Article Text

Download PDFPDF

Impaired cerebral autoregulation: measurement and application to stroke
  1. Li Xiong1,
  2. Xiuyun Liu2,
  3. Ty Shang3,
  4. Peter Smielewski2,
  5. Joseph Donnelly2,
  6. Zhen-ni Guo4,
  7. Yi Yang5,
  8. Thomas Leung1,
  9. Marek Czosnyka2,
  10. Rong Zhang3,
  11. Jia Liu5,
  12. Ka Sing Wong1
  1. 1 Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong, China
  2. 2 Department of Clinical Neurosciences, Brain Physics Laboratory, Division of Neurosurgery, University of Cambridge, Cambridge, UK
  3. 3 Department of Neurology and Neurotherapeutics, University of Texas Southwestern Medical Center, Dallas, Texas, USA
  4. 4 Department of Neurology, Neuroscience Center, The First Norman Bethune Hospital of Jilin University, Changchun, China
  5. 5 Chinese Academy of Sciences, Shenzhen Institutes of Advanced Technology, Shenzhen, China
  1. Correspondence to Jia Liu, Chinese Academy of Sciences, Shenzhen Institutes of Advanced Technology; jia.liu{at} and Dr Ka Sing Wong, Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong, China; ks-wong{at}


Cerebral autoregulation (CA) is a protective mechanism that maintains cerebral blood flow at a relatively constant level despite fluctuations of cerebral perfusion pressure or arterial blood pressure. It is a universal physiological mechanism that may involve myogenic, neural control as well as metabolic regulations of cerebral vasculature in response to changes in pressure or cerebral blood flow. Traditionally, CA has been represented by a sigmoid curve with a wide plateau between about 50 mm Hg and 170 mm Hg of steady-state changes in mean arterial pressure, defined as static CA. With the advent of transcranial Doppler, measurement of cerebral blood flow in response to transient changes in arterial pressure has been used to assess dynamic CA. However, a gold standard for measuring CA is not currently available. Stroke has been the leading cause of long-term adult disability throughout the world. A better understanding of CA and its response to pathological derangements can help assess the severity of stroke, guide management decisions, assess response to interventions and provide prognostic information. The objective of this review is to provide a comprehensive insight about physiology of autoregulation, measurement methodologies and clinical applications in stroke to help build a consensus for what should be included in an internationally agreed protocol for CA testing and monitoring, and to promote its translation into clinical bedside practice for stroke management.

  • Cerebral autoregulation
  • ischaemic stroke
  • intracerebral haemorrhage
  • cerebral blood flow
  • transcranial Doppler

Statistics from


The ability of the brain to regulate its own blood supply is termed cerebral autoregulation (CA), which maintains an adequate and stable cerebral blood flow (CBF) despite changes in cerebral perfusion pressure (CPP).1 CA is a universal physiological mechanism, depicted by the Lassen curve (Figure 1), with arterial blood pressure (ABP) or CPP along the x-axis and CBF along the y-axis. A decrease in CBF to ischaemic levels may be tolerated for a short period, depending on the severity of the ischaemia. If CBF ceases completely, brain cell death occurs within minutes. A variety of conditions are encountered clinically, such as in stroke or traumatic brain injury, an actual or potential alteration in CBF puts the brain at risk for ischaemia and infarction. Stroke has been the leading cause of long-term adult disability throughout the world.2 Therefore, a better understanding of CA and its response to pathological derangements can help assess the severity of stroke, guide management decisions, assess response to interventions and provide prognostic information.

Figure 1

Lassen autoregulatory curve. It has been obtained from 3.5 days of recording of thermodilution CBF and radial artery blood pressure in a patient after poor-grade haemorrhagic stroke. Natural variations of ABP from 60 mm Hg to 120 mm Hg provoked changes in CBF, which enabled plotting of curve and visualising the upper (ULA) and lower (LLA) limit of autoregulation. ABP, arterial blood pressure; CBF, cerebral blood flow.

However, a gold standard for measuring CA is not currently available. Research is also ongoing to assess and validate the impact of CA measures on clinical outcomes under a variety of conditions.3–6 In this paper, we provide a comprehensive and educational review from physiology of autoregulation, measurement methodologies and clinical applications in stroke to help build a consensus for what should be included in an internationally agreed protocol for CA testing and monitoring, and to promote its translation into clinical bedside practice for stroke management. It consists of two parts. The first part contains technical description of methodologies. The second shows the most important evidenced clinical applications of CA testing in stroke.

