Pathophysiological and clinical considerations in the perioperative care of patients with a previous ischaemic stroke: a multidisciplinary narrative review

Summary

With an ageing population and increasing incidence of cerebrovascular disease, an increasing number of patients presenting for routine and emergency surgery have a prior history of stroke. This presents a challenge for pre-, intra-, and postoperative management as the neurological risk is considerably higher. Evidence is lacking around anaesthetic practice for patients with vascular neurological vulnerability. Through understanding the pathophysiological changes that occur after stroke, insight into the susceptibilities of the cerebral vasculature to intrinsic and extrinsic factors can be developed. Increasing understanding of post-stroke systemic and cerebral haemodynamics has provided improved outcomes from stroke and more robust secondary prevention, although this knowledge has yet to be applied to our delivery of anaesthesia in those with prior stroke. This review describes the key pathophysiological and clinical considerations that inform clinicians providing perioperative care for patients with a prior diagnosis of stroke.

Editor's key points.

• •The challenge of increasing numbers of patients with previous stroke presenting for surgery is growing.
• •This multidisciplinary narrative review provides key recommendations based on current evidence from clinical and physiological studies.

Long-term survival after first stroke in the era before widespread use of pharmacological reperfusion therapy was 86.1% at 30 days and 47% at 5 yr.¹ A population-based study between 1975 and 1989 showed that independent predictors of mortality included old age, congestive cardiac failure, persistent atrial fibrillation, ischaemic heart disease, and recurrent stroke.¹ Since the late 1980s, the care of patients after stroke has advanced significantly with a revolution in both pharmacological and mechanical reperfusion therapies resulting in an increasing number of stroke survivors. Stroke is a leading cause of death and the most serious neurological disease affecting those within the UK (incidence of 115–150 per 100 000 population),^2,3 However, there are ongoing challenges in the delivery of newer therapies, particularly endovascular therapy (EVT) to eligible patients.⁴ Mortality from stroke in those <65 yr of age has declined.⁵ Nevertheless, the prevalence of stroke disease in elderly patients is increasing and is associated with higher risk-adjusted mortality, longer duration of hospitalisation, and associated cognitive deficits after stroke.^6,7 The risk of recurrent stroke is highest within the first month after a stroke.⁸ Prior stroke disease is a key risk factor for perioperative stroke though there are several complicating factors to consider when anaesthetising these patients. Timing after stroke, comorbidities, arterial pressure, preoperative medications, and premorbid function are some of the many confounding factors. Baroreceptor function is altered in the first few weeks after stroke and this predisposes to increased risks of cardiovascular instability during anaesthesia and surgery.⁹ Consequently, physicians and anaesthetists are often consulted on optimal timing for both routine and emergency surgery, but limited evidence is available to support robust decision making. The increased vulnerability to cardiovascular instability and recurrent stroke should be weighed against the likely mortality benefit from urgent procedures and this can be difficult. It is widely considered beneficial to delay non-urgent surgery in patients who have suffered a recent stroke, although this is debated.¹⁰ The Society for Neuroscience in Anaesthesiology and Critical Care (SNACC) have produced consensus guidelines on perioperative stroke, amalgamating evidence on this topic and identifying significant gaps in knowledge.^11,12 Evidence is lacking on the pathophysiological changes that occur after stroke or transient ischaemic attack and their relevance to perioperative management. In addition, the cerebral vasculature is susceptible to intrinsic factors (carbon dioxide [CO₂], changes in arterial pressure, and glycaemic variability), and extrinsic factors (timing, anaesthetic agents and medications), all of which should be considered more carefully in this patient subgroup. This review will provide an understanding of the key pathophysiological and clinical considerations to help decision-making for clinicians providing perioperative care for patients with a prior ischaemic stroke, the largest stroke sub-type.

Pathophysiology, treatment, and recovery after stroke

Acute ischaemic stroke (AIS) manifests secondary to thrombotic occlusion (large or small vessel), embolic phenomena (with or without confirmed cardiac or carotid source) or secondary to systemic hypoperfusion (watershed infarct) or more rarely a venous occlusion.¹³ This produces a core of infarcted tissue surrounded by a penumbra of tissue with potentially reversible ischaemic injury. The extent of the damage is affected by the extent of the local collateral circulation. Within minutes to hours, cytotoxic oedema develops, which although reversible, pertains to release oxygen-derived free radicals.¹³ Unfortunately, over hours to days the irreversible vasogenic oedema peaks and can precipitate raised intracranial pressure (ICP).¹³ Acute intracerebral haemorrhage (ICH) can occur as a complication of AIS in the acute (immediate to 5 days) and sub-acute (5 days to weeks) phases. ICH manifests most often as haemorrhagic transformation of the cerebral infarction with haemorrhagic areas adjacent to acute ischaemic lesions.¹³ Arterial pressure management differs depending on the time from ischaemic insult, evidence supports necessity for control through endovascular and intravenous thrombolysis (IVT), and supportive management (acute), days to weeks (sub-acute), and greater than 3 months (long term) after stroke.

