space and net uptake of water from plasma, also Positron emission tomography (PET) take place in the penumbra. Originally, PET measurements were undertaken with As cerebral blood flow falls from its normal mean, the oxygen-15 steady-state technique. This quantitative around 50 mL/100 g per min, to less than 20 mL/100 g imaging maps the main physiological variables involved per min, a critical threshold is reached when the in tissue ischaemia, namely perfusion (or CBF), and oxy- electrical activity of neurones is suppressed and gen consumption (CMRo ) together with oxygen extrac- 2 neuronal function is impaired. As flow falls further, tion ration (OEF) and cerebral blood volume (CBV). 3,11,45 another threshold is reached when cellular integrity However, because these determinations of flow and begins to break down. Cells falling in between these energy metabolism required arterial blood sampling two thresholds make up the ‘ischaemic penumbra’: and complex logistics, they were difficult to perform they may not be functioning, but they are still viable in patients with acute stroke and impossible when and could either recover function, with early planning thrombolytic therapy. This led to the devel- reperfusion or early interruption of the adverse opment of noninvasive measurement of the extent of metabolic or neurochemical cascade, or die. Survival the penumbra by means or tracers (that indicate tissue of the penumbra is the main determinant of clinical integrity or hypoxia) combined with semiquantitative recovery. measures of perfusion. 11 Because both the core and the penumbra can [ C]Flumazenil (FMZ) is a central benzodiazepine contribute to neurological deficits, it is not possible receptor ligand labelled with C-11. It detects neuronal to clinically determine their relative effects; imaging damage in the cortex in the first few hours after is required. ischaemic stroke and is therefore a marker of tissue integrity (but not so much in the white matter and basal ganglia which have low concentrations of benzodiaze- Imaging the ischaemic penumbra 60 pine receptors). 18 Imaging studies, with various techniques, have established [ F]fluoromisonidazole (FMISO) is another tracer and the clinical importance of penumbral salvage, showing it indicates hypoxic but viable tissue, and can also be a clear association between the volume of penumbra used to detect the ischaemic penumbra when the results not progressing to infarction and the improvement in are directly calibrated by conventional PET measure- neurological scores of impairments and function. 57 To ments. 57 However, because reliable detection of FMISO be accepted as penumbra, the affected tissue must fulfil uptake is delayed (at least 2 h between tracer injection several well-defined operation criteria (Table 12.1). The and imaging) its value is limited in guiding treatment main methods of imaging the ischaemic penumbra decisions in acute ischaemic stroke. in vivo are by positron emission tomography (PET), mag- Based on validated thresholds, PET can classify affected netic resonance imaging (MRI), CT perfusion and single- tissue as core, penumbra, oligaemia and hyperper- fused. 3,19,60 Oligaemia displays mild misery perfusion – i.e. a moderately reduced cerebral blood flow (above the penumbra threshold) and a high oxygen extraction Table 12.1 Currently accepted operational criteria defining the ischaemic penumbra. 19 fraction. A high oxygen extraction fraction in itself does not equate with penumbra as this fraction rises as soon Hypoperfusion < 20 mL/100 g brain per min as CBF decreases. Early spontaneous hyperperfusion, Abnormal neuronal function documented by a correlation present in about one-third of patients within 18 h with acute clinical deficit of onset of ischaemic stroke, almost invariably predicts Physiological and/or biochemical characteristics consistent preserved cerebral oxygen metabolism (CMRo ) and 2 with cellular dysfunction but not death tissue integrity, suggesting this as a marker of prior Uncertain fate (i.e. ischaemic brain tissue may die or recover recanalization that salvages the penumbra. 19 depending on timing and degree of reperfusion) Although PET is arguably still the ‘gold standard’ for Salvage of this tissue is correlated with better clinical 59 measuring the ischaemic penumbra, particularly given recovery its superior quantification, it is limited by the expensive .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 646 646 Chapter 12 Specific treatments for acute ischaemic stroke Table 12.2 Main physiological imaging techniques, with the parameters they map, the criteria they use to identify the penumbra, and their main advantages and limitations. 19 Technique Parameters Definition of penumbra Advantages Limitations CT Perfusion CBF Relative CBF < 66%* Combined with Limited brain coverage, insensitive in CBV CBV > 2.5 mL/ plain CT, posterior circulation, indirect MTT 100 g brain available, fast visualization of core, iodinated contrast Xenon (Xe) CBF CBF 7–20 mL/ Quantitative, Not fully validated, provides CBF only, 100 g brain combined with pharmacological effects of Xe plain CT MRI DWI-PWI CBF Relative TPP Fast, best sensitivity, Limited availability, mismatch concept CBV (or MTT) delay no radiation, not validated, CBF values not accurate, MTT > 4s* and normal directly visualizes patient’s cooperation required, TTP DWI core frequent contraindications ADC Arterial spin CBF Not validated No contrast needed Provides CBF only, poor sensitivity to labelling low flows MRI-based CBF Not validated Non-invasive Validity unclear (and PWI) OEF mapping of OEF CMRo 2 and CMRo 2 Spectroscopy NAA Elevated lactate and Biochemically Not validated, poor resolution Lactate normal NAA characterizes tissue PET 15 Multi-tracer O CBF CBF 7–22 mL/ Quantitative, validated Complex, time-consuming, not widely CBV 100 g brain/min available, expensive MTT and OEF > 0.70 CMRo 2 OEF 11 [ C]FMZ Tracer binding Relative binding Based on physiological As above and only suitable for cortex, (+ H 2 15 O) > 3.4** and CBV neuronal integrity basis not validated < 14 mL/ 100 g brain/min 18 [ F]FMISO Tracer uptake Uptake ratio > 1.3* Produces a direct positive As above and validation incomplete, image of viable long imaging time tissue hypoxia SPECT [ 99m Tc]-labelled CBF Relative CBF < 65%* Cheap and relatively Provides perfusion only so thresholds HMPAO or ECD available uncertain, limited spatial resolution, slow brain kinetics *relative to mean contralateral hemisphere; **relative to contralateral healthy white matter; ADC, apparent diffusion coefficient; CBF, cerebral blood flow; CBV, cerebral blood volume; DWI, diffusion-weighted imaging; ECD, ethyl-cysteinate dimer; FMISO, fluoromisonidazole; FMZ, flumazenil; HMPAO, hexamethylpropyleneamine oxime; MTT, mean transit time; NAA, N-acetylaspartate; OEF, oxygen extraction fraction; PWI, perfusion-weighted imaging; SPECT, single photon emission computed tomography; TTP, time to peak. equipment and the complex logistics of a multidiscip- MRI DWI-PWI linary team. A more widely applicable method is the magnetic reson- PET is able to differentiate between normal, penumbral, ance imaging (MRI) technique of combined diffusion- and infarcted tissue in the acute stage of ischaemic weighted imaging (DWI) and perfusion-weighted imaging stroke. However, because of its logistic and practical (PWI) as a measure of tissue integrity and cerebral limitations, its clinical role in acute stroke is limited. 12,13,17,19,61–63 perfusion respectively (section 5.5.2). .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 647 12.1 Pathophysiology of acute ischaemic stroke 647 The apparent diffusion coefficient (ADC) on DWI is an core). 17,19,69 Typically, a PWI > DWI lesion is associated ‘apparent’ or ‘relative’ measure of the diffusion of free with subsequent infarct enlargement but, because PWI is water in one area compared with other areas during the very sensitive in detecting perfusion defects, the PWI/ DWI sequence. 61 It is decreased in areas of ischaemic DWI mismatch region may comprise not only tissue at brain which have a critically low CBF. 64,65 A decrease in risk but also hypoperfused tissue with CBF values above the ADC is seen as an area of hyperdensity on diffusion- the critical viability thresholds. Further, the reliability weighted images (due to a restriction in the diffusional of assessing the degree of diffusion-perfusion mismatch movement of water), as soon as 11 min after stroke. 66 is not established. 70 Because DWI abnormalities typically evolve into infarc- The diffusion-perfusion mismatch concept of the tion without reperfusion or neuroprotection. 62 It has ischaemic penumbra (hypoperfusion volume > DWI been suggested that the DWI abnormality corresponds lesion) has now been modified with the recognition that to the ischaemic core, but this is not always the case. 12,63,67 a portion of hypoperfused brain reflects benign oli- PWI, on the other hand, provides information on gaemia, and that the so-called DWI core may in fact the haemodynamic status of the tissue by using para- be partially salvageable. The validity and reliability of magnetic contrast agents, for example, gadolinium-based DWI and PWI evidence of mismatch as a marker of the chelates. On the basis of magnetic resonance perfusion ischaemic penumbra, as compared with PET, has been imaging data, maps of relative CBF can be calculated, supported in recent small studies. 63 However, its validity and these demonstrate impaired perfusion of both the in predicting lesion expansion and response to therapy ischaemic core and the surrounding brain regions, is uncertain and is being evaluated in ongoing stud- thereby complementing the information derived from ies. 12,17,19,71–73 Table 12.3 summarizes the evidence base DWI. 68 During the first few hours of stroke evolution, for the DWI-PWI mismatch hypothesis. 19 PWI typically demonstrates regions with abnormal T2-based blood oxygen level-dependent (BOLD) MR perfusion that are larger than the DWI abnormalities. imaging can visualize deoxyhaemoglobin and may serve It has been postulated that this mismatch reflects the as a valid non-invasive indicator of the oxygen extrac- ischaemic penumbra (i.e. functionally impaired ‘tissue at tion fraction and thus the metabolic state of threatened risk’ surrounding the irreversibly damaged ischaemic brain tissue. 39 Table 12.3 Validity of the DWI-PWI Operational criteria DWI-PWI mismatch mismatch hypothesis to identify the ischaemic penumbra. 19 (see Table 12.1) Hypoperfusion < 20 mL/ Absolute CBF values unreliable with PWI; but 100 g brain per min TTP > 4 s delay shown to correspond to PET-derived CBF < 20 mL/100 g per min Abnormal neuronal function Volume of DWI lesion and of mismatch documented by a correlation significantly correlate with acute-stage stroke with acute clinical deficit severity Physiological and/or In humans, acutely the ADC starts to decline biochemical characteristics for CBF consistent with cellular dysfunction but not death, values corresponding to the penumbra threshold. ADC declines represent cytotoxic oedema and consistently correspond to reduced CMRO in humans. In 2 the rat, ADC declines shown to correspond with impaired aerobic glycoslysis and ATP production Uncertain fate Without early reperfusion, tissue-at-risk progresses to full infarction Salvage of this tissue is correlated Salvage of tissue-at-risk transfers to better with better clinical recovery outcome and correlates with clinical recovery ADC, apparent diffusion coefficient; CBF, cerebral blood flow; CMRo , cerebral 2 metabolic rate of oxygen; NIHSS, National Institute of Health Stroke Scale; TTP, time to peak. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 648 648 Chapter 12 Specific treatments for acute ischaemic stroke Meanwhile, other incremental refinements in the defini- 12.1.4 Phases and mediators of cell death tion of the ischaemic penumbra are likely, including measurement of biochemical thresholds with spectro- With sudden, severe and prolonged reductions in cere- scopy, and imaging of a ‘molecular penumbra’ by means bral blood flow (CBF) below about 10 mL/100 g brain per of measuring gene markers of protein expression. 1 minute, ischaemic necrosis (infarction/cell death) pro- gresses within minutes in the core of the affected region Although early studies suggested that various MRI because of low ATP levels and energy stores, ionic disrup- techniques such as perfusion- and diffusion-weighted tion and metabolic failure. 77 However, not all brain cells imaging could differentiate between penumbra and die immediately after stroke. In the ischaemic penumbra infarct, more recent studies have shown that this may that surrounds the anoxic core, cells can be rescued by be an oversimplification. rapid reperfusion. Ischaemic brain cells die by the convergence of three main mechanisms: excitoxicity and ionic imbalance, oxid- CT perfusion 77–82 ative stress and apoptotic-like cell death (Fig. 12.7). Measures of CBF and CBV obtained from CT perfusion can be sensitive and specific for infarction, and may Excitotoxicity have utility for differentiating between infarct and penumbra. 34 Penumbral regions are characterized by a When the blood supply to a part of the brain is inter- mismatch between CBF (which is reduced) and CBV rupted, neurones deprived of their supply of oxygen (which is maintained), whereas infarcted areas show a and glucose rapidly (within minutes) fail to generate matched decrease in both parameters. sufficient ATP and energy to maintain ionic gradients + across membranes because of failure of the Na , + Contrast-enhanced CT may be able to delineate K -ATPase pump. This causes an increase in extracellular – + + infarcted and penumbral tissue in acute stroke. K as well as an influx of Na , Cl and Ca ++ into the cells. The initial increase in extracellular K + may spread, triggering depolarizations of neurones and reversal of Reperfusion and brain damage the amino acid transporters. In these conditions, both Reperfusion within the revival times of ischaemic voltage-operated and receptor-operated calcium channels tissues may salvage cells and aid recovery (by restoring are recruited, thus provoking an elevation of free cyto- ++ oxygen and nutrient delivery to ischaemic brain tissue). solic Ca . The depolarized presynaptic neurones release However, it may also be detrimental, causing so-called the neurotransmitter glutamate into the synaptic cleft. ‘reperfusion injury’ due to the resupply of glucose (which Reuptake of glutamate is impaired, leading to an increase may increase lactic acidosis), oxygen (which may trigger in extracellular concentration of the excitatory amino production of injurious free radicals), water and osmotic acid glutamate. Glutamate activates postsynaptic re- equivalents (which may exacerbate vasogenic oedema), ceptors, including the ionotropic N-methyl-D-aspartate and blood-borne cells (such as neutrophils) which may (NMDA), 2-amino-3-(3-hydroxy-5-methylisoxazol-4-yl) exacerbate ischaemic damage. 74,75 proprionate (AMPA) and kainate receptors. 83 Upon their activation, these open their associated ion channel to + Rescue of the penumbra, either by restoration of blood allow the influx of Ca ++ and Na ions. 2,84 Glutamate- + supply or by interruption of the adverse metabolic or evoked Na increase in astrocytes evokes the transmis- + neurochemical cascade, is the basis of acute stroke sion of intercellular Na and Ca ++ metabolic waves and therapy. an increase in uptake of glucose. 85 ++ Under physiological conditions, calcium ions (Ca ) As discussed below, leucocytes migrate into the injured govern a multitude of cellular processes, including region within hours of reperfusion and may cause tissue cell growth, differentiation and synaptic activity. Con- injury by occluding the microvasculature, generating sequently, there are energy-dependent homeostatic oxygen-free radicals, releasing cytotoxic enzymes, alter- mechanisms to maintain a low intracellular Ca ++ con- ing vasomotor reactivity and increasing cytokine and centration (about 100 times lower than its extracellular chemo-attractant release. 76 Although, as ever, for every concentration) so that Ca ++ signals remain spatially and potential benefit of treatment there are potential harms, temporally localized. This permits multiple independent in humans the empirical evidence suggests that the Ca-mediated signalling pathways to occur in the same benefits of reperfusion are greater than the hazards (sec- cell. In excitotoxicity, the excessive synaptic release ++ tion 12.5). of glutamate which leads to excessive influx of Ca , .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 649 12.1 Pathophysiology of acute ischaemic stroke 649 Reperfusion Reduced CBF Ischaemia Hypoglycaemia Cytokine release Reduced ATP formation Adhesion Failure of Na-K-ATPase pump Acidosis molecule expression Reduced tissue pH, Increased extracellular K+ and increased Na+ ions Leucocyte adhesion to endothelium Depolarization of Na+ ions into presynaptic neurons the cell Release of excitatory amino acid (glutamate) Opening of voltage-dependent calcium channels and influx of Activation of NMDA calcium and AMPA receptors Phospholipase Increase in intracellular calcium activation NO Increase NO Protein Activation of production synthase malfolding proteases, lipases, and endonucleases Free radical formation DNA damage Membrane damage Mitochondrial injury Energy depletion Activation of pro- apoptotic proteins Fig. 12.7 Schematic diagram of some Cytochrome C potential mechanisms of ischaemic brain damage and reperfusion injury. CBF, Caspase activation cerebral blood flow; ATP, adenosine triphosphate; AMPA, D-amino-3- DNA fragmentation hydroxyl-5-ethyl-4-isoxazole propionate; NMDA, N-methyl-D-aspartate; NO, nitric Apoptotic cell death Necrotic cell death oxide. together with any Ca ++ release from intracellular com- species (see below). 6,84 Mitochondrial calcium accumula- partments, can overwhelm Ca ++ regulatory mechanisms tion and oxidative stress can then trigger the assembly and this leads to disregulation of Ca ++ homeostasis, and (opening) of a high-conductance pore in the inner mito- ultimately to cell damage and death. The cell damage chondrial membrane. The mitochondrial permeability associated with a non-physiological unregulated rise in transition (MPT) pore leads to a collapse of the electro- + intracellular cytoplasmic Ca ++ is caused mainly by chemical potential for H , thereby arresting ATP pro- changes in total calcium influx. 6 duction and triggering further production of reactive The initial calcium load is sequestered, at least in part, oxygen species. by the mitochondria. Mitochondrial injury is associated An increase in total calcium influx also contributes to ++ with failure of mitochondrial functions such as oxida- cell death by activation of Ca -ATPase (which results tive phosphorylation, and release of reactive oxygen in further consumption of cellular ATP); activation of .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 650 650 Chapter 12 Specific treatments for acute ischaemic stroke ++ Ca -dependent phospholipases, synthases, proteases this system. The free radicals which are elevated in cere- – and endonucleases (which degrade key cytoskeletal and bral ischaemia include O , hydroxyl (OH) and nitric 2 enzymatic proteins such as phospholipids, proteins oxide (NO). Nitric oxide is generated primarily by and nucleic acids); and alteration of protein phosphory- neuronal and inducible NO synthases (NOS). 84 Like lation, which secondarily affect protein synthesis and other free radicals, the free radicals which are elevated genome expression. 2,84,86 in cerebral ischaemia react with and damage proteins, ++ In addition, the augmented intracellular Ca enhances nucleic acids and lipids, particularly the fatty acid com- the increase in extracellular glutamate, thus propagat- ponent of membrane phospholipids, producing changes ing excitotoxicity. 84 Although Ca ++ disregulation is in the fluidity and permeability of the cellular mem- paramount to neurodegeneration, the exact mechanism branes (lipid peroxidation). 80,84 The free radicals also cause by which Ca ++ ions actually mediate excitotoxicity is less microvascular dysfunction and disrupt the blood–brain ++ clear. 2 One hypothesis suggests that Ca -dependent barrier, leading to brain oedema. In addition, the conver- neurotoxicity occurs following the activation of distinct sion of xanthine dehydrogenase to xanthine oxidase signalling cascades downstream from key points of promotes the cellular formation of toxic oxygen free Ca ++ entry at synapses, and that triggers of these cascades radicals such as the superoxide anion which further are physically colocalized with specific glutamate breaks down membrane, cytoskeletal and nuclear struc- receptors. 2 tures. An important source of oxidative stress-mediated brain damage is the oxidant reactions due to the forma- Calcium influx is mediated directly and tion of peroxynitrite, a powerful oxidant that results predominantly by NMDA receptors, but it is also from the interaction between NO and superoxide. This + triggered secondarily by Na influx through AMPA-, anion has been shown to cause cell damage by several kainate- and NMDA-receptor-gated channels. The mechanisms that include lipid peroxidation, tryrosine resultant excessive Ca ++ influx leads to elevated nitration, sulphydryl oxidation and nitrosylation, and intracellular and intramitochondial Ca ++ DNA breakage. 84 Neurones undergoing oxidative stress- concentrations and lethal metabolic derangements, related injuries typically display a biphasic or sustained which include calcium-dependent activation of pattern of extracellular signal regulated protein kinase intracellular enzyme systems, failure of oxidative (ERK1/2) activation. 87 phosphorylation and the generation of free radicals With reperfusion, reactive oxygen radicals may be that further compromise cells by attacking proteins, generated as by-products of the reactions of free arachi- lipids and nucleic acids. donic acid (released from membrane phospholipids during ischaemia) to produce prostaglandins and leukotrienes, which perhaps lead to reperfusion injury Oxidative stress (production of free radicals) to the brain and its microvessels. A free radical is any atom, group of atoms or molecule with an unpaired electron in its outermost orbital. 80 Free Apoptosis radicals are produced in small quantities by normal cel- lular processes in all aerobic cells; for example, ‘leaks’ in Apoptosis is a mode of programmed cell death in which mitochondrial electron transport allow oxygen to accept the cell synthesizes proteins and plays an active role – single electrons, forming superoxide (O ). However, in its own demise. It occurs both physiologically and 2 they are inherently toxic – they can react with and dam- pathologically. For example, during normal human age proteins, nucleic acids, lipids and other classes of embryonic development, apoptosis results in the loss of molecules such as the extracellular matrix glycosamino- the interdigital webs required for normal formation of glycans (e.g. hyaluronic acid). The sulphur-containing the fingers and toes. Likewise, tadpoles lose their tails amino acids and the polyunsaturated fatty acids (found as they develop into frogs. The normal turnover of cells in high concentrations in the brain) are particularly vul- in the intestinal villi is also apoptotic, as is the turn- nerable. Fortunately, cells possess appropriate defence over of normal lymphocytes. Indeed, inhibition of the mechanisms in the form of free radical scavengers apoptotic death of lymphocytes may lead to B-cell and enzymes which metabolize free radicals or their lymphoma. precursors. Although necrosis is the predominant mechanism During severe brain ischaemia, insufficient oxygen is of cell death after stroke (mediated by excitotoxicity, available to accept electrons passed along the mito- as described above), growing evidence indicates that chondrial electron transport chain, leading to eventual hypoxic/ischaemic cell death also continues to some reduction (‘electron saturation’) of the components of extent hours to days after the onset of ischaemia, .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 651 12.1 Pathophysiology of acute ischaemic stroke 651 particularly within the peri-infarct zone or ischaemic 200% of baseline), cortical spreading depression exacer- penumbra, by apoptosis as a consequence of a genetic- bates cellular injury and augments tissue damage by ally regulated programme that allows cells to die with causing profound oligaemia that persists for up to 3 days, minimal inflammation or release of genetic material. thereby imposing an energy burden of re-establishing Families of executioner enzymes (e.g. caspases, apop- ionic equilibrium in hypoxic tissue. 93 Cortical spread- 88 tosis inducing factor ) are expressed in neurones and ing depression also alters blood–brain permeability by become activated during and after ischaemic stroke. activating brain matrix metalloproteinases, 92 and is asso- Active caspases kill cells by cleaving critical cell repair ciated with changes in immediate early genes, growth and homeostatic proteins as well as cytoskeletal pro- factors and inflammatory mediators such as interleukin- teins. 82 In other words, they dismantle multiple cell 1β and tumour necrosis factor-α. 92 Proteolysis of the processes in the cytoplasm and nucleus to promote cell neurovascular matrix by plasminogen activator and death by suicidal mechanisms resembling apoptosis. 81 metalloproteinases is believed to be linked with hae- The c-Jun N-terminal protein kinase (JNK) signalling morrhagic transformation of the brain infarction and pathway is implicated in ischaemia-induced neuronal brain oedema. However, treatment with matrix metallo- apoptosis. 89 Apoptosis is characterized morphologically proteinase inhibitors has not been shown to protect by coarse, regularly shaped chromatin condensation, the brain. Indeed, a recent study suggested that such loss of cell volume and extrusion of membrane-bound inhibitors might contribute to brain damage. 94,95 cytoplasmic fragments (apoptotic bodies), and biochem- ically by DNA fragmentation in neurones and glia after inflammation ischaemic injury. Necrosis may proceed by analogous Cerebral ischaemia triggers an inflammatory reaction, programmed pathways recently revealed in the nema- by means of an active gene expression, within hours to tode Caenorhabditis elegans. 90 days that may last up to several months. 76,96 Important transcription factors are activated and/or synthesized, including NF-κB, hypoxia-inducible factor 1, interferon Other mediators of ischaemic cell death 97 regulatory factor 1 and STAT3. Several mediators are Although the original model of excitotoxicity emphas- released or activated, such as adhesion molecules, pro- izes calcium influx through glutamate receptor-coupled teolytic enzymes belonging to the family of metallo- ion channels, ionic imbalance (and cell death) may also proteinases (including matrix metalloproteinases), and proceed via other routes. 81 inflammatory cytokines. 98,99 Cytokines might activate the expression of inflammation-related genes, such as iNOS and cycloxygenase-2. 98 The result is infiltration of acidosis Acidosis arises as a result of energy deprivation by sus- leucocytes into the brain parenchyma, and the activa- tained tissue ischaemia. It may contribute to tissue dam- tion of resident microglial and astroglial cells. 98,99 The age and prevent or retard recovery during reoxygenation effects of inflammation are both detrimental (e.g. exacer- by several mechanisms: activation of novel classes of bation of oedema) and potentially beneficial. acid-sensing ion channels that further perturb sodium and calcium homeostasis, 91 and inhibition of mito- altered gene expression chondrial respiration and lactate oxidation. 4,5 Cellular Ischaemia not only decreases protein and mRNA syn- acidosis may also promote intracellular oedema forma- thesis, but it also induces at least 100 genes (and thus – + tion by inducing Na and Cl accumulation in the cell via protein synthesis) which include immediate early – + – + coupled Na /H and Cl /hydrogen carbonate (HCO ) genes, stress proteins, growth factors, adhesion proteins, 3 exchange. cytokines, kinases and genes directly regulating apoptosis. 87,89 peri-infarct depolarization + As stated above, ischaemia-induced failure of the Na , + K -ATPase pump causes an increase in extracellular K + Cell death occurs by a necrotic pathway characterized by either ischaemic cell change or oedematous cell which may spread, triggering a propagating wave of change. Death also occurs via an apoptotic-like neuronal and glial depolarization (cortical spreading pathway that is characterized by DNA laddering, a depression) associated with depressed neuronal bioelec- dependence on caspase activity, characteristic protein trical activity, transient loss of membrane ionic gradients, and phospholipid changes, and morphological and massive surges in extracellular potassium, neuro- attributes of apoptosis. Death may also occur by transmitters and intracellular calcium (150 times the basal level). 