Basic physiology of CA

The underlying physiological mechanisms of CA may involve myogenic, neural control as well as metabolic regulations in response to changes in CPP or CBF.1 However, relative contributions of each of these mechanisms to CA under specific physiological or pathophysiological conditions are complex and poorly understood. Furthermore, with poorly delineated border between the conductive and the resistive part of the cerebral arterial tree, CA may involve both large and small cerebral arteries and arterioles, and it has been shown that large extracranial arteries and intracranial pial blood vessels may contribute about 50% of cerebrovascular resistance, and the rest are from the penetrating parenchymal arteries and arterioles.7 8 In addition, the arteries and arterioles embedded in the brain parenchyma may possess unique and distinct properties from that of pial arteries in regulating cerebrovascular resistance. Finally, CA is likely to be heterogeneous segmentally, regionally and temporally in the regulation of CBF to meet brain metabolic demand.

It has been well recognised that neurogenic and myogenic mechanisms play important roles in CA. Neurogenic regulation includes sympathetic and cholinergic mechanisms.9 Using pharmacological blockage methods, studies have shown that sympathetic, cholinergic and myogenic mechanisms in pial arteries accounted for about 62% of changes in total cerebrovascular resistance in response to moderate changes in arterial pressure.10 However, the relative contribution from neurogenic and myogenic mechanism to CA may vary in different regions of autoregulation curve. The neurogenic mechanism may be responsible for CA during moderate changes in arterial pressure, whereas myogenic mechanism may be predominant during large changes in arterial pressure, which may lead to hypoperfusion or hyperperfusion injury out of the range of effective CA.10

There is also segmental heterogeneity in neurogenic regulation. One of the manifestations of the complexity in CA is that the neurogenic reactivity varies as cerebral blood vessel branches down from the pial arteries into parenchymal arteries and arterioles.9 11 Both the pial and parenchymal arteries and arterioles are likely to contribute equally to the total cerebrovascular resistance and thus CA. However, differences exist between the pial and parenchymal arteries and arterioles in terms of anatomy, physiology and neurogenic reactivity. The pial artery receives perivascular innervation from the peripheral autonomic nervous system originated from the superior cervical ganglion, sphenopalatine ganglion, otic ganglion and trigeminal ganglion (extrinsic innervation).9 11 On the other hand, the parenchymal arteries and arterioles are innervated mainly by the intrinsic nerves originated from the subcortical neurons, such as those located in the locus coeruleus, raphe nucleus, basal forebrain or local cortical interneurons, that project to the perivascular space surrounding the parenchymal arteries and arterioles (intrinsic innervation).9 11 Parenthetically, it should be highlighted that parenchymal arterioles are a crucial part of neurovascular unit that control local perfusion to meet brain metabolic demands, linking CA with the brain neuronal activity.12 The heterogeneity in neurogenic reactivity in the pial and parenchymal artery is manifested also by the different expression levels of neurotransmitter receptors and different functional responses to similar stimuli. For example, the α-adrenoceptor reactivity is absent in the parenchymal artery due to a shift from α-adrenoceptor to β-adrenoceptor in the parenchymal artery.13 Similar heterogeneity has been shown with serotonin receptor.13 As the consequence of these changes, the marked vasoconstrictive effects of the pial arteries to serotonin and norepinephrine are absent in the parenchymal arteries or arterioles or even cause parenchymal artery dilation.14 This segmental heterogeneity in neurogenic regulation may provide the brain with the capability to efficiently change CBF locally to meet the metabolic demand.

CA also possesses segmental heterogeneity in myogenic regulation in the pial and parenchymal arteries and arterioles. The smooth muscle cells (SMC) in the cerebral resistance vessels are responsible for the myogenic tone and regulation. For example, the important role of SMC in myogenic regulation is seen in cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) disease,15 showing a degree of diffuse SMC degeneration in small cerebral arteries. In animal model of and human with CADASIL, myogenic regulation is impaired, which leads to impaired CA.16 17 Notably, the myogenic tone and regulation are not homogeneous along the resistance vessels. An interesting characteristic of cerebral parenchymal artery is that the parenchymal artery possesses greater basal tone compared with the pial arteries.14 Under this basal tone, pressure-induced myogenic reactivity in the parenchymal artery is reduced compared with that observed in the pial arteries.14 This mechanism may buffer effects of upstream rapid changes in ABP on CBF and attenuate the transmission of pulsatile mechanical stress into the brain microcirculation.