Arterial pressure and acute endovascular therapy, intravenous thrombolysis, and supportive management

The dawn of mechanical clot retrieval has prompted consideration for the best anaesthetic option for EVT after AIS. Some of the available evidence suggests that adverse neurological outcome is more likely after EVT with general anaesthesia (GA) as compared with conscious sedation. Firstly, several studies have demonstrated decreases in blood pressure during EVT and association with unfavourable outcome.¹⁴ However, in one study, arterial pressure <70–90 mm Hg, increased mean arterial pressure variability, and use of vasopressors were all higher in those undergoing EVT under GA compared with conscious sedation.¹⁵ In contrast, a recent RCT demonstrated no greater improvement in neurological status at 24 h or indeed mortality at 3 months for those undergoing GA compared with conscious sedation during thrombectomy for AIS.¹⁶ Importantly, post-hoc data from the Multicentre Randomized Clinical Trial of Endovascular Treatment of Acute Ischaemic Stroke in The Netherlands (MR CLEAN) study showed a U-shaped correlation between baseline pre-EVT systolic arterial pressure and functional outcome, although post-EVT arterial pressure data and association with outcomes are lacking.^17,18 Overall, the 2018 American Heart Association (AHA)/ASA stroke guidelines reflect the conflicting evidence and suggest ‘it is reasonable to select an anaesthetic technique during EVT for AIS on the basis of individualised assessment of patient factors, technical performance of the procedure, and other clinical characteristics’.¹⁹ Additionally, a SNACC Expert Consensus Statement advises systolic arterial pressure should be maintained between >140 and <180 mm Hg and diastolic <105 mm Hg through the periprocedural period.²⁰ Arterial pressure targets should then be reviewed with vascular neurologist and neurointerventionalist input, particularly if successful recanalisation is achieved.²⁰

For patients eligible for IVT with alteplase, guidelines state arterial pressure should be carefully lowered to systolic <185 mm Hg and diastolic <110 mm Hg before administration of the lytic agent.¹⁹ Unfortunately, patients with severe hypertension (>220/120 mm Hg) were largely excluded from clinical trials assessing arterial pressure lowering in AIS. Therefore, in those undergoing a supportive care strategy without IVT or EVT, guidelines state lowering arterial pressure in the first 24 h by 15% is probably safe.¹⁹

Post-stroke arterial pressure management (sub-acute and long term)

Hypertension is a crucial factor in the development of AIS.²¹ Hypertension after stroke can be transient and usually decreases with time. The potential benefits of improved cerebral perfusion in AIS²¹ need to be balanced against the potential for symptomatic ICH (via haemorrhagic transformation of an infarct) or cerebral oedema. In AIS, uncontrolled arterial pressure predicts unfavourable clinical outcomes at 3 months.^22,23 Therefore, it is crucial to consider preoperative arterial pressure control in patients with a history of stroke, particularly within the first 3 months. Systolic arterial pressure in AIS is usually much closer to the longer term pre-morbid level.²⁴ Well-managed hypertension to guideline-recommended levels within 90 days of index stroke decreases 1 yr stroke recurrence rates,²⁵ and higher post-stroke arterial pressure within the first year after index stroke is associated with higher risk of recurrent stroke in those aged >65 yr.²⁶

Recovery and risk of recurrence after ischaemic stroke

There is a paucity of long-term data on AIS outcomes. However, recent data demonstrate a concerning prognostic picture with two out of three AIS patients dead or dependent at 5 yr.²⁷ The likelihood of suffering death or dependence is increased in those with existing intracranial disease. Leucoaraiosis is common amongst elderly patients (>75 yr), the large majority of the stroke patient population. The larger the burden of leucoaraiosis or white matter hyperintensity, the worse the stroke outcomes in AIS, though this relationship is likely governed by stroke subtype with non-cardioembolic aetiology AIS being most affected (i.e. small or large vessel occlusions).²⁸

There are several determinants of early recurrence after AIS, including atherothrombotic infarction, history of diabetes and hypertension, and diastolic hypertension and hyperglycaemia during stroke admission.²⁹ After a stroke episode, neuroplasticity mediates recovery of motor function and this process can involve needing to relearn motor skills for example.³⁰ Unfortunately, aside from improvement in function, deterioration can also occur, the latter being a transient worsening of previously improved or residual post-stroke focal neurological deficits.³¹ Those more likely to experience such phenomena are patients with diabetes, dyslipidaemia, and leucoaraiosis, and smokers. The use of hydromorphone for analgesia,³² and using midazolam and fentanyl for sedation³³ have been implicated in transient neurological worsening of existing stroke disease. This is particularly pertinent as many patients receive conscious sedation (often by proceduralists) for transoesophageal echocardiogram or percutaneous gastrostomy tube placement within the 3 months after a stroke. Ultimately, data in patients undergoing cardiac surgery highlight previous stroke as a major risk factor for recurrent ischaemic stroke and major adverse cardiovascular events (MACE) if time elapsed between previous stroke and surgery is less than 3 months.³⁴ Supplementary Table S1 summarises the key considerations during treatment and recovery after stroke.

Secondary prevention after stroke and anaesthesia

Secondary prevention of recurrent cerebrovascular events after stroke includes a range of medications to control: risk of thrombosis, blood pressure, total and low-density lipoproteins (LDL) cholesterol, and to prevent hyperglycaemia.

