Perioperative stroke after noncardiac, nonneurosurgical procedures is more common than generally acknowledged. It is reported to have an incidence of 0.05-7% of patients. Most are thrombotic in origin and are noted after discharge from the postanesthetic care unit. Common predisposing factors include age, a previous stroke, atrial fibrillation, and vascular and metabolic diseases. The mortality is more than two times greater than in strokes occurring outside the hospital. Delayed diagnosis and a synergistic interaction between the inflammatory changes normally associated with stroke, and those normally occurring after surgery, may explain this increase. Intraoperative hypotension is an infrequent direct cause of stroke. Hypotension will augment the injury produced by embolism or other causes, and this may be especially important in the postoperative period, during which monitoring is not nearly as attentive as in the operating room. Increased awareness and management of predisposing risk factors with early detection should result in improved outcomes.
STROKE is an important cause of morbidity and mortality, particularly in patients more than 65 yr old. In cardiac, neurologic, and carotid surgery, the incidence is known to be high (2.2–5.2%).1However, little is known regarding perioperative stroke following other types of surgery including general, urologic, orthopedic, thoracic, and gynecologic procedures. The aims of this article are to review the incidence, pathophysiology, risk factors, and outcomes associated with perioperative stroke following noncardiac, nonneurologic, and vascular surgery. Suggestions regarding the timing of elective surgery after stroke and ways in which one can reduce the incidence and improve outcomes are also outlined.
The World Health Organization definition of stroke is a “focal or global neurologic deficit of cerebrovascular cause that persists beyond 24 h or is interrupted by death within 24 h.” Transient ischemic attack is acute loss of focal cerebral or ocular function with symptoms lasting less than 24 h and is usually presumed to be embolic or thrombotic in origin. In addition, a third type of cerebrovascular event has recently attracted much attention in the nonsurgical setting. Covert stroke is an asymptomatic ischemic event usually only detected by advanced neuroimaging techniques, such as diffusion-weighted magnetic resonance imaging sequences.2Although the diagnosis is often missed at the time of the event, covert stroke has been associated with an adverse effect on cognitive function and quality of life. Currently other than in cardiac and carotid artery surgery, there is no study evaluating the incidence, impact, and risk factors of covert stroke in the general surgical population.3This review therefore concentrates on perioperative overt stroke that is commonly defined as one that occurs during and up to 30 days after surgery.
Incidence, Morbidity, and Mortality of Perioperative Stroke in Noncardiac and Nonneurosurgical Procedures
The reported incidence of stroke following procedures other than cardiac, neurosurgical, and carotid artery surgery ranges between 0.05 and 7.4% (table 1).4,–,17Differences in patient population, changing clinical practice over 40 yr (1967–2009), study design, diagnostic tests, and duration of follow up may account for the large variance in reported preoperative stroke rates. The majority of the older and current studies are retrospective reviews of administrative databases. Although most major strokes are usually revealed in these databases,18it is likely that minor strokes, covert strokes, and transient ischemic attacks are missed because they are not noted by the caregivers or the clinical notes are not correctly interpreted by the coding officers.