CA was also demonstrated to have regional heterogeneity. Differences in regional sympathetic innervation and CA measured in the anterior and posterior circulation were observed. Specifically, sympathetic nerve fibre endings were denser in the anterior circulation arteries than those in the posterior circulation arteries originated from the vertebrobasilar arteries.18 In addition, CA was found to be more effective in the brainstem than in the anterior circulation during severe hypertension.19 For example, CBF was increased markedly in the anterior circulation in contrast to an only modest increase in the brainstem during severe hypertension. This has been attributed to the observation that the resistance of small vessels in the anterior circulation was decreased, whereas that in the brainstem was increased during severe hypertension, indicating regional CA heterogeneity.19 It is interesting to note that this regional heterogeneity in sympathetic neurogenic mechanism may play a role in developing posterior reversible encephalopathy syndrome (PRES).20 PRES is characterised radiographically by transient bilateral subcortical vasogenic oedema mostly common in the posterior circulation territory.

Methodologies of the assessment of CA

Table 1 contains a brief description of the methods used in patients with stroke and patients with carotid stenotic disease. Selected methods of assessment of CA listed in table 1 are illustrated in Figure 2. According to the definition of autoregulation, the principle of assessing CA is to examine changes in CBF in response to changes in ABP assuming other confounding factors such as arterial CO2 or PaO2 (pCO2) remains unchanged. Intuitively, changes of ABP ‘must’ be induced so as to examine the corresponding dynamics of CBF and then determine the status of autoregulation. In the early study, the pressure change was often induced by vasoactive drug infusion, for example, phenylephrine.21 At different pressure levels, the corresponding blood flow was measured to produce a CA curve showing that within a certain pressure range (normally 50–170 mm Hg in a healthy subject) CBF may remain relatively constant. This pressure–flow relationship is referred to as ‘static’ autoregulation.1

Figure 2

(A–C)Example of evaluation of static rate of autoregulation. Arterial blood pressure has been increased from 95 mm Hg to 114 mm Hg with vasopressor. Changes in CBF were recorded using transcranial Doppler ultrasonography (FV). CBF, cerebral blood flow.

Table 1

Methods for CA assessment and monitoring in patients with stroke

With the advent of continuous measurements of blood flow (ie, transcranial Doppler for CBF velocity) and blood pressure (vascular unloading techniques in the fingers), allowing for recordings of instantaneous response of CBF velocity to changes of ABP, the concept of ‘dynamic’ autoregulation was coined.22 A number of approaches for manipulation of ABP have been proposed to investigate dynamic autoregulation, including thigh-cuff release, body tilt, handgrip, lower-body negative pressure, squat-stand manoeuvres, paced breathing, transient compression of carotid artery, and others.23 However, these approaches of ABP might not be acceptable in the assessment of CA in patients with stroke. For example, a thigh-cuff test can result in a sudden drop of blood pressure up to 30 mm Hg.24 This may potentially impose risks of secondary injury of the brain for patients with ischaemic stroke, especially with autoregulation compromised. Moreover, alterations in ABP using these methods may induce responses from other physiological systems (eg, sympathetic activation or vagal withdrawal) or changes in arterial pCO2, which can profoundly affect blood flow independent of changes in ABP. It is therefore more desirable, for patients with stroke, to have alternative methods that avoid these problems.23 Recently, a series of studies reported external counterpulsation (ECP) may benefit patients with acute ischaemic stroke, which is a non-invasive method to augment CBF by elevation of blood pressure.25 26 Cerebral augmentation index, the increase in the percentage of the middle cerebral artery (MCA) mean flow velocity during ECP compared with baseline using transcranial Doppler monitoring method, is calculated to evaluate the augmentation effects of ECP. The relationship between the cerebral augmentation index and blood pressure changes induced by ECP may reflect the status of dynamic CA in ischaemic stroke.27 However, this speculation should be further validated as ECP has never been previously used for assessment of CA.