Antiplatelet agents

Aspirin and clopidogrel are the most commonly used antiplatelet agents for secondary prevention of stroke. Observational studies suggest an increase in mortality with withdrawal of aspirin therapy perioperatively for cardiac surgery,³⁵ whereas the occurrence of bleeding and thrombosis during antiplatelet therapy in noncardiac surgery (OBTAIN)³⁶ study showed no clinical benefit in noncardiac surgery patients with prior percutaneous coronary intervention (PCI) on aspirin monotherapy. In the OBTAIN study, treatment with dual antiplatelet therapy was associated with a trend towards increased MACE risk (odds ratio [OR]=1.93; 95% confidence interval [CI], 0.93–3.88) and significantly increased bleeding risk (OR=6.55; 95% CI, 3.2–17.96).³⁶ The POISE-2 study in patients undergoing noncardiac surgery reported an increased risk of bleeding with no improvement in vascular outcomes or mortality.³⁷ However, it should be noted that the POISE-2 study excluded those with prior PCI/coronary stents and the duration of the study intervention for those already receiving aspirin was limited to 7 days.³⁷ As the use of clopidogrel is associated with a significant increase in bleeding complications, perioperative cessation is usually advised. Aspirin may be substituted perioperatively in case of high thrombotic risk and low bleeding risk.³⁸ Routine procedures should be delayed while the risk of perioperative stroke remains significantly elevated (∼9 months).³⁹ However, the decision is often based on a balance of risks of continuing anticoagulation and antiplatelets (bleeding) against stopping (thrombosis—venous, arterial, or stent related).⁴⁰ Other factors to be considered are drugs and combinations, type of surgery and bleeding risk, indication of drugs, and type of anaesthesia.^36,41,42 In those with cardiac stents, bare metal or drug-eluting, consensus guidelines advise at least 4 weeks and 12 months dual antiplatelet therapy (DAPT), respectively.⁴³ This is important because between 4% and 8% of patients who have received PCI subsequently require surgery within 1 yr of stenting, and perioperative MACE being greatest within the first 6 months after PCI is an important issue.⁴³ Lastly, periprocedural neuroendovascular antiplatelet strategies have largely been extrapolated from data in cardiac patient populations.⁴⁴ Early after intracranial stent insertion for example, despite DAPT, incidence of thrombotic rates is as high as 12%.⁴⁴ Crucially, patients with intracranial stents are distinct from those receiving antiplatelet therapy for AIS as thrombosis rates are significantly higher when antiplatelet therapy is stopped in those with stents. Newer strategies being investigated in this particular high-risk cohort include testing for genetic polymorphisms affecting bioactivation of clopidogrel and use of cilostazol, an older antiplatelet agent.⁴⁴ Overall, for those established indefinitely on antiplatelet therapy for secondary prevention of stroke, aspirin monotherapy can be continued for most invasive noncardiac procedures, except if bleeding risk is high, in which case stopping for 7 days can be carefully considered.⁴³ A proposed antiplatelet management strategy is summarised in Supplementary Table S2.

Anticoagulants

Anticoagulants are highly effective at reducing the risk of cardioembolic stroke, usually in the context of atrial fibrillation. The National Institute for Health and Care Excellence (NICE) recommends standard dose-adjusted warfarin or one of four direct oral anticoagulants (DOAC) to minimise the high risk of associated stroke.⁴⁵ Factors influencing perioperative management are summarised in Supplementary Table S3, and consequent clinical decision-making options are shown in Table 1. As the pharmacokinetics of DOAC are relatively stable, the timeline of anticoagulant effect is predictable, and specific recommendations for perioperative management of DOAC from the European Society of Cardiology are summarised in Table 2. In summary, patients with normal renal function undergoing routine procedures should not take a DOAC for 24 h before the procedure. Furthermore, in low-risk procedures, anticoagulation should be recommenced 6–8 h after the procedure though in high-risk procedures re-initiation should be delayed until at least 48 h after the procedure.⁴³

Table 1.

Proposed strategy for perioperative anticoagulant use. AF, atrial fibrillation; VTE, venous thromboembolism; DOAC, direct oral anticoagulant; LMWH, low molecular weight heparin; PCC, prothrombin complex concentrate; CCF, congestive cardiac failure; INR, international normalised ratio

Preoperative anticoagulation			Perioperative cover (bridging)	Postoperative anticoagulation
Continued for minor procedures (simple dental procedures; cataract extraction; endoscopy without intervention; superficial surgery)			Based on bleeding risk, there are three options: None, prophylactic LMWH or full-dose LMWH.	Anticoagulants should be resumed once haemostasis has been achieved.
Stopped for a few days based on days to loss of anticoagulant effect, and specific patient characteristics.			The BRIDGE 246 study reported non-inferiority between bridging and non-bridging for perioperative interruption of warfarin, with an increased risk of bleeding complications (absolute risk increase 1.9%; [95% CI 1.3–3.2%]; p=0.001). Routine bridging is thus not advised.	Standard advice
For emergency procedures, delay 12–24 h if possible. For dabigatran, use antidote.			If there are high thrombotic risk factors, bridging with prophylactic or full dose LMWH is advised on a case-by-case basis.	Simple procedures—restart DOAC at 6–8 h
For other, use PCC or Andexanet in discussion with Haematologist.				Complex procedures
Agent and patients specific factors and duration of stopping needed:				a)Start prophylactic LMWH at 6–8 h b)Restart DOAC at 48–72 h
Anticoagulant based	Patient based	Duration
Warfarin		5 days
Warfarin with other factors	Elderly, CCF, concomitant medications increasing anticoagulation effect; higher INR target	≥6 days Adjust on an individual basis
Acenocoumarol		3–4 days
DOAC		See Table 2

Table 2.