92 After an initial transient hyperaemia (up to autophagocytosis. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 652 652 Chapter 12 Specific treatments for acute ischaemic stroke those tested only in animal models. 100 The quality of the The four major stages of cell death studies was highly variable: The cell death process has four major stages: 78 • poor trial study design; • The induction stage includes several changes initiated • inadequate sample size; by ischaemia and reperfusion that are very likely to • unreliable outcome evaluation; play major roles in cell death. These include inhibition • anatomical and aetiological hetereogeneity in the (and subsequent reactivation) of electron transport, study populations; ++ decreased ATP, decreased pH, increased cell Ca , • extrapolating results from young rats to old humans release of glutamate, increased arachidonic acid, and with comorbidities; also gene activation leading to cytokine synthesis, syn- • dose ceilings imposed by adverse effects of the drug; thesis of enzymes involved in free radical production, • inadequate drug penetration to the site of the lesion; and accumulation of leucocytes. These changes lead to • possible differences in responsiveness between isch- the activation of five damaging events, termed per- aemic grey and white matter; petrators: the damaging actions of free radicals and • drug delivery after the therapeutic time window had ++ their product peroxynitrite; the actions of the Ca - closed; dependent protease calpain; the activity of phospholi- • publication bias in the basic science literature; pases; the activity of poly-ADPribose polymerase • misleading hypotheses from the basic scientific (PARP); and the activation of the apoptotic pathway. researchers linking NMDA-receptor-mediated excito- • The second stage of cell death involves the long-term toxicity to hypoxic-ischaemic neuronal death. changes in macromolecules or key metabolites that are All this questions whether the most efficacious drugs caused by the perpetrators. have been selected for stroke clinical trials, and suggests • The third stage involves long-term damaging effects that greater rigour in the conduct, reporting and analysis of these macromolecular and metabolite changes, of animal data is required to improve the translation and of some of the induction processes, on critical cell of scientific advances from bench to bedside. 100 More- functions and structures that lead to the defined end over, new approaches and targets for cytoprotection are stages of cell damage. These targeted functions and required. 84,101–105 There seem to be so many pathways structures include the plasmalemma, the mitochon- leading from ischaemia to neuronal cell death that dria, the cytoskeleton, protein synthesis and kinase blocking just one of them is probably fruitless; perhaps activities. this is like ligating a feeding artery to an arteriovenous • The fourth stage is the progression to the morpho- malformation (there are always others that will take over logical and biochemical end stages of cell death. Of to keep up its blood supply), or trying to extinguish a these four stages, the last two are the least well under- large blazing fire with just a small garden hose. stood. Rather little is known of how the perpetrators affect the structures and functions and whether and 12.1.5 Ischaemic cerebral oedema how each of these changes contribute to cell death. The key step in ischaemic cell death is adequate activa- Cerebral ischaemia not only causes loss of neuronal tion of the perpetrators. 78 function but also cerebral oedema. Within hours of onset, cytotoxic cerebral oedema arises as a result of cell membrane damage allowing intracellular accumulation Implications for future neuroprotective therapies of water. The grey matter is affected more than the white The neurovascular unit – a conceptual model comprised matter. The CT and MRI scan appearances are of well- of cerebral endothelial cells, astrocytes, neurones and circumscribed altered signal density involving the cortex the extracellular matrix – provides a framework for and subcortex (sections 5.4.2 and 5.5.2). Cytotoxic understanding the pathophysiology of ischaemic stroke oedema reaches a maximum volume at 2–4 days and and developing treatments based on an integrative and then subsides over 1–2 weeks. dynamic view of evolving tissue damage. Drugs which Initially, endothelial tight junctions are maintained aim to protect the neurovascular unit (neuroprotectants) and the blood–brain barrier remains intact. However, have been evaluated preclinically and clinically in a large after several days of ischaemia, breakdown of the blood– number of trials (section 12.8.1). However, a systematic brain barrier occurs, leading to vasogenic cerebral oedema review of 1026 experimental drug treatments in acute as plasma constituents enter the brain extracellular stroke (clinical studies of 114 drugs, animal models of space, affecting the white matter more than the grey 912 drugs) found that there was no evidence that drugs matter. Blood–brain barrier breakdown may be initiated used clinically were more effective experimentally than and regulated by several proinflammatory mediators .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 653 12.1 Pathophysiology of acute ischaemic stroke 653 (oxidative mediators, adhesion molecules, cytokines, chemokines). These mediators not only regulate the magnitude of leucocyte extravasation into brain par- enchyma, but also act directly on brain endothelial cells causing loosening of junction complexes between endothelial cells, increasing brain endothelial barrier permeability, and causing vasogenic oedema. 106 The brain scan appearance of vasogenic oedema includes the characteristic finger-like projections of altered signal density in the white matter, which characteristically accompany cerebral tumours. Imaging studies indicate that fluid volume after infarction peaks in 7–10 days and oedema remains detectable for about 1 month. The effects of cerebral oedema are to compromise blood flow even further by increasing pressure in the extra- vascular space, thus causing vascular congestion and sometimes haemorrhagic transformation, and to cause mass effect, brain shift and eventually brain herniation (Figs 12.8 and 12.9). The main danger of herniation is Fig. 12.9 Central transtentorial herniation. Diffuse or that it can initiate vascular and obstructive complications multifocal swelling of the cerebral hemispheres (or bilateral subdural or extradural haematomas) compresses and elongates the diencephalon from above. The mammillary bodies are displaced caudally. The cingulate gyrus is not herniated. Modified from Plum F, Posner JB. The Diagnosis of Stupor and Coma. Philadelphia, PA: F.A. Davis & Co., 1985. 319 C which aggravate the original expanding lesion by com- pressing important vessels and tissues and causing even more cerebral ischaemia, congestion and oedema which, in turn, enhance the expanding process. In addition, the herniating brain can compress the aqueduct and P subarachnoid spaces and so interfere with cerebro- spinal fluid circulation, leading to hydrocephalus and U increased cerebrospinal fluid pressure. There are three anatomical patterns of supratentorial brain shift which can be identified by their end stages: cingulate herniation, central transtentorial herniation and uncal herniation. Fig. 12.8 Uncal and cingulate herniation: a mass such as a Cingulate herniation large cerebral haemorrhage or oedematous infarct displaces the diencephalon and mesencephalon horizontally and Cingulate herniation occurs when the expanding cere- caudally. The cingulate gyrus (C) on the side of the lesion bral hemisphere shifts across the intracranial cavity, herniates under the falx cerebri. The uncus (U) of the ipsilateral forcing the ipsilateral cingulate gyrus under the falx temporal lobe herniates under the tentorium cerebelli and cerebri, compressing and displacing the internal cerebral becomes grooved and swollen and may compress the ipsilateral vein and the ipsilateral anterior cerebral artery. This oculomotor (third cranial) nerve causing pupillary dilatation may lead to additional infarction in the territory of the (Hutchinson’s sign). The cerebral peduncle (P) opposite the anterior cerebral artery (Fig. 12.8). supratentorial notch becomes compressed against the edge of the tentorium, leading to grooving (Kernohan’s notch) 107 and causes a paresis ipsilateral to the cerebral mass lesion. Central Central transtentorial herniation downward displacement also occurs but is less marked than in Fig. 12.9. Modified from Plum F, Posner JB. The Diagnosis of Central transtentorial herniation of the diencephalon Stupor and Coma. Philadelphia, PA: F.A. Davis & Co., 1985. 319 is the end result of displacement of the cerebral .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 654 654 Chapter 12 Specific treatments for acute ischaemic stroke hemispheres and basal nuclei, compressing and even- all correspondingly be reduced in survivors. For patients tually displacing the diencephalon and the adjoining with a large volume of ischaemic brain, minimizing the midbrain rostro-caudally through the tentorial notch volume which infarcts should also reduce the risk of (Fig. 12.9). The great cerebral vein is compressed, which early death, particularly from ischaemic cerebral oedema raises the hydrostatic pressure of the entire deep territory and transtentorial herniation (section 12.1.5). The it drains. In addition, downward displacement of the pathophysiology of acute cerebral ischaemia is complex midbrain and pons stretches the medial perforating (section 12.1.3), and many potential treatments may branches of the basilar artery (the artery cannot shift have more than one mechanism of action; for some, the downward because it is tethered to the circle of Willis), precise mechanism is unknown. However, the main leading to paramedian brainstem ischaemia (and hae- targets of acute treatment are to restore and then main- morrhage if perfusion continues). tain blood flow, and simultaneously keep alive as much ischaemic brain tissue as possible while the blood supply is restored, either spontaneously or therapeutically. Uncal herniation Uncal herniation characteristically occurs when expand- 12.2.1 How to evaluate evidence about different ing lesions in the temporal fossa or temporal lobe shift treatments the inner, basal edge of the uncus and hippocampal gyrus toward the midline so that they bulge over the Evidence-based medicine incisural edge of the tentorium, and push the adjacent midbrain against the opposite incisural edge (Fig. 12.8). 107 We try to base our decisions on whether or not to use a The third cranial nerve and the posterior cerebral artery particular treatment in clinical practice on the evidence on the side of the expanding temporal lobe are often from randomized trials and systematic reviews of ran- caught between the overhanging swollen uncus and the domized trials. This approach is part of what is now free edge of the tentorium or the petroclinoid ligament, known as ‘evidence-based medicine’, a style of medical leading to a third nerve palsy and occipital and medial practice which Sackett described as ‘a life-long process of temporal lobe infarction and swelling, which further self-directed learning in which caring for our patients compounds the problem. creates the need for clinically important information Transtentorial herniation is the most common cause about diagnosis, prognosis and therapy’. 108 The book by of death during the first week after acute stroke, account- Sackett and co-authors on the subject is short and clear ing for about 80% of deaths in cerebral infarction and and provides succinct advice on how to find the best evid- 90% in intracerebral haemorrhage. The risk of death ence, appraise it critically and then apply it in everyday peaks within 24 h for intracerebral haemorrhage, but clinical practice (Table 12.4). 108 This section summarizes later at 4–5 days for cerebral infarction as ischaemic cere- what the best forms of evidence are (and why) and where bral oedema develops. Brainstem compression, with sub- up-to-date evidence can be found. sequent haemorrhage and infarction within it, accounts for the serious morbidity and mortality associated with Why are randomized trials the best way to evaluate herniation. treatments? Transtentorial herniation is the most common There is now little disagreement that the randomized cause of death within the first week of onset of both controlled trial (RCT) is the best way to evaluate most ischaemic stroke and intracerebral haemorrhage. treatments 109 (Table 12.4). Weaker research designs, i.e. case series without any controls, series with retrospective controls, and series with concurrent controls but non- random treatment allocation, may introduce far too many biases into the assessment of treatment effect. 108–110 12.2 General treatment considerations The Medical Research Council trial of streptomycin for pulmonary tuberculosis, designed by the statistician Sir Austin Bradford Hill, is probably the strongest contender The primary aims of the specific treatments considered for the title ‘first clinical trial to use strictly random in this chapter are, broadly speaking, to minimize the allocation’, although there is some debate about this volume of brain damaged by ischaemia. The assump- claim. 111,112 The trial not only established the benefits of tion is that, if this can be achieved without excess hazard, the treatment beyond all reasonable doubt, but also neurological impairment, disability and handicap should served as the most ethical and equitable way of utilizing .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 655 12.2 General treatment considerations 655 Table 12.4 Methodological questions in the critical assessment Strictly random allocation and proper concealment of an article evaluating treatment (modified from Sackett). 108,109 minimize selection bias. 110,113 If the randomizing clinician can find out what treatment the next patient Was the assignment of patients to treatment really randomized?* will be allocated (e.g. by holding a sealed trial alloca- Was the similarity between groups documented? tion envelope up to a bright light), he may be tempted Was prognostic stratification used in treatment allocation? not to enter the patient in the trial. Allocation conceal- Was foreknowledge of the randomly allocated treatment ment reduces or avoids this happening. 113 A central possible? telephone randomization system offers complete con- Were all clinically relevant outcomes reported? Was case fatality as well as morbidity reported? cealment: the clinician telephones a randomization Were deaths from all causes reported? centre and gives the patient’s details, these are then Were quality-of-life assessments conducted? recorded on the central computer during the call, and Was outcome assessment blind to treatment allocation? the computer generates the next treatment allocation Were the study patients recognizably similar to your own? only after all the baseline data are checked and entered; Were reproducibly defined exclusion criteria stated? the patient is then irrevocably in the trial and selec- Was the setting primary, secondary or tertiary care? 113 tion bias cannot occur. Studies where patients are Was the type of patient included in the study clearly entered into a trial by selecting a numbered drug pack described? in the participating hospital, administering the drug/ Were both statistical and clinical significance considered? placebo contained in the pack, and then telephoning If statistically significant, was the difference clinically a central office to notify the trial administrators that important? If not statistically significant, was the study big enough to a patient has been entered into the trial, are prone to show a clinically important difference if one really exists? all sorts of bias. In the worst case, the pack could be Is the therapeutic manoeuvre feasible in your practice? opened and given but the patient details then not Available, affordable, sensible? notified to the trial office if there was an early adverse Were contamination† and co-intervention avoided? event – the pack could just be recorded as ‘opened and Was the manoeuvre administered blind? discarded’. The pack drug might not have been adminis- Was compliance measured? tered when the hospital said it was, but possibly earlier Were all patients who entered the study accounted for at its so as to look better on the records. There is no possibility conclusion? of balancing the treatment allocation on important (i.e. were drop-outs, withdrawals, non-compliers and those key prognostic variables resulting in imbalances between who crossed over handled appropriately in the analysis?) treatment and control arms which could skew the *Methods of allocation such as alternation, use of hospital results. Unfortunately, with this system of ‘randomiza- number or date of birth all provide foreknowledge of the tion’ any of these scenarios is possible, and probably has next treatment allocation, and so allow the trial allocation happened. process to be subverted which may lead to selection bias It is important to reduce observer bias, if possible, between the two treatment groups. by ‘blinding’ both the patients and the observers who †In this context, contamination implies that some collect the outcome data. 110,113 patients allocated active treatment either did not receive Random error is reduced by recruiting a sufficiently the treatment or were given the treatment which the other large sample of patients. However, RCTs need to be sur- treatment group should have received, or some non-trial prisingly large to provide really reliable evidence on the treatment with similar properties to one of the trial balance of risk and benefit for a particular treatment. 110 interventions under test. If the measure of outcome is an uncommon but serious Note: reports of RCTs should conform to the requirements of the CONSORT statement as far as possible. 127 event, such as death, and the effect of the treatment on that outcome is only moderate, then the trial may need to recruit several tens of thousands of patients. 110 To the very limited supply of the drug that was available recruit such large numbers often involves clinicians in in Britain at the time. It also prevented the treatment many countries, and hence the trial design needs to be ‘creeping’ into routine clinical practice without being simple, with minimal data collection and audit (though properly evaluated. such simplicity is more difficult to achieve in the current over-regulated and bureaucratic research climate). 110,114 We could answer many of the therapeutic questions The effort of conducting such large RCTs is substantial, posed in this book a great deal faster if a larger and so many of the ‘mega trials’ have used a factorial proportion of stroke patients were entered in design to assess two or three treatments at the same time appropriate randomized controlled trials. 110 without loss of statistical efficiency. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 656 656 Chapter 12 Specific treatments for acute ischaemic stroke is not going to go away, but there are now guidelines Good trial design seeks to reduce bias and random on the relation between sponsors and investigators in error in the assessment of treatment effect: strict stroke trials that may improve the situation. 124 randomization with concealment reduces selection There is therefore a continuing need to increase the bias; blinding reduces observer bias; and recruitment number, size and quality of randomized trials in stroke, of large numbers of patients reduces random error. to increase non-commercial sources of funding and to improve the quality of the reports of those trials in the Problems with randomized controlled trials in acute stroke literature. Unfortunately, the regulatory hurdles that from 1956 to the present must now be overcome before a trial is initiated are very Eight years after the streptomycin trial was published great. 114 So much so, that they may stifle this vital form in 1948, Dyken and White wrote of the many methodo- of clinical research, making the ideal of ‘most stroke logical problems they identified in evaluating treatments patients get their treatment in the context of a random- for acute stroke; not the least was to know, in an indi- ized trial’ a rather distant dream. 125,126 vidual patient, whether or not treatment had been effect- ive. 115 They wrote this in the report of the first quasi- Steps to improve the quality of stroke trials and reports of randomized study of a medical treatment for acute trials in the literature stroke. 115 Thirty-six patients were alternately (but not randomly) assigned to cortisone or control. Thirteen of The CONSORT group have produced guidelines to improve the cortisone patients and 10 of the controls died; a trend the quality of reports of RCTs. 127 Their statement is avail- against treatment, but an inconclusive result. Sadly, a able in several languages and has been endorsed by review of the quality of the final reports of RCTs of drug prominent medical journals such as the Lancet, Annals of therapies for acute stroke conducted over the next 40 Internal Medicine and Journal of the American Medical years (i.e. between 1956 and 1996) makes depressing Association. It comprises a checklist and flow diagram to reading. 116 Many trials had weaknesses which included: help improve the quality of reports of RCTs. The checklist over-complex eligibility criteria; incomplete descrip- includes the items that need to be discussed in the report; tions of the methods used (particularly the method of the flow diagram provides a clear picture of the progress randomization); inadequate sample size; inappropriate of all participants in the trial, from the time they are ran- measures of outcome; poor standards of execution (i.e. domized until the end of their involvement. The intent poor trial discipline); an unacceptably high proportion is to make the experimental process clearer, flawed or not, of patients ‘lost’ to follow-up; inappropriate statistical so that users of the data can more appropriately evaluate methods (often with inappropriate subgroup analysis); their validity. The CONSORT website provides evidence and failing to account for all patients randomized. 116 to support each aspect of the checklist and is very useful On the other hand, trial quality did appear to improve (http://www.consort-statement.org). One can only hope in parallel with an increase in trial size. 116 While this that, in future, stroke triallists will adhere to these guide- is encouraging, a more recent review of reports of the lines when designing trials and reporting their results. randomized acute stroke trials completed before 2001 showed that most trials were underpowered, i.e. power Describing the effects of treatment in numbers <0.90, used inappropriate assumptions for event rates, and were grossly overoptimistic in their expectation of The effects of a treatment can be expressed in a number treatment effect. 117 These deficiencies will together have of ways. Relative treatment effects can make a treatment resulted in trials being far too small and reduced their appear to have impressive benefits; ‘drug X reduced the chance of being able to detect real treatment effects. 117 risk of event Y by a half’. Table 12.5 describes how the Most trials in acute stroke to date have been directly most common numerical measures of relative and abso- sponsored (or financially supported by) the pharmaceu- lute treatment effects are calculated. The absolute benefit tical industry. 118,119 However, of some concern is that is more important for clinical decisions, and is greatly research funded by pharmaceutical companies is more influenced by the frequency of events in the control likely to report outcomes favouring the sponsor than group. If events are rare in the control group, the abso- research funded from other sources. 120,121 Although lute benefit – however great the relative treatment effect – pharmaceutical company research is certainly not of will be small. In this case, the number-needed-to-treat poorer quality than other research, there is a tendency to prevent one event (NNT) will be correspondingly to delay publishing unfavourable results (or even not large, so the treatment may have little clinical value in to publish them at all). 122,123 The number of industry- low-risk patients. For example, in a randomized trial, if sponsored trials in stroke is increasing, 119 so this problem fatal pulmonary embolism occurred in 0.5% of stroke .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 657 12.2 General treatment considerations 657 Table 12.5 Describing the numerical results of a clinical trial in different ways. Treated Control No. originally randomized in this group 1000 1000 No. who were dead at the end of the trial 80(a) (8%) 100(b) (10%) No. alive at the end of the trial 920(c) (92%) 900(d) (90%) Relative risk of death (RR) = 8%/10% = 0.80*† Relative risk reduction of death (RRR) = (1 – RR) × 100 = (1 – 0.80) × 100 = 20% Odds ratio (OR) = (a/c) ÷ (b/d) = ad/bc = 0.78* 95% Confidence interval of OR = 0.58 – 1.06 Relative odds reduction = (1 – OR) × 100 = (1 – 0.78) × 100 = 22%† Absolute risk of death in control group = 100/1000 = 10% Absolute risk of death in treated group = 80/1000 = 8% Absolute risk difference in death (ARD) = 10% – 8% = 2%, or 20 deaths avoided or postponed per 1000 patients treated Number-needed-to-treat to avoid one event (NNT) = 100/ARD = 100/2 = 50†‡ need to be treated to prevent (or at least postpone) one death Notes: The Cochrane Handbook is the most up-to-date source of advice on this topic, and is available online through the Cochrane Library. 316 This table is referred to as a 2 × 2 table. The four ‘cells’ in the table are labelled a, b, c and d. *The odds ratio and relative risk both provide an estimate of relative treatment effect. The estimates will be similar if the absolute risk in the control group is low. However, if the absolute risk in the controls is high, then the odds ratio will be larger than the relative risk. For example, in the Cochrane review of the effects of stroke units (compared with general medical wards), about 58% of patients treated on general medical wards were dead or dependent at the end of follow-up, compared with 54% of patients treated on a stroke unit; this gives an odds ratio of 0.80 and a relative odds reduction of 20% in favour of stroke units. When the same results are expressed as relative risks, the relative risk of 0.91 and the relative risk reduction of 9% associated with the benefits of stroke units appear more modest. The odds ratio is easy to calculate, but difficult to understand, the relative risk (and its 95% CI) are harder to calculate, but easier to understand. The Cochrane Handbook gives useful advice on the advantages and disadvantages of each measure. 316 †It is important to calculate the confidence intervals for these estimates of effect. In this example, the upper confidence interval exceeds unity, which is equivalent to ‘not significant at the P < 0.05 level’. While the effect may not be statistically significant, a 20% reduction in the risk of death is potentially clinically significant; however, it would require a trial with a much larger sample size of at least 10 000 patients to detect such an effect reliably. 110 To determine reliably whether the effect on death varied between different subgroups would require an even larger trial with several tens of thousands of patients. 