With a number of mathematical methods proposed, induced changes of ABP may not necessarily be a ‘must’. Instead, the use of spontaneous fluctuations of blood pressure for dynamic CA assessment has been increasingly employed.23 Although it is clinically appealing to record spontaneous changes in ABP and CBF velocity, assessment of CA at rest may not reveal full capacity of autoregulation because of relatively low variations in spontaneous ABP. In addition to the transcranial Doppler methods, other neuroimaging techniques, such as MRI or positron emission tomography (PET), that measure  regional CBF or blood oxygen level-dependent signal have also been explored in assessing autoregulation. Understanding the advantages as well as the limitations of these methods will be helpful for clinicians to choose appropriate methods for their studies of CA in stroke diagnosis or management.

Most studies of CA in stroke relied on linear (cross-spectral or correlation-based) methods to assess the integrity of autoregulation. If the pressure–flow relation displays a low coherence (ie, if it lacks linear dependence), the relation between pressure and CBF cannot be quantified via linear analyses. In other words, the method itself creates uncertainty in its linear estimates. This is a problem of the analytical technique, not the input signal (spontaneous oscillations vs induced blood pressure manipulation). Limitation can be overcome by using a non-linear approach.

Autoregulation index

Autoregulation index (ARI) is a gauging system to quantify the status of CA.22 The index consists of 10 levels (0–9) of autoregulatory status from damaged to intact. A second-order differential equation simulates 10 possible CBF velocity responses to an ideal step change of ABP by giving 10 sets of predefined parameters including damping factor, time constant and gain parameter. By comparing the recorded CBF velocity with the 10 simulated CBF velocities, the index number is determined by finding the best match. This grading method is simple and easy to implement and interpret. However, this method presumes a predefined linear and stationary relationship between ABP and CBF velocity, which is generally not true. The actual responses of CBF velocity are more diverse and dispersed. Therefore, the 10 predefined cases are not enough to explain all situations, resulting in inappropriate assignments of the index number in some cases and have large individual variability.

Transfer function analysis

Transfer function analysis (TFA) is based on a linear and stationary model with no preconstraint levels on the status of autoregulation. The principle is that autoregulation is supposed to attenuate the influence of ABP on CBF velocity by preventing a direct transfer of the waveform at a low frequency range (normally <0.2 Hz). Two parameters, gain and phase-shift, can be derived from the transfer function at each frequency. Gain quantifies the compression of the relative changes in amplitude of CBF velocity to ABP, whereas phase-shift indicates the time lag (given a specific frequency) between ABP and CBF velocity. Gain is a continuous value. For example, a value of 0.5 suggests that 50% of the relative amplitude of CBF velocity is attenuated with respect to a unit of changes in ABP. Phase-shift is denoted in degree or radian. The larger the phase of CBF velocity is shifted from ABP, the better the autoregulation is taking effect.23 It is important to note that TFA can only explain linear relationship between ABP and CBF velocity. Coherence is thus normally accompanied to test the linearity of the two variables. A value above 0.5 is considered acceptable if TFA is considered to apply, as it can explain at least 50% of the linear relationship. A series of studies of the performance and implementation of TFA by multicentres was recently published by the Society of Cerebral Autoregulation Research Network.23 Please refer to these works for more comprehensive details concerning the method of TFA.

Time correlation

Time correlation method allows continuous estimation of CA through a moving linear correlation between slow waves of ABP and ICP (PRx), or CBF velocity and ABP (Mxa) or near-infrared spectroscopy (NIRS)-derived tissue oxygenation index (TOI) and ABP (TOxa).28 These parameters can reflect the compliant ability of cerebral vessels in response to changes in ABP or CPP. The parameters are always calculated as the correlation coefficient between 30 consecutive, 10 s averaged values of input (ABP, CBF velocity) and corresponding output signals (ICP, ABP or TOI). Averages over 10 s were used to suppress the influence of the pulse and respiratory frequency wave components. A positive correlation is indicative of passive cerebral vasculature and impaired autoregulation. Zero or negative correlation is indicative of reactive vasculature and intact autoregulation. They have been widely used for continuous monitoring CA in patients requiring neurointensive care predominantly after haemorrhagic stroke.29 30 One of the limitations of this method is that calculation of the correlation coefficient may be influenced by the time scales used. Another is that the method assumes intact autoregulation if there is lack of correlation. However, poor data quality also leads to poor correlation and may be misinterpreted as intact autoregulation.