When to stop DOAC for routine surgical interventions adapted from European Heart Rhythm Association Practical Guide.⁴⁷ Low risk: with a low frequency of bleeding, minor impact of a bleeding, or both; high risk: with a high frequency of bleeding, important clinical impact, or both

Renal function (creatinine clearance, ml min−1)	Dabigatran		Factor Xa inhibitors (apixaban, edoxaban, rivaroxaban)
	Low risk	High risk	Low risk	High risk
≥80	≥24 h	≥48 h	≥24 h	≥48 h
50–80	≥36 h	≥72 h	≥24 h	≥48 h
30–50	≥48 h	≥96 h	≥24 h	≥48 h
15–30	Dabigatran contraindicated		≥36 h	≥48 h
<15	DOAC contraindicated
	Bridging therapy usually not indicated

Antihypertensives

In general, most chronic antihypertensive medication should be continued throughout the perioperative period, although dose adjustments may be needed in response to perioperative haemodynamic changes caused by haemorrhage, hypovolaemia, hypotension, or where renal perfusion is reduced. Recommendations for specific drug agents/classes are summarised in Table 3.

Table 3.

Recommendations for specific classes of antihypertensive medications, adapted from the Canadian Society Guidelines⁴⁸ and the European Society of Cardiology Guidelines.⁴⁹ ACEI, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor blockers; CCB, calcium channel blockers

Drug class	Evidence	Recommendation
Beta blockers
Chronic beta blocker therapy (except metoprolol)	Risk of coronary ischaemia with abrupt withdrawal, and theoretically useful in countering effects of catecholamine release, as well as suppressing arrhythmia.	Should generally be continued.
De novo beta blocker therapy		Not routinely indicated, unless new diagnosis of coronary artery disease.
Metoprolol	Adverse outcomes with perioperative and intraoperative use of 100–200 mg.50,51	Do not use de novo high-dose (>100 mg) metoprolol.
Other beta blockers	Observational evidence suggests no adverse effects from beta blockers other than metoprolol.	Short-acting agents such as esmolol are advised where required in a dose-titrated fashion.
CCB	No effect on outcomes.	De novo treatment with CCB not advised.
Alpha blockers	Risk of rebound hypertension with withdrawal of alpha blockers.	Should generally be continued.
ACEI and ARB	Perioperative use is associated with a higher rate of hypotensive episodes, but no clear increase in adverse outcomes.	Withhold 24 h before. Restart 24 h after.
Diuretics	Individualised based on assessment of volume status.	Withhold on day of procedure. Intravenous diuretic for fluid overload.

Statins

A small study in patients with acute stroke suggested that withdrawal of statins may be associated with worse outcome.⁵² However, there is a paucity of evidence with regard to perioperative statin use, and statin therapy should generally be continued. De novo statin initiation is not routinely indicated, but should be considered in the setting of vascular surgery.⁴⁹

Anaesthetic agents and cerebral blood flow

The problem

Anaesthetic agents can have a profound and varied influence on the balance of cerebral oxygen delivery (CDO₂) and utilisation. The former is determined by the product of cerebral blood flow (CBF) and oxygen content (i.e. CDO₂=CBF×CaO₂), and the latter is determined largely by local metabolism (i.e. the cerebral metabolic rate of oxygen [CMRO₂]). Under normal situations, there is a tight coupling between CBF and CMRO₂; however, as illustrated in Figure 1, based on the selective inclusion of studies in humans, various anaesthetic agents can result in a differential influence on CBF and CMRO₂ and hence lead to an uncoupling of flow and metabolism (Fig. 2). For example, indicative of uncoupling of flow and metabolism, volatile anaesthetics (isoflurane, sevoflurane, desflurane) and nitrous oxide have been reported to increase CBF and elicit reductions in CMRO₂.54, 55, 56, 57, 58^,73, 74, 75, 76 However, these findings do not seem to be universal (Fig. 2).⁵⁹ In contrast, indicative of maintained flow–metabolism coupling, many other anaesthetic and sedative agents (i.e. propofol, etomidate, dexmedetomidine, benzodiazepines, and barbiturates) can result in reductions in both CBF and CMRO₂,⁵⁸ whereby these concurrent reductions are highly correlated with each other.⁶⁰ Consistent with this flow–metabolism coupling, ketamine and fentanyl have been reported to increase both CBF and CMRO₂.^77,78 In addition to the variety of methods used to assess CBF, CMRO₂, or both, variability in the mismatch between CBF and oxidative metabolism also likely arises from the direct effects of volatile anaesthetics on cerebral vascular resistance and direct/indirect effects on local metabolism.⁵³ Importantly, there is marked regional heterogeneity in these responses.^61,62

Fig 1.