110 Formulae to calculate confidence intervals are available in many statistical textbooks. The calculations can be done on a simple calculator or with standard statistical software or with tools available on statistical websites. The book and program published by Altman et al. are simple, clear, easy to use and – most importantly – error-free! 317 ‡One should not calculate NNTs if the estimate of effect is not statistically significant. If the data from a systematic review of several trials are used to calculate the NNT, interpret the results with caution, since the ‘overall’ NNT will be very heavily influenced by the event rates in the control groups. It is preferable to take the overall estimate of relative treatment effect and apply it to the expected event rate in the group of patients to which the treatment will be applied; 316 a convenient online calculator for this is available at http://www.dcn.ed.ac.uk/csrg/resources.asp patients allocated to aspirin alone, and in 0.25% of those What are systematic reviews and why do we need them? allocated to the combination of aspirin with a second antithrombotic agent (e.g. low-molecular-weight hep- The biomedical literature is so enormous that clinicians arin) the relative risk reduction would be an impressive are faced with an unmanageably large amount of infor- 50%, but the absolute risk difference would be only mation to assimilate. For example, the Cochrane Stroke 0.25%. In other words, if we treated 1000 patients, only Group’s specialized register (published in the Cochrane two or three would avoid fatal pulmonary embolism Library issue 3 for 2007) now includes 11 000 reports of and the NNT to prevent one person having pulmonary stroke trials. The stroke trials literature is growing fast embolism would be 400. However, if fatal pulmonary too: on average, about 900 new reports of stroke trials embolism occurred in 5% of controls, and the relative have been added to the register each year for the last few risk reduction was still 50%, the absolute risk difference years (Brenda Thomas, personal communication). Faced would be 2.5% and the NNT would be just 40 patients. with such a vast, confusing and rapidly changing array .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 658 658 Chapter 12 Specific treatments for acute ischaemic stroke Table 12.6 Methodological questions in the critical assessment with choosing estimates derived from subsets of trials, of a systematic review (adapted from Sackett et al., 1991). 109 and are more precise than the estimate from any one trial (i.e. less subject to random error) because they are Were the question(s) and methods clearly stated? based on more outcome data. Were the search methods used to locate relevant studies comprehensive? Meta-analysis is simply the best way to obtain the least Were explicit methods used to determine which articles to biased and most precise estimate of treatment effect include in the review? from a group of similar trials of the same intervention Was the methodological quality of the primary studies assessed? in the same type of patients and using the same type Were the selection and assessment of the primary studies of outcome measures. reproducible and free from bias? Were differences in individual study results adequately Hazards of inappropriate subgroup analysis in trials (and explained? systematic reviews) Were the results of the primary studies combined appropriately? Subgroup analysis is popular with clinical triallists and Were the reviewers’ conclusions supported by the data cited? people who like to generate hypotheses to explain the ‘negative’ or ‘positive’ overall results of particular trials. Comprehensive details on the design, conduct, analysis and It is, however, a dangerous sport, since even apparently reporting of systematic reviews are available in the Cochrane large effects observed in subgroups can merely be due to Handbook, which is regularly updated, and also available online. 316 the play of chance and not to the treatment itself. 110,129 Claims for the benefits of a treatment based on a sub- group analysis of a single trial, or of a meta-analysis, need to be viewed with caution and should be seen as of information about treatment, what should the busy hypothesis generating. To test such subgroup hypothe- clinician do? Focus on a few selected reports published in ses reliably, generally requires further very large trials well-known English-language journals? Such selection with appropriate and pre-specified hypotheses. 110,129 could easily lead to a biased assessment of the effects of Inappropriate subgroup analyses may have disastrous the treatment. Rely on visiting pharmaceutical company effects. The report of the Canadian aspirin sulphinpy- representatives and being flown to exotic locations to razone trial concluded that, overall, among people with suffer sponsored symposia? Hopefully not. The best way threatened stroke, aspirin was associated with a signi- to minimize bias is to review systematically all of the ficant 30% reduction in the risk of stroke or death. 130 evidence from all of the relevant RCTs, both published A subgroup analysis suggested that the benefit was con- and unpublished. 108 Briefly, a systematic review: fined to males and was not seen in females. As a result, • defines the question to be answered; the United States Food and Drug Administration (FDA) • uses a defined search strategy to identify relevant studies; licensed aspirin for stroke prevention but only in males, • selects studies and extracts data from them using not in females. However, the results of the Antiplatelet explicit criteria; Trialists’ Collaboration systematic review in 1994 of all • synthesizes the evidence in a quantitative manner of the relevant randomized trials showed that, among whenever possible (i.e. provides an overall estimate of high-risk individuals, the benefits of antiplatelet drugs the treatment effect). were similar in males and females, so refuting the results The reader should consider the checklist in Table 12.6 of the Canadian trial aspirin subgroup analysis. 131 The when reading a report of a systematic review. For those FDA did not license aspirin for stroke prevention in women preparing systematic reviews, the Quality of Reporting until 1998, so in the years between 1980 and 1998, it of Meta-analyses (QUOROM) Group has defined the seems likely that many women worldwide at high risk criteria for an adequate report of a systematic review. 128 of stroke were not treated with aspirin and consequently had strokes which might have otherwise been avoided. Likewise, undue emphasis on the results of a single posi- What is meta-analysis? tive trial in a meta-analysis may result in misleadingly Meta-analysis is the numerical technique which derives optimistic conclusions. For example, the most widely an overall estimate of the treatment effect from all of the cited study of therapeutic thrombolysis in acute ischaemic trials included in a systematic (or non-systematic) review. stroke, the NINDS trial, may have overestimated the Such overall estimates avoid the selection bias inherent benefits of this treatment (section 12.5.2). 132,133 .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 659 12.2 General treatment considerations 659 available, we have searched for a high-quality systematic Translating the results of trials and systematic reviews into review. The Cochrane Library includes the Database of improving clinical practice Abstracts of Reviews of Effectiveness (DARE), a series of There are now many books and articles on how to use quality-assured reviews (and their abstracts) which is a the results of trials and systematic reviews to improve very useful source of such reviews. If we could not find a one’s own clinical practice. However, it requires a great systematic review, we either resorted to asking a known deal of time and resources to take the next steps, i.e. expert in the field or to a search of the Cochrane Central to develop and implement evidence-based guidelines Register of Controlled Trials (CENTRAL); in the Cochrane or policy documents and then consistently apply them Library (issue 2 for 2006), CENTRAL contained over to an entire clinical service over a long period. Many 473 000 references to reports of randomized trials and strategies to improve clinicians’ performance, ranging controlled clinical trials. from financial incentives, to guidelines, to continuing There are a number of other regularly updated sources medical education, have all been suggested. The of evidence. Some of the ones we often use are: Cochrane Effective Practice and Organization of Care • The Internet Stroke Center at the University of (EPOC) Group is undertaking systematic reviews to Washington, St Louis, an independent web resource identify which interventions are most likely to achieve about stroke care and research which includes an an improvement in the clinical practice of healthcare up-to-date register of current stroke trials (http://www. professionals. 134 If we are to spend some of our limited strokecenter.org/); health service resources on getting healthcare profes- • Clinical Evidence is a compendium of the best available sionals to improve their standards of care, it is important evidence on effective healthcare, published online that we spend them only on those methods which at http://www.clinicalevidence.com/ceweb/conditions/ can bring about a measurable and sustainable change index.jsp and in various other paper and electronic in clinical practice (and do so efficiently and cost- formats (including a version for your personal digital effectively) (Table 17.22). Our efforts to improve stand- assistant (PDA) or palmtop organizer); ards of care must not be wasted on uncritical application • The Royal College of Physicians of London (http:// of interventions – such as audit without feedback – that www.rcplondon.ac.uk); take great effort, but achieve little. 135 • The Scottish Intercollegiate Guidelines Network (http:// www.sign.ac.uk/); • The Cochrane Stroke Group website is building up a Where we look for up-to-date reports of randomized list of links to stroke guidelines (http://www.dcn.ed. trials, systematic reviews and clinical practice ac.uk/csrg); guidelines • The National Library for Health (NLH) seeks to make Evidence-based medicine requires one to use up-to-date relevant up-to-date information on effective forms of reports, so, with each year that passes after the publica- healthcare available to patients and doctors (http:// tion of this book, more and more of the evidence we www.library.nhs.uk/). have cited will be superseded. However, there are several sources of regularly updated evidence that can fill this 12.2.2 Organization of acute stroke care ‘information gap.’ 108 For example, the Cochrane Col- laboration Stroke Review Group is coordinating a series The way acute stroke care is organized within a particular of systematic reviews of different forms of healthcare hospital will depend very much on what resources are for the treatment and prevention of stroke which are available and what treatments are planned (Chapter 17). updated as new information becomes available. 136 Its The clinical and radiological diagnosis of acute stroke website includes a link to the Cochrane Collaboration is dealt with in Chapters 3, 4 and 5. The other aspects of website, where abstracts of all the Cochrane systematic general management in the acute phase are dealt with in reviews are available free of charge (http://www. Chapters 10, 11 and 17. dcn.ed.ac.uk/csrg). Where possible therefore we have in this book cited 12.2.3 Which treatments to use routinely, stroke reviews from the Cochrane Database of Systematic selectively or not at all Reviews, since they are often – but not always – of higher quality and are more frequently updated than many There is strong evidence to support the routine use corresponding reviews published in conventional paper of aspirin in almost all patients with acute ischaemic journals. 137 Where there is no relevant Cochrane review stroke (section 12.3). For anticoagulants, the substantial .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 660 660 Chapter 12 Specific treatments for acute ischaemic stroke amounts of available data do not support their routine tion) 144 (section 16.6.2). Similarly, there is substantial use, but it may be justifiable to use them in a relatively variation in the use of thrombolytic therapy for acute small proportion of patients for specific reasons (sec- ischaemic stroke in the USA 145,146 and Europe. 147 There tion 12.4). Similarly, there is reasonable evidence (based are many possible reasons for these variations, but the on relatively small trials) that thrombolytic treatment lack of really reliable evidence from appropriately large is indicated for the few, very carefully selected cases RCTs and the lack of widely disseminated systematic that meet the very restrictive conditions of the current reviews of existing trial evidence must play at least some regulatory approvals; large-scale RCTs will be needed part. For thrombolytic treatment, concern about the risk before its place in routine clinical practice worldwide of bleeding, emergency physicians’ perception of lack of becomes clear (section 12.5). There are many other inter- evidence of benefit and variation in provision of neuro- ventions which have been tested to some extent, but for logical support in the emergency department are some of which the evidence remains inconclusive. The remain- the physician-related factors in the variation. 147–149 The der of this chapter explores these broad conclusions barriers to effective delivery of thrombolytic therapy are in greater detail. Some treatments, such as intravenous discussed in section 17.4.4. thrombolysis, should only be given in the context of a highly organized specialist acute stroke service, whereas Substantial variation in the use of particular others, such as aspirin, could be used very widely, even treatments sometimes indicates lack of really reliable where healthcare resources are limited. evidence on the effects of that treatment; sometimes it merely represents reluctance of doctors or health systems to change. Whatever the cause, major 12.2.4 Variation in treatment worldwide variation in clinical practice or in the delivery of There are enormous variations in the general manage- a service is inequitable and ethically indefensible. ment of (and outcome after) acute stroke both between and within different countries. 138–141 There are also 12.2.5 Treatment of any specific underlying cause large variations in the use of specific agents such as anticoagulants, both as an acute treatment and for Although the great majority of ischaemic strokes are in secondary prevention. 141 For example, a survey of 36 US one way or another related to atheroma and its throm- academic medical centres showed that 29% of a sample boembolic complications (section 6.3), intracranial small of 497 patients with acute ischaemic stroke were treated vessel disease (section 6.4) and embolism from the heart with intravenous heparin, but the proportion treated (section 6.5), a small proportion are due to other, rarer varied enormously between centres; from ‘not at all’ in conditions. These unusual causes and their treatment are seven centres to 88% in one centre. 142 The causes of this discussed in Chapter 7. It is important to emphasize that, variation was explored in a further survey of opinion although conditions like vasculitis are infrequent, failure among 280 neurologists from the United States and 270 to recognize and treat them appropriately may lead to neurologists from Canada. US neurologists were signi- a poor outcome or even death. A systematic approach ficantly more likely than Canadian neurologists to use to history taking, examination and investigation will intravenous heparin for patients with stroke in evolu- minimize the risk of missing a potentially treatable cause tion (51% vs 33%), vertebrobasilar stroke (30% vs 8%), of ischaemic stroke. carotid territory stroke (31% vs 4%), and multiple tran- sient ischemic attacks (47% vs 9%). 143 US neurologists more often cited medicolegal factors as a potential influ- ence on the decision-making process than Canadian neurologists. 143 Despite the publication of four clinical 12.3 Routine use: aspirin trials, which have not shown any long-term benefit from heparin for patients with acute stroke, both US and Canadian neurologists would use intravenous heparin in 12.3.1 Rationale large numbers for this condition. 143 There is clearly a need to understand why some clinicians persisted in Potential benefits using a treatment which is not supported by reliable evid- ence (intravenous heparin in acute ischaemic stroke), yet On the arterial side, aspirin may act in several ways others for many years underused a proven treatment to reduce the volume of brain tissue damaged by ischae- (oral anticoagulants for long-term secondary prevention mia. It may prevent distal and proximal propagation after ischaemic stroke associated with atrial fibrilla- of arterial thrombus, and may prevent re-embolization .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 661 12.3 Routine use: aspirin 661 and platelet aggregation in the microcirculation. It also patients. In the IST, patients were allocated, in an open reduces the release of thromboxane and other neuro- factorial design, to treatment policies of: 300 mg aspirin toxic eicosanoids and so may even be neuroprotect- daily, heparin, the combination, or to ‘avoid both ive. 150,151 In the venous circulation, among patients at aspirin and heparin’ for 14 days. In the CAST, patients high risk (chiefly as a result of general or orthopaedic were allocated, in a double-blind design, to 1 month of surgery) antiplatelet drugs significantly reduce the risk of 160 mg aspirin daily or matching placebo. deep venous thrombosis and pulmonary embolism. 152 There are two main sources of evidence on aspirin in acute ischaemic stroke. The first is a Cochrane systematic review that included all the completed RCTs of any Potential risks antiplatelet drug in acute stroke and examined their However, antiplatelet drugs do have significant anti- effects on a variety of short- and long-term clinical out- haemostatic effects. In a systematic review of the comes. 156 The second is a meta-analysis of individual randomized trials, antiplatelet drugs were consistently patient data from CAST and IST that examined the associated with a small but definite excess of both effects of aspirin in particular categories of patient dur- intracranial and extracranial haemorrhages. 131 ing the scheduled treatment period. 157 12.3.2 Evidence Effects on major outcomes: recurrent stroke, intracranial haemorrhage, death and functional status Data available The effects of aspirin on various outcomes are summar- The benefit of aspirin as treatment for acute myocardial ized in Table 12.7. Aspirin significantly reduces the odds infarction was first reliably established in 1988. 153 The of recurrent ischaemic stroke during the treatment lack of data about the effects of antiplatelet drugs as period by 30% (95% CI 20–40%) from 2.3% in controls a treatment for the acute phase of ischaemic stroke led to 1.6% in treated patients, i.e. avoiding seven events to two large-scale RCTs of aspirin, the International per 1000 patients treated. 157 There is a small excess of Stroke Trial (IST) 154 and the Chinese Acute Stroke symptomatic intracranial haemorrhages with aspirin Trial (CAST), 155 which together randomized over 40 000 (including symptomatic transformation of an infarct); Table 12.7 Effects of aspirin, started within 24 h of an acute ischaemic stroke. a Events avoided/ caused per 1000 patients Benefits within the acute treatment period b Fatal or non-fatal pulmonary embolism 2 Death with no further stroke 4 Death from any cause 5 Fatal or non-fatal recurrent ischaemic stroke 7 Further stroke or death 9 Harms within the acute treatment period b Fatal or non-fatal intracranial haemorrhage 2 Major extracranial haemorrhage c 4 If not receiving heparin 2 If receiving heparin 9 Benefits at final follow-up b Death from any cause 8 Dead or dependent 13 Full recovery from the stroke 10 a Estimates derived from an individual patient data meta-analysis of the CAST and IST trials, and from a Cochrane systematic review; 156,157 the events are not mutually exclusive, so the various events cannot be summed. b The acute treatment period was 14 days in IST and 1 month in CAST, and final follow-up was at 6 months in the IST and 1 month in CAST. 157 c Major was defined as severe enough to require transfusion or to lead to death. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 662 662 Chapter 12 Specific treatments for acute ischaemic stroke in 0.8% of controls vs 1.0% of treated patients, a non- or the presence of thrombophilia, then the question is significant 21% relative increase in odds (95% CI 1% ‘what to add to aspirin?’ Graded compression stockings reduction to 49% increase), or an absolute increase of are one option and low-dose subcutaneous heparin about two per 1000 patients treated. 157 Aspirin signi- another; both have advantages and disadvantages, ficantly reduces the relative odds of death (without which are discussed in detail in section 11.13. further stroke) during the treatment period by 8% (95% CI 1–16%) from 5.4% in controls to 5.0% in treated Subgroup analyses patients, i.e. avoiding four deaths for every 1000 patients treated. 157 ‘Recurrent (further) stroke or death during the The individual patient data meta-analysis did not iden- treatment period’ conveniently summarizes the overall tify any group in which the benefits – or the risks – of balance of benefit and risk since it encompasses both aspirin were significantly greater than or less than the benefits and harms: recurrent ischaemic stroke, recur- averages reported above. 157 The recurrence risk among rent stroke of unknown type, symptomatic intracranial control patients was similar in all 28 subgroups, so the haemorrhage, symptomatic haemorrhagic transforma- absolute reduction of seven per 1000 did not differ sub- tion of the infarct and death from any cause. We refer stantially with respect to age, sex, conscious level, atrial to this outcome as ‘further stroke or death’. Aspirin fibrillation, CT findings, blood pressure, stroke subtype significantly reduces the relative odds of this outcome of or concomitant heparin use. There was no good evid- further stroke or death by 11% (95% CI 5–15%) from ence that the 11% reduction in relative odds of death 9.1% to 8.2%; for every 1000 patients treated, nine avoid without further stroke was reversed in any subgroup, or further stroke or death during the treatment period that in any subgroup the increase in haemorrhagic (Fig. 12.10). 157 At the end of follow-up, aspirin avoids stroke was much larger than the average of about two per about eight deaths for every 1000 patients treated, 156 1000, and there was no heterogeneity between the reduc- and also significantly reduces the odds of being dead or tions in further stroke or death during the scheduled dependent by 6% (95% CI 2–8%). The risk difference treatment period (Fig. 12.10). was 13 extra patients alive and independent for every 1000 patients treated. Aspirin also significantly increases Aspirin in haemorrhagic stroke and in patients without the odds of making a complete recovery by 6% (95% CI CT or MR brain scans 1–11%), an extra 10 patients making a complete recov- ery for every 1000 patients treated. 156 Among the 9000 patients randomized without a prior CT scan in the IST and CAST, aspirin appeared to be of net benefit, with no unusual excess of haemorrhagic stroke, Effect on deep venous thrombosis and pulmonary and among the 800 who had inadvertently been ran- embolism domized after a haemorrhagic stroke, there was no evid- Aspirin significantly reduced the relative odds of pul- ence of net hazard (further stroke or death 67 control vs monary embolism (PE) by 29% (95% CI 4–47%), from 63 aspirin). 157,159 The RCT data are broadly reassuring, 0.5% in controls to 0.3% in treated patients, i.e. for every in that they establish that the patients inadvertently 1000 patients treated, two avoided PE. 157 Though under- entered in the trials with a haemorrhagic stroke (who ascertainment of events in both groups may mean that then received only one or two doses of aspirin), were not, the absolute benefit has been underestimated, the pro- on average, harmed as a result. However, they do not portional reduction is not likely to be. If the true rate of establish the safety of continued aspirin treatment in PE were 3% in the controls, and the same proportional patients with intracerebral haemorrhage, nor do they reduction applied, then for every 1000 patients given establish the safety of giving aspirin to patients who are aspirin, 12 might avoid PE. These benefits are consistent not CT scanned at all. with those seen in the systematic review of aspirin in the prevention of PE in surgical patients 152 and in the 12.3.3 Who to treat Pulmonary Embolism Prevention Trial. 158 These data therefore strengthen the rationale for the Early aspirin is of benefit for a wide range of patients. The routine use of aspirin in the acute phase of a stroke and recommendation is therefore that all patients with sus- continuing it long-term; aspirin is likely to be adequate pected acute ischaemic stroke – irrespective of lesion loca- thromboprophylaxis for patients at low and moderate tion or presumed aetiology (e.g. cardioembolic or not) – risk of deep venous thrombosis (DVT) and PE. For receive aspirin unless there is a clear contraindication. 160 patients at high risk of DVT, perhaps because of a his- However, implementing a hospital policy or guideline of tory of a previous episode of venous thromboembolism ‘immediate aspirin for all patients with acute ischaemic .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 663 12.3 Routine use: aspirin 663 Recurrent stroke or death Odds ratio and during the treatment period Confidence interval Categorization Aspirin Control Aspirin better Aspirin worse Fig. 12.10 Aspirin in acute ischaemic Hours from stroke onset to randomization: stroke: effects on recurrent stroke or 0–6 323 (11.4%) 358 (12.9%) death in different subgroups during 7–12 379 (11.0%) 367 (10.3%) the scheduled treatment period in the 13–24 435 (7.7%) 505 (9.1%) 25–48 498 (6.2%) 589 (7.2%) Chinese Acute Stroke Trial (CAST) and Age (years): International Stroke Trial (IST). Fatal and <65 411 (5.3%) 441 (5.6%) non-fatal events are included. For each 65–74 498 (7.4%) 565 (8.5%) particular subgroup the number of events 75+ 727 (13.2%) 811 (14.7%) Sex: among aspirin- and no-aspirin-allocated Male 872 (7.5%) 997 (8.4%) patients, and the odds ratio (dark purple Female 769 (9.1%) 826 (10.1%) square, with area proportional to the total Level of consciousness: Alert 817 (5.0%) 954 (5.8%) number of patients with an event) and its Drowsy/coma 820 (23.0%) 865 (24.2%) 99% confidence interval (horizontal line) Atrial fibrillation: are given. A square to the left of the solid Present 421 (18.2%) 420 (18.7%) Absent 1151 (6.7%) 1337 (7.8%) vertical line of no treatment difference CT findings (odds ratio 1.0) suggests benefit, Infarct visible 839 (7.7%) 939 (8.6%) significant at 2P < 0.01 only if the whole Infarct not visible 408 (8.7%) 457 (9.8%) No prior CT 385 (8.6%) 421 (9.5%) 99% confidence interval (CI) (horizontal Systolic BP (mmHg): line) is to the left of the solid vertical <130 216 (9.4%) 249 (10.8%) line. The overall result and its 95% CI is 130–159 595 (7.8%) 632 (8.5%) 160–189 559 (7.8%) 637 (8.8%) represented by an open diamond. A 190+ 268 (9.1%) 302 (10.1%) indicates aspirin; H, heparin. Here and Stroke syndrome: elsewhere, results for those with missing Lacunar 168 (3.1%) 215 (4.0%) information on particular characteristics Non-lacunar 1470 (10.0%) 1606 (11.0%) Allocated trial heparin: are not listed separately (except for Yes (IST: A+H vs H) 531 (10.9%) 604 (12.4%) computed tomography findings), but No (IST: A vs nil) 565 (11.6%) 605 (12.4%) numerators and denominators for them No (CAST: A vs nil) 545 (5.3%) 614 (5.9%) Overall prognostic index (three equal groups): can be obtained by subtraction of the Good 167 (2.6%) 189 (2.9%) subgroup results from the total (e.g. the Average 336 (5.2%) 404 (6.3%) numbers with no prognostic index Poor 1060 (16.4%) 1155 (17.8%) Days from randomization to further stroke or death: calculated were 16/638 aspirin vs 18/638 0–1 327 (1.6%) 320 (1.6%) control). Reproduced from ‘Indications 2–7 757 (3.8%) 884 (4.4%) for early aspirin use in acute ischemic +8 557 (2.8%) 619 (3.1%) stroke: a combined analysis of 40 000 Proportional randomized patients from the Chinese reduction 11% SD 3 Acute Stroke Trial and the International Total 1641 (8.2%) 1823 (9.1%) (2P = 0.001) Stroke Trial’. Stroke 2000; 31:1240–9 with permission from the authors 99% or 95% Cls 0.5 0.75 1.0 1.25 1.5 (Chen et al.). 157 stroke’ requires considerable effort; it is of course part of allocated to aspirin. The one-third reduction in the a well-organized stroke service (Chapter 17). Several dif- relative odds of recurrent ischaemic stroke with aspirin ferent strategies may be required to maintain a high level was no different to that seen in patients without AF; 157 of compliance with the policy (Table 17.22). all such patients should be started on aspirin. Atrial fibrillation Patients already on antiplatelet drugs The IST and CAST trials included about 4500 patients The priority is to perform a brain CT or magnetic reson- who were in atrial fibrillation (AF) at the time of random- ance (MR) scan to determine whether the stroke is ization. In these patients, the risk of recurrent stroke in haemorrhagic or ischaemic. If the scan shows that the hospital was 2.9% in the controls and 2.0% in patients stroke was due to haemorrhage, the aspirin should .