Multimodal pressure flow (MMPF) or wavelet analysis

It is a model based on Hilbert Huang transform.31 The principle is to calculate the phase-shift between ABP and CBF velocity without assuming a linear relationship a priori. The ABP and CBF velocity are empirically decomposed into a series of modes from fast to slow waves, respectively. The mode that is considered the most relevant to autoregulation is then identified, and the instantaneous phase of ABP and CBF velocity is computed by Hilbert transform. The phase difference between the variables can then be calculated readily. A number of studies have shown that it is capable of dealing with non-linear relationship of ABP and CBF velocity. Wavelet analysis is a time frequency approach that can be used to quantify dynamic CA under non-stationary conditions. Both time-varying coherence and phase at different time scales between changes in ABP and CBF velocity or brain tissue oxygenation using NIRS can be obtained.32 However, the validity and effectiveness of these methods need to be further tested or replicated in multicentre studies.

Advanced mathematical models

There is increasing evidence showing that the variation of blood flow is a non-linear, non-stationary (changing over time) and multivariate (eg, the influence of CO2) process. A number of advanced mathematical models have therefore been proposed to deal with these characteristics. Several groups have attempted to apply the Laguerre-Volterra expansion of kernels to model the dynamics of CA with non-linearity considered.33 Additionally, multivariate models were designed to model the influence of covariates, for example, CO2, on CBF velocity, and more recently non-stationary property was investigated by using moving windows or adaptive methods.34 Analytical techniques to measure CA should include projection pursuit regression. This approach has been shown to predict delayed cerebral ischaemia after subarachnoid haemorrhage (SAH) on an individual patient basis.35 36Given more degrees of freedom, it is not surprising that these advanced models may significantly reduce estimation error of CBF velocity. For example, higher order models can always achieve better (at least the same) estimation than the ones with lower order. This does not necessarily mean that the complex models are always better than the simple ones, as the advanced models may be overfitting the data, which may result in a worse prediction. We therefore suggest that in the future study one should have a clear and proper justification before choosing one of these models. For example, a multivariate model can help in assessing autoregulation when significant CO2 reactivity also presents.34

Advanced neuroimaging techniques

Of the greater potential imaging techniques, MRI, PET and CT perfusion (CTP), appear suitable because of their availability in acute care situations, high-spatial resolution and excellent safety record. Through these techniques, CBF, blood volume, oxygen extraction fraction and oxygen consumption rate (CMRO2) can be measured in multiple regions. The relationship between changes in blood volume and changes in CBF velocity or changes in CMRO2 was analysed. Studies showed that the result of using advanced neuroimaging techniques to assess CA matched with traditional methods (such as transcranial Doppler).37 38 However, these methods in general are expensive and have poor temporal resolution (approximately in minutes) and can be challenging to be used in bedside studies. Advanced neuroimaging techniques may be useful and are available in acute care situations, but it is important to recognise that they may not be generally applicable and there are disadvantages in the assessment of a critically ill patient in this way.

Clinical applications of CA studies in stroke

The presence or absence of CA in acute stroke is critical for the maintenance of stable blood flow in the ischaemic penumbra and for avoidance of excessive hyperperfusion. A widely applicable method of measuring CA in patients with acute stroke is needed to allow detailed investigation of the relation between altered CA following stroke and clinical outcome, and may ultimately be relevant in the treatment of blood pressure in the acute period following stroke. Therefore, in this section, we identified and reviewed articles applying different methods of measuring CA in the stroke-related clinical studies during the past two decades. These original articles were searched by the keywords ‘cerebral autoregulation’ and ‘stroke’ in PubMed (provided by the National Center for Biotechnology Information, USA), with additional criteria of date from 1996 and studies of humans. We then chose the studies of stroke or cerebrovascular diseases with autoregulatory assessment. This results in more than 40 articles as listed from 950 searched items (table 2), which revealed that there is inconsistency in the application of autoregulatory methods, with altered CA over the infarcted ipsilateral, contralateral or bilateral side and its time course effects after stroke onset.