Outcome after perioperative stroke is usually devastating.16In contrast to the 12.6% mortality rate associated with strokes in the nonsurgical setting,19mortality from perioperative stroke ranges from 26% after general surgery10to 87% in patients who have had a previous stroke.9,16Early mortality from major stroke may result from delayed recognition and diagnosis on surgical services, cerebral edema, and intracranial hypertension, whereas late mortality may be caused by aspiration, pneumonia, metabolic derangement, sepsis, or myocardial infarction.7
Extensive evidence implicates inflammation in multiple phases of stroke pathophysiology, and there is increasing awareness that inflammatory events outside the brain have an important impact on stroke susceptibility and outcome, perhaps by amplification of those pathways activated in stroke (fig. 1).20,21It is probable that the acute systemic inflammatory response triggered by surgery initiates or exacerbates ischemic cerebral injury. Laboratory studies using sepsis models in conjunction with neurologic injury22,23and clinical studies demonstrate more severe neurologic deficits, especially when preceded by respiratory and urologic infections.24,–,29Various cytokines have been implicated in the postoperative inflammatory response including interleukin-1, interleukin-6, and tumor necrosis factor α,30and an increased C-reactive protein level.31Interleukin-6 appears to be the major mediator of the stress response after surgery. Interestingly, the peak plasma concentration of interleukin-6 correlates significantly with computed tomography brain infarct volume and clinical outcome assessed by modified Rankin scale score at 3 months in patients who suffer an acute ischemic stroke. A peak plasma interleukin-6 concentration greater than 30 pg/ml was associated with increased mortality at 12 months.32The association between interleukin-6 measured on admission and in-hospital mortality in patients with acute ischemic stroke has been demonstrated in another study, which found that a 1-unit increase in interleukin-6 predicted an 18% higher risk of dying during hospitalization for acute ischemic stroke.33
Strokes are either ischemic or hemorrhagic. Ischemic stroke may be classified according to presenting symptoms and signs, using the Oxfordshire Community Stroke Project categories, or etiologically, according to the definition in the Trial of Org 10172 in Acute Stroke Treatment.34,35The latter identifies the pathophysiologic mechanism of vessel occlusion, including large artery thrombosis, artery to artery embolism, cardioembolic, and small vessel (lacunar) occlusion. Watershed infarction is another important mechanism of ischemic stroke. Cortical and internal (white matter, centrum ovale) watershed areas occur at the junction of the distal fields of two or more nonanastomosing arterial systems and are vulnerable to ischemia because of their position, where perfusion pressure is the lowest.36The pathophysiology of watershed infarcts is controversial, with substantial evidence supporting both low flow and multiple microembolic mechanisms.37An embolic component is most frequent with cortical watershed infarcts, especially when associated with atheromatous stenosis of inflow vessels. Hemodynamic factors seem most important for internal watershed infarcts or infarcts associated with complete occlusion of inflow vessels. It has been postulated that hypoperfusion, or hypotension, and embolism play a synergistic role: the decreased perfusion reduces clearance of microemboli, and the blocked vessels extend the hypoperfused area.38
The majority of perioperative strokes occur after the second postoperative day.7,10Of all the studies that reported the details of stroke, only 5.8% (14 of 242) of cases were thought to have occurred during surgery,4,–,11,13,15suggesting that postoperative events rather than intraoperative mechanisms are important.7,11It is estimated that more than 60% of strokes in the cardiothoracic population have an embolic origin; about 12–15% are ischemic because of hypoperfusion, lacunar infarction, and thrombosis; 1% are hemorrhagic; 10% have multiple etiologies; and 15% have unknown etiology.39The pathophysiology of perioperative stroke occurring in patients undergoing noncardiac, nonneurosurgical procedures is less well defined. Currently only nine studies have reported the mechanisms of stroke after noncardiac and nonneurologic surgery (301 strokes, table 1).6,–,12,15In contrast to the cardiothoracic patients, the majority, or 68%, of strokes were because of cerebrovascular thrombosis. About 16% of strokes were thought to be embolic in nature. Another 5% of strokes were because of intracerebral hemorrhage.
It is unclear why patients would have higher incidences of thrombotic strokes after noncardiac and nonneurosurgical procedures. However, postoperative endothelial dysfunction may be a major contributory factor. In this regard, the vascular endothelium is important in the regulation of vascular tone, thrombosis, and inflammation through the release of nitric oxide, prostacyclin, and a series of endothelial-derived hyperpolarizing factors.40,41Vessels with endothelial dysfunction are prone to plaque rupture, reactive vasospasm, and thrombus formation. General anesthetics, in particular nitrous oxide, impair endothelial function.42,43In conjunction with the neuroendocrine, or “stress,” response after surgery, it is plausible that endothelial dysfunction predisposes patients to cerebrovascular thrombosis although there is no clinical evidence currently to support this. In addition, withholding antiplatelet agents or anticoagulants in the perioperative period may aggravate surgery-induced hypercoagulability and increase the risk of stroke.44,–,47The lower incidence and mortality in hip arthroplasty patients found by Bateman et al. may reflect the early anticoagulation of these patients.16Fourteen percent of strokes after general surgery are associated with atrial fibrillation, again highlighting the importance of embolism and the hypercoagulable state.7,–,9
Who Is at Risk?