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 664 664 Chapter 12 Specific treatments for acute ischaemic stroke generally be stopped. If the scan shows that the stroke those randomized early (within the first 0–6 h), 157 so was ischaemic, but there is some petechial haemorrhagic aspirin should be started as soon as a CT or MR scan has transformation, it is probably reasonable to continue the excluded intracranial haemorrhage as the cause of the aspirin. However, there is no good evidence to support stroke. A policy of ‘CT scan immediately and give aspirin these suggestions for management! The question of what if haemorrhage is excluded’ is the most cost-effective antiplatelet regimen to use for long-term secondary strategy in this setting. 163 Moreover, if the doctor who prevention in patients who have a stroke while already admits the patient to hospital writes the prescription for on aspirin or other antiplatelet drugs is discussed in aspirin immediately, the patient is more likely to receive section 16.5.4. long-term aspirin. If writing the aspirin prescription is left until later, it is easily forgotten. If CT scanning is not immediately available and the clinician is confident, 12.3.4 Who not to treat based on the patient’s clinical features, that the stroke is From first principles, it is not logical to treat a bleeding unlikely to be haemorrhagic (i.e. there is no history of an disorder (intracerebral haemorrhage) with aspirin or any ‘apoplectic onset’ of the stroke with early headache or other antiplatelet drugs (Chapter 12). There is however, vomiting, the patient is fully conscious, etc.) (section little reliable evidence on the effects of aspirin and 5.3.1), then aspirin can be started while the CT or MR other antiplatelet agents in such patients. As mentioned scan is being organized. This policy does not appear to above, there was no clear evidence of harm from a few reduce the benefits from aspirin. 157 doses of aspirin while waiting for CT scanning to be per- If the patient is being considered for thrombolysis, it formed. 157,159 A retrospective observational study has may be necessary to delay the start of aspirin treatment. suggested that, in patients who have an intracerebral For thrombolysis with streptokinase, the risk of intra- haemorrhage, being on aspirin at the time of the event cranial haemorrhage is increased if it is given together increases the chances of haematoma expansion (and, with aspirin. 133,164 There is no good-quality RCT evidence in this series, that often led to surgical evacuation of on whether aspirin increases the risk of bleeding when the haematoma). 161 A prospective cohort study in 207 recombinant tissue plasminogen activator (rt-PA) is the survivors of intracerebral haemorrhage, of whom 46 (22%) thrombolytic agent, though there is weak evidence from were prescribed antiplatelet agents (most commonly a non-randomized case series suggesting that giving for prevention of ischaemic heart disease), found that rt-PA in patients already on aspirin is not unduly haz- antiplatelet use was not significantly associated with ardous. 165 Despite the lack of good RCT evidence about recurrent intracerebral haemorrhage; in survivors of the combination of aspirin and rt-PA, UK and US guide- either lobar haemorrhage (hazard ratio [HR] 0.8; 95% lines recommend, in patients to be treated with rt-PA, CI 0.3–2.3) or of deep haemorrhage (HR 1.2; 95% CI that aspirin treatment is only started 24 h after throm- 0.1–14.3). 162 The authors concluded that, although bolysis. 166,167 antiplatelet use is relatively common following intra- cerebral haemorrhage, it did not appear to be associated Start aspirin immediately even if there is likely to be with a large increase in risk of recurrent intracerebral some delay before a CT or MR scan can be performed haemorrhage. Therefore, patients with intracerebral and if you think intracranial haemorrhage is an haemorrhage who have a very clear and pressing indica- unlikely cause of the stroke. However, delay starting tion to continue aspirin (e.g. unstable angina) and are aspirin in patients to be treated with thrombolysis thought – for whatever reason – to have a low risk of until the day after the treatment has been given. further intracerebral bleeding can probably continue aspirin. 159,162 Furthermore, it is not clear, if aspirin is to 12.3.6 Agent/dose/route be used after intracerebral haemorrhage, whether it is best to stop it for a few days or to just continue without We use aspirin in a dose of at least 150 mg daily in the interruption. Patients with a history of definite aspirin acute phase of stroke. The initial dose has to be high (and sensitivity (e.g. wheeze or skin rash on exposure to certainly higher than is required for long-term second- aspirin) should not be given aspirin. ary prevention) to inhibit thromboxane biosynthesis as quickly and completely as possible. 168,169 For patients who can swallow safely (section 11.17), aspirin can be 12.3.5 When to start given by mouth; for the remainder, it can be given rect- There is no clear evidence of a ‘time window’ for the ally by suppository, 170 by nasogastric tube, or by intra- benefit of aspirin; the relative benefits among those ran- venous injection (as 100 mg of the lysine salt, infused domized late (24–48 h after stroke onset) are as great as over 10 min). .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 665 12.4 Selective use: anticoagulants 665 or defibrinogenating agent could therefore theoretically 12.3.7 Adverse effects increase the tendency to intracranial bleeding and this Major extracranial haemorrhage (defined as bleeding might offset some, or all, of any benefits. serious enough to cause death or require transfusion) The clinical impact of haemorrhagic transformation is the most frequent serious adverse effect. In the trials, over and above the effects of the original infarct (i.e. the the relative increase in odds was large (68%; 95% CI independent contribution of haemorrhagic transforma- 34–109%), but the absolute excess was small – four tion to the likelihood of a poor outcome) is difficult to additional major extracranial haemorrhages for every assess in an individual patient. Minor degrees of trans- 1000 patients treated. 156 The excess was greater among formation can occur without any clinical deterioration, patients allocated heparin (0.9% heparin alone vs 1.8% whereas the development of a large parenchymatous allocated aspirin plus heparin; excess nine per 1000) haematoma may be fatal. However, even if a specific than among other patients (0.5% among those allocated category of patient were at especially high risk of sym- no aspirin and 0.7% allocated aspirin; excess two per ptomatic haemorrhagic transformation, which might 1000). 157 The risk of adverse events with aspirin, both in be exacerbated by an antihaemostatic agent, this does the short and long term, is best kept to a minimum by not necessarily mean that the net balance of risk and avoiding combined treatment with anticoagulants 171 benefit will be adverse for that type of patient. By ana- (section 16.6.2 on the use of combined treatment in logy with carotid endarterectomy for carotid stenosis, patients with prosthetic heart valves). some patients are at high risk of stroke with surgery, but even higher risk without it, so that the net balance of risk and benefit may be favourable (section 16.11.6), even in high-risk individuals; this may be true for throm- bolysis as well (section 12.5). 12.4 Selective use: anticoagulants As well as increasing the risk of haemorrhagic trans- formation, anticoagulants could both increase the risk of symptomatic intracranial haemorrhage arising de novo (as intracerebral, subarachnoid or subdural bleeding) and 12.4.1 Rationale the risk of bleeding at other, extracranial sites. The key question is therefore ‘do the benefits of anticoagul- Potential benefits ants in the acute phase of ischaemic stroke treatment The rationale for using anticoagulants applies to both outweigh the risks?’ Randomized trials provide the best the arterial and venous circulations. In large arteries, and means to assess this balance. in the perforating arteries involved in lacunar infarction, the aim of treatment is to prevent local propagation of 12.4.2 Evidence any occluding thrombus (or embolus), to tip the balance in favour of spontaneous lysis of the occlusion, and to Data available prevent early re-embolization from any proximal arterial or cardiac sources. In small arteries and the microvessels, By 2004, 22 RCTs comparing anticoagulants with control anticoagulation might also prevent sludging which may had been completed and were available for review. 172 contribute to ischaemia in the penumbral zone around Although the review has not been updated since then, the infarct core. In the venous circulation, anticoagula- no further RCTs of anticoagulants vs control eligible tion should reduce the risk of deep venous thrombosis for inclusion in the review have been published. The and pulmonary embolism which are common complica- trials included in the review tested standard unfraction- tions of immobility after stroke (section 11.13). ated heparin, low-molecular-weight heparin, heparinoid, direct thrombin inhibitors and two tested heparin given for just 24 h followed by oral anticoagulation. 172 Most Potential risks of the data came from trials in which unfractionated As a cerebral infarct evolves over the first few days, red heparin was administered by subcutaneous injection in cells can leak from microvessels. Minor degrees of leak- medium (12 500 IU twice daily) or low dose (5000 IU age appear on CT or MR brain scanning as petechiae, but twice daily). In total, the trials included 23 547 patients if bleeding is more extensive, a parenchymal haematoma with acute presumed ischaemic stroke. 172 Most patients may form. This pathophysiological process is known as had a CT scan to exclude intracerebral haemorrhage haemorrhagic transformation of cerebral infarction (sec- before treatment was started. Patients were generally tion 5.7). Any anticoagulant, antiplatelet, thrombolytic randomized within 48 h of stroke onset, and treatment .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 666 666 Chapter 12 Specific treatments for acute ischaemic stroke continued for about 2 weeks. There were fewer trials heparin with aspirin; 180 and argatroban, a direct thrombin directly comparing one agent with another. 173 Another inhibitor, with control (in which all patients received review focused on the trials comparing low-molecular- 81–350 mg aspirin within the first 48 h). 181 In our view, weight heparins or heparinoids with control in acute the results of these later studies do not alter the overall ischaemic stroke. 174 There have been a smaller number conclusions about the effects of anticoagulants in acute of trials comparing anticoagulants with antiplatelet ischaemic stroke. agents; a review included data from four trials on 16 558 patients comparing either unfractionated heparin Effects on major outcomes: recurrent stroke, intracranial or low-molecular-weight heparin with aspirin. 175 Since haemorrhage, death and functional status those reviews were completed, a number of small trials have been published, comparing: intravenous full-dose The Cochrane review, based on nine trials comparing heparin with aspirin; 176 intravenous full-dose heparin anticoagulants with control (22 570 patients) showed no with low-dose heparin; 177 low-molecular-weight heparin evidence of net benefit within the scheduled treatment with unfractionated heparin; 178,320 different doses of period (Fig. 12.11) or at the end of follow-up (Fig. 12.12). low-molecular-weight heparin; 179 low-molecular-weight Although anticoagulants were associated with about Events/Patients Odds ratio & CI Odds Anticoagulant regimen Anticoagulant Control reduction Recurrent ischaemic (or unknown) stroke Unfractionated heparin s.c. 283/9781 370/9785 (2.9%) (3.8%) Unfractionated heparin i.v. 0/24 2/21 (0.0%) (9.5%) Low-molecular-weight 6/227 6/135 heparin s.c. (2.6%) (4.4%) Heparinoid s.c. 1/78 0/54 (1.3%) (0.0%) Heparinoid i.v. 7/646 7/635 (1.1%) (1.1%) Thrombin inhibitor i.v. 1/69 0/69 (1.4%) (0.0%) Oral anticoagulant 2/41 3/40 (4.9%) (7.5%) Fig. 12.11 Anticoagulants in acute Total 24% SD 7 (2P > 0.0003) ischaemic stroke: relative effects within Symptomatic intracranial haemorrhage the treatment period. Results of a Unfractionated heparin s.c. 120/9816 41/9815 systematic review of the 21 randomized (1.2%) (0.4%) Unfractionated heparin i.v. 0/136 0/134 trials comparing heparin with control (0.0%) (0.0%) groups in patients with acute ischaemic Low-molecular-weight 28/825 9/466 heparin s.c. (3.4%) (1.9%) stroke: effects on recurrent ischaemic Heparinoid s.c. 1/78 0/54 stroke, symptomatic intracranial (1.3%) (0.0%) Heparinoid i.v. 10/646 30/635 haemorrhage and death, all within the (1.5%) (0.5%) scheduled treatment period. The estimate Thrombin inhibitor i.v. 1/69 0.69 (1.4%) (0.0%) of treatment effect for each heparin Oral anticoagulant 3/26 1/25 regimen is expressed as an odds ratio (11.5%) (4.0%) Total –152% SD 23 (solid square) and its 99% confidence (2P > 0.0001) adverse) interval (horizontal line). The size of the Death in the scheduled treatment period square is proportional to the amount of Unfractionated heparin s.c. 889/9872 916/9871 (9.0%) (9.3%) information available. An odds ratio of 1.0 Unfractionated heparin i.v. 2/136 3/134 corresponds to a treatment effect of zero, (1.5%) (2.2%) Low-molecular-weight 33/324 16/231 an odds ratio of less than 1.0 suggests heparin s.c. (10.2%) (6.9%) treatment is better than control, and an Heparinoid s.c. 7/78 6/54 (9.0%)) (11.1%) odds ratio of greater than 1.0 suggests Heparinoid i.v. 12/646 9/635 treatment is worse than control. The (1.9%) (1.4%) Thrombin inhibitor i.v. 1/173 2/121 overall result and its 95% CI is represented (4.6%) (1.7%) Oral anticoagulant 12/14 10/40 by an open diamond. The figures given to (29.3%) (25.0%) the right are relative odds reductions (SD). Total 1% SD 5 (Drawn from data from Sandercock, (2P > 0.1;NS) 2003.) 156 NS, not significant; SD, standard 0.1 0.25 0.5 1.0 2.0 4.0 10.0 99% or 95% confidence interval Anticoagulant Anticoagulant deviation; s.c., subcutaneous; better worse i.v., intravenous. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 667 12.4 Selective use: anticoagulants 667 Events/Patients Odds ratio & CI Odds Anticoagulant regimen Anticoagulant Control reduction Death at the end of follow-up Unfractionated heparin s.c. 2196/9861 2129/9879 (22.3%) (21.6%) Unfractionated heparin i.v. 17/112 8/113 (15.2%) (7.1%) Low-molecular-weight 178/723 88/355 heparin s.c. (24.6%) (24.8%) Heparinoid s.c. 4/50 4/25 (8.0%) (16.0%) Heparinoid i.v. 42/646 38/635 (6.5%) (6.0%) Thrombin inhibitor i.v. not not evaluated evaluated Oral anticoagulant 8/26 7/25 (30.8%) (28.0%) Total –5% (2P > 0.1:NS Death or dependency at the end of follow-up adverse) Fig. 12.12 Anticoagulants in acute Unfractionated heparin s.c. 6063/9717 6062/9718 ischaemic stroke: relative effects at the (62.4%) (62.4%) Unfractionated heparin i.v. not not end of follow-up. Results of a systematic evaluated evaluated review of the five randomized trials Low-molecular-weight 400/723 210/355 heparin s.c. (55.3%) (59.2%) comparing anticoagulants with control Heparinoid s.c. 13/28 15/29 groups in patients with acute ischaemic (46.4%) (51.7%) Heparinoid i.v. 159/641 167/635 stroke reporting long-term outcome: (24.8%) (26.3%) effects on death from all causes, and Thrombin inhibitor i.v. not not evaluated evaluated death or dependency. Same conventions Oral anticoagulant not not as Fig. 12.11. (From Gubitz et al. evaluated evaluated Total 1% ‘Anticoagulants for acute ischaemic (2P > 0.1:NS) stroke’, Cochrane Database Systematic Reviews 2004: (3) 172 copyright Cochrane 0.0 0.5 1.0 1.5 2.0 Collaboration, reproduced with 99% or 95% confidence interval Anticoagulant Anticoagulant permission.) better worse nine fewer recurrent ischaemic strokes per 1000 patients comparisons of different dosing regimens showed treated (OR 0.76; 95% CI 0.65–0.88), they were also consistently higher bleeding risks with higher dose regi- associated with a similar sized nine per 1000 increase in mens. In the IST, patients allocated to subcutaneous symptomatic intracranial haemorrhages (OR 2.52; 95% unfractionated heparin were randomized to high dose CI 1.92–3.30) (Fig. 12.11). There was no evidence that (12 500 IU twice daily) or to low dose (5000 IU twice anticoagulants reduced the odds of death from all causes daily) and the proportions with symptomatic intra- at the end of follow-up (OR 1.05; 95% CI 0.98–1.12) cranial haemorrhage were 1.8% and 0.7%, respectively – (Fig. 12.12); 172 if anything, there was a non-significant a highly significant 11 per 1000 excess with the higher excess of deaths with anticoagulants. Similarly, there dose (2P < 0.00001). 154 was no evidence that anticoagulants reduced the odds As far as extracranial bleeds are concerned, for of being dead or dependent at the end of follow-up every 1000 acute ischaemic stroke patients treated with (OR 0.99; 95% CI 0.93–1.04) (Fig. 12.12). anticoagulants, about nine have a major extracranial haemorrhage. The indirect comparisons of different agents showed that the bleeding risks were higher with higher Bleeding risks with different regimens dose regimens. In the IST, the risk of major extracranial In the Cochrane review, each of the regimens tested, bleeds was 2% among patients allocated high-dose and when compared with control, appeared to increase the 0.6% among those allocated low-dose heparin – a highly risks of both symptomatic intracranial haemorrhage and significant 14 per 1000 excess with the higher dose extracranial haemorrhage. 172 The relative increase in (2P < 0.00001). symptomatic intracranial haemorrhage was consistent A systematic review of all trials directly comparing across the different regimens, although (because of small high- with low-dose anticoagulants in acute ischaemic numbers) it was only statistically significant for sub- stroke supported the finding that the bleeding risks were cutaneous unfractionated heparin (Fig. 12.11). Indirect dose-dependent for both intra-and extracranial bleeds. 182 .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 668 668 Chapter 12 Specific treatments for acute ischaemic stroke The RCTs published since then comparing different doses by a higher risk of early recurrent ischaemic stroke. of low-molecular-weight heparin in acute ischaemic In patients with AF, the absolute risk of early recurrent stroke also showed that intracranial bleeding was stroke is low (in the IST, about 5% of patients in AF dose-dependent. 179,183 A recent open RCT comparing full- allocated to ‘avoid heparin’ had a recurrent stroke of dose intravenous heparin with low-dose subcutaneous ischaemic or unknown type within 14 days), and there heparin, started within 3 h of stroke onset, also showed was no net advantage to treatment with heparin. 184 The a higher risk of intracranial haemorrhages with the Cochrane review showed no statistically significant dif- higher dose (6.2% vs 1.4%). 177 ference in the effect of treatment on death or depend- ence at final follow-up if a cardiac source of embolism was suspected (OR 1.00; 95% CI 0.85–1.18) or not sus- Deep venous thrombosis and pulmonary embolism pected (OR 1.00; 95% CI 0.94–1.06) to be the cause of the Data on the effects of anticoagulants on deep venous stroke. 172 The small Heparin in Acute Embolic Stroke thrombosis (DVT) are only available for 916 acute stroke Trial (HAEST) compared a low-molecular-weight heparin patients. There was heterogeneity between the trials, with aspirin in 449 patients with acute ischaemic stroke which makes it harder to give a reliable overall estimate of suspected cardioembolic origin and did not show any of treatment effect. Overall, symptomatic or asymp- advantage to low-molecular-weight heparin. 185 These tomatic DVT occurred in 43% of controls and 15% of data do not support the widespread use of either inten- treated patients, a highly significant 79% reduction in sive (i.e. full-dose intravenous) unfractionated or low- relative odds with anticoagulants (95% CI 61–85%). molecular-weight heparin regimens in the acute phase Thus, for every 1000 patients treated, 280 avoid DVT. of ischaemic stroke associated with AF. 160 Fatal or non-fatal pulmonary embolism (PE) was not systematically sought in the trials. It was reported in In patients with suspected acute cardioembolic only 0.9% of controls and 0.6% of treated patients, a ischaemic stroke, there is no evidence to support significant 39% reduction in relative odds with anti- the routine use of immediate anticoagulation with coagulants (95% CI 17–55%), i.e. for every 1000 patients heparin, heparinoid or low-molecular-weight treated, three avoid PE. It is difficult to judge whether heparin. Aspirin is a safe and effective alternative the reductions in DVT or PE are dose-dependent from to anticoagulants, which reduces the risk of early the indirect comparisons. In the IST, fatal or non-fatal recurrent stroke in patients with acute ischaemic PE occurred in 0.4% of those allocated high dose and stroke and atrial fibrillation. 0.7% of those allocated low dose, a non-significant difference. 154 A systematic review of all of the direct ran- Progressing hemispheric stroke domized comparisons confirmed the greater reduction in PE with higher doses, but the absolute benefit was very By far the most important step is to find out why the small. 182 If we allow for the likely under-ascertainment stroke is ‘progressing’. There are numerous causes besides of pulmonary embolism (section 12.3.2), and assume propagating cerebral arterial thrombosis for a worsening that the true risk in the controls was 3%, and then apply neurological deficit (section 11.5) (Table 11.7). These the same 39% proportional reduction (i.e. from 3% to factors should be identified and – if possible – treated. For 1.85%), for every 1000 patients treated about 12 would patients in whom all other possible causes for progres- avoid PE. However, even if the benefit is that large, it will sion have been excluded, many textbooks and reviews still be substantially offset, since an extra nine patients recommend immediate intravenous heparin. However, will have a major extracranial haemorrhage associated there have not been any trials of intravenous heparin with anticoagulants. specifically in patients with progressing stroke, and neither the trials of intravenous unfractionated heparin nor of intravenous heparinoid in patients with acute Suspected cardioembolic ischaemic stroke 172 ischaemic stroke showed net benefit overall. There is All patients with atrial fibrillation (AF) and ischaemic therefore no direct or indirect evidence to support the stroke should have a CT brain scan, since about 5–10% use of heparin (either intravenous or subcutaneous) in turn out to have intracerebral haemorrhage. A detailed this particular type of patient. 160 analysis of the 3169 patients in IST with acute ischaemic stroke and AF confirmed that such stroke patients have a Basilar thrombosis higher risk of early death than those in sinus rhythm. 184 This higher death rate was accounted for by the older The IST included over 2000 patients with posterior circu- age and larger infarcts among those with AF but not lation ischaemic infarcts, and there was no evidence .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 669 12.4 Selective use: anticoagulants 669 that the effects of treatment in this subgroup were any 12.4.3 Who to treat different to those seen in the trial overall. 154 However, it is likely that only a small proportion had occlusion Although there is no indication to use anticoagulants of the basilar artery. Further trials of anticoagulants routinely in patients with acute ischaemic stroke, 160,166,191 in ‘vertebrobasilar territory infarcts’ are probably not there are still some occasions where they may be warranted; but a trial focused on patients with proven appropriate; these are listed in Table 12.8 and discussed basilar occlusion might be justified, although trials below. The use of heparin after thrombolysis to prevent which seek to recruit a type of patient only rarely re-occlusion of the opened cerebral artery is discussed encountered in clinical practice are notoriously difficult in section 12.5.2. to do. This lack of evidence led a recent guideline to state that anticoagulants were ‘not recommended’ for this Heparin, when used for the immediate treatment indication. 160 of patients with acute ischaemic stroke, has risks (haemorrhage in the brain and elsewhere) which exactly cancel out the benefits (fewer recurrent Acute stroke in patients with acute myocardial ischaemic strokes and less venous thromboembolism). infarction We therefore do not use this treatment routinely. (See also section 7.10.) The first step in management is to determine whether the stroke was due to intra- Selected patients at high risk of deep venous thrombosis cerebral haemorrhage, either spontaneous or iatrogenic (e.g. caused by thrombolytic or anticoagulant treatment). Factors which increase the risk of deep venous thro- Some patients may have iatrogenic ischaemic strokes mbosis (DVT) and pulmonary embolism (PE) after stroke due to particulate emboli reaching the brain as a com- include: hemiplegia or immobility; dehydration; a his- plication of some invasive cardiological procedure such tory of previous episodes of venous thromboembolism; as coronary angiography or angioplasty. The value of and a known thrombophilia or active cancer. 191 Simple, anticoagulants in these latter patients is not clear, as reliable methods to predict which patients are most the embolic material is often atheromatous debris from likely to develop DVT would be very useful. the large arteries rather than fresh thrombus or platelet Guidelines vary in their recommendations about aggregates. Patients with full-thickness anterior myo- whether or not heparin should be used for DVT preven- cardial infarction (MI) have a higher than average risk tion. The Scottish Intercollegiate Guidelines Network of developing left ventricular (LV) thrombus (section recommended aspirin for all patients with ischaemic 7.10). Although two small trials 186,187 showed that stroke and that low-dose subcutaneous heparin should medium-dose subcutaneous heparin (12 500 IU twice only be added to aspirin in patients at high risk of DVT and daily) reduced the frequency of LV thrombus, an over- PE. 191 The Joint American Stroke Association/American view of the 26 trials of anticoagulants in acute MI Academy of Neurology (ASA/AAN) guideline states (including 73 000 patients) found little evidence of any ‘Subcutaneous unfractionated heparin, low molecular significant further net clinical benefit (in terms of major weight heparins, and heparinoids may be considered for clinical events) from adding either subcutaneous or DVT prophylaxis in at-risk patients with acute ischaemic intravenous heparin to the treatment of patients who stroke, recognising that non-pharmacologic treatments are given aspirin. 188,189 The value of anticoagulants in a for DVT prevention also exist. A benefit in reducing the patient with an acute MI complicated by acute ischaemic incidence of PE has not been demonstrated. The relative stroke is therefore unclear. benefits of these agents must be weighed against the risk Oral anticoagulants, given to stroke-free MI survivors of systemic and intracerebral haemorrhage.’ 160 for several months, probably reduce the risk of having It is not clear whether low-dose subcutaneous heparin a first stroke in the longer term, but given the cost and can add useful benefit to aspirin for the prevention of inconvenience of oral anticoagulants and the uncer- DVT and PE without causing an increase in bleeding tainty about the optimal intensity of treatment, aspirin complications (Table 12.7 gives an estimate of the is probably the long-term antithrombotic agent of increase in risk of extracranial bleeding when subcut- choice in most patients. 189,190 However, 6 months of oral aneous heparin is added to aspirin). 154 Other strategies anticoagulants may be worthwhile in patients with MI for prevention of DVT and PE (e.g. rehydration, early complicated by acute ischaemic stroke thought to be mobilization, graded compression stockings) and the due to embolism from the left ventricle, especially if management of stroke patients who develop symptoms congestive heart failure, AF or extensive LV dysfunction and signs suggestive of DVT or PE after admission are are present. 189 discussed in section 11.13. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 670 670 Chapter 12 Specific treatments for acute ischaemic stroke Table 12.8 Patients with acute ischaemic Anticoagulants worthwhile stroke of suspected cardioembolic origin Transient ischaemic attack or mild ischaemic stroke with complete recovery within who should be considered for immediate 1–2 days and in atrial fibrillation (start as soon as possible) anticoagulation, and those who should Non-disabling ischaemic stroke, with or without minor (petechial) haemorrhagic have anticoagulants deferred (or avoided transformation and in atrial fibrillation (best time to start unclear) completely). Anticoagulants probably worthwhile Acute myocardial infarction within past few weeks and confirmed ischaemic stroke or transient ischaemic attack Disabling ischaemic stroke and atrial fibrillation Anticoagulants not worthwhile Low risk of recurrent cardioembolic stroke without anticoagulants: Low-risk cardiac lesion (e.g. hypertrophic cardiomyopathy, minor degree of mitral leaflet prolapse) Other, non-cardiac lesion, a more likely cause of the ischaemic stroke Little to gain from long-term anticoagulation: Already severely disabled before the stroke Moribund High risk of cerebral haemorrhage on anticoagulants: Infective endocarditis Large cerebral infarct with midline shift Likely to comply poorly with anticoagulation Severe, uncontrolled hypertension Major haemorrhagic transformation on brain CT Absolute or relative contraindication to anticoagulants (Table 12.9) not recommended for the prevention of early recurrent The choice of method for deep venous thrombosis ischaemic stroke in acute ischaemic stroke presumed prophylaxis largely depends on its perceived risk, the due to cardioembolic stroke. 160 To help decide whether likely duration of immobility, and any predisposing (and when) to use anticoagulants in the acute phase, one factors. For patients with mild strokes who mobilize should therefore take into account the factors that are within a day or so, and have no other predisposing likely to increase the risk of haemorrhagic transforma- factors for deep venous thrombosis, aspirin is tion of the infarct (section 12.4.5), the likely risk of recur- probably adequate. For immobile patients with rent ischaemic stroke within the first week or so, and the predisposing factors (and who do not have bad severity of the stroke (sections 12.4.5 and 16.6.6). If anti- arterial disease or sensory loss in the legs), graded coagulants are indicated, see section 12.4.5 on when to compression stockings may provide additional start and section 12.4.6 for the choice of agent. benefit. If stockings are not acceptable, then low-dose subcutaneous unfractionated heparin in addition to aspirin is an alternative. 12.4.4 Who not to treat Relative contraindications and cautions about the use Atrial fibrillation of anticoagulants are given in Table 12.9. Any patient Patients with atrial fibrillation (AF) who have had a with acute stroke due to established intracranial hae- stroke or transient ischaemic attack are likely to benefit morrhage, or who has not had a CT brain scan to exclude from long-term oral anticoagulants as secondary preven- intracranial haemorrhage, should not be given anticoagu- tion, provided there are no contraindications (Table 12.8) lants. The evidence on the effects of heparin in patients (see section 16.6 for further details on the selection with haemorrhagic stroke is scant but suggests – as one of patients with AF most likely to benefit). However, as might expect – it may well be associated with some harm, discussed above, both fixed-dose subcutaneous unfrac- though the confidence interval is wide. 159 Warfarin tionated heparin, adjusted-dose intravenous unfrac- should not be used during pregnancy (Table 12.9 and tionated heparin and low-molecular-weight heparin are section 16.6.3). .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 671 12.4 Selective use: anticoagulants 671 Table 12.9 Contraindications to, and cautions with, full-dose would wait at least 1–2 weeks before starting. We would anticoagulants (these depend on individual circumstances and then consider the relative contraindications listed in are seldom absolute) (adapted from Scottish Intercollegiate Table 12.9 and, if necessary, delay the start of antico- Guidelines Network, 1999). 193 agulants further (or abandon the idea altogether). When oral anticoagulants are started in this way (i.e. without Uncorrected major bleeding disorder aggressive loading doses) some time after the acute Thrombocytopenia event, concomitant heparin (to overcome any transient Haemophilia prothrombotic state associated with the start of war- Liver failure Renal failure farin) is probably not needed. 193 In patients with tran- Uncontrolled severe hypertension sient ischaemic attack or with an ischaemic stroke which Systolic blood pressure over 200 mmHg resolves within a day or so, the risk of haemorrhagic Diastolic blood pressure over 120 mmHg transformation is probably negligible; if such a patient Potential bleeding lesions has a clear indication for the use of long-term antico- Active peptic ulcer agulants (most commonly AF), we would start treatment Oesophageal varices immediately. Unruptured intracranial aneurysm (especially if large and untreated) In a patient with a disabling acute ischaemic stroke Proliferative retinopathy and atrial fibrilation, the risk of stroke recurrence Recent organ biopsy within the first 2 weeks is not high, so we would start Recent trauma or surgery to head, orbit or spine Confirmed intracranial or intraspinal bleeding aspirin immediately, avoid anticoagulants for the Evidence on MR or CT of previous intracranial bleeding first week or two, or longer if we felt the risk of Recent stroke, but patient has not had brain CT scan or MRI haemorrhagic transformation of the infarct remained Recent stroke, with large infarct on brain CT or MRI (i.e. high (e.g. large infarct on CT with mass effect). high risk of haemorrhagic transformation) Patients with transient ischaemic attack or very minor History of heparin-induced thrombocytopenia or thrombosis ischaemic stroke and atrial fibrilation can be started (if heparin planned) on both aspirin and oral anticoagulants immediately. If warfarin planned If aspirin is given with anticoagulants, it should be Pregnancy stopped once the international normalized ratio is Homozygous protein C deficiency (risk of skin necrosis) in the therapeutic range. History of warfarin-related skin necrosis Uncooperative/or unreliable patients (long-term therapy) Risk of falling 12.4.6 Agent/dose/route/adverse effects Unable to monitor anticoagulation intensity Agent Oral anticoagulants do not act sufficiently fast to have a major effect within the first few hours of onset of an 12.4.5 When to start ischaemic stroke. Anticoagulants must be given by injec- The available trial data do not provide reliable evidence tion if they are to achieve a rapid effect. Most of the on the best time to start anticoagulants. So, in the occa- evidence on heparin in acute stroke comes from trials sional circumstance where we do feel compelled to use which used unfractionated heparin. 172 Though other them in acute stroke (section 12.4.3), the decision must agents have been tested, systematic reviews of the direct be based on the likely risk of events without treat- and indirect comparisons of different regimens did not ment, and the risk of haemorrhagic transformation with provide evidence that any one agent was better than treatment. unfractionated subcutaneous heparin. 172,173 In choosing In the IST, the frequency of recurrent ischaemic or between the different heparin regimens and aspirin, the undefined stroke within 14 days was not significantly ASA/AAN guideline states ‘Intravenous unfractionated different in those with and without AF: 3.9% among heparin or high-dose low molecular weight heparin those with AF vs 3.3% without. 154,184 Symptomatic or heparinoids are not recommended for any specific haemorrhagic transformation of the infarct occurs most subgroup of patients with acute ischaemic stroke that is commonly during the first 2 weeks and the risk is highest based on any presumed stroke mechanism or location in patients with large infarcts 192 and in patients given (e.g., cardioembolic, large vessel atherosclerotic, vertebro- more intensive heparin regimens. 154,177,182 Therefore, basilar, or ‘progressing’ stroke) because data are insuf- in a patient with a major ischaemic stroke and AF, we ficient. Although the low molecular weight heparin, .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 672 672 Chapter 12 Specific treatments for acute ischaemic stroke dalteparin, at high doses may be efficacious in patients Table 12.10 Adverse effects of heparin. with AF, it is not more efficacious than aspirin in this setting. Because aspirin is easier to administer, it, rather Local minor complications of subcutaneous heparin at injection site than dalteparin, is recommended for the various stroke Discomfort subgroups (Grade A).’ 160 The results of the trials pub- Bruising Local complications of intravenous heparin at cannula site lished since that guideline 176,178,179,320 have not materi- (or elsewhere) ally altered those conclusions. Pain at cannula site Infection at cannula site (sometimes with severe systemic Dose and route infection) Reduced patient mobility because of infusion lines and The risk of bleeding with heparin is clearly dose- pump dependent; 154,179,182,183 the higher the dose, the higher Systemic complications the risk of intracranial and extracranial haemorrhage Intracranial bleeding and there is no evidence to support the use of full-dose Haemorrhagic transformation of cerebral infarct adjusted regimens with intravenous unfractionated (potentially disabling or fatal) heparin or heparinoid. 160 The Rapid Anticoagulation Intracerebral haemorrhage Subarachnoid haemorrhage Prevents Ischaemic Damage (RAPID) trial in 67 patients Subdural haematoma with non-lacunar ischaemic stroke within 12 h of onset Extracranial haemorrhage compared full doses of intravenous unfractionated Subcutaneous (can sometimes be massive) heparin with aspirin 300 mg daily; showed no evidence Visceral (haematemesis, melaena, haematuria) of an advantage of full-dose intravenous heparin over Thrombocytopenia aspirin, 176 although a more recent RCT of full-dose intra- Type I: dose and duration related, reversible, mild, usually venous heparin vs low-dose subcutaneous heparin within asymptomatic, not serious and often resolves 3 h of stroke onset suggested net benefit from the higher spontaneously heparin dose. 177 Hence, since bleeding is dose-related, Type II: idiosyncratic, allergic, severe (may be complicated if anticoagulants are to be used, low-dose subcutaneous by arterial and venous thrombosis). Affects 3–11% of patients treated with intravenous heparin and less than regimens are preferable (e.g. 5000 IU unfractionated 1% of patients treated with subcutaneous heparin heparin two or three times daily) since they are simpler, Osteoporosis do not require complex monitoring and are associated Skin necrosis with lower bleeding risks. The options for the treatment Alopecia of established deep venous thrombosis and pulmonary embolism are discussed in section 11.13. Reversal of warfarin must be done in consultation with the local haematology specialist (Table 12.13). Adverse effects Table 12.10 lists the most important adverse effects 12.4.7 Acute stroke in a patient already and cautions to be considered when using heparin. The receiving anticoagulants most life-threatening risks are of massive intra- or extra- cranial haemorrhage. Management of severe extracranial Urgent reversal of anticoagulation in patients with haemorrhage consists of stopping any heparin adminis- intracranial bleeding on anticoagulants tration; estimation of the clotting time; and reversal with intravenous protamine sulphate in a dose of 1 mg for If a patient on oral anticoagulants presents with sym- every 100 IU of heparin infused in the previous hour. 193 ptoms suggestive of a stroke or develops any of the neuro- Protamine should be given slowly over 10 min and the logical symptoms outlined in Table 12.11 (any of which patient monitored for hypotension and bradycardia. may indicate intracranial bleeding), they must have a Not more than 40 mg of protamine should be admin- CT or MR scan performed immediately. 193 If the CT or MR istered in any one injection. For advice on reversal of scan shows extradural, subdural, subarachnoid or intra- low-molecular-weight heparin or heparinoids, consult cerebral haemorrhage, then anticoagulants should be the manufacturer’s data sheet. Patients on oral anti- stopped and urgently reversed. If the scan does not show coagulants who suffer a head injury or develop any of intracranial bleeding, consider the possibility that the the neurological symptoms listed in Table 12.11, or who neurological deficit might be due to a spinal haematoma, require a diagnostic lumbar puncture, should be investi- especially if the neurological deficit spares the face gated and treated as described in Tables 12.11–12.13. 193 and is associated with neck pain (the diagnosis is best .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 673 12.4 Selective use: anticoagulants 673 Table 12.11 Advice on management of problems in patients Table 12.12 Strategies for reversal of oral anticoagulation. 193 on anticoagulants. Life-threatening haemorrhage (intracranial or major Indication for immediate CT or MR brain scan if patient has a gastrointestinal bleeding) Significant head injury (i.e. any loss of consciousness or Stop warfarin prescription and amnesia or the presence of a scalp laceration) Give intravenous vitamin K1 (5 mg, repeated if necessary) Headache (especially of recent onset, or increasing in and either severity) Intravenous factor IX complex concentrate (also contains Acute neck pain with neurological deficit in limbs b factors II and X) (50 IU/kg body weight) plus Drowsiness Factor VII concentrate (50 IU/kg body weight) (if available) Confusion or Onset of focal neurological symptoms or signs Recombinant factor VIIa or If the scan shows intracranial bleeding Fresh frozen plasma (15 mL/kg – approximately 1 L Reverse anticoagulants (Table 12.12) for adult) If scan shows extradural or subdural bleeding or Less severe haemorrhage (e.g. haematuria, epistaxis) hydrocephalus, consider surgical decompression after Stop warfarin for 1–2 days reversal Give vitamin K1, 0.5–2 mg intravenously or 5–10 mg orally If scan shows only subarachnoid bleeding, consider CT High international normalized ratio (INR) but no haemorrhage or MR angiography to exclude underlying aneurysm Stop warfarin, monitor INR, restart warfarin when INR < 5.0 Is the patient being considered for a diagnostic or therapeutic procedure involving the spine? c Lumbar puncture of PCC, was associated with a better outcome. 196 RCTs Epidural or spinal anaesthesia of the different options are clearly needed. 195,196 After Contrast myelography successful reversal of the haemostatic deficit, further treatment may be needed, depending on the location of a The CT or MR scan must be performed as soon after the haematoma and any underlying causative lesion, for admission as possible. Delay may reduce the potential example: evacuation of a subdural haematoma, clip- benefits of reversing the anticoagulants if intracranial ping or coiling of a ruptured intracranial aneurysm, or bleeding is found. 196 b In patients with acute neck pain with a neurological deficit evacuation of an intracerebral or spinal haematoma. in the limbs, but no evidence of intracranial bleeding on the brain scan, MR scanning of the spine may be needed to Risk of arterial thromboembolism after stopping exclude spinal epidural haematoma. 194 anticoagulants c Lumbar puncture, epidural anaesthesia and myelography in patients on anticoagulants can cause spinal epidural Patients with a high risk of arterial thromboembolism haemorrhage, spinal cord compression and paraplegia. (usually because they either have a mechanical prosthetic The need for the procedure should be discussed with a heart valve or AF) who develop intracranial bleeding neurologist or neurosurgeon before anticoagulants are while on oral anticoagulants create particular manage- reversed, since reversal carries a risk of thromboembolism. ment difficulties. 197–199 There are competing risks to be Alternative investigation (e.g. MR scanning) or alternative balanced: of valve thrombosis and re-embolization if methods of anaesthesia may be preferable. anticoagulants are permanently withdrawn, and of further intracranial bleeding if anticoagulants are re- confirmed by MR scanning of the spine); a spinal epidural instituted. 197–201 A retrospective non-randomized observ- haematoma may require neurosurgical treatment. 194 ational study of 141 patients with a high risk of A narrative review found no reliable evidence from ischaemic stroke who had an intracranial haemorrhage RCTs to select the best method for rapid reversal, but the while taking warfarin examined these competing risks. 198 options are: vitamin K, prothrombin complex concen- The indication for anticoagulants fell into one of three trates (PCC), recombinant factor VIIa, or fresh frozen categories: prosthetic heart valve in 52 patients, AF in plasma (FFP) 195 (section 12.2.4) (Tables 12.12–12.13). 53 and a recurrent TIA or an ischaemic stroke in 36. The There is weak evidence from a retrospective non- median time off warfarin after the bleed was 10 days. randomized observational study in 55 patients with Three patients had an ischaemic stroke within 30 days of anticoagulant-related intracerebral haemorrhage that stopping warfarin. In the three groups, the Kaplan-Meier haematoma growth within 24 h was significantly lower estimate of the probability of having an ischaemic stroke in patients receiving PCC compared with FFP or within 30 days of stopping warfarin was: prosthetic heart vitamin K. 196 There was some evidence that achieving valve 2.9% (95% CI 0–8.0%); AF 2.6% (95% CI 0–7.6%); a normal INR within 2 h, generally through the use and TIA or an ischaemic stroke 4.8% (95% CI 0–13.6%) .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 674 674 Chapter 12 Specific treatments for acute ischaemic stroke Table 12.13 Options for reversal of warfarin therapy. 193 Method and dosage Advantages Disadvantages Stop warfarin Simple May take several days for INR to normalize, particularly with liver disease Vitamin K1 Safe if given slowly 2–6 h to take effect 0.5–5 mg i.v. 12–24 h to take effect 5–10 mg orally Factor IX complex concentrate Acts immediately Exposes patients to blood product which rarely may 50 IU/kg body weight Small volume transmit hepatitis B (but safe from hepatitis C and HIV) Can be given quickly Effect is temporary (8–24 h), may need to be repeated Should be combined with vitamin K1 Risk of thromboembolism Factor VII concentrate Should be given with factor IX As for factor IX complex 50 IU/kg body weight complex concentrate Recombinant factor VIIa No risk of blood-borne infection Expense Fresh frozen plasma 1L for adult Acts immediately Large intravenous volume load Allergic reactions Not as efficacious as factor IX complex concentrate Difficult to give repeat infusions because of large volume Virally inactivated plasma is preferred respectively. Another retrospective review reported that However, there are many factors one has to consider in none of the 16 patients with a prosthetic heart valve decision-making for an individual patient. Consider the who had intracranial haemorrhage and then had their (fairly typical) case of an elderly patient in AF who has warfarin discontinued (for a median of 7 days) suffered been on anticoagulants for a number of years because of a recurrent thromboembolic event while off warfarin. 201 a prior ischaemic stroke who presents with an intracerebral It therefore seems from these very limited data that the haemorrhage. A number of questions immediately spring risk of recurrent thromboembolism from stopping oral to mind: was the bleed due to over-anticoagulation (and anticoagulants for a week or two after an intracranial if so, is there a level of international normalized ratio haemorrhage is low. (INR) which clearly separates ‘excessive’ from ‘accept- able’)?; if the INR was in the therapeutic range, how far should investigations to identify the cause be pur- Is it possible to restart anticoagulation after intracranial sued?; and, is the patient still at high risk from recurrent bleeding, and if so, in whom, and when? thrombembolism (and if so, should the target INR be After an intracerebral haemorrhage on anticoagulants, lowered)? one needs to estimate the risk of recurrent thromboem- bolism (section 16.6.2 for risk prediction in patients If the scan shows infarction, what are the management with AF; assessing the risk in patients with mechanical options? prosthetic heart valves requires expert cardiological advice). In the study of 141 patients by Phan et al., of If the CT or MR is normal or shows infarction with no the 35 who had warfarin therapy restarted, none had evidence of any haemorrhage, the reason for the infarc- recurrence of intracranial bleeding during the same tion must be sought. Often, the cause is inadequate hospital admission. 198 Butler has reported 2-year follow- warfarin dose, but infective endocarditis must be ruled up data on 13 patients with mechanical prosthetic heart out. If the INR was below target range before – or at the valves who restarted anticoagulation after an intracranial time of onset of – the stroke, more careful supervision to haemorrhage. Of the four patients with intracerebral maintain the INR within the target range for that patient haemorrhage as the initial bleed, none suffered recurrent should reduce the risk of further ischaemic strokes. If intracranial bleeding and two suffered thromboembolic INR control had been poor despite apparent compliance events. On reintroduction of oral anticoagulants, the with the prescribed warfarin dose, one must make care- target INR was lowered in 9 of 13 patients. The authors ful enquiry about recent changes in other prescribed concluded that careful reintroduction of oral anticoagu- drug treatment, the use of non-prescribed medication lation is appropriate in these patients. (including herbal and complementary therapies) or .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 675 12.5 Selective use: reperfusion with thrombolytic drugs and other methods 675 changes in diet, since all of these can interact with war- onset of confirmed ischaemic stroke. Individual patient farin. However, if a patient suffers a recurrent ischaemic data meta-analyses of the rt-PA 203 and streptokinase stroke despite an adequate INR, one can consider adding trials 204 have also been published. Since then a number low-dose aspirin; while this may reduce the risk of of small-scale RCTs have been published: a dose-finding recurrent stroke, it also increases the risk of intracranial study of intravenous tenecteplase; 205 two phase II studies haemorrhage 171,202 (section 16.6.2). of intravenous desmoteplase; 52,53,206 and, of intra-arterial urokinase in acute posterior circulation stroke. 207 Intra- venous glycoprotein (Gp) IIb/IIIa inhibitors are discussed in section 12.6.1. There have not been any large-scale trials of mechanical clot retrieval or ultrasound-assisted 12.5 Selective use: reperfusion with reperfusion; the limited data on these interventions are thrombolytic drugs and other methods reviewed in section 12.6.3. Fewer data than for acute myocardial infarction 12.5.1 Rationale Over 58 000 patients have been randomized in trials of intravenous thrombolysis vs control in patients with Potential benefits 208 acute myocardial infarction. Compared with this Restoration of blood flow, to reperfuse the ischaemic very large amount of data, there are relatively few on brain as soon as possible after a cerebral vessel has been the effects of thrombolysis for ischaemic stroke (espe- occluded, should lessen the volume of brain damaged cially when one recalls the size of the global burden of by ischaemia, reduce the likelihood of major cerebral stroke and that the incidence of acute stroke and acute oedema and result in a better clinical outcome (sec- MI are similar in Western populations). Furthermore, it tion 12.1.3). Therefore, therapeutic attempts to hasten required a systematic review of these data to establish reperfusion by removing any occluding thrombus with that, though early treatment was the most effective, for thrombolytic drugs, mechanical methods, ultrasound or patients with ST elevation or bundle branch block, the a combination of these approaches ought to be helpful. ‘time window’ for worthwhile benefit was up to 12 h (certainly much longer than had originally been thought plausible). 208 Potential risks Thrombolytic drugs will also lyse haemostatic plugs Intracranial haemorrhage and thus may increase bleeding into the brain in the area of ischaemia, in areas of the brain remote from the There is a very consistent trend across all the stroke trials ischaemia or in extracranial sites (into the skin, joints, for an excess of fatal intracranial haemorrhage with the gastrointestinal or urinary systems). Mechanical thrombolysis and the proportional excess is similar methods require intra-arterial instrumentation, which with all the agents. The absolute excess depends on the can cause a number of complications, including arterial risk of haemorrhage in the control group, and this risk dissection and rupture. was somewhat different between trials. For intravenous rt-PA, the only licensed thrombolytic agent, treatment increases the risk of fatal intracranial haemorrhage about 12.5.2 Evidence fourfold from 1% in controls to about 4% (P < 0.00001), equivalent to 33 extra fatal intracranial haemorrhages per Data available for reperfusion with thrombolytic drugs 1000 patients treated (Fig. 12.13). rt-PA increases fatal and mechanical methods or non-fatal symptomatic intracranial haemorrhages Eighteen RCTs including 5727 patients are included in threefold (P < 0.00001). 133 Many of the haemorrhages the most up-to-date systematic review available in the occur within the ischaemic area, but others occur at sites Cochrane Library. 133 About half the data relate to trials of remote from the ischaemia. These remote bleeds may intravenous recombinant tissue plasminogen activator represent bleeding from an undetected pathological vas- (rt-PA) and the remainder to intravenous streptokinase cular lesion (small arteriovenous malformation, vessel (42%), intravenous urokinase (3%) and intra-arterial affected by amyloid angiopathy, or a previous micro- recombinant pro-urokinase (5%). Ninety per cent of the bleed). 209 However, data are also accumulating to suggest data are derived from trials of intravenous thrombolysis that the presence of microbleeds does not materially with either rt-PA or streptokinase, given within 6 h of influence the risk of bleeding with thrombolysis. 210 .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 676 676 Chapter 12 Specific treatments for acute ischaemic stroke Thrombolysis Control Odds ratio & Odds n/N n/N 95% CI reduction Intravenous urokinase vs control Atarashi 1985 1/192 0/94 Chinese UK 2003 8/317 0/148 Subtotal (95% CI) Intravenous streptokinase vs control MAST-E 1996 26/156 2/154 MAST-I 1995 8/157 0/156 Morris 1995 2/10 0/10 ASK 1996 14/174 2/166 Subtotal (95% CI) Intravenous rt-PA vs control Haley 1993 0/14 1/13 ECASS 1995 19/313 7/307 NINDS 1995 9/312 1/312 ECASS II 1998 18/409 4/391 ATLANTIS B 1999 8/307 1/306 ATLANTIS A 2000 8/71 0/71 Subtotal (95% CI) Intravenous streptokinase + oral aspirin vs oral aspirin MAST-I 1995 13/156 2/153 Subtotal (95% CI) Intra-arterial pro-urokinase + intravenous heparin vs intravenous heparin PROACT 1998 1/26 1/14 Subtotal (95% CI) –343%* Total (95% CI) (2P<0.00001) 100 0.01 0.1 110 Thrombolysis Thrombolysis better worse Fig. 12.13 Thrombolysis in acute ischaemic stroke: effects on confidence intervals are 95%. (From Wardlaw et al. fatal intracranial haemorrhage. Results of a systematic review of ‘Thrombolysis for acute ischaemic stroke’, Cochrane Database 13 trials. *A negative odds reduction indicates an adverse effect of Systematic Reviews 2003; (3) 133 copyright Cochrane of treatment. Similar conventions as Fig. 12.11 except all the Collaboration, reproduced with permission.) communication); if confirmed in a large study this could Massive infarct swelling be of substantial clinical significance. Massive swelling of infarction is a common cause of early death after acute ischaemic stroke, and is more likely to Death occur in patients where the occluded artery does not spontaneously reopen (section 5.7). It is reasonable Overall, thrombolysis significantly increased the odds to assume then that reopening the occluded artery of death within the first 2 weeks, though, by the end of might reduce the risk of this event, reduce the risk of follow-up (3 months in most trials), the difference between death and increase the chance of a good recovery from the thrombolysis and control was smaller (Fig. 12.14). 