Table 2

Summary of studies using different methods to measure CA in patients  with stroke

Global or focal impairment of CA?

CA may become impaired after ischaemic stroke. Focal impairment of static autoregulation in the reperfused ischaemic area itself has been demonstrated.1 Moreover, some studies indicated a more global impairment of static autoregulation in the affected and also in the unaffected hemisphere,39 40 whereas another study found static autoregulation in the unaffected hemisphere to be generally preserved.41 Because assessment of static autoregulation requires considerable manipulation of ABP, it is not routinely applicable in acute stroke treatment. Therefore, the so-called dynamic CA approach has evolved. It has been postulated that dynamic CA may be more sensitive to cerebral haemodynamic impairment.22

Using the method of ARI estimated from thigh-cuff test or spontaneous transient pressor and depressor blood pressure stimuli, a global bihemispheric impairment of dynamic CA in acute stroke within 24–72 hours of symptom onset with preserved static autoregulation has been found,42–44 which is not related to stroke subtype classified as total/partial anterior circulation syndrome, lacunar syndrome and posterior circulation syndrome.42 44 Dawson et al also found dynamic CA, as assessed by ARI with thigh-cuff test, is globally impaired within 96 hours of ischaemic stroke onset and remains abnormal for at least 1–2 weeks post-ictus.45 However, using TFA approach to derive the autoregulatory parameters of gain and phase, Immink et al reported dynamic CA is impaired ipsilaterally in the MCA territory ischaemic stroke but bilaterally in lacunar ischaemic stroke.46 With the same measurement method, a recent study further demonstrated the similar findings that dynamic CA is impaired ipsilaterally in stroke of large artery atherosclerosis but bilaterally in stroke of small artery occlusion,47 which might be attributed to the varied pathological changes of cerebral blood vessels. Assessing dynamic CA on days 0–2, 3–6 and ≥7 days after acute large vessel ischaemic stroke in the MCA territory, one recent study showed that dynamic CA is impaired in the affected hemisphere throughout the first week, and then normalises by week 2.48

Changes in CA over time after stroke

Kwan et al assessed dynamic CA in acute MCA territory patients with ischaemic stroke at <7 days, 6 weeks and 3 months after stroke, and found that bilateral impairment of dynamic CA might improve up to 3 months after ischaemic stroke.49 However, using a different modified MMPF method, Hu et al reported that the impaired dynamic CA may persist up to 6 months after ischaemic stroke.50 In contrast to what were found in these studies, Reinhard et al found dynamic CA, as assessed by both time correlation and TFA methods, might not be disturbed in minor MCA stroke within 22 hours of symptom onset but slightly impaired at the subacute stage within 134 hours of ictus.51 In one study to investigate the effect of recombinant tissue plasminogen activator (rtPA) on dysautoregulation in 16 patients with acute ischaemic stroke based on both TFA and time correlation methods, dynamic CA is found to be increasingly impaired, mainly on the affected side, over the first 5 days of major ischaemic stroke after unsuccessful rtPA thrombolysis and bilaterally preserved in minor stroke after successful rtPA thrombolysis.52 These suggest that development of cerebral dysautoregulation may be particularly critical during the early stage of reperfusion in acute ischaemic stroke. This is probably also the most interesting time period with regard to functional brain reorganisation. Cerebral dysautoregulation may therefore play a significant role within the detrimental effect of reperfusion and hyperperfusion. As supposed for malignant MCA infarction,53 a vicious circle may start in the peri-infarct area by spreading local acidosis with consequent dysautoregulation, hypoperfusion or hyperperfusion, oedema, and further infarction. Studies with large sample size emphasising the role of reperfusion in the evolution of autoregulatory failure are needed.

CA and clinical outcome

There are two studies to investigate the relationship between dynamic CA and clinical outcome after ischaemic stroke using MMPF and combined time correlation and TFA method, respectively. One study indicated that better bilateral dynamic CA is associated with less atrophy and better long-term functional status in older adults with chronic ischaemic infarctions.3 The other showed that impairment of dynamic CA ipsilateral to acute ischaemic stroke is associated with larger infarction, and dysautoregulation tends to worsen and spread to the contralateral side over the first days poststroke and is associated with poor clinical outcome.4 However, these observations were obtained from relatively small sample size studies. It remains largely unclear whether impairment of dynamic CA is related to the long-term clinical outcome in stroke survivors. Multicentre studies with greater statistical powers are needed to address the prognostic significance of perturbations in CA, which may benefit patients with stroke in the future.