Simplified summary of cerebral blood flow (CBF) changes with common anaesthetic agents. Simplified illustration of the impact of common anaesthetic agents on both CBF and the cerebral metabolic rate of oxygen (CMRO₂). The text in bold indicates anaesthetic agents where the reductions (or increases) in CMRO₂ are generally reflected in reciprocal changes in CBF (i.e. flow–metabolism coupling). The non-bolded text indicates agents that result in an uncoupling of this flow and metabolism. See text and recent review⁵³ for human-based references.

Fig 2.

Influence of different anaesthetics on cerebral blood flow and metabolism. The top row of figures presents data on cerebral blood flow whereas the bottom row shows the corresponding changes in metabolism for each anaesthetic agent. Studies utilising transcranial Doppler ultrasound are coloured red, those utilising the Kety and Schmidt¹⁴⁸ technique are coloured green, those utilising positron emission tomography are coloured blue, those utilising xenon inhalation are coloured orange, and those using magnetic resonance imaging are coloured purple. As depicted, isoflurane, propofol, and N₂O lead to relative consistent changes in cerebral blood flow and metabolism, whereas the blood flow response of sevoflurane is highly variable. Also important to note is that although these studies represent global changes in flow and metabolism, regional heterogeneity will exist.56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72

The following question arises: Is there a clinical consequence of coupling or uncoupling of this flow–metabolism matching? For example, neuroprotection may be facilitated when delivery is greater than utilisation (i.e. CDO₂>CMRO₂).⁷⁹ Conversely, potential neurological damage may ensue in a situation when the anaesthetic agents result in an increased utilisation that is not met by an appropriate delivery (i.e. CMRO₂>CDO₂). It should be noted, however, that the impact of many other anaesthetic agent properties, as well their dose responsiveness, and the implications for use before operation and during recovery after ICH/AIS remain unknown.

The evidence

Anaesthetic agents and carbon dioxide

Carbon dioxide is likely the most potent regulator of CBF and an important clinical target during mechanical ventilation of anaesthetised patients. By altering extracellular/perivascular pH, the cerebral vasculature is highly sensitive to increases (i.e. hypercapnia) and decreases (i.e. hypocapnia) in the partial pressure of arterial CO₂ (Pa_CO2); these changes result in marked increases and decreases in CBF, respectively.⁸⁰ This responsiveness to changes in CO₂ is termed ‘cerebrovascular reactivity’ (CVR), and is a crucial mechanism in the regulation of neuronal haemostasis. Moreover, controlling Pa_CO2 to reduce cerebral blood volume and thus ICP is a common practise in neuro-anaesthesia and neuro-critical care. Consequently, it is important to understand how CVR is influenced by different anaesthetic agents.

Anaesthesia with propofol reduces CVR.⁸¹ Comparatively, sevoflurane leads to a greater CVR than propofol at equipotent doses (determined via bispectral index) in elderly patients.⁸² However, consideration of potential differences in CVR between young and old is imperative, especially considering propofol⁸³ and sevoflurane⁸⁴ may reduce CVR to a greater extent in the elderly. Another important consideration is that CVR is influenced by blood pressure.⁸⁰ With regard to the anaesthetised patient, a higher blood pressure is related to greater CO₂ reactivity during anaesthesia with either sevoflurane or propofol.⁸⁵ Therefore, careful consideration of the impact of these and other anaesthetics,⁵³ on CVR is required.

How do we optimise?

When utilising volatile anaesthetics, isoflurane leads to a lower MCA velocity (MCAv) than sevoflurane, but a much greater CO₂ reactivity.⁸⁶ Therefore, sevoflurane may be suited for use when CO₂ mediated control of cerebral blood volume and ICP is needed.^74,87, 88, 89 However, this suggestion comes with the caveat that volatile anaesthetics convey a comparatively greater detriment to autoregulation than other anaesthetics such as propofol^85,90 and fentanyl.⁸⁵ Indeed, re-analysis of data by McCulloch and colleagues⁹⁰ (reported in Table 1 of their study), indicates an interaction effect between Pa_CO2 and choice of anaesthetic (sevoflurane vs propofol) on the autoregulation index (ARI) (Fig. 3) (see section on control of blood flow for further discussion of anaesthetics and cerebral autoregulation).

Fig 3.

Interaction between different anaesthetics, Pa_CO2, and cerebral autoregulation. A clinical study by McCulloch and Colleagues⁹⁰ in 2000 aimed to investigate the influence of anaesthesia on what level of hypercapnia was needed to impair autoregulation (n.b. Impaired static autoregulation was arbitrarily determined as an autoregulation index [ARI] ≤0.4, where 0 is the absence of autoregulation and 1 is ‘perfect’ autoregulation). They determined that under sevoflurane anaesthesia the threshold Pa_CO2 for less efficient autoregulation was 56(4) mm Hg, whereas under propofol anaesthesia it was 61(4) mm Hg.⁹⁰ Interesting in and of itself, these findings are further strengthened by re-analysis of their data with linear mixed modelling demonstrating an interaction effect between choice of anaesthetic and Pa_CO2 on cerebral autoregulation (ARI). Fixed factors were drug (sevoflurane vs propofol) and Pa_CO2. Subjects were included as a random effect. The statistical output is displayed on the graph, with solid lines representing the derived regression line for each anaesthetic. Under sevoflurane anaesthesia, increases in Pa_CO2 led to greater reductions (i.e. impairments) in autoregulation than under propofol anaesthesia.⁹⁰