133 the stroke. However, an early non-randomized study In the trials of intravenous rt-PA, there was a non- suggested that rt-PA increased the risk of massive fatal significant 14% increase (95% CI 4% reduction to 37% brain swelling. 211 A more recent, but still uncontrolled, increase) in the risk of death by the end of follow-up (at case series of patients, all treated with rt-PA, suggested around 3 months in most trials). 133 There was statistic- that, while ‘malignant’ fatal brain oedema was more ally significant heterogeneity of treatment effect among common that expected, overall ‘massive infarct swell- the trials of rt-PA which makes the interpretation of this ing’ was less frequent. 212 This outcome has not been overall estimate difficult. 133 In the individual patient extensively studied in the trials to date, but a preliminary data meta-analysis of the rt-PA trials, the hazard ratio analysis of the limited data from four trials suggests (HR) for death, adjusted for baseline stroke severity, was rt-PA showed a trend to a non-significant reduction not significantly different from 1.0 for the 0–90, 91–180 in this outcome by about a fifth (Wardlaw J, personal and 181–270 min intervals but for 271–360 min it was .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 677 12.5 Selective use: reperfusion with thrombolytic drugs and other methods 677 Thrombolysis Control Odds ratio & Odds n/N n/N 95% CI reduction Intravenous urokinase vs control Abe 1981 1/54 1/53 Atarashi 1985 7/192 4/94 Ohtomo 1985 3/169 6/181 Chinese UK 2003 33/317 10/148 Subtotal (95% CI) Intravenous streptokinase vs control MAST-E 1996 73/156 59/154 MAST-I 1995 44/157 45/156 Morris 1995 3/10 3/10 ASK 1996 63/174 34/166 Subtotal (95% CI) Intravenous rt-PA vs control Mori 1992 2/19 2/12 Haley 1993 1/14 3/13 JRSG 1993 3/51 4/47 ECASS 1995 69/313 48/307 NINDS 1995 54/312 64/312 ECASS II 1998 43/409 42/391 ATLANTIS B 1999 33/307 21/306 ATLANTIS A 2000 16/71 5/71 Subtotal (95% CI) Intravenous streptokinase + oral aspirin vs oral aspirin MAST-I 1995 68/156 30/153 Subtotal (95% CI) 156 153 Intra-arterial pro-urokinase + intravenous heparin vs intravenous heparin PROACT 1998 7/26 6/14 PROACT 2 1999 29/121 16/59 Subtotal (95% CI) Total (95% CI) –33%* 2P = 0.0002 10 0.1 0.2 0.5 1 25 Thrombolysis Thrombolysis better worse Fig. 12.14 Thrombolysis in acute ischaemic stroke: effects on as Fig. 12.13 (from Wardlaw et al. ‘Thrombolysis for acute death at the end of follow-up. Results of a systematic review of ischaemic stroke’, Cochrane Database of Systematic Reviews 2003; 18 trials reporting this outcome. *A negative odds reduction (3) 133 copyright Cochrane Collaboration, reproduced with indicates an adverse effect of treatment. Same conventions permission.) significant at the 5% level with an HR of 1.45 (1.02– 5–25%; P = 0.004) (Fig. 12.15). This would be clinically 2.07). 203 It is not known reliably whether the survival important – if confirmed by further trials – as it is equi- disadvantage from thrombolysis diminishes or increases valent to about 43 fewer dead or dependent patients per with longer-term follow-up, since only one trial with 624 1000 treated. However, there was heterogeneity of treat- patients has reported follow-up beyond 6 months. 213 ment effect between the trials, which makes the overall estimate of effect less reliable. Among the six trials of intravenous rt-PA (2830 patients), thrombolysis reduced Dead or dependent, or fully recovered at the end of trial the risk of death or dependency by 20% (95% CI 7–31%; follow-up P = 0.003) equivalent to 55 fewer patients being dead or Despite the excess of deaths from intracranial haemor- dependent. Again, there was significant heterogeneity rhage, patients who survived the treatment were on of treatment effect. The precise definition of dependency average less disabled. Most of the trials reported follow- did not materially alter these conclusions, i.e. it did not up at 3 months. For patients treated with any agent make much difference whether ‘dependent’ was defined within 6 h, there was a significant 16% reduction in the as modified Rankin/Oxford Handicap Score (OHS) 3, 4, risk of death or dependency with thrombolysis (95% CI 5 or 2, 3, 4, 5 (section 17.12.5). The proportion of .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 678 678 Chapter 12 Specific treatments for acute ischaemic stroke Thrombolysis Control Odds ratio Odds n/N n/N 95% CI reduction Intravenous urokinase vs control Chinese UK 2003 127/317 61/148 Subtotal (95% CI) Intravenous streptokinase vs control MAST-E 1996 124/156 126/154 MAST-I 1995 97/157 106/156 Morris 1995 6/10 5/10 ASK 1996 84/174 74/166 Subtotal (95% CI) Intravenous rt-PA vs control Mori 1992 11/19 10/12 ECASS 1995 171/313 185/307 NINDS 1995 155/312 192/312 ECASS II 1998 187/409 211/391 ATLANTIS B 1999 141/307 135/306 ATLANTIS A 2000 64/71 56/71 Subtotal (95% CI) Intravenous streptokinase + oral aspirin vs oral aspirin MAST-I 1995 99/156 94/153 Subtotal (95% CI) Intra-arterial pro-urokinase + intravenous heparin vs intravenous heparin PROACT 1998 18/26 11/14 PROACT 2 1999 73/121 44/59 Subtotal (95% CI) 16% (2P = 0.004) 0.1 0.2 0.5 1 25 10 Thrombolysis Thrombolysis better worse Fig. 12.15 Thrombolysis in acute ischaemic stroke: effects ‘Thrombolysis for acute ischaemic stroke’, Cochrane Database on death or dependency at the end of follow-up. Results of a of Systematic Reviews 2003; (3) 133 copyright Cochrane systematic review of the 13 trials which reported this outcome. Collaboration, reproduced with permission.) Same conventions as Fig. 12.13. (From Wardlaw et al. patients making a complete recovery from their stroke in patients with and without the feature in question. In (OHS zero) was also increased with thrombolysis. Only statistical terms, if a variable does influence the size or three trials, with a total of 1556 patients, reported direction of the treatment effect, there is an ‘interaction’ follow-up data at 6 months and only one has reported between that variable and the treatment effect. 129,216,217 follow-up at 12 months. The benefit seen at 3 months The most common type of interaction is where the persisted to 12 months in that one trial, but further long- treatment is effective in people with and without the term follow-up data are clearly needed. 213,214 variable, and only the size of the treatment effect is somewhat different (e.g. a treatment is effective in both men and women, but the benefit is greater in men); this Subgroups: predicting prognosis is not the same as is called a quantitative interaction. A less commonly predicting response to thrombolytic treatment encountered scenario is when the direction of treatment There is an enormous literature on the clinical and radio- effect is different in the two groups (e.g. harmful or in- logical factors that predict outcome after ischaemic effective in men, but beneficial in women); this is called stroke 215 though the number of simple, well-validated a qualitative interaction. 129,216,217 Subgroup analyses look- and robust predictive models that are useful for clinical ing for such interactions are often very underpowered practice is surprisingly small (section 10.2.7). 215 However, and multiple underpowered exploratory subgroup ana- in the context of identifying which patients to treat with lyses are particularly likely to lead to inappropriate inter- thrombolytic therapy, it is important to look for vari- pretations. 110,129,217 This is the case for the RCT evidence ables that predict a good response to treatment rather than for thrombolysis, where there are insufficient data for reli- those which just predict a good outcome. One needs to able analyses of the impact of almost all of the different be clear whether the treatment effect really is different features which have been said to influence the response .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 679 12.5 Selective use: reperfusion with thrombolytic drugs and other methods 679 to treatment. There are also a large number of non- Subgroups: clinical features to identify who benefits? randomized case series purporting to demonstrate factors which modify the response to thrombolysis; but studies The rt-PA study group individual patient data meta- with non-randomized controls simply cannot provide an analysis focused on the impact of onset to treatment unbiased estimate of interaction with treatment effect. time on the odds of a favourable outcome with rt-PA. 203 However, the authors also developed a model to exam- ine the influence of the most important prognostic Subgroups: time since stroke onset factors that might confound the relationship between The individual patient data meta-analyses of rt-PA 203 pro- the benefit of treatment and time from onset to treatment: vides evidence that (as in acute myocardial infarction), age, weight, sex, systolic and diastolic blood pressure the earlier treatment is given, the greater the benefit. (mmHg) at start of treatment, blood glucose concentra- However, though the curve of decreasing benefit from tion at baseline, National Institutes of Health Stroke rt-PA with increasing time to treatment appears smooth, Scale (NIHSS) score on admission, various items on pre- it is somewhat artificial, since there is a significant dis- vious history of vascular events, previous or current continuity in the underlying data at 90 and 180 min, atrial fibrillation and whether the patient was on aspirin because of clustering of randomizations at these time at the time of the index stroke (but not the influence of points; this clustering reduces the reliability of the ana- baseline CT scan appearances). These analyses suggested lysis. 218 The data for streptokinase show a similar, though that stroke severity did not influence response to treat- not quite statistically significant, trend for decreasing ment and that the likelihood of significant intracerebral benefit with time. 204 Though there is a statistically sig- haemorrhage was not importantly modified by time to nificant interaction between treatment benefit with treatment. Beyond those two conclusions, these analyses rt-PA and time to treatment, different people draw differ- do not shed light on which patients are most likely to ent conclusions from the data in Fig. 12.16 on exactly benefit. The RCT data for rt-PA are particularly scanty in what the therapeutic ‘window’ is (i.e. the time in which older people, since only 42 stroke patients over the age there is clear evidence of net benefit): just 90 min; up of 80 have been included in randomized trials of the to 3 h (the regulatory approval in most countries, but drug. 133,203 While a recent series from Canada has shown only for selected patients); up to 6 h; or, longer than 6 h that such patients can be treated with rt-PA, 219 we would in some individuals. The interpretation is made more argue that only randomized trial evidence can provide a difficult because time is just one of four inextricably proper assessment of the balance of risk and benefit in linked factors: stroke severity (severe strokes arrive quick- this increasingly large group of patients. est at hospital, mild strokes come later); the appearance of the baseline CT scan (the more severe the stroke, the Subgroups: radiological features to identify who benefits? more likely ‘infarction’ will be visible on an early scan); the time from onset (the longer the time, the more likely There has been much debate about whether ‘early infarct infarction will be visible on CT); and reduction in the signs’ on a pre-treatment scan should preclude the use benefit from treatment with time. The only way to of thrombolysis 220,221 (the definition and observer reli- decide which patients can be treated outside the current ability of these signs is discussed in section 5.4.2). 222 A very restrictive approved criteria is to obtain further data systematic review of the literature published between from further large-scale RCTs. 1990 and 2003 assessed the evidence. In 15 studies of 4.0 3.5 Odds ratio estimated by model 95% CI for estimated odds ratio Fig. 12.16 Thrombolysis in acute 3.0 ischaemic stroke. Model estimating 2.5 odds ratio for a favourable outcome at 3 months compared with control, by Adjusted odds ratio 2.0 onset to treatment time (OTT) in minutes. 1.5 Odds ratio adjusted for age, baseline glucose, baseline diastolic pressure, 1.0 previous hypertension and interaction 0.5 between age and baseline stroke severity. 0 (From the rt-PA study group 2004 with permission from rt-PA study group 203 60 90 120 150 180 210 240 270 300 330 360 and the Lancet.) OTT (min) .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 680 680 Chapter 12 Specific treatments for acute ischaemic stroke early infarction signs and outcome (including seven number of small-scale RCTs are underway, evaluating: thrombolysis trials) in 3468 patients, any ‘early infarc- newer thrombolytic agents (tenecteplase, reteplase, micro- tion sign’ increased the risk of poor outcome regardless plasmin); intravenous vs intra-arterial thrombolytic of treatment. 222 However, the only two studies that therapy; thrombolytic agent combined with intravenous sought an interaction between early infarction signs and Gp IIb/IIIa inhibitors; supplementing thrombolytic drug thrombolysis found no evidence that thrombolysis, given therapy with transcranial Doppler ultrasound. 231 in the presence of early infarction signs, resulted in worse outcome than that due to early signs alone. 222 A Is any one dose better than another? re-analysis of the scans from the NINDS and ECASS-2 studies, applying the Alberta Stroke Program Early CT For intravenous thrombolytic therapy, there is a sug- scale (ASPECTS), found that early ischaemic change was gestion that low-dose urokinase and low-dose rt-PA more frequent than originally reported, but there was no have a lower risk of symptomatic intracranial haemor- evidence that the presence of early ischaemic change rhage, but the comparisons are confounded by a number modified the response to rt-PA. 223,224 The authors con- of factors, such as baseline severity of stroke, race and cluded that further work was required to determine time to randomization, so no firm recommendation whether any early infarction sign should influence on the ‘best’ dose can be given. 133,230 rt-PA in dose of decisions concerning thrombolysis. The presence of a 0.9 mg/kg (maximum dose 90 mg) is the most widely perfusion-diffusion ‘mismatch’ (assessed either with tested regime. MR or CT) is a much debated radiological criterion 225,226 but unfortunately only one ongoing RCT, the EPITHET Concomitant aspirin or heparin? study, has an appropriate design to assess the interaction of ‘mismatch’ with treatment effect, and – since it will In the Multicentre Acute Stroke Trial (MAST-I) there was include only 120 subjects – it may well lack power to an interaction between streptokinase and aspirin such examine the interaction reliably. 227 that the risk of symptomatic intracranial haemorrhage was significantly higher for the combination of aspirin and streptokinase than for either drug separately. 164,232 Is any one agent better than another? In all of the other trials, aspirin was not randomly allo- There are a large amount of data from trials which directly cated, so it is very difficult to determine whether any randomized patients with acute myocardial infarction differences in the frequency of symptomatic intracranial (AMI) to different thrombolytic regimens. 228 A meta- haemorrhage were due to concomitant aspirin treatment analysis of the 14 trials (total number of randomized or to other factors (such as thrombolytic agent, dose, subjects 142 907) showed no significant differences in delay, etc.). However, there was a strong tendency for the mortality at 30–35 days, but total stroke and hae- effect of thrombolysis to be more favourable in trials morrhagic stroke rates were higher for rt-PA than for where antithrombotic therapy was avoided for the first streptokinase (total stroke, OR 1.29; 95% CI 1.1–1.5; 24 h (and the more antithrombotic therapy that was haemorrhagic stroke OR 1.8; 95% CI 1.1–2.9). 228 The given, the less favourable the outcome. 133 authors concluded that in AMI all thrombolytic drugs The need for concomitant anticoagulants may be dif- appeared to be of similar efficacy in reducing mortality, ferent when intra-arterial thrombolysis is used, but there and the differences in stroke favoured streptokinase over are no trials of intra-arterial thrombolysis which directly newer drugs (tenecteplase, reteplase, rt-PA). 228,229 randomized patients to ‘intravenous heparin’ vs ‘no Unfortunately, there have not been any comparably intravenous heparin’. In the Prolyse in Acute Cerebral large-scale RCTs directly randomizing large numbers Thromboembolism (PROACT)-I study of intra-arterial of ischaemic stroke patients between streptokinase, rt-PA pro-urokinase, at the time of angiography, the first 16 and urokinase to allow reliable comparison of these patients were given a 100 IU bolus of unfractionated agents. 230 The indirect comparisons available are not a heparin intravenously, followed by an infusion of 1000 reliable way to compare treatment effects but, for what IU/h. However, because of the high frequency of intra- it is worth, they did not provide clear evidence of differ- cranial haemorrhage, the heparin dose was reduced ences in efficacy between the three agents once the time for subsequent patients. 233 In the PROACT-II study, all to treatment was accounted for. 133,229,230 patients received an intravenous bolus of 2000 IU fol- In summary, systematic reviews of the indirect and the lowed by an intravenous infusion of 500 IU/h of unfrac- direct randomized comparisons of different intravenous tionated heparin, and intravenous heparin was given and intra-arterial thrombolytic agents for stroke have with intra-arterial thrombolysis. 234 Unfortunately these not established that any one agent is clearly superior to data do not reliably establish whether concomitant another (they may be, but we cannot be sure). 133,230 A aspirin or heparin therapy is necessary with intra-arterial .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 681 12.5 Selective use: reperfusion with thrombolytic drugs and other methods 681 or intravenous thrombolysis, and if so, what the best effect in reducing the overall burden of stroke (section antithrombotic regimen is. 18.3). The concomitant heparin treatment used in the PROACT trials was – as might be expected – associated with quite a high risk of symptomatic intracranial hae- Blood pressure management morrhage in the control groups (14% and 4% of controls The blood pressure entry criteria and management in PROACT-1 and -2, respectively), and an even higher algorithms that were used in the NINDS thrombolysis rate in the treated patients (15% and 10%, respectively). trial were included in the ASA Guidelines for stroke The Interventional Management of Stroke Study was patients receiving thrombolysis. 235 However, an inde- a non-randomized feasibility study of intravenous fol- pendent re-analysis of the NINDS trial identified several lowed by intra-arterial thrombolysis (the latter was given problems with blood pressure measurement and man- only if the target vessel remained occluded); the authors agement in the study: 112 patients (18%) were found concluded a randomized trial comparing intravenous to have identical admission and baseline (randomiza- with intravenous plus intra-arterial thrombolysis was tion) blood pressure readings; 22 patients had missing warranted. 239 baseline blood pressure data; an unknown number of patients had high blood pressure treated both before Treatment of ischaemic stroke following angiography, arrival in the emergency department and in the emer- interventional radiology or cardiac catheterization gency department by non-study physicians; and no data were recorded to indicate whether patients had There are many possible causes of stroke after various high blood pressure treated after randomization. 218,236 invasive radiological procedures (sections 6.8.5 and Based on these observations, the authors concluded that 7.18). The first step – as ever – is an urgent CT to exclude the effect of blood pressure and its management on intracranial haemorrhage (since such strokes can be clinical outcome in acute ischaemic stroke patients due to haemorrhage). 240 However, if the intra-arterial treated in the NINDS trial could not be assessed. Con- catheter is left in place (or a sheath has been inserted), sequently, all the blood pressure variables were excluded or if the cerebral lesion proves to be ischaemic, the from the statistical models. 236 Hence, any baseline blood radiologist may then inject contrast material to see if pressure selection or exclusion criteria for thrombolysis, an occlusive cerebral arterial lesion can be visualized. or recommendations on management of blood pressure Unfortunately, it may be difficult to ascertain whether during thrombolytic treatment, are not based on reli- the vessel has occluded because of local thrombosis, able evidence (in section 11.7 we discuss the general intimal dissection, arterial spasm, emboli dislodged by management of high and low blood pressure in acute the catheter from proximal arterial sites or even metallic stroke). fragments of the interventional tool itself. 240 And even if the vessel is occluded by emboli, it may not be possible to determine whether the emboli consist of atheroma- Intra-arterial thrombolysis for acute spontaneous tous debris or fresh thrombus. Nonetheless, inter- occlusion of a cerebral artery ventional cardiologists and radiologists like to intervene, There have been five small RCTs in which occlusion of and there are several anecdotal reports of thrombolysis the relevant cerebral artery was demonstrated before to clear such obstructions. An uncontrolled study in treatment was given. In two of the trials, angiographic 11 patients who had acute ischaemic strokes during confirmation was available in most patients 237,238 and interventional neuroradiological procedures (10 of the in the two PROACT trials it was a prerequisite for trial 11 were having coils or balloons inserted into inoperable entry. 233,234 The two PROACT trials compared pro- intracranial aneurysms) suggested that thrombolysis urokinase plus intravenous heparin vs intravenous within 4 h of symptom onset, with an unspecified intra- heparin alone. The Australian Urokinase Stroke Trial arterial dose of urokinase, was not catastrophic. 241 A (AUST) comparing intra-arterial thrombolysis plus heparin retrospective review reported the use of intra-arterial with heparin alone in basilar artery thrombosis was rt-PA or urokinase in 21 patients with post-coronary terminated after only 16 patients were randomized, angiography stroke. 242 Without a control group for because of difficulties recruiting. 207 The benefits of intra- comparison, it is hard to judge the clinical outcome: arterial thrombolysis appear to be at least as great as with however, 48% achieved a modified Rankin Score of 2 intravenous rt-PA (Fig. 12.15), but the treatment can be or better, but symptomatic intracranial haemorrhage given only in a few highly specialized centres, so is only occurred in three (14%), and four (19%) died. A review available to a very small proportion of all stroke patients. of case series of thrombolytic treatment for strokes Treatments which are highly effective, but applicable to after cardiac catheterization similarly suggested that only a small proportion of all patients have almost no thrombolytics are often used, but the balance of risk and .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 682 682 Chapter 12 Specific treatments for acute ischaemic stroke benefit is unclear. 240 An alternative ‘rescue’ treatment www.ist3.com. For details of the other current trials see for peri-procedural stroke due to vessel occlusion by http://www.strokecenter.org/trials/index.aspx. thrombus is to give a Gp IIb/IIIa inhibitor intra-arterially (section 12.6.1). 12.5.3 Who to treat There is general agreement that thrombolysis for stroke Treatment of suspected cardioembolic stroke should only be given in centres that have a well- A small trial comparing rt-PA with controls in 98 patients established acute stroke care system, sufficient numbers with suspected cardioembolic ischaemic stroke showed of appropriately trained staff and adequate facilities. a non-significant trend towards benefit with throm- The clinical diagnosis of stroke in the hyperacute phase bolysis. 238 There is some suggestion that an embolus relevant to clinical triage for thrombolysis is discussed in obstructing a cerebral artery may be more susceptible to section 3.3, the imaging strategies to select patients for lysis if it is due to embolism from the heart rather than thrombolysis in sections 5.4 and 5.5, and the organiza- from a lesion in the extracranial arteries. tion of the acute stroke service that is needed to support safe and effective thrombolytic treatment (including support via telemedicine) in section 17.5. The need for further large-scale RCTs of intravenous If thrombolysis is to be given safely, a great deal of thrombolysis preparation and training are required for staff in the Although the limited trial data have led to regulatory local teams and departments involved in each stage of approval for the use of rt-PA, many questions remain the chain: emergency call response centre, ambulance about how this treatment should be used in routine service, emergency department, radiology department clinical practice. If more patients are to have access to and the acute stroke unit. Such training needs to be this treatment and if it is to be delivered more equitably, repeated at intervals to allow for staff turnover. And then we need reliable answers to a number of important training sessions must be backed up with clearly written questions: Is there an upper age limit for treatment? simple protocols. Useful materials are available over the Which patients are most likely to suffer early hazard? Internet on many different websites; this list gives a Which patients will gain the greatest long-term bene- selection with particular relevance to thrombolysis: fit? How wide is the time window – 3, 4, 5, 6, 7, 8 or even • The IST-3 trial website has a number of resources 9 h for some cases? Is the window the same for all isch- (monitoring for, and management of complications, aemic stroke subtypes? Is there worthwhile net benefit training in image interpretation): www.ist3.com when thrombolysis is given in non-specialist centres? • The NIH website offers a web-based training tool for To answer these questions reliably, we need to design administering the NIHSS and providing certification of and conduct appropriately large trials. 110,114 Large trials competence: http://www.professionaleducationcentre. are essential because, if the treatment really works as well americanheart.org as we suspect, the time window is longer than just 2 or • The Internet Stroke Centre has a variety of educational 3 h, or the selection criteria can be broadened, it should materials about stroke: http://www.strokecenter.org/ be used more widely. If, on the other hand, large-scale prof/index.html trials show it to be less effective than expected, health services should focus resources on the treatments that Clinical and radiological selection criteria yield greater net health gain for the population (e.g. stroke units) (sections 17.6 and 18.4). 243,244 The clinical and radiological selection criteria for treat- The rt-PA study group estimated that a randomized ment with intravenous thrombolytic therapy set by the controlled trial with about 6000 patients would be regulatory authorities and laid out in national guidelines needed to confirm or refute the estimates of effect from vary somewhat between countries. Details of the eligibil- their pooled analysis. 203 The Third International Stroke ity criteria for the approved use in Europe are detailed in Trial (IST-3) began its start-up phase in April 2000. The the summary of product characteristics for rt-PA. 245 The main UK Medical Research Council phase started in items common to many guidelines are summarized in 2005 and will continue until 2009. The trial seeks to Table 12.14. randomize 6000 patients with acute ischaemic stroke within 6 h of onset to intravenous rt-PA or control. It has Consent an appropriately streamlined and efficient design, as is appropriate for such a large-scale study. 110,114 Full The risk of fatal intracranial haemorrhage with throm- details of the trial protocol and progress are available at bolysis deters a large number of patients, emergency .