The risk factors most consistently linked to perioperative ischemic stroke are shown in table 2. Age, a previous history of stroke, and atrial fibrillation are among the most important. There are conflicting data regarding whether stroke occurs more frequently in men than in women.48,49
The role of extracranial carotid artery stenosis on perioperative stroke remains controversial.16,49Carotid bruit does not correlate with the severity of the underlying carotid artery stenosis, per se , and has not been shown to increase the risk of perioperative stroke.7,50Even in patients with significant carotid artery stenosis who are undergoing cardiac surgery, strokes are often contralateral to the affected carotid artery and therefore cannot be attributed to the stenosis alone.51,52However, similar data in noncardiac, nonneurosurgical patients are currently lacking. The risk of perioperative stroke in patients with intracranial cerebral artery stenosis is unknown, but these patients carry a much higher risk (15% each yr) of stroke in the nonsurgical setting, and therefore possibly in the perioperative period, too.53
Type of Surgery
The nature of surgery also has some bearing on perioperative stroke.1,16,49For example, hip arthroplasty and peripheral vascular surgery are associated with a higher incidence than knee arthroplasty or general surgery.1,16,49Head and neck surgery increases the risk for stroke to 0.2–5%.54,–,57Patients undergoing neck dissection for cancer typically present with multiple medical comorbidities that already increase their risk for stroke. In addition, external beam radiation accelerates atheromatous changes in arteries.58It is therefore not surprising that manipulation of these diseased vascular structures could result in plaque rupture, embolism, and vasospasm.59,–,61
Other procedures that potentially compromise cerebral perfusion may also increase the risk of stroke. Pohl and Cullen reported four cases of strokes and ischemic spinal cord injury in patients after shoulder surgery in the beach-chair (nearly 90 degrees upright) position.62Two patients had posterior fossa infarcts, one had disseminated cerebral and temporal infarcts, and one had a unilateral watershed infarct. The authors could only speculate on mechanisms and suggested that postural hypotension and extreme rotation, or hyperflexion of the neck, resulted in a decrease in cerebral blood flow and potentially aggravated thromboembolic mechanisms.62,–,64A study measuring cerebral oximetry in the sitting position found a high incidence, 80% of patients, with a more than 20% decrease in saturation, but no adverse neurologic consequences. This may reflect the small sample size and the need for cerebral saturation to be reduced for a relatively long time, perhaps up to 50 min at less than 50% saturation to produce an adverse outcome.65Despite studies using surrogate endpoints, the risk of stroke after shoulder surgery in beach-chair position remains undefined.64
β-Blockers and Hypotension
A meta-analysis by Bangalore et al. found an increased risk of nonfatal stroke, hypotension, and bradycardia with the use of β-blockers in patients undergoing noncardiac surgery.66The recent PeriOperative Ischemic Evaluation trial (POISE) carried the greatest weight in the analysis.67. Although there is an association among β-blocker use, hypotension, and stroke, this cannot be assumed to be a direct cause and effect.66Many aspects remain unclear. There are substantial variations in drug choice, drug dose, and regimen; definitions of outcomes including definition of hypotension; and patients' surgical and medical risks. Previous studies have shown that hypotension may not increase the risk of perioperative stroke even in patients with extracranial carotid artery stenosis.7,51,68Importantly, it is unclear in studies, including POISE, whether the hypotension occurred immediately before the stroke or after. POISE reports “clinically significant hypotension,” defined as a systolic blood pressure of less than 100 mmHg for an unspecified time, as an “intraoperative and postoperative predictor” of stroke (see table 4), but the results do not state the temporal relationship between hypotension and the stroke, and also fail to specify if the hypotension was intraoperative or postoperative.67Postoperative hypotension occurring on the ward is likely to be prolonged and therefore potentially more detrimental than intraoperative hypotension, which is usually easily detected. There is consensus that patients chronically on β-blockers should be kept on their drugs through the perioperative period, and that caution is needed when these agents are begun immediately before surgery.
It is unclear whether the stroke rate is increased with other antihypertensive or sympatholytic agents, such as clonidine and dexmedetomidine. The introduction of these drugs in the perioperative period could increase vulnerability to cerebral ischemia, as the compensatory mechanisms to cope with hypotension are attenuated because of hypovolemia and blood loss. Clinical trials will be needed to address this. The POISE 2 trial, which is expected to be completed in 2014, includes a clonidine arm and more detailed reporting on intra- and postoperative blood pressure (personal communication, April 2010, P. J. Devereaux, M.D., Ph.D., Associate Professor, McMaster University, Hamilton, Ontario, Canada).