Anaesthesia considerations and control of cerebral blood flow

In ischaemic stroke, the risk of secondary injury is increased in the presence of hypo- or hyperperfusion, which could lead to an extension of the penumbra, oedema, or both, and capillary damage. These extreme conditions are more likely to occur if control of CBF against changes in arterial pressure, termed cerebral autoregulation (CA), is impaired, as reported in 32% of patients with AIS, with a strong association with the severity of injury.^91,92

When considering the timing of routine surgery in patients with stroke, it is also relevant to consider the temporal course of changes in dynamic CA in these patients. In the ultra-acute stage, dynamic CA might not be affected, but an increasing number of studies suggest that CA is impaired 5–10 days later, with a gradual recovery in the following weeks.93, 94, 95, 96, 97

In addition to pathology, anaesthetic agents are also known to influence CA.^53,85,98, 99, 100, 101, 102 In general, volatile anaesthetics, such as isoflurane, sevoflurane, and desflurane, lead to impairment of CA in a dose-dependent manner, notable above 1.5 MAC with isoflurane or desflurane, but above 1.0 MAC with sevoflurane.^53,85,101 In contrast, intravenous agents do not seem to interfere with CBF control.^53,98 With propofol, CA remains intact in doses up to 200 μg kg⁻¹ min.⁸⁵ Ketamine increases CBF, and it has been proposed to also provide neuroprotection.^53,79 A number of other intravenous agents that do not disturb CA, such as etomidate, dexmedetomidine, opioids (fentanyl, morphine, remifentanil), benzodiazepines, or barbiturates, have also been suggested to provide neuroprotection, but in a recent review, Slupe and Kirsch⁵³ concluded that none of these agents have unequivocal evidence for this role.

Clinical application

Assessment of CA before surgery could provide valuable information for surgical planning, including the choice of anaesthetic agent(s) and the need for closer monitoring, but evidence of impact on outcome with this practice has not been demonstrated by clinical trials. Assessment of CA can be performed over different timescales. Static CA involves measurements of CBF and arterial pressure averaged over several minutes,103, 104, 105 whereas dynamic CA assesses the response of CBF caused by rapid changes in arterial pressure, usually observed over a few seconds.^105,106 Recent development of methods to estimate changes in CBF with higher temporal resolution, such as transcranial Doppler (TCD) ultrasound and near-infrared spectroscopy (NIRS), and concerns about the methodological limitations of static CA methods¹⁰⁷ have led to the current preference for the dynamic approach to assess CA in patients with cerebrovascular conditions.¹⁰⁸ In particular, the use of transfer function analysis, to estimate dynamic CA parameters, based on spontaneous fluctuations in arterial pressure and CBF, involves minimal physiological disturbance and discomfort to patients, and obviates the need for large fluctuations in arterial pressure as a stimulus to induce a strong dynamic CA response, which has been of concern.109, 110, 111

Continuous monitoring of dynamic CA during surgery may provide a means of protecting cerebral haemodynamics in patients with prior AIS or ICH. However, owing to the need for continuous measurement of CBF, and concurrent model fitting,¹⁰⁸ no commercial systems are available for this purpose yet. The risks of simply monitoring arterial pressure, aiming to maintain normotension, should be appreciated. In the absence of simultaneous monitoring of CBF and ICP, it is not possible to make reliable estimates of cerebral perfusion. Even in patients where cerebral hypertension can be ruled out, intra-surgical changes in CBF cannot be predicted simply from parallel changes in arterial pressure, because of the non-linear nature of cerebral vascular resistance. In particular, the critical closing pressure of the cerebral circulation determines the effective perfusion pressure, and can be highly variable with changes in Pa_CO2, ICP, anaesthetic agents, and vasomotor tone.^99,112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126 For these reasons, intra-surgical continuous measurement of CBF, with either TCD or NIRS, requires further prospective evaluation before specific recommendations can be made on their potential value for preventing secondary injury in patients undergoing surgery after ischaemic or haemorrhagic stroke.

Intraoperative care

The intraoperative care of patients who have already suffered a stroke is poorly studied, with very little prospective evidence to support a particular approach. It remains unclear whether the process of extrapolating risk modification from studies of those patients who develop perioperative stroke is adequate for those with a previous stroke. However, a number of controversies dominate decision making in this area, which are addressed below.

What is the best time to undertake routine surgery?