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 683 12.5 Selective use: reperfusion with thrombolytic drugs and other methods 683 Table 12.14 Essential aspects of therapeutic thrombolysis. a Preparation and maintenance of service organization: History of bleeding disorder (e.g. haemophilia) Audit existing service, identify delays History of recent bleeding Draw up ‘fast-track’ pathway of care in consultation with all Arterial puncture at non-compressible site within past relevant disciplines and departments 14 days Train relevant staff Major surgery within past 14 days Inform general public and primary care teams Previous stroke or serious head injury within past 3 months Assess patients immediately: Neurological deficit trivial and likely to recover within next Ambulance crew perform basic assessment and radio ahead to few hours hospital to warn of arrival Evidence of accelerated hypertension Immediate assessment on arrival at hospital by trained ‘triage’ Blood glucose is < 3 or > 22 mmol/L nurse or paramedic Seek expert advice before giving thrombolysis if any of the following Local acute stroke protocol activated including: are true: c Systematic but brief clinical assessment: Unruptured intracranial aneurysm Number of hours since onset of stroke symptoms Menstruation Focal neurological symptoms and signs Pregnancy Vital signs (pulse, blood pressure, respiration, temperature) Decision to give thrombolysis or not must be made by senior Intravenous cannula inserted and blood samples taken for member of stroke team basic blood tests (blood glucose measurement essential) If decision is to give thrombolysis, transfer to place where Immediate transfer to neuroimaging thrombolysis can be given and monitored closely: Results of preliminary neuroimaging conveyed to stroke Establish intravenous (i.v.) infusion if not already done so team Estimate weight, draw up infusion with a dose of 0.9 mg/kg Trained stroke physician reviews diagnosis, neuroimaging rt-PA and other information Infuse rt-PA i.v.: 10% as a bolus over 1–2 min, remainder over Consent/assent sought from patient and/or relative where 60 min feasible Monitor pulse and BP: Do not give thrombolysis if any of the following are true: b Every 15 min during infusion Time of stroke onset not clearly known Every 30 min for next 2 h, then Neuroimaging shows: Hourly for 5 h Acute intracranial haemorrhage Neurological observations hourly That the symptomatic infarct is much older than the Postacute management: history suggests Transfer to stroke unit (if not already there) A non-stroke lesion as the cause of the symptoms (e.g. Start aspirin 160–300 mg daily 24 h after infusion: cerebral tumour) If able to swallow safely, by mouth On oral anticoagulants and therapeutically anticoagulated If not able to swallow safely, per rectum or i.v. (i.e. INR > 1.4) (section 12.3.6) a See www.ist3.com for useful documents, protocols and training resources for thrombolysis in stroke. b This list of contraindications is based on the UK summary of product characteristics, i.e. the European licence for the use of alteplase for stroke. In the IST-3 trial, the protocol states that ‘If the patient has had a stroke within the previous 14 days or has had treatment for acute ischaemic stroke with thrombolytic therapy within the past 14 days’ this is an absolute contraindication to treatment; patients with a history of an ischaemic stroke more than 2 weeks previously (with or without thrombolytic treatment) may – at the treating physician’s discretion – be considered for treatment or inclusion in the trial. c See section 12.5.4 for details. INR, international normalized ratio; rt-PA, tissue plaminogen activator. physicians, neurologists and stroke physicians from giv- to accept the risks of treatment is often influenced by ing the treatment. 149,246 Although clinicians worry that their own perception of the severity of the stroke. If the intracranial bleeding due to treatment may lead to litiga- patient has neglect or other perceptual difficulties, this tion, in practice, failure to administer the treatment in can further complicate the consent procedure. 249 an otherwise eligible patient is the more common reason for litigation, in North America at least. 246 However, when A patient presented with witnessed abrupt onset of complete asked, many people (who have not yet had a stroke) left hemiplegia, 1 h before arrival at the hospital, had a CT would be prepared to accept the short-term risks (of severe consistent with early ischaemic change in the right cerebral haemorrhage or death) in order to have the chance of hemisphere, and met all the criteria for thrombolytic surviving free of disability. 247,248 A patient’s willingness treatment within the current approval. However, because .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 684 684 Chapter 12 Specific treatments for acute ischaemic stroke she also had anosognosia, she was completely unable to Treatment recognize that she had had a stroke, and hence saw no reason to be exposed to the risks associated with thrombolysis. The The aim should be to treat – without delay – patients stroke team felt it inappropriate to treat her against her who meet all of the following criteria: express wishes. • the time of symptom onset is known with certainty; • the clinical diagnosis is definitely an acute stroke of The scenario above is drawn from a published case sufficient severity to merit the risks of treatment; report; 249 a comment on the case stated that it would • no clear clinical or radiological contraindications are have been reasonable to give the treatment without the identified; patient’s consent. 250 This is supported by the fact that • consent (or relative’s assent) is documented in accord- the patient in question was left with substantial residual ance with local requirements; neurological impairment requiring major assistance in • passage through the above system has been fast all activities of daily living. 249 The risks of thrombolysis enough to allow treatment well within 3 h of are definite and not trivial, so fully informed consent is onset; desirable, whether thrombolysis is to be used in routine • there is a vacant bed and sufficient staff available clinical practice or in an RCT. But, in practice, getting on the acute stroke unit to monitor the patient dur- fully informed consent is often a barrier to early treat- ing the infusion and the immediate post-treatment ment: 251 the patient’s acute neurological impairment period. may impair communication and writing; a close relative Patients who fulfil the above criteria, but do not or partner may not be available at the critical moment; exactly meet the strict terms of the current approval, and and the pressure of time may make decision-making for whom thrombolysis is considered ‘promising but difficult for all concerned. 252–257 However, one must seek unproven’ should be considered for inclusion in IST-3 to achieve whatever degree of consent is required to or one of the other relevant RCTs. meet local clinical governance and research practice ethical standards. Simple materials (adaptable for The urgent priorities are to establish, by means of routine clinical or trial use) have been developed with appropriate large-scale randomized trials, whether the input from patients and lay people, that can facilitate the balance between early hazard and long-term benefit is process of consent (see www.ist3.com). 248,257 However, still favourable when thrombolysis is used in a wider the rigorous standards for informed consent required variety of patients. These data will help decide which in research do not obtain when thrombolysis is used in categories of patient are most likely to benefit and routine practice. A survey of the documentation of con- whether the benefits really persist for more than just sent for routine clinical use of thrombolysis for stroke in a few months. Connecticut hospitals found that a substantial percent- age of patients who received rt-PA for stroke had no con- 12.5.4 Who not to treat sent documented. 258 Surrogates often provided consent when the patients had the capacity to give consent; con- The absolute contraindications to thrombolysis are versely, patients with diminished capacity sometimes listed in Table 12.14. In addition, there are many relative provided their own consent. 258 contraindications reported in the literature, but which are not supported by good evidence. There are case There should be no double standards for randomized reports of thrombolysis being used for the treatment of trials and clinical practice, yet patients given acute ischaemic stroke in patients who do have these treatments in routine practice are often not given any 259 relative contraindications, including: pregnancy; explanation or information. Many patients refused 260 261 menstruation; recent surgery; unruptured intracra- randomization in the first International Stroke Trial 262 nial aneurysms and anticoagulation with an INR > when they were told that aspirin and heparin might 263,264 1.7. We are aware of cases with other contraindica- cause intracranial bleeding, yet how many tens of tions being treated (remote past history of intracranial thousands of patients with acute stroke have been haemorrhage but current ischaemic stroke, history of treated with these two drugs (before and since the ischaemic stroke within past 3 months, etc.); in our view, large-scale trials began) without being asked for their it would be more informative to consider including consent after a full explanation of the risks and patients who have some of these features of uncertain benefits? Similar double standards seem to apply to significance in relevant RCTs in order to determine thrombolysis in randomized trials and routine clinical reliably whether or not they should influence patient practice. selection. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 685 12.5 Selective use: reperfusion with thrombolytic drugs and other methods 685 The decision about treatment often requires a synthesis of clinical and radiological information. A 70-year-old man who lived alone was brought by ambulance to the emergency department at 9 am, following an emergency call at 8 am. He had aphasia and profound weakness of his right arm. The CT scan showed an area of ischaemia in the left middle cerebral artery territory appropriate to his symptoms. However, the lesion showed marked hypoattenuation and quite well-defined edges, suggesting it had been present for much longer than the 1 h suggested by the ambulance crew. Thrombolysis was not given. The patient’s niece later confirmed that her uncle’s speech had been ‘muddled’ when she had telephoned him at 10 pm the evening before, suggesting his ischaemic stroke lesion had been present for at least 11 h by the time he presented to hospital. At present, age over 80 is a contraindication to treat- ment within the European approval, whereas it is not in the USA and Canada. The risk of intracranial bleeding with rt-PA rises with age, but so too does the risk of a poor outcome without treatment. 203,265 Hence, given the very limited published RCT data in this age group (just 42 patients), while we accept that such patients can be treated, 219 we often prefer to enrol them in IST-3. We do this, because we – and the patients – are usually genuinely uncertain of the balance of risks and benefits. Fig. 12.17 ‘Thrombolysis box’ used on the acute stroke unit at the Western General Hospital, Edinburgh. The top drawer 12.5.5 When to start treatment – as soon as contains the infusion pump and written protocols (on possible! laminated sheets) for nursing management, monitoring, and management of suspected complications. The lower drawers Since the benefit (but perhaps not the risk) of treatment contain the other items required for drug administration. is very time dependent, the sooner it can be given, the Photograph by Dr I Kane (reproduced with permission). better (whether as part of routine clinical practice, or in a randomized trial). A systems approach is needed to ensure every step in the process of care is completed as efficiently as possible; from the moment the patient 12.5.6 Agent/dose/route/concomitant develops their stroke symptoms to the safe completion treatment of the treatment in hospital. There are many barriers to the delivery of efficient care, but a variety of interven- Agent tions to overcome these barriers have been evaluated (section 17.5). 214,251,266 Intravenous rt-PA is the only agent that has wide regula- Many centres have found it useful to create a ‘stroke tory approval for use in stroke. In Europe, at the time of thrombolysis box’ (Fig. 12.17). The design and the selec- writing, the approval was provisional on two conditions tion of its contents will vary according to local circum- being met: stances, but it should be portable, contain everything • Satisfactory evidence of the safety of rt-PA in routine needed to carry out a rapid assessment of a potential practice. Up to the end of April 2006 clinicians were stroke thrombolysis patient and then initiate therapy required to record (in the SITS-MOST registry: http:// (i.e. all the information, printed protocols, drugs, speci- www.acutestroke.org/) the use of, and the clinical men bottles, request forms, intravenous infusion pumps, outcome in, patients treated with rt-PA when treated drug prescription, drug administration charts, clinical within the terms of the approval. After that date, regis- monitoring records, etc.). If everything is in one place, tering the data has been voluntary. valuable time can be saved. • Completion of the ECASS-3 RCT. 267 .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 686 686 Chapter 12 Specific treatments for acute ischaemic stroke Dose and route All treatment protocols should have clear arrangements on how to monitor for, and then treat, these complica- The approved regimen is 0.9 mg/kg of rt-PA intra- tions if they arise. Examples of the sort of monitor- venously; 10% of the dose as a bolus over 1 min and the ing and management guides (which can be adapted as remainder as an infusion over 1 h, and the maximum needed) are available at the IST-3 website (www.ist3.com); dose is 90 mg. 133,230 they can be laminated and kept in the ‘thrombolysis Intra-arterial thrombolysis does not have regulatory box’. Tables 12.15 and 12.16 outline advice on the approval, but a variety of regimens have been used: in management of suspected intracranial or extracranial the PROACT-II trial, patients were given 9 mg of recom- haemorrhage. binant pro-urokinase intra-arterially; 234 in AUST, intra- arterial urokinase was given in increments of 100 000 IU to a maximum dose of 1 000 000 IU; and, in a recent large case-series, up to 1 000 000 IU of urokinase was Table 12.15 Advice on management of suspected intracranial given, and if recanalization was not achieved at that bleeding in patients treated with rt-PA. dose, gentle mechanical disruption of the clot with the catheter tip was attempted. 268 Suspect intracranial bleeding if Intra-arterial thrombolysis has a number of potential Neurological deterioration disadvantages: delay in initiation of fibrinolysis while New headache Fall in conscious level the diagnostic angiogram is performed and the micro- Acute hypertension catheter positioned (start of the thrombolytic infusion Seizure typically occurs 50–90 min later than the start of Nausea or vomiting an intravenous infusion); the procedure is labour- and Initial actions capital-intensive; and the intervention can only be Stop infusion of rt-PA performed at tertiary and secondary hospitals capable Arrange an urgent CT scan of acute endovascular therapy. 231 Check fibrinogen, PT, APTT, full blood count and send blood for ‘group and save’ Support circulation with i.v. fluids if needed Concomitant antithrombotic treatment If the scan shows intracranial bleeding Current US and UK guidelines recommend that anti- If scan shows extradural or subdural bleeding or hydrocephalus, consider surgical decompression after thrombotic therapy with aspirin or heparin should be reversal of any haemostatic defcit avoided for at least 24 h after intravenous thrombolysis If scan shows only subarachnoid bleeding, consider CT with rt-PA. 166,167 The situation is less clear for patients or MR angiography to exclude underlying aneurysm who have had intra-arterial thrombolysis. Some centres If the scan shows no intracranial bleeding follow one of the anticoagulant regimes employed in the If no intracranial bleeding on CT or MR, consider other RCTs of intra-arterial thrombolysis, 207,233,234 whereas causes of deterioration others – in the light of the results of the CAST and IST trials If the patient has a neurological deficit consistent with a – use aspirin instead of heparin as the post-treatment spinal location, sparing the face, especially if there is antithrombotic regimen. 268 acute neck or back pain, consider MR scan of spine to exclude epidural haematoma Blood pressure management for patients having The manufacturers of rt-PA suggest that most patients thrombolytic treatment with severe bleeding can be managed with volume replacement. 245 It is rarely necessary to replace the clotting The North American Stroke Thrombolysis Guidelines 235 factors because of the short half-life of the drug and the recommended the NINDS trial treatment algorithm, 132 minimal effect on the systemic coagulation factors. In those but the more recent version on acute stroke did not who fail to respond, transfusion of cryoprecipitate, fresh make such firm recommendations 167 (section 11.7 for frozen plasma and platelets should be considered; the general guidance on blood pressure monitoring and British Society of Haematology guidelines for the use of management). thrombolysis suggest for severe life-threatening bleeding after rt-PA, that has not responded to initial measures, giving a fibrinolytic inhibitor such as aprotinin or 12.5.7 Adverse effects tranexamic acid and replacement of clotting factors depending on the results of a coagulation screen (see also Intracranial and major extracranial haemorrhage are the Table 12.16). 318 most feared adverse effects of thrombolytic treatment. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 687 12.6 Unproven value: other reperfusion strategies 687 Table 12.16 Advice on management of suspected extracranial Tissue plasminogen activator can cause anaphylactoid bleeding in patients treated with rt-PA. reactions which may require prompt supportive treat- ment: adrenaline (0.5–1 mL 1 in 1000 i.m. or s.c., but Suspect extracranial bleeding if not i.v.) – the dose depends on the severity of the Drop in blood pressure reaction, and it may be necessary to repeat the dose; Clinical shock hydrocortisone 200 mg i.v.; chlorpheniramine 10 mg Evidence of blood loss, e.g. melaena, haematuria i.v.; salbutamol nebulizer 5 mg; and fluid resuscitation Initial actions if shocked. It is important to be aware of an uncommon Stop infusion of rt-PA Use mechanical compression, if possible, to control anaphylactoid reaction, orolingual angioneurotic oed- bleeding ema, which occurs in about 2–5% of patients treated Check fibrinogen, PT, APTT, full blood count and send with rt-PA, and is more likely if the patient is on an blood for ‘group and save’ or cross-match depending ACE inhibitor. 271,272 It can become life-threatening if the on situation rapid tongue swelling causes upper airway obstruction, Support circulation with fluids and blood transfusion as which may then require endotracheal intubation. 272 appropriate Discuss results with local haematology department For severe life-threatening bleeding not responding to fluid replacement Consider giving a fibrinolytic inhibitor (depending on the 12.6 Unproven value: other reperfusion results of a coagulation screen), either: 500 000 kallikrein inactivator units of aprotinin i.v. over strategies 10 min followed by 200 000 units over 4 h or tranexamic acid i.v. 1 g over 15 min repeated every 8 h as necessary or 12.6.1 Gp IIb/IIIa inhibitors fresh frozen plasma and/or cryoprecipitate Delay any surgery until haemostatic deficit corrected Rationale The manufacturers of rt-PA suggest that most patients The platelet glycoprotein (Gp) IIb/IIIa receptor is the final with severe bleeding can be managed with volume replacement. 245 It is rarely necessary to replace the clotting common pathway of platelet aggregation. Agents which block this receptor have antithrombotic and throm- factors because of the short half-life of the drug and the bolytic actions; in patients undergoing primary coronary minimal effect on the systemic coagulation factors. In those who fail to respond, transfusion of cryoprecipitate, fresh stenting for the treatment of acute myocardial infarc- frozen plasma and platelets should be considered; the tion, the Gp IIb/IIIa inhibitor abciximab improved coro- British Society of Haematology guidelines for the use of nary patency before stenting, the success of the stenting thrombolysis suggest for severe life-threatening bleeding procedure, coronary patency at 6 months, left ventricu- after rt-PA, that has not responded to initial measures, lar function and clinical outcomes. 273 A meta-analysis giving a fibrinolytic inhibitor such as aprotinin or showed that, in patients with acute myocardial infarc- tranexamic acid and replacement of clotting factors tion, adjunctive abciximab significantly reduced 30-day depending on the results of a coagulation screen. 318 a The options listed are based on expert opinion and not on and long-term mortality in patients treated with primary angioplasty but not in those receiving fibrinolysis. 274 reliable RCT evidence; it is prudent to discuss the options Abciximab and a number of other similar agents are with your local haematologist. therefore now being tested in acute ischaemic stroke. Evidence If a patient develops intracranial bleeding, early guide- lines suggested consideration of surgical evacuation of Abciximab is the agent of this class that has been most any intracranial clot. 235 This may not be wise because the widely tested as a treatment for acute ischaemic stroke. A STICH trial and a systematic review have not been able to 400-patient phase II study showed promising results, 275 demonstrate net benefit from evacuation of intracerebral so an 1800 patient efficacy RCT was initiated (AbESTT-2), haematoma. 269,270 However, if the patient has subdural but it was halted prematurely after 808 patients had been or extradural bleeding but no major intracerebral hae- enrolled, because of an unfavourable risk : benefit ratio matoma, surgical drainage may be indicated. Surgery for (largely a higher than anticipated risk of intracranial hae- any type of intracranial haemorrhage should be delayed morrhage). 276 In the light of the AbESTT-2 trial results, until the haemostatic deficit is fully reversed. abciximab is unlikely to become an established treatment .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 688 688 Chapter 12 Specific treatments for acute ischaemic stroke for acute stroke. 276 The evidence for the use of tirofiban more reliable conclusions could be drawn. 284 The ESTAT in acute stroke is more limited. 277 Two Gp IIb/IIIa inhibi- trial has recently been published. It was a randomized tors, eptifibatide and tirofiban, are still being evaluated, placebo-controlled trial of ancrod which included 1222 either alone, or in combination with thrombolytic patients with an acute ischaemic stroke. The primary agents, in ongoing small-scale RCTs, and it is possible outcome was functional success at 3 months (survival, that one of these will prove effective in acute stroke. Barthel Index of 95 or 100, or return to prestroke level). There are a large number of uncontrolled case series Functional success at 3 months did not differ between of intra-arterial Gp IIb/IIIa inhibitors (most relating to patients given ancrod (42%) and those given placebo abciximab or tirofiban) for ‘rescue’ treatment for the (42%) (P = 0.94; OR 0.99, 95% CI 0.76–1.29). 285 cerebral embolic complications of percutaneous coro- nary interventions 278 and interventional neuroradiology Conclusion procedures (e.g. coiling of intracranial aneurysms, 279 carotid angioplasty and stenting 280,281 ). A retrospective At present, there is not enough evidence to justify the use review of 1373 consecutive patients who underwent of defibrinogenating agents in routine clinical practice. neuroendovascular procedures identified 29 (2%) where the procedure was complicated by acute cerebral throm- 12.6.3 Interventional neuroradiology and boembolic events and were then treated with abcix- ultrasound augmentation of clot lysis imab. 282 Angiographic improvement was achieved in 29 (81%) of 36 arteries. Three intracerebral haemorrhages Rationale (10%) occurred with abciximab when administered with concurrent mechanical clot disruption; in two of For patients with acute ST-segment elevation myocardial these haemorrhages, recombinant tissue plasminogen infarction (AMI), primary percutaneous transluminal activator was also administered. 282 A non-randomized coronary angioplasty (PTCA) is often used as the primary study comparing abciximab with emboli prevention revascularization method in preference to intravenous devices suggested the use of emboli prevention devices thrombolytic therapy. A systematic review of the trials may be safer. 281 comparing PTCA with thrombolysis in AMI, involving 7739 patients, found that primary PTCA was signi- ficantly better than thrombolytic therapy at reducing 12.6.2 Defibrinogenation early death, non-fatal reinfarction, stroke, and the com- bined outcome cluster of death, non-fatal reinfarction Rationale 286 and stroke. The advantage to primary PTCA-allocated There are several defibrinogenating agents. The most patients persisted during long-term follow-up, and was widely tested is ancrod, a serine protease derived from independent of both the type of thrombolytic agent the venom of the Malayan pit viper. It causes a fall in the used, and whether or not the patient was transferred for levels of plasma fibrinogen, plasminogen, plasminogen primary PTCA. 286 activator inhibitor and antiplasmin. 283 Large quantities In centres that regularly perform neuro-interventional of circulating fibrinogen and fibrin degradation prod- procedures in acute ischaemic stroke, if intravenous or ucts are generated and tissue plasminogen activator is intra-arterial thrombolytic drug therapy does not reopen released from the vascular endothelium. Ancrod also the occluded artery, a number of techniques can now reduces plasma and whole-blood viscosity. 283 Defibrino- be tried to restore flow, including: direct mechanical genation might therefore improve perfusion in the balloon angioplasty of the thrombus; mechanical ischaemic brain and so have a net beneficial effect, pro- removal of clot; intravascular stenting of the underlying vided it is not associated with a substantial excess of occlusive atherosclerotic lesion; suction thrombectomy; major bleeding. and laser-assisted thrombolysis of emboli. 167,231 Trans- cranial Doppler ultrasound (with or without supple- mentary microbubble injection) on its own, or as an Evidence adjunct to thrombolytic therapy, may also improve clot A Cochrane systematic review of defibrinogenating lysis. 231,287 agents in acute ischaemic stroke included five trials involving 2926 patients. 284 Four trials tested ancrod Evidence and one trial defibrase. The reviewers concluded that, although fibrinogen-depleting agents appeared promis- The process for approval of the use of devices and proced- ing, the data from the ESTAT trial were needed before ures for acute stroke treatment is controversial because .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 689 12.7 Unproven value: treatment of cerebral oedema and raised intracranial pressure 689 devices are not subject to the same stringent approval onset of ischaemia, might be reduced by neuroprotective processes as drugs (i.e. there is no requirement for phase agents which can prevent influx of sodium ions and III efficacy trials). 288 In acute ischaemic stroke, case series water into cells (section 12.1.5). have reported the use of mechanical clot disruption, 289 and the use of the MERCI clot retrieval device. 290 The Evidence FDA has approved the MERCI device for the removal of clot from brain arteries on the basis of a non-randomized There have been at least 17 apparently randomized trials study, 290 (though not as treatment for acute ischaemic in acute stroke, but unfortunately only seven (involv- stroke), a decision regarded by some as premature. 288,291 ing 453 people) were of sufficient quality to merit inclu- The device is, however, being tested in the MR-RESCUE sion in a Cochrane systematic review. 295 There was no trial. 