Stroke Prevention – Risk Modification
Timing Elective Surgery after a Recent Stroke
Acute stroke impairs cerebral autoregulation so that cerebral blood flow becomes passively dependent on perfusion pressure. The injured brain is therefore vulnerable to even modest hypotension.69,70The impairment in cerebral autoregulation is not limited to the affected stroke hemisphere, but appears to be an overall phenomenon that occurs to the same extent after both anterior and posterior circulation strokes.70Failure of cerebral autoregulation and vasomotor response to carbon dioxide has been demonstrated within 8 h after ictus and lasting for 2–6 months.70,–,73Ideally, enough time should be allowed for autoregulation to be restored before elective surgery, as well as abatement of the inflammatory response. Previous investigators have recommended delaying nonurgent surgery for 1–3 months after stroke.9,74If surgery needs to take place sooner, blood pressure should be meticulously monitored, and one could make a reasonable argument for monitoring for cerebral ischemia using transcranial Doppler or neurophysiology (e.g. , electroencephalography and evoked potentials).
Preoperative Carotid Artery Revascularization
In patients with high-grade (more than 70% stenosis) symptomatic carotid artery stenosis, revascularization in the form of carotid artery stenting or endarterectomy should be offered before nonurgent elective surgery. However, in accordance with the 2009 European Society for Vascular Surgery guideline, carotid endarterectomy is contraindicated for patients with less than 50% stenosis.75Revascularization for patients with asymptomatic carotid artery stenosis is much more controversial. Some evidence suggests that intensive medical therapy offers superior secondary stroke prevention compared with carotid endarterectomy or stenting.76This includes smoking cessation, blood pressure control, anticoagulation for atrial fibrillation, and administration of lipid lowering and antiplatelet therapy. In contrast, the recently published 10-yr follow-up of the European Asymptomatic Carotid Surgery Trial found a benefit favoring surgery in asymptomatic patients with more than 60% stenosis. The implications for preoperative carotid endarterectomy were not addressed.77
Perioperative Management of Atrial Fibrillation
Patients with preexisting atrial fibrillation who have received antiarrhythmic or rate-controlling agents should continue therapy throughout the perioperative period. Intravenous formulations should be used if necessary. Correction of postoperative electrolyte imbalances and fluid volume is important, as electrolyte disturbances and dehydration increase atrial activity and predispose the arrhythmia.49Although no randomized controlled trial is available to specifically address the problem of anticoagulation for postoperative atrial fibrillation, the American College of Chest Physicians recommends that heparin therapy be considered for high-risk patients or those with a previous history of stroke or transient ischemic attack, and to continue anticoagulation therapy for 30 days after the return of normal sinus rhythm.78The Effectiveness of Bridging Anticoagulation for Surgery study, which compares dalteparin with a placebo, is expected to be completed by 2013. For patients with new onset atrial fibrillation, echocardiography and cardioversion are appropriate.
Perioperative Management of Patients on Oral Anticoagulant Therapy
Patients who are on oral anticoagulant therapy pose a common and challenging perioperative problem. There is a need to balance the risk of a thromboembolic event during interruption of oral anticoagulant therapy against the risk of bleeding. However, there is a paucity of data to inform decisions regarding perioperative antithrombotic therapy.79Consequently, there is little consensus on the optimal management strategy. Nevertheless, three approaches to oral anticoagulation management have been suggested: (1) to continue warfarin therapy, (2) to withhold warfarin therapy for some time before and after the procedure, or (3) to temporarily withhold warfarin therapy while also providing a short acting, or bridging, anticoagulant during the perioperative period.80
For procedures that carry a low risk of bleeding, continuation of warfarin therapy is probably safe.79However, in patients undergoing major surgery or invasive procedures, interruption of antithrombotic therapy is typically required to avoid bleeding.81,–,83The risk stratification and bridging anticoagulant approaches adopted by the 2008 guidelines of the American College of Chest Physicians should advise clinical practice (table 3).
Perioperative Management of Patients on Antiplatelet Drugs
Withdrawing antiplatelet therapy given for secondary prevention of stroke exposes patients to an increased risk of recurrent ischemic stroke. This may be related to the rebound or prothrombotic effect associated with antiplatelet withdrawal in addition to the prothrombotic effect of surgery.84,85On the other hand, continuing antiplatelet therapy may increase the risk of major bleeding. Unfortunately, there are currently no data to guide the management, and clinical practice has been variable. The situation is more problematic when antiplatelet drugs are given for previous placement of intracranial stents. Withholding antiplatelet in these patients may confer higher risks of stroke. The POISE 2 trial, a large randomized control trial evaluating the risks and benefits of continuing antiplatelet therapy, has recently started recruiting patients, and should be completed by 2014. This study may provide important guidelines for patients taking aspirin in the perioperative period.
Intraoperative Blood Pressure Management
Hypotension in the presence of severe internal carotid artery stenosis, occlusion, and/or an incompetent circle of Willis, the circulus arteriosus cerebri, are risk factors for watershed infarcts (see Pathophysiology).38,86Only 45–50% of the population has the typical complete or normal polygon configuration of the circle of Willis.87Maintenance of an adequate cerebral perfusion pressure is thus important, especially in patients with these risk factors. Patients who suffer a cerebral embolic event are also dependent on cerebral perfusion to maintain collateral flow to reduce infarct size. Unfortunately, there are few data to inform on optimal intraoperative blood pressure targets, and this is compounded by the lack of a standardized definition of hypotension.88Below the cerebral blood flow penumbra threshold, which is approximately a 50% reduction in cerebral blood flow, neuronal function is impaired, and the tissue is at risk of injury.89It is unknown whether prolonged reductions in cerebral blood flow above the penumbral threshold may also result in injury.86,90Studies in healthy, awake individuals suggest that the average lower limit of autoregulation is approximately 70 mmHg, i.e. , higher than usually stated in textbooks. It is commonly assumed that one should be safe keeping blood pressure within the individual's autoregulatory range.91There is, however, enormous individual variation in the lower limit of autoregulation, 41–113 mmHg, to the extent that it is inappropriate to assume that any one target arterial pressure may apply to a broad population of patients.91In addition, the lower limit of autoregulation is not static, being increased by hypertension; returned to “normal” with treatment of hypertension; and rapidly shifting with changes in sympathetic tone.92The latter presumably accommodates the 20% or more decrease in blood pressure that occurs during nonrapid eye movement sleep.93
Given such complexity, the absence of outcome data defining a blood pressure threshold for stroke in general surgical patients, and the relative infrequency of purely hemodynamic strokes, it is not surprising that there is no consensus on appropriate perioperative blood pressure targets.88A common practice is to maintain mean or systolic blood pressures within 20% of baseline, which is usually the blood pressure measured immediately before entry to the operating room.88,94This reference value is above the patient's usual awake blood pressure and even further increased above their sleep values.94,95Such a relatively conservative approach to intraoperative blood pressure management is justifiable in patients at high risk for perioperative strokes. In healthy patients, blood pressures that approximate sleep levels are likely quite acceptable, i.e. , 25–35% decrease from immediate preoperative baseline.95
Other Contributing Factors
Beyond avoidance of severe hyperglycemia and hypoglycemia, both of which are detrimental,96,97there is no consensus on the perioperative target glucose values to enhance clinical outcomes. Although an earlier study of patients in the surgical intensive care unit found a survival benefit following strict glucose control, or glucose concentrations 4.4–6.1 mM, compared with traditional management, or glucose concentrations 10–11.1 mM,98evidence from recent randomized controlled trials99and meta-analyses suggest that there is no benefit to overall mortality with strict glycemic control compared with traditional management.100,101In fact, an increased incidence of stroke was found with strict intraoperative glucose control in cardiac surgery patients.99The American Diabetes Association and the American Association of Clinical Endocrinologists have published a joint consensus statement defining a target range of 7.8–10 mM among critically ill patients. Pending results from prospective trials in patients undergoing major surgery, this seems a reasonable aim for all surgical patients.102
Patients with previous stroke are often taking statin and other lipid-lowering drugs for secondary stroke prevention. Discontinuation of statin therapy can be harmful and may acutely impair vascular function. Statin withdrawal in patients with acute stroke was associated with an 8.7-fold increase in the risk for early neurologic deterioration.103Thus, statin therapy should be continued throughout the perioperative period in at-risk patients. Instituting statins de novo preoperatively or immediately at stroke onset has yet to be convincingly shown in randomized trials to be of benefit.104
Genetic predisposition to perioperative stroke is not well understood or studied. A limited number of studies in cardiac surgery have found that a few genetic polymorphisms, including apolipoprotein E4 allele, glycoprotein Ib, and C-reactive protein contributed to worse cognitive outcome and poorer recovery after stroke in cardiac surgery.105,106Similar data in noncardiac surgery are currently lacking. Despite the advances in genomic research, currently no single locus has been identified in genome-wide association studies.
Early Management of a Perioperative Stroke
Important elements in early management are identifying at-risk patients and early diagnosis (table 4). Nursing staff initially detect the vast majority of perioperative strokes, as most strokes are not acute intraoperative events but are diagnosed during the postoperative period and later hospital stay.109However, routine surgical nursing care does not include a useful neurologic examination. There is extensive research and clinical experience with the use of the National Institutes of Health Stroke Scale and the Canadian Neurologic Scale.110These do not take long to do, but require some training, making it unlikely that they would be widely adopted. Wide integration of simpler scales would be a step in the direction of earlier detection.111There is also a need to shorten the time from nursing diagnosis to neurologist's assessment, without waiting for the surgical team to complete the day's operating room activities.112An increasing number of medium- to large-sized medical centers have an “Acute Stroke Team” that is managed by neurologists and who respond rapidly if a stroke is suspected. Emergency noncontrast scanning of the brain is the primary diagnostic brain-imaging study for evaluation of patients with suspected stroke, and the goal is to complete the computer tomography examination within 25 min of the provisional diagnosis being made.113Noncontrast computer tomography accurately discriminates ischemic stroke from intracranial hemorrhage and nonvascular causes of neurologic symptoms, such as tumors. It is, however, relatively insensitive in detecting acute and small cortical or subcortical infarctions, especially in the posterior fossa.114Multimodal computer tomography and magnetic resonance imaging may provide additional information that will improve diagnosis of ischemic stroke. However, emergency treatment of stroke should not be delayed to obtain multimodal imaging studies.113
General supportive care and prevention of complications are also important to patients with perioperative stroke. This may be best achieved by moving the surgical patient to an Acute Stroke Unit, where management is coordinated by stroke neurologists. Partial airway obstruction, hypoventilation, aspiration pneumonia, and atelectasis are common causes of hypoxia that may worsen the brain injury. Both hypertension and hypotension are associated with poor outcome after stroke. Urgent correction of common postoperative causes of hypotension, including volume depletion, blood loss, myocardial ischemia, or arrhythmias, may improve neurologic outcomes. Although there are data from small clinical trials suggesting that drug-induced hypertension could be used for the treatment of selected patients with acute ischemic stroke, there are no data from large clinical trials, and current consensus does not recommend this treatment.113Intentional hemodilution does not improve functional outcome and is not recommended for the treatment of acute stroke.113The source of any fever should be ascertained and treated, as fever is associated with poor neurologic outcome after acute stroke.115,116Both myocardial ischemia and cardiac arrhythmias are potential complications, and the current American Heart Association guidelines recommend that patients with acute ischemic stroke should have cardiac monitoring for at least the first 24 h and that any serious arrhythmia should be actively treated.113
Unfortunately, much of the emergency advanced treatment of acute stroke, such as pharmacologic thrombolysis, mechanical recanalization of occluded arteries, and heparin administration, are not suitable for patients after surgery. The current American Heart Association guidelines consider major surgery within 14 days of stroke a contraindication to intravenous thrombolysis. However, given the benefit to be gained, on a case-by-case basis, patients who suffer a perioperative ischemic stroke may be eligible for intravenous thrombolysis. Intraarterial thrombolysis is another option for treatment, either alone or in conjunction with intravenous thrombolysis, and may be safely administered to patients within 6 h of symptom onset. Currently, aspirin is the only oral antiplatelet agent that has been found beneficial in the treatment of acute ischemic stroke, and should be used in the perioperative period whenever deemed safe.113
In conclusion, perioperative stroke is more common than generally acknowledged. Increased awareness and management of predisposing risk factors with early detection should result in improved outcomes. Prospective studies of overt and covert stroke in the perioperative period are needed in order to inform incidence, pathophysiology, prevention, and treatment.