The risk of recurrent stroke without surgery decreases over time in the first year after a stroke. However, it is unclear as to what extent surgery adds to that risk within the first year. The most comprehensive study to date was a retrospective review of 481 183 patients undergoing routine noncardiac surgery; 7137 patients previously having had an ischaemic stroke. The adjusted risk of mortality, MACE, or further ischaemic stroke within 30 days was increased markedly in those within 3 months of a stroke, and remained elevated but showed a stepwise decline at 3–6, 6–12, and >12 months.³⁹ Further analysis compared the 7137 patients who had surgery after a stroke to a matched cohort of 72 007 patients who had a stroke, but did not undergo surgery. It found that the risk of suffering a further ischaemic stroke was significantly higher in those patients undergoing surgery than those not (in the first 3 months after stroke, surgery [11.95%] vs no surgery [3.54%]; at >12 months, surgery [1.43%] vs no surgery [0.32%]).^39,127 These results led the authors to conclude that the risk of mortality, MACE, and further ischaemic stroke plateaus at 9 months, although risk remained increased relative to those who have never suffered a stroke (at >12 months, OR [95% CI]; 30 day all-cause mortality, 1.12 [0.74–1.69]; MACE, 2.47 [1.66–3.68]; 30 day ischaemic stroke, 9.6 [3.82–24.11]). Balancing the risks of surgical timing against the risks of delaying surgery will remain a complex decision, to be taken on an individual basis. However, it would seem prudent to delay non-urgent routine surgery for at least 9 months after an ischaemic stroke unless there is sufficient benefit from earlier surgery to justify it or accept the risks and what we can to mitigate them by careful attention to physiological variables. The data support a significant decline in risk of a poor outcome at 12 months, as compared with the 3–6 month period, with a plateau beginning at 9 months.

What is the best time to undertake emergency surgery?

Christiansen and colleagues¹²⁸ carried out a retrospective analysis of 146 694 patients undergoing non-vascular emergency surgery. Overall, 5.4% of patients included had previously had an ischaemic stroke. In these patients, risk of death in the 30 days after surgery was increased for at least 9 months after the index stroke,¹²⁸ with risk greatest in the first 3 months (<3 months: OR=1.65; 95% CI, 1.45–1.88; 3–9 months: OR=1.20; 95% CI, 0.98–1.47; >9 months: OR=1.20; 95% CI, 1.08–1.34). A similar pattern was observed in the risk of perioperative stroke, particularly in the 3 months after the index stroke (OR=23.36; 95% CI, 19.24–28.37), remaining elevated at >9 months (OR=5.16; 95% CI, 4.03–6.62).

Importantly, the study found that the risk of MACE was in fact lower in patients who underwent emergency surgery in the first 3 days after a stroke, than those who underwent surgery 3–14 days after, and in comparing this immediate vs early time frame, 30 day mortality risk was similar, indicating that if surgery is urgent, there may be no benefit, or indeed harm in delaying.¹²⁸ This study has several limitations including retrospective design, lack of clarification for recurrent stroke aetiology, and lack of clarification as to whether these postoperative events were indeed recurrent or new infarcts.¹²⁹ Despite the lack of immediate biological plausibility, we have previously provided data on dynamic CA being unaffected in the ultra-acute stage after stroke, with CA impairment beginning 5–10 days later, with a gradual recovery in the following weeks. Ultimately, the decision for any procedure depends on the recovery from stroke, functional disability and mobility, quality of life, type and urgency of surgery, and overall capacity to benefit.

Is there a benefit to choosing a regional technique?

In certain surgical procedures, particularly lower limb orthopaedic surgery, central neuraxial blockade offers an alternative to GA. Importantly though, this necessitates withdrawal of anticoagulation and certain antithrombotic drugs.

No data were available on the interaction between having a history of stroke, and the risk of death or stroke subsequent to regional or GA, and there are no prospective trials examining stroke risk in relation to anaesthetic technique. However, there is evidence that in routine hip or knee replacement, neuraxial blockade is associated with lower adjusted risk of death or risk of a cerebrovascular event than GA.¹³⁰ The benefits of central neuraxial blockade, necessitating withdrawal of clopidogrel must be balanced against the increased risk of cerebrovascular thrombosis.^131,132

Control of intraoperative blood pressure

A recent systematic review and meta-analysis of a large undifferentiated sample of patients undergoing noncardiac surgery by Wesselink and colleagues¹³³ found that there were no statistically significant associations between intraoperative hypotension and risk of postoperative stroke. A recent analysis of 7457 patients undergoing cardiac surgery found that the risk of perioperative stroke was increased in line with time spent with a mean arterial pressure <65 mm Hg.¹³⁴

Very few of the patients in the included studies had a history of stroke, and there is no good evidence describing the effect of intraoperative hypotension on the risk of perioperative stroke in those patients who have already suffered a stroke. However, it is well known that CA is impaired for at least a month after stroke.^97,135 The aetiology of perioperative stroke in patients with no history of stroke is predominantly embolic or thrombotic, with little contribution demonstrated by watershed ischaemia, at least in cardiac surgery.^136,137