292 Future studies will need to evaluate the overall significant difference in the odds of death within 1 year efficacy when compared with intra-arterial or intraven- (odds ratio 1.08; 95% CI 0.68–1.72). And nor did treat- ous treatment, overall safety and the major complication ment appear to improve functional outcome in sur- rate (e.g. vessel rupture). 288,289,293 vivors. The results were inconsistent between individual The CLOTBUST RCT has shown that, in patients trials. presenting within 3 h of acute ischaemic stroke onset who were treated with intravenous rt-PA, 2 h of con- Conclusion tinuous transcranial Doppler ultrasound increased the frequency of cerebral arterial recanalization and this These data do not support the routine or selective use was associated with a trend to better clinical out- of corticosteroids in acute ischaemic stroke. We do not comes. 294 Another phase II ultrasound trial, TRUMBI, is use corticosteroids as a routine treatment for unselected ongoing, 287 and CLOTBUST-2, evaluating microbubble- patients with ischaemic stroke, nor do we use them in augmented sono-thrombolysis is planned. 231 patients who have massive cerebral oedema complicated by transtentorial herniation. Our only indication is in patients with inflammatory vascular disorders (e.g. Conclusion giant-cell arteritis, polyarteritis nodosa, systemic lupus Further RCTs of interventional radiology (preferably on erythematosus; section 7.3). a cardiological scale) will be needed to assess its place in routine clinical practice. 288,291,293 As with intra-arterial 12.7.2 Glycerol and mannitol thrombolysis, the number of centres able to deliver such highly specialized treatments will need to increase Rationale dramatically for such trials to become feasible, and enormous investment would be required to minimize A 10% solution of glycerol is a hyperosmolar agent the average time to treatment. Ultrasound augmentation which is said to reduce cerebral oedema and possibly of thrombolysis is simpler, and potentially more widely increase cerebral blood flow, and mannitol is another applicable. osmotically active agent. These treatments aim to reduce intracranial pressure by reducing infarct-related oedema and thereby improve perfusion in and around infarcts. Evidence 12.7 Unproven value: treatment of cerebral oedema and raised intracranial pressure A Cochrane systematic review included 11 randomized trials comparing intravenous glycerol with control (482 glycerol treated patients were compared with 463 control patients). 296 Glycerol was associated with a significant 12.7.1 Corticosteroids reduction in the odds of death during the scheduled treatment period (OR 0.65; 95% CI 0.44–0.97). However, Rationale at the end of the scheduled follow-up period, there was Vasogenic cerebral oedema, which tends to develop about no significant difference in the odds of death (OR 0.98; 24–48 h or more after stroke onset, particularly in and 95% CI 0.73–1.31). Functional outcome was reported in around large infarcts, may be reduced by corticosteroids only two studies and there were non-significantly more (such as dexamethasone) (section 12.1.5). Cytotoxic patients who had a good outcome at the end of sched- oedema, that develops almost immediately after the uled follow-up in the glycerol group (OR 0.73; 95% CI .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 690 690 Chapter 12 Specific treatments for acute ischaemic stroke 0.37–1.42). Haemolysis seemed to be the only relevant included in DESTINY, 302 DECIMAL 303 and HAMLET 304 adverse effect of glycerol treatment. and treated within 48 h after stroke onset were pooled A Cochrane systematic review of mannitol in stroke for analysis. 321 The protocol for these analyses was included three small trials with 226 randomized pa- designed prospectively when the trials were still recruit- tients. Neither beneficial nor harmful effects of mannitol ing patients and outcomes were defined without know- could be proved. 297 The number of included patients ledge of the results of the individual trials. The primary was small, and the follow-up was short. Case fatality, the outcome measure was the score on the modified Rankin proportion of dependent patients, and adverse effects scale (mRS) at 1 year dichotomized between favourable were not reported. A similar Cochrane review of mannitol (0–4) and unfavourable (5 and death) outcome. A total in traumatic head injury found no evidence of benefit of 93 patients were included in the pooled analysis. More in patients with raised intracranial pressure who did not patients in the decompressive surgery group than in t have an operable intracranial haematoma. 298 he control group had an mRS ≤ 4 (75% vs 24%; pooled absolute risk reduction 51% [95% CI 34–69]), an mRS ≤ 3 (43% vs 21%; 23% [95% CI 5–41]), and survived (78% Conclusion vs 29%; 50% [95% CI 33–67]), indicating number- These data do not support the routine or selective use of needed-to-treat of two for survival with mRS ≤ 4, four for glycerol or mannitol in acute ischaemic stroke, and so we survival with mRS ≤ 3, and two for survival irrespective generally do not use them in our own clinical practice of functional outcome. The effect of surgery was highly (though our colleagues in intensive care may use man- consistent across the three trials. The authors concluded nitol in some of our patients with severe elevations of ‘In patients with malignant MCA infarction, decom- intracranial pressure after ischaemic stroke). pressive surgery undertaken within 48 h of stroke onset reduces mortality and increases the number of patients with a favourable functional outcome. The decision to 12.7.3 Surgical decompression perform decompressive surgery should, however, be made on an individual basis in every patient. 321 Posterior fossa decompression or shunting for massive cerebellar infarcts Conclusion: decompressive surgery Surgical decompression (or ventricular shunting) of mas- sive cerebellar infarcts may improve cerebral perfusion, Even with these results, the decision about whether relieve any obstructive hydrocephalus and so prevent and when to consider decompressive surgery therefore early death. Decompressive surgery has not been evalu- remains very difficult; leave it too late and you ‘miss the ated by adequately designed trials, so it is still contro- boat’ and irreversible damage occurs. Operate too soon, versial as to which patients with cerebellar infarction and one risks worsening the ischaemic brain insult. We should have decompressive surgery, shunting or medical do occasionally refer younger patients with no com- therapy. orbidity and a ‘malignant’ course for surgery, and hope to recruit patients in the HAMLET trial. Hemicraniotomy for massive supratentorial infarcts 12.7.4 Hyperventilation Decompressive surgical techniques that attempt to relieve high intracranial pressure due to oedema have been Hyperventilation is used to lower intracranial pres- described, 299 but their efficacy in reducing case fatality sure, but there is no randomized evidence on its effects and disability has, until recently, been very uncertain. A in patients with stroke available on the Cochrane Cochrane systematic review did not identify any com- Controlled Trials Register. A Cochrane review of hyper- pleted RCTs, five observational studies and some small ventilation in severe head injury concluded the data series and single case reports, so no clear conclusions available were inadequate to assess any potential benefit could be drawn. 300 At least five RCTs have since been ini- or harm that might result from hyperventilation. 305 tiated: HAMLET, HEADDFIRST, DESTINY, HEMMI and We therefore would not recommend hyperventilation DECIMAL, of which three (HEADDFIRST, 301 DESTINY 302 to reduce raised intracranial pressure, though we accept and DECIMAL 303 ) have been completed (recruitment that our colleagues in intensive care may sometimes into HAMLET continues). A prospectively pooled ana- administer hyperventilation to some of our patients with lysis of individual data for patients aged between 18 years severe elevations of intracranial pressure after ischaemic and 60 years, with space-occupying MCA infarction, stroke. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 691 12.8 Unproven value: other interventions 691 gain a licence for use in acute ischaemic stroke in the 12.8 Unproven value: other interventions next few years. It is probable that the extremely large reductions in cerebral infarct volumes achieved with 12.8.1 Neuroprotective agents neuroprotective agents in experimental animal models will translate into only moderate reductions, Rationale if any, in disability when used in humans; There are many – perhaps too many – points in the patho- neuroprotective therapy is unlikely to prove a panacea physiological cascade between vessel occlusion and for acute stroke. irreversible cell death where pharmacological inter- vention might be beneficial (sections 12.1.3 and 12.1.4), 12.8.2 Haemodilution and the pharmaceutical industry has worked through a very large number of compounds for clinical develop- Rationale ment and testing. There is no generally agreed definition of a neuroprotective drug, but in the context of acute Haemodilution aims to reduce whole-blood viscosity by ischaemic stroke, the aim of this class of agents is to limit an amount that increases cerebral blood flow yet does the volume of brain damaged by ischaemia. There are a not reduce cerebral oxygen delivery. The hypothesis is number of steps to be taken between first identifying that this should reduce ischaemic damage in and around a promising agent during preclinical testing in animals the infarct. Haemodilution is generally performed by and man before any large-scale clinical trials are giving an infusion of dextran, hydroxyethyl starch or mounted. 306,307 And systematic reviews of the available albumin. Such infusions increase blood volume (hyper- animal data offer the prospect of reducing bias in the volaemic haemodilution). In patients with acute stroke assessment of preclinical data. 308–311 Greater attention to in whom an increase in total blood volume may be achieving these milestones might increase the chances undesirable, haemodilution can be achieved isovolaemic- of a successful clinical development and licensing of a ally by simultaneously removing several hundred milli- neuroprotective drug than hitherto. litres of blood. If the optimum haematocrit is achieved, Many trials have been completed, but no neuro- any increase in cerebral oxygen delivery achieved in protective drug has been licensed for clinical use. The practice might be neuroprotective and so lessen infarct ever-increasing number of compounds withdrawn from volume. clinical development in recent years has been a dis- appointment. This has certainly tempered the opinions Evidence of even the diehard optimists who said, just a few years back, that the neuroprotective drugs would make stroke A Cochrane systematic review of haemodilution started rehabilitation units redundant. It appeared in early 2006 within 72 h of stroke onset included 18 trials (including that an effective agent had been found, when the SAINT- 2766 patients). Haemodilution did not significantly I trial of the agent NXY-059 in 1700 patients with an reduce deaths within the first 4 weeks (OR 1.09; 95% CI acute ischaemic stroke showed a significant reduction 0.86–1.38). Similarly, haemodilution did not influence in post-stroke disabilities (modified Rankin Scale) among deaths within 3–6 months (OR 1.01; 95% CI 0.84– patients treated with the study drug (P = 0.038). 312 1.22), or death and dependency or institutionalization Though the full results of a second phase III trial, SAINT- (OR 0.98; 95% CI 0.84–1.15). No statistically significant 2, are awaited, a press release in October 2006 stated benefits were documented for any particular type of that ‘the SAINT-2 trial did not demonstrate a statistically haemodiluting agent. The overall results of this review significant reduction on the primary endpoint of stroke- were compatible both with modest benefit and moderate related disability in patients treated with NXY-059, as harm of haemodilution for acute ischaemic stroke, but it assessed versus placebo using the modified Rankin Scale has not been proven to improve survival or functional (mRS) (P = 0.33, odds ratio 0.94).’ 313 These disappointing outcome. 314 results led the company to discontinue clinical develop- ment of the compound. Conclusion Despite these unpromising results, the situation may change, and it is conceivable that one of the neuro- Given this information, we can see no reason to use this protective compounds currently undergoing trials may treatment in routine clinical practice, although it is used .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 692 692 Chapter 12 Specific treatments for acute ischaemic stroke quite widely in Austria and in some parts of Eastern 5 Siesjo BK. Pathophysiology and treatment of focal cerebral Europe. ischemia. Part II: Mechanisms of damage and treatment. J Neurosurg 1992; 77(3):337–54. 6 Kristian T, Siesjo BK. Calcium in ischemic cell death. Stroke 12.8.3 Calcium antagonists 1998; 29(3):705–18. 7 Pellerin L, Magistretti PJ. Neuroenergetics: calling upon Rationale astrocytes to satisfy hungry neurons. Neuroscientist 2004; 10(1):53–62. The influx of calcium through voltage-sensitive channels 8 Magistretti PJ. Neuron-glia metabolic coupling and has a role in neuronal death from ischaemia; this influx plasticity. J Exp Biol 2006; 209(Pt 12):2304–11. can be inhibited by a number of agents, so-called cal- 9 Bouzier-Sore AK, Voisin P, Canioni P, Magistretti PJ, cium channel blockers or antagonists (section 12.1.4). Pellerin L. Lactate is a preferential oxidative energy substrate over glucose for neurons in culture. J Cereb Blood Flow Metab 2003; 23(11):1298–306. Evidence 10 Leenders KL, Perani D, Lammertsma AA, Heather JD, A systematic review included 29 RCTs (7665 patients). 315 Buckingham P, Healy MJ et al. Cerebral blood flow, blood No effect of calcium antagonists on poor outcome at the volume and oxygen utilization: normal values and effect of age. Brain 1990; 113 ( Pt 1):27–47. end of follow-up (relative risk, 1.04; 95% CI 0.98–1.09) 11 Baron JC. Mapping the ischaemic penumbra with PET: or on death at end of follow-up (relative risk, 1.07; 95% a new approach. Brain 2001; 124(Pt 1):2–4. CI 0.98–1.17) was found. Indirect comparisons of the 12 Guadagno JV, Warburton EA, Aigbirhio FI, Smielewski P, effects of intravenous and oral administration showed Fryer TD, Harding S et al. Does the acute diffusion- an increase in the number of patients with poor outcome weighted imaging lesion represent penumbra as well as with intravenous administration. Comparisons of differ- core? A combined quantitative PET/MRI voxel-based ent doses of nimodipine suggested that the highest doses study. J Cereb Blood Flow Metab 2004; 24(11):1249–54. were associated with poorer outcome. Administration 13 Guadagno JV, Donnan GA, Markus R, Gillard JH, Baron JC. within 12 h of onset was associated with an increase in Imaging the ischaemic penumbra. Curr Opin Neurol 2004; the proportion of patients with poor outcome, but this 17(1):61–7. effect was largely due to the poor results associated with 14 Baron JC. Stroke research in the modern era: images versus dogmas. Cerebrovasc Dis 2005; 20(3):154–63. intravenous administration. 15 Gibbs JM, Wise RJ, Leenders KL, Jones T. Evaluation of cerebral perfusion reserve in patients with carotid-artery Conclusion occlusion. Lancet 1984; 1(8372):310–14. 16 Warach S, Baron JC. Neuroimaging. Stroke 2004; The trials do not provide any evidence to support the 35(2):351–3. routine use of oral calcium antagonists in patients 17 Warach S, Wardlaw J. Advances in imaging 2005. Stroke presenting within 12 h, or intravenous administration. 2006; 37(2):297–8. We do not routinely use calcium antagonists in acute 18 Derdeyn CP. Positron emission tomography imaging ischaemic stroke patients. of cerebral ischemia. Neuroimaging Clin N Am 2005; 15(2):341–xi. 19 Muir KW, Buchan A, von Kummer R, Rother J, Baron JC. Imaging of acute stroke. Lancet Neurol 2006; 5(9):755–68. 20 Johnston AJ, Steiner LA, Balestreri M, Gupta AK, Menon References DK. Hyperoxia and the cerebral hemodynamic responses to moderate hyperventilation. Acta Anaesthesiol Scand 2003; 47(4):391–6. 1 Sharp FR, Lu A, Tang Y, Millhorn DE. Multiple molecular 21 Symon L, Branston NM, Strong AJ. Autoregulation in penumbras after focal cerebral ischemia. J Cereb Blood Flow acute focal ischemia: an experimental study. Stroke 1976; Metab 2000; 20(7):1011–32. 7(6):547–54. 2 Arundine M, Tymianski M. Molecular mechanisms of 22 Eames PJ, Blake MJ, Dawson SL, Panerai RB, Potter JF. calcium-dependent neurodegeneration in excitotoxicity. Dynamic cerebral autoregulation and beat to beat blood Cell Calcium 2003; 34(4–5):325–37. pressure control are impaired in acute ischaemic stroke. 3 Baron JC. Mapping the ischaemic penumbra with PET: J Neurol Neurosurg Psychiatry 2002; 72(4):467–72. implications for acute stroke treatment. Cerebrovasc Dis 23 Dawson SL, Panerai RB, Potter JF. Serial changes in static 1999; 9(4):193–201. and dynamic cerebral autoregulation after acute ischaemic 4 Siesjo BK. Pathophysiology and treatment of focal cerebral stroke. Cerebrovasc Dis 2003; 16(1):69–75. ischemia. Part I: Pathophysiology. J Neurosurg 1992; 24 Pulsinelli W. Pathophysiology of acute ischaemic stroke. 77(2):169–84. Lancet 1992; 339(8792):533–6. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 693 References 693 25 Yam AT, Lang EW, Lagopoulos J, Yip K, Griffith J, metabolic description of tissue at risk in acute stroke Mudaliar Y et al. Cerebral autoregulation and ageing. patients. Stroke 2006; 37(7):1778–84. J Clin Neurosci 2005; 12(6):643–6. 40 Roc AC, Wang J, Ances BM, Liebeskind DS, Kasner SE, 26 Aaslid R, Lindegaard KF, Sorteberg W, Nornes H. Detre JA. Altered hemodynamics and regional cerebral Cerebral autoregulation dynamics in humans. Stroke 1989; blood flow in patients with hemodynamically significant 20(1):45–52. stenoses. Stroke 2006; 37(2):382–7. 27 Panerai RB, Dawson SL, Eames PJ, Potter JF. Cerebral blood 41 Markus HS. Transcranial Doppler ultrasound. flow velocity response to induced and spontaneous sudden J Neurol Neurosurg Psychiatry 1999; 67(2):135–7. changes in arterial blood pressure. Am J Physiol Heart Circ 42 Derdeyn CP. Is the acetazolamide test valid for Physiol 2001; 280(5):H2162–H2174. quantitative assessment of maximal cerebral 28 Derdeyn CP, Videen TO, Yundt KD, Fritsch SM, autoregulatory vasodilation? Stroke 2000; 31(9):2271–2. Carpenter DA, Grubb RL et al. Variability of cerebral 43 Yamauchi H, Kudoh T, Sugimoto K, Takahashi M, Kishibe blood volume and oxygen extraction: stages of cerebral Y, Okazawa H. Pattern of collaterals, type of infarcts, and haemodynamic impairment revisited. Brain 2002; haemodynamic impairment in carotid artery occlusion. 125(Pt 3):595–607. J Neurol Neurosurg Psychiatry 2004; 75(12):1697–701. 29 Powers WJ. Cerebral hemodynamics in ischemic 44 Markus H, Cullinane M. Severely impaired cerebrovascular cerebrovascular disease. Ann Neurol 1991; 29(3):231–40. reactivity predicts stroke and TIA risk in patients with 30 Schellinger PD, Latour LL, Wu CS, Chalela JA, Warach S. carotid artery stenosis and occlusion. Brain 2001; The association between neurological deficit in acute 124(Pt 3):457–67. ischemic stroke and mean transit time: comparison of four 45 Baron JC. Using PET to identify carotid occlusion patients different perfusion MRI algorithms. Neuroradiology 2006; at high risk of subsequent stroke: further insights. J Neurol 48(2):69–77. Neurosurg Psychiatry 2004; 75(12):1659–60. 31 Muir KW, Santosh C. Imaging of acute stroke and transient 46 Derdeyn CP, Grubb RL, Jr., Powers WJ. PET screening of ischaemic attack. J Neurol Neurosurg Psychiatry 2005; 76 carotid occlusion. Adm Radiol J 2001;20(1):20–5. Suppl 3:iii19–iii28. 47 Derdeyn CP, Gage BF, Grubb RL, Jr., Powers WJ. Cost- 32 Pindzola RR, Balzer JR, Nemoto EM, Goldstein S, Yonas H. effectiveness analysis of therapy for symptomatic carotid Cerebrovascular reserve in patients with carotid occlusive occlusion: PET screening before selective extracranial-to- disease assessed by stable xenon-enhanced CT cerebral intracranial bypass versus medical treatment. J Nucl Med blood flow and transcranial Doppler. Stroke 2001; 2000; 41(5):800–7. 32(8):1811–17. 48 Htun P, Fateh-Moghadam S, Tomandl B, Handschu R, 33 Mahagne MH, David O, Darcourt J, Migneco O, Dunac A, Klinger K, Stellos K et al. Course of platelet activation and Chatel M et al. Voxel-based mapping of cortical ischemic platelet-leukocyte interaction in cerebrovascular ischemia. damage using Tc 99 m L,L-ethyl cysteinate dimer SPECT Stroke 2006; 37(9):2283–7. in acute stroke. J Neuroimaging 2004; 14(1):23–32. 49 Warach S. Tissue viability thresholds in acute stroke: the 34 Murphy BD, Fox AJ, Lee DH, Sahlas DJ, Black SE, Hogan MJ 4-factor model. Stroke 2001; 32(11):2460–1. et al. Identification of penumbra and infarct in acute 50 Heiss WD, Rosner G. Functional recovery of cortical ischemic stroke using computed tomography perfusion- neurons as related to degree and duration of ischemia. derived blood flow and blood volume measurements. Ann Neurol 1983; 14(3):294–301. Stroke 2006; 37(7):1771–7. 51 Marchal G, Beaudouin V, Rioux P, de la Sayette V, Le Doze 35 Koroshetz WJ, Lev MH. Contrast computed tomography F, Viader F et al. Prolonged persistence of substantial scan in acute stroke: ‘You can’t always get what you want volumes of potentially viable brain tissue after stroke: a but . . . you get what you need’. Ann Neurol 2002; correlative PET-CT study with voxel-based data analysis. 51(4):415–16. Stroke 1996; 27(4):599–606. 36 Eastwood JD, Alexander MJ, Petrella JR, Provenzale JM. 52 Hacke W, Albers G, Al Rawi Y, Bogousslavsky J, Davalos A, Dynamic CT perfusion imaging with acetazolamide Eliasziw M et al. The Desmoteplase in Acute Ischemic challenge for the preprocedural evaluation of a patient Stroke Trial (DIAS): a phase II MRI-Based 9-hour window with symptomatic middle cerebral artery occlusive disease. acute stroke thrombolysis trial with intravenous Am J Neuroradiol 2002; 23(2):285–7. desmoteplase. Stroke 2005; 36:66–73. 37 Hellier KD, Hampton JL, Guadagno JV, Higgins NP, 53 Furlan AJ, Eyding D, Albers GW, Al-Rawi Y, Lees KR, Antoun NM, Day DJ et al. Perfusion CT helps decision Rowley HA et al. Dose Escalation of Desmoteplase for making for thrombolysis when there is no clear time of Acute Ischemic Stroke (DEDAS): evidence of safety and onset. J Neurol Neurosurg Psychiatry 2006; 77(3):417–19. efficacy 3 to 9 hours after stroke onset. Stroke 2006; 38 Ziyeh S, Rick J, Reinhard M, Hetzel A, Mader I, Speck O. 37(5):1227–31. Blood oxygen level-dependent MRI of cerebral CO 2 54 Fibrinolytic Therapy Trialists’ (FTT) Collaborative Group. reactivity in severe carotid stenosis and occlusion. Indications for fibrinolytic therapy in suspected acute Stroke 2005; 36(4):751–6. myocardial infarction: collaborative overview of early 39 Geisler BS, Brandhoff F, Fiehler J, Saager C, Speck O, mortality and major morbidity results from all randomised Rother J et al. Blood-oxygen-level-dependent MRI allows trials of more than 1000 patients. Lancet 1994; 343:311–22. .. ..

9781405127660_4_012.qxd 10/16/07 10:47 AM Page 694 694 Chapter 12 Specific treatments for acute ischaemic stroke 55 Baron JC. How healthy is the acutely reperfused ischemic multimodal magnetic resonance imaging. Stroke 2003; penumbra? Cerebrovasc Dis 2005; 20 (Suppl 2):25–31. 34(11):2729–35. 56 Ginsberg MD. Adventures in the pathophysiology of brain 72 Butcher KS, Parsons M, MacGregor L, Barber PA, Chalk J, ischemia: penumbra, gene expression, neuroprotection: Bladin C et al. Refining the perfusion-diffusion mismatch the 2002 Thomas Willis Lecture. Stroke 2003; hypothesis. Stroke 2005; 36(6):1153–9. 34(1):214–23. 73 Kucinski T, Naumann D, Knab R, Schoder V, Wegener S, 57 Markus R, Reutens DC, Kazui S, Read S, Wright P, Pearce Fiehler J et al. Tissue at risk is overestimated in perfusion- DC et al. Hypoxic tissue in ischaemic stroke: persistence weighted imaging: MR imaging in acute stroke patients and clinical consequences of spontaneous survival. Brain without vessel recanalization. Am J Neuroradiol 2005; 2004; 127(Pt 6):1427–36. 26(4):815–19. 58 Baron JC, Warach S. Imaging. Stroke 2005; 36(2):196–9. 74 Coutts SB, Hill MD, Hu WY. Hyperperfusion syndrome: 59 Heiss WD. Best measure of ischemic penumbra: positron toward a stricter definition. Neurosurgery 2003; emission tomography. Stroke 2003; 34(10):2534–5. 53(5):1053–8. 60 Heiss WD, Kracht LW, Thiel A, Grond M, Pawlik G. 75 Bright R, Raval AP, Dembner JM, Perez-Pinzon MA, Penumbral probability thresholds of cortical flumazenil Steinberg GK, Yenari MA et al. Protein kinase C delta binding and blood flow predicting tissue outcome in mediates cerebral reperfusion injury in vivo. J Neurosci patients with cerebral ischaemia. Brain 2001; 2004; 24(31):6880–8. 124(Pt 1):20–9. 76 Price CJ, Warburton EA, Menon DK. Human cellular 61 Latchaw RE, Yonas H, Hunter GJ, Yuh WT, Ueda T, inflammation in the pathology of acute cerebral Sorensen AG et al. Guidelines and recommendations ischaemia. J Neurol Neurosurg Psychiatry 2003; for perfusion imaging in cerebral ischemia: a scientific 74(11):1476–84. statement for healthcare professionals by the writing 77 Lo EH, Dalkara T, Moskowitz MA. Mechanisms, group on perfusion imaging, from the Council on challenges and opportunities in stroke. Nat Rev Neurosci Cardiovascular Radiology of the American Heart 2003; 4(5):399–415. Association. Stroke 2003; 34(4):1084–104. 78 Lipton P. Ischemic cell death in brain neurons. Physiol Rev 62 Warach S. Measurement of the ischemic penumbra with 1999; 79(4):1431–568. MRI: it’s about time. Stroke 2003; 34(10):2533–4. 79 Graham SH, Chen J. Programmed cell death in cerebral 63 Sobesky J, Zaro WO, Lehnhardt FG, Hesselmann V, ischemia. J Cereb Blood Flow Metab 2001; 21(2):99–109. Neveling M, Jacobs A et al. Does the mismatch match the 80 Chan PH. Reactive oxygen radicals in signaling and penumbra? Magnetic resonance imaging and positron damage in the ischemic brain. J Cereb Blood Flow Metab emission tomography in early ischemic stroke. Stroke 2005; 2001; 21(1):2–14. 36(5):980–5. 81 Lo EH, Moskowitz MA, Jacobs TP. Exciting, radical, suicidal: 64 Guadagno JV, Jones PS, Fryer TD, Barret O, Aigbirhio FI, how brain cells die after stroke. Stroke 2005; 36(2):189–92. Carpenter TA et al. Local relationships between restricted 82 Moskowitz MA, Lo EH. Neurogenesis and apoptotic cell water diffusion and oxygen consumption in the ischemic death. Stroke 2003; 34(2):324–6. human brain. Stroke 2006; 37(7):1741–8. 83 Bergersen LH, Magistretti PJ, Pellerin L. Selective 65 Rivers CS, Wardlaw JM. What has diffusion imaging in postsynaptic co-localization of MCT2 with AMPA receptor animals told us about diffusion imaging in patients with GluR2/3 subunits at excitatory synapses exhibiting AMPA ischaemic stroke? Cerebrovasc Dis 2005; 19(5):328–36. receptor trafficking. Cereb Cortex 2005; 15(4):361–70. 66 Hjort N, Christensen S, Solling C, Ashkanian M, Wu O, 84 Lizasoain I, Cardenas A, Hurtado O, Romera C, Mallolas J, Rohl L et al. Ischemic injury detected by diffusion imaging Lorenzo P et al. Targets of cytoprotection in acute ischemic 11 minutes after stroke. Ann Neurol 2005; 58(3):462–5. stroke: present and future. Cerebrovasc Dis 2006; 67 Rivers CS, Wardlaw JM, Armitage PA, Bastin ME, 21 (Suppl 2):1–8. Carpenter TK, Cvoro V et al. Do acute diffusion- and 85 Bernardinelli Y, Magistretti PJ, Chatton JY. Astrocytes + perfusion-weighted MRI lesions identify final infarct generate Na -mediated metabolic waves. Proc Natl volume in ischemic stroke? Stroke 2006; 37(1):98–104. Acad Sci USA 2004; 101(41):14937–42. 68 Shih LC, Saver JL, Alger JR, Starkman S, Leary MC, 86 Aarts MM, Arundine M, Tymianski M. Novel concepts in Vinuela F et al. Perfusion-weighted magnetic resonance excitotoxic neurodegeneration after stroke. Expert Rev Mol imaging thresholds identifying core, irreversibly infarcted Med 2003; 2003:1–22. tissue. Stroke 2003; 34(6):1425–30. 87 Chu CT, Levinthal DJ, Kulich SM, Chalovich EM, 69 Wardlaw JM, Farrall AJ. Diagnosis of stroke on DeFranco DB. Oxidative neuronal injury. The dark side neuroimaging. Br Med J 2004; 328(7441):655–6. of ERK1/2. Eur J Biochem 2004; 271(11):2060–6. 70 Coutts SB, Simon JE, Tomanek AI, Barber PA, Chan J, 88 Lipton SA, Bossy-Wetzel E. Dueling activities of AIF in cell Hudon ME et al. Reliability of assessing percentage death versus survival: DNA binding and redox activity. of diffusion-perfusion mismatch. Stroke 2003; Cell 2002; 111(2):147–50. 34(7):1681–3. 89 Okuno S, Saito A, Hayashi T, Chan PH. The c-Jun N- 71 Kidwell CS, Alger JR, Saver JL. Beyond mismatch: evolving terminal protein kinase signaling pathway mediates Bax paradigms in imaging the ischemic penumbra with activation and subsequent neuronal apoptosis through .. ..