CA in intracerebral haemorrhage

So far, management of hypertension during acute intracerebral haemorrhage (ICH) is controversial. Haematoma expansion may be attenuated by acute blood pressure reduction, but concern persists that dynamic CA may be impaired after ICH, making perfusion of the brain passively dependent on blood pressure.6 Perihaematoma tissue is moderately hypoperfused and therefore may be vulnerable to blood pressure reduction. However, in contrast, a CTP study in 20 patients within 72 hours of ICH indicated that CBF remained stable after acute blood pressure reduction, suggesting some preservation of CA which was assessed by linear regression analysis between systolic blood pressure and relative CBF measured by CTP.54 Using the TFA method, 26 patients with ICH were studied on days 1, 3 and 5 after ictus to evaluate the time course of dynamic CA in acute ICH and its relationship with clinical outcome.5 Dynamic temporal characteristics of CA are not generally altered in acute ICH, but poorer individual phase values are associated with larger ICH volume, lower blood pressure and worsened outcome.5 Blood pressure reductions in acute ICH should thus be viewed with caution.

CA after SAH

In SAH studies using TFA or time correlation method, impaired dynamic CA is observed in the first 5 days post-SAH, before evidence of vasospasm can be found, is strongly predictive of delayed cerebral ischaemia29 55 and is independently associated with an unfavourable outcome after 6 months of SAH onset.56 Including autoregulation as part of the initial clinical and radiographic assessment may enhance our ability to identify patients at a high risk for developing secondary complications after SAH.

CA and blood pressure management after stroke

Considering the complexity of CA, it is not surprising that significant controversies exist in blood pressure management in patients suffering from acute stroke. It is believed that both very high and very low blood pressure after acute stroke are harmful and are associated with worsening outcome. The ideal blood pressure range after acute stroke is unknown. Current guideline recommended permissive hypertension after acute ischaemic stroke. In patients who do not receive thrombolysis, it is recommended that blood pressure treatment should be on hold unless systolic blood pressure is >220 mm Hg or diastolic blood pressure is >120 mm Hg within the initial 24 hours after stroke.57 In patients who receive intravenous thrombolysis, the systolic blood pressure should be below 185 mm Hg and the diastolic blood pressure should be below 110 mm Hg before initiating thrombolysis. After initiating thrombolysis, the systolic blood pressure should be below 180 mm Hg and the diastolic blood pressure should be below 105 mm Hg within 24 hours after intravenous thrombolysis.58 In patients with spontaneous ICH and elevated blood pressure, the 2010 American Heart Association (AHA) guideline recommends keeping systolic blood pressure below 160 mm Hg if no other contraindication.59 Overall, an optimal ABP range likely exists during acute stroke but probably depends on an individual CA variability, temporal and spatial heterogeneity of stroke pathophysiology, and stroke subtype. Unfortunately, such an ideal blood pressure range has not yet been determined based on randomised, controlled trials. Recently, several large randomised clinical trials were performed to address the controversies in acute blood pressure management after acute stroke, including ICH and ischaemic stroke. Most trials suggested that lowered blood pressure in acute stroke has acceptable safety profile but did not improve functional outcome. For example, INTERACT2 is the first large randomised trial of rapid blood pressure lowering in acute ICH, which suggests that aggressive lowering of blood pressure <140 mm Hg within 1 hour is safe in acute ICH within 6 hours from symptom onset.60 Accordingly, the 2015 AHA guideline states that for patients with ICH presenting with systolic blood pressure between 150 mm Hg and 220 mm Hg and without contraindication to acute blood pressure treatment, acute lowering of systolic blood pressure to 140 mm Hg is safe.61 However, harmful effects (renal adverse events) of aggressive blood pressure were reported in another trial (ATACH-2).62 The ongoing ENCHANTED trial will evaluate whether early intensive blood pressure lowering (systolic blood pressure target to 130–140 mm Hg) is superior to guideline-recommended systolic blood pressure target of <180 mm Hg) in patients with ischaemic stroke who receive intravenous thrombolysis in terms of efficacy and safety.63 It should be noted that CA study was not performed or even considered in these trials; thus, its potential effects on clinical outcome measures are not known. Furthermore, hypertension is one of the major risk factors for stroke. De novo hypertension is common in patients with acute stroke. It is unclear whether de novo hypertension should be treated or prestroke antihypertensive medications should be continued or stopped in acute stroke. The recent ENOS trial did not show any benefit to continue prestroke antihypertensive treatment in patients with acute stroke.64 It is possible that a subpopulation of patients with acute stroke with relative intact CA may benefit from aggressive blood pressure treatment to improve clinical outcome and decrease harmful side effects. It is also not known if similar to acute stroke, an ‘optimal ABP’ which assures best CA exists and can be assessed using continuous monitoring.30 This critical knowledge gap further supports the need to understand the complex mechanisms of CA on an individual basis and apply to patient care in an era of precision medicine.

Comparison of different methodologies

Several studies have investigated the comparisons among the current methods of measuring dynamic CA. Time correlation and TFA are comparable in assessing autoregulation in carotid artery occlusive disease.65 For patients with stroke, TFA needs no provocation and adverse patient effects are minimal,66 and time correlation provides a non-invasive and continuous assessment of autoregulation.67 A transfer function filter may reduce ARI variability in controls, resulting in more pronounced group differences between patients with stroke and healthy subjects.68 Moreover, MMPF might be better than ARI in terms of reproducibility estimated in controls, traumatic brain injury and patients with stroke.50 It should be borne in mind that the results from a single episode of induced hypotension, for example, a thigh-cuff based ARI, may represent distinct processes from spontaneous fluctuation of ABP, and their impairment and recovery after stroke may be dissociated. This issue is also addressed in our methodology section with more explanation.

Currently, there is no widely accepted ‘best choice’ for the autoregulatory assessment. The decision should be made according to the status of the targeted patients and the research questions to be answered. For example, for patients with stroke at acute stage, TFA might be a good option, as it requires minimal cooperation with no stimulus of blood pressure and the procedure of data recording can be performed at bedside; for patients with stroke at chronic stage, paced breathing or thigh-cuff tests might become acceptable to provide visual inspections of blood flow responses after passive ABP challenges and for patients in the intensive care unit, a long-term monitoring of CPP and CBF velocity is possible, and therefore time correlation method is more suitable to understand the temporal course of autoregulation, which can be helpful in predicting outcomes.

Clinical trials are needed. Although ‘autoregulation-oriented therapy’ has been addressed in traumatic brain injury (although there is still no evidence that it is able to improve outcome), in stroke there is very little room for manoeuvre. Statins proved to improve autoregulation and short-term outcome in haemorrhagic stroke in a phase II study, but a phase III multicentre trial did not show any benefit.


This review shows that TFA and time correlation are the most frequently used assessments across multicentres. These methods can differentiate patients with disturbed haemodynamics due to stroke or cerebrovascular diseases from healthy controls by the phase-shift derived from TFA and (or) correlation coefficient indices. It is generally accepted that CA is disturbed both in ischaemic and haemorrhagic stroke, and in ICH or severe cerebrovascular diseases. CA may change with pathological progression. The status of autoregulation at a certain temporal stage might be associated with the outcome.6 29 Further investigation on autoregulation follow-ups with large sample size is required to confirm this. Moreover, it shows that the individual status of autoregulation cannot always be determined by the current methods accurately.40 Although a few attempts have been made to improve the current assessments by including non-linear and multivariate properties in the mathematical models, these studies are still limited to single centres. Clinical study across multi-centres of advanced mathematical models and the neuroimaging techniques is desirable to improve assessment of individuals.



  • Contributors LX, XL and TS performed the literature search and wrote the manuscript. JL and KSW conceived and designed the review outlines. JL, MC and RZ helped revise the manuscript. PS and JD proposed professional advice on the methodologies of the assessment of cerebral autoregulation. ZG, YY and TL proposed professional advice on the clinical research related to cerebral autoregulation in ischaemic stroke.

  • Funding This work was supported by the Health and Medical Research Fund (HMRF, Project No 02130836) and the General Research Fund from Research Grants Council (GRF, Reference No. 14100215) in Hong Kong. 

  • Competing interests None declared.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.