Given the findings of Wesselink and colleagues¹³³ that although not associated with increased stroke risk, intraoperative hypotension is associated with increased perioperative mortality, risk of myocardial infarction, and acute kidney injury, and given that patients after stroke have impaired cerebral autoregulatory function for at least a month after a stroke, we recommend intraoperative blood pressure should be maintained as close as practical to the patient's preoperative, awake blood pressure, and should not be allowed to decrease below a MAP of 80 mm Hg based on the point at which cumulative risk begins to accrue in non-stroke patients. The caveat to this recommendation is its basis on observational data alone with acknowledgement for the difficulties and indeed controversy surrounding thresholds for intraoperative arterial pressure management.¹³⁸ Furthermore, the paucity of data available limits the ability to provide specific individualised MAP goals for vulnerable sub-types of ischaemic stroke including large vessel occlusion and watershed territory infarction. These recommendations align with current SNACC guidelines which recommend avoidance of intraoperative hypotension in those at high risk of perioperative stroke. In addition, SNACC recommends an individualised patient-specific definition of intraoperative hypotension calculated as a percent reduction from baseline arterial pressure rather than an absolute value.¹¹ Careful consideration of the effects of hypertension on autoregulation should be given to invasive or noninvasive continuous arterial pressure monitoring to facilitate this, especially in patients who suffered stroke <3 months previously, to allow early recognition of haemodynamic compromise.

Glucose control

There is insufficient evidence to make recommendations on perioperative glycaemic control for patients who have previously suffered a stroke other than those already recommended for surgical patients more generally.^11,137 However, the AHA/ASA guidelines state that it is reasonable to treat hyperglycaemia to achieve glucose levels in the range of 140–180 mg dl⁻¹.¹⁹ Blood glucose should be closely monitored to prevent hypoglycaemia (blood glucose <60 mg dl⁻¹).¹⁹ Recent data from the Stroke Hyperglycaemia Insulin Network Effort (SHINE) randomised trial demonstrated no improvement in 90 day functional outcomes between intensive (target blood glucose concentration of 80–130 mg dl⁻¹) compared with standard (target glucose concentration of 80–179 mg dl⁻¹) glucose control.¹³⁹

Transfusion

There is insufficient evidence to make recommendations on when to provide transfusion of red cells to patients who have suffered a stroke. However, studies in both cardiac and noncardiac surgery have demonstrated a link between receiving a blood transfusion and the risk of a first stroke after operation.^140,141 In critically ill patients with AIS, it is thought that both high and low haematocrit are dangerous; high compromising CBF potentially caused by viscosity changes though conflicting data does exist, and low compromising oxygen delivery because of inadequate oxygen carriage.^142,143 As such, the decision to transfuse must be made on a case-by-case basis, weighing surgical risks, patient factors, and time since stroke with the benefits of blood transfusion.

Ventilation and oxygenation

Owing to the alterations in CA, the adequacy of CBF is persistently at risk of compromise because of hypocapnia in patients who have suffered a stroke.¹⁴⁴ As such, ventilation should be carefully titrated to ensure normocapnia is maintained. With reference to oxygenation, a study of 16 037 patients across 25 RCT assessing oxygen therapy in acute illnesses including stroke, demonstrated increased in-hospital mortality (relative risk [RR]=1.21; 95% CI, 1.03–1.43) during liberal oxygen therapy without improvement in patient-important outcomes.¹⁴⁵ The only stroke-specific RCT included within this study demonstrated prophylactic use of low-dose oxygen therapy did not reduce death or disability at 3 months.¹⁴⁶ These findings are reflected in current AHA/ASA guidelines which recommend the provision of oxygen for maintenance of saturations at a lower limit of >94%, with no upper limit.^19,147

Postoperative care

Care should be taken to maintain the same arterial pressure targets as those used intraoperatively, and particular care should be paid to the timing of postoperative antithrombotic and anticoagulant drugs on a case-by-case basis.

Recommendations

This review has outlined the concerns around anaesthetising those with a prior diagnosis of stroke disease by outlining the key pathophysiological and clinical considerations. There exists a lack of definitive evidence to support previous recommendations of a minimum period of 3 months between the stroke and surgery. The key recommendations from this review, based on extrapolated data from observational data, suggest:

• •Delay non-urgent surgery for at least 9 months after an ischaemic stroke unless the benefits of earlier surgery outweigh the increased risks of perioperative stroke during this time.
• •There are few prospective data on which to base recommendations for intraoperative care of patients with a previous stroke. However, if surgery is performed within 9 months of a stroke, arterial pressure should be controlled as far as possible with appropriate monitoring and interventions throughout the perioperative period.
• •Concurrent medications should be managed according to existing guidelines, though these should be considered on a case-specific basis and the individual risks and benefits acknowledged.
• •The benefits of central neuraxial blockade must be balanced against the increased risk of cerebrovascular thrombosis related to the withdrawal of clopidogrel therapy.

Authors' contributions

Study concept: JSM, TGR

Design, literature searching, initial drafting, and final version of the article: all authors

Declarations of interest

JSM, WR, RBP, RLH, PNA, JPT, TGR declare no relevant conflicts of interest. AKM has received honoraria for non-promotional educational talks from companies manufacturing direct oral anticoagulants (Bayer, Boehringer-Ingelheim, Daiichi-Sankyo, and BMS-Pfizer); conference travel grants from Boehringer-Ingelheim; and funding for an investigator-initiated trial from Novo-Nordisk. JSM, is an NIHR Clinical Lecturer in Older People and Complex Health Needs at the Department of Cardiovascular Sciences, University of Leicester. TGR is an NIHR Senior Investigator.

Handling editor: Jonathan Hardman

Appendix A. Supplementary data

The following is the Supplementary data to this article: