Dynamic Cerebral Autoregulation in Acute Intracerebral Hemorrhage
Background and Purpose—Cerebral autoregulation (CA) is not universally impaired in acute intracerebral hemorrhage (ICH); however, the dynamic components of CA are probably more vulnerable. This study, therefore, evaluates the time course of dynamic CA in acute ICH and its relationship to clinical outcome.
Methods—Twenty-six patients with ICH were studied on days 1, 3, and 5 after ictus. Dynamic CA was measured from spontaneous fluctuations in blood pressure and middle cerebral artery flow velocity by transfer function phase (reflecting rapidity of CA) and gain (reflecting damping characteristics of CA) in the low frequency range. Results were compared with those from 55 controls and related with clinical factors and 90-day outcome (modified Rankin scale).
Results—Phase did not fluctuate significantly over time, nor did it differ between sides or differ from controls. Gain was always higher in patients than in controls but showed no significant association with outcome or other clinical factors. At day 1, poorer ipsilateral phase was associated with lower blood pressure and higher ICH volume. Poorer phase always coincided with lower Glasgow Coma Scale values. Poorer ipsilateral phase on day 5 was related with poorer clinical outcome according to multivariate analysis (P=0.013).
Conclusions—Dynamic temporal characteristics of CA (phase) are not generally altered in acute ICH. Poorer individual phase values are, however, associated with larger ICH volume, lower blood pressure, and worsened outcome. Dampening characteristics of CA (gain) are generally impaired in acute ICH but not related to clinical factors or outcome.
Optimal blood pressure treatment of patients with intracerebral hemorrhage (ICH) is still under debate. Antihypertensive treatment prevents secondary enlargement of ICH,1 but large reductions in blood pressure can also cause hypoperfusion and worsen long-term outcome. This may be a reason for the missing reduction in poor outcome with early blood pressure reduction, despite the prevention of secondary ICH enlargement.
Intact cerebral autoregulation protects the brain from hypoperfusion and hyperperfusion when cerebral perfusion pressure (CPP) changes. Previous studies on ICH have mainly applied the concept of static autoregulation,2 whereby steady-state correlations between cerebral blood flow and arterial blood pressure (ABP) are assessed during artificial, gradual changes of ABP. Examining static autoregulation means quantifying the completed response of the regulatory system to an artificial stimulus. The examination of autoregulation with transcranial Doppler in ICH showed that indices reflecting static and dynamic characteristics of autoregulation (index Mx) are not generally impaired in ICH. However, patients with comparably poor autoregulation several days after ICH onset experienced a poorer clinical outcome.3
Dynamic autoregulation describes the rapid autoregulatory properties that compensate for sudden changes in CPP within a few seconds.4 Such characteristics may be more vulnerable than static autoregulatory mechanisms and might even represent a different regulatory entity. Dynamic autoregulation is often described by the transfer function analysis parameters: phase (a parameter reflecting the speed of the autoregulatory response) and gain (a parameter probably reflecting the damping characteristics of dynamic regulation).5 These parameters can be monitored via slow spontaneous oscillations in blood pressure ≈0.1 Hz, which represent the continuously repeating upward and downward stimuli for the dynamic autoregulation mechanism. Recently, a study showed that patients with ICH had significant higher gains than healthy controls a few days after onset,6 whereas phase shift was unaltered. The time course of these dynamic autoregulation parameters during the acute stage of ICH and their relation to clinical factors and outcome after ICH is not known.
This study, therefore, uses serial measurements to investigate: (1) whether dynamic autoregulation (transfer function phase and gain) is impaired during the acute stage of ICH; and (2) how it associates with clinical characteristics and outcome.
Subjects and Methods
Patient and Controls
This prospective study comprised 26 nonconsecutive patients with spontaneous ICH who were admitted to the neurocritical care unit of the Neurocenter, University Hospital Freiburg, during a period of 15 months. Results of other, more static autoregulation characteristics (index Mx) from this study cohort have been reported previously.3 The ICH in the patients studied was confirmed by computed tomography or MRI. The study was approved by the local ethics committee, and written informed consent was obtained either from the patients or their relatives.
Exclusion criteria included severe stenosis of either the common or the internal carotid artery (≥70%) on ultrasound or MRI-/computed tomography-angiography and the absence of a sufficient bilateral transtemporal bone window for transcranial Doppler sonography. Patients undergoing palliative care in accordance with their own declared or presumed wishes were not included for ethical reasons.
A control group of 55 age- and sex-matched subjects (64±8 years of age, 44 men) was randomly chosen from our data pool of healthy volunteers who had neither a history of cardiopulmonary nor any other vascular disease or carotid stenosis on ultrasound.
All patients were continuously monitored, including blood pressure, heart rate, oxygen saturation, and temperature, on a neuro-intensive care unit. Blood glucose level and neurological status were verified at regular intervals. Blood pressure was adjusted according to ICH management guidelines7 by independent physicians who were not aware of the autoregulation data acquired by our study.
During 30 of the 78 recordings, different vasodilating drugs were used either alone or in combination to reduce blood pressure (dihydralazine, n=9; urapidile, n=19; clonidine, n=7). Sedation and ventilation using propofol (doses, 40–60 mg/h), midazolam (4.8–19.2 mg/h), or fentanyl (0.025–0.05 mg/h) were performed in 4 patients. Eight patients with ventricular hemorrhage or hydrocephalus received ventricular drainage, and 8 patients with large hematoma underwent frameless stereotactic puncture and drainage. In 8 cases of large hematoma or intraventricular bleeding, urokinase was applied for intraventricular or direct intraclot thrombolysis. None of the patients underwent craniectomia. Furthermore, all subjects received routine treatment with various combinations of antihypertensive, antidiabetic, antithrombotic, lipid-lowering, and antibiotic agents.
The clinical history of all patients, including previous reports and laboratory results, was scrutinized for the presence of cerebrovascular risk factors, such as hypertension, diabetes mellitus, hyperlipidemia, or smoking. Assessment of the Glasgow Coma Scale (GCS) was performed at each study point in all patients not receiving sedative drugs. A blinded interviewer assessed the modified Rankin Scale 90 days after ictus with patients or their relatives as a parameter of midterm outcome via telephone.8
Patients underwent initial computed tomography or MRI at an average of 8±10 hours after ictus. In case of a regularly shaped hematoma, ICH volume was estimated by dividing the product of length, width, and depth by 2; for irregularly shaped bleedings, the diameter product was divided by 3.9 These measurements were performed by a blinded examiner. On day 4 or 5, 10 of 26 patients received follow-up neuroimaging. These data were not used because of a potential selection bias.
Assessment of Cerebral Hemodynamics and Autoregulation
Measurements were performed on the first, third, and fifth day (ie, 12–24, 48–72, and 96–120 hours) after onset of symptoms. We used 2-MHz transducers (DWL-Multidop-X, Germany) for transcranial Doppler recording of cerebral blood flow velocity (CBFV) in each middle cerebral artery. A finger plethysmograph (Finapres, Ohmeda) permanently measured ABP. Mean absolute blood pressure values were assessed at the beginning of each measurement by oscillometric and direct invasive measurements. End-tidal CO2 partial pressure of nasal expiration was detected by infrared capnometry (Normocap Datex, Finland). After stabilization of hemodynamic parameters, spontaneous fluctuations were recorded during a period of 8 to 10 minutes. As described previously, we analyzed ABP and ipsilateral CBFV waveforms for noninvasive estimation of intracranial pressure10; noninvasive estimation of intracranial pressure was subtracted from ABP to obtain mean noninvasive CPP. Pulsatility index was calculated as the difference between systolic and diastolic flow velocity divided by mean flow velocity.
Transfer function analysis was used to determine the dynamic autoregulatory parameters, phase and gain. As described previously, the power spectra of ABP and CBFV and their cross-spectrum were estimated by transforming the time series of ABP and CBFV with discrete Fourier transformation to the frequency domain. From the cross-spectrum, the phase and gain spectra were derived. Phase and gain were determined for both middle cerebral artery sides from the low frequency range (0.06–0.12 Hz) by averaging the values at all frequency bins of significant (ie, >0.49) coherence.11
The Shapiro–Wilk test was used to test for normal distribution of continuous variables. Comparisons of phase, gain, and pulsatility index were performed using Mann–Whitney U tests between control subjects and patients. The Holm–Bonferroni method was applied to counteract the problem of multiple comparisons. The time course of clinical and hemodynamic parameters within the patient group was examined by Friedman test for non-normally distributed data. For normally distributed data, repeated-measures ANOVA was used. If this test showed significant changes, an additional Student t test for paired samples was applied. Explorative linear regression was used to assess the correlation of absolute phase and gain values (and their changes between days) with independent factors including initial ICH volume and current GCS, noninvasive CPP, and end-tidal CO2 partial pressure or ABP. Mann–Whitney U tests were used to analyze phase and gain in respect of their correlation to discrete factors as sex, history of hypertension, diabetes mellitus, the existence of ventricular hemorrhage and hemorrhage location. Univariate linear regression was applied to assess how much of the variation in clinical outcome (modified Rankin Scale at day 90) can be explained by the autoregulation parameters: phase, gain, and other factors. Multivariate linear regression analyses were later applied to assess the association of phase, gain, and clinical outcome controlling for other cerebral hemodynamic data such as CBFV in the middle cerebral artery, noninvasive CPP, and clinical factors such as age, sex, GCS (admission or day of autoregulation measurement), ICH volume, and presence of ventricular hemorrhage. A P value of <0.05 was considered statistically significant.
Table 1 shows the clinical characteristics of included patients. Overall, 11 of 78 scheduled measurements could not be analyzed because of signal artifacts, low coherence between signals, patient death, or referral to another hospital. Four examinations could only be analyzed on 1 middle cerebral artery side.
Dynamic Autoregulation in Acute ICH Versus Controls
At a group level, phase did not differ between patients (ipsilateral and contralateral sides) and controls across all study points (Table 2). Gain was significantly higher compared with controls on ipsilateral and contralateral sides at each study point (Figure 1). There were no appreciable intraindividual changes of phase over the first days after ICH. Gain decreased by trend within the first 5 days on the affected side (not statistically significant).
Clinical Factors and Dynamic Autoregulation in Acute ICH
A lower mean ABP was significantly associated with lower ipsilateral phase (indicating poorer autoregulation) on the initial examination but not on days 3 and 5. A similar association was also observed for ipsilateral phase values and noninvasive CPP at day 1. Interestingly, a positive correlation between phase and blood pressure was also found in the control group. Furthermore, a higher ICH volume was associated with lower ipsilateral phase on day 1. There was also a significant positive association between ipsilateral phase and GCS at all time points measured (Table 3).
We found no significant relationship between phase and each of the following parameters: age, sex, diabetes mellitus, hemorrhage location (deep versus lobar), presence of ventricular hemorrhage, body temperature, end-tidal CO2 partial pressure, and pulsatility index of CBFV. Changes in individual phase values between days (ie, trend over the first days) were also not associated with any clinical or hemodynamic factors. Gain showed no specific association with clinical factors apart from a weak correlation between ipsilateral gain and age on the first day after onset (P=0.047; coefficient=0.024; adjusted R2=0.124). Intraindividual changes in gain over the first days were not correlated with other hemodynamic or clinical factors. Administration of various antihypertensive drugs or statins did not correlate with phase and gain results. Hematocrit slightly decreased from days 1 to 5, probably because of infusion therapy.
Dynamic Autoregulation and Clinical Outcome
Univariate analysis revealed that on day 5, a correlation existed between low phase (ipsilateral and contralateral sides) and poor clinical outcome (Table 3). Gain was not associated with 90-day outcome. Of note, ventricular hemorrhage was one of the strongest independent predictors for poor clinical outcome. Furthermore, the daily GCS, the initial hemorrhage volume, and the initial body temperature each correlated significantly with outcome. In a multivariate analysis, ipsilateral phase on day 5 was a significant independent predictor for clinical outcome (Table 4, top). When GCS on day 5 was considered instead of admission GCS, ventricular hemorrhage was the only remaining significant independent predictor (Table 4, bottom). This was caused by the close correlation between phase and GCS on day 5.
The present study shows that dynamic dampening aspects of cerebral blood flow (gain) are generally impaired in ICH patients but are not related to any clinical factor or clinical outcome. In contrast, the rapidity of the dynamic autoregulatory process (phase) is poorer in patients with a large ICH and low GCS on day 1. Furthermore, when measured on day 5, phase associates with poor clinical outcome.
Autoregulation Parameter Phase in Acute ICH
Although phase did not differ from controls on a group level, individual patients with lower phase on ipsilateral sides had a lower GCS on each examination. Dynamic autoregulatory dysfunction might have led to sporadic cerebral perfusion and, in turn, an increase in brain damage, thus leading to a poorer clinical status. However, there may also be a common underlying mechanism that successively corrupts cerebral autoregulation and higher cerebral functions, namely, a greater mass effect with higher ICH volume that ultimately leads to local reductions in CPP.3 In fact, we also observed that a low phase on affected sides significantly associated with higher ICH volumes and lower CPP. Of interest, however, phase on day 5 after ICH onset was still a significant predictor of clinical outcome independently of other hemodynamic factors, ICH volume, ventricular hemorrhage, and admission GCS. Phase on day 5 was, however, not an independent predictor of clinical outcome when individual GCS values from day 5 were included; this was most likely because of the close correlation between autoregulation and GCS values already during the early course of ICH.
Taken together, we primarily assume that a large ICH volume leads to reductions in CPP with consecutive impairment of dynamic autoregulation and simultaneously causes poorer GCS values and poorer outcome. Impairment of phase might also independently contribute to poorer GCS and poorer outcome (or even early enlargement of ICH volume).
Another hypothesis for impaired autoregulation after ICH is an inflammatory reaction with excessive production of reactive oxygen species,12 which could directly impair myogenic activity (similar to ischemic stroke).13 Regarding the more static autoregulatory index Mx, we previously described a secondary decline in patients with significant impact on clinical outcome in the same patient group.3 In the present analysis of phase, such a relationship could not be observed. Within the limits of the small sample size, it seems that the dynamic characteristics of phase are less affected over time during acute ICH.
Relationship Between the Autoregulation Parameter Phase and Blood Pressure in Acute ICH
There was a significant association between lower ABP (and lower CPP) and lower phase values on ipsilateral sides. This finding suggests that lower ABP might indeed have a detrimental effect on dynamic autoregulatory ability in acute ICH. Of interest, we also observed a less steep correlation between absolute ABP values and phase on contralateral sides, as well as in healthy controls. This relationship has not been analyzed in detail previously. In an animal model, phase shift clearly decreases when, but not before, a lower limit of static autoregulation is reached.14 The current sample size is too small to establish whether the observed strong relationship between ABP and phase on ipsilateral sides in patients could be better represented by a nonlinear function resembling a lower limit of autoregulation. Looking at individual data, however, there is a clear trend toward lower phase values starting from mean ABP values ≈100 mm Hg (Figure 2). A previous study that challenged the autoregulatory plateau in acute ICH found autoregulation in both hemispheres to be globally impaired, mainly in patients with a mean ABP reduction of >20% (without providing absolute ABP values).15 Others found no changes in cerebral blood flow when reducing mean ABP by ≤15% (down to 90–133 mm Hg).2
With regard to the clinical management of acute ICH patients, the consequence of the observed association of ABP and dynamic autoregulation should be to maintain ABP at a level at which phase shift is still normal. Whether an optimal ABP for phase shift also exists (according to optimal CPP calculated with the correlation index, Mx, or pressure reactivity index, PRx)16,17 cannot be answered by this study.
Autoregulation Parameter Gain in Acute ICH
In contrast to phase, the dampening parameter gain was significantly higher in patients from the first day onward. These results are supported by a previous study that showed a significantly higher gain, both in low and high frequency ranges, in patients with ICH compared with healthy subjects.6 Moreover, gain was related to hematoma size, and the authors presumed a causal link among hematoma size, intracranial pressure, and gain. We found no such correlation in our cohort. Furthermore, in contrast to phase, gain showed neither an influence on modified Rankin Scale at day 90 nor a significant link to current clinical status. It is not obvious why patients with ICH had a higher average gain than healthy subjects. A previous study showed that there was no primary difference in terms of gain when ischemic stroke patients were compared with healthy controls.18 Because we do not know the preictal autoregulatory status of the studied patients, we cannot rule out that patients with ICH already had a pre-existing higher gain. Nearly all of the studied patients had hypertension. Hypertension, however, may only lead to clearly elevated gain values in case of malignant hypertension.19 A higher gain, that is, impaired dampening of ABP, could increase pulsatile transmural shear forces, damage arteries, and thereby promote a rhexis mechanism. In contrast, a global increase in gain just after the onset of ICH might have occurred via the disturbance of globally effective systems, such as the autonomic nervous system.
Although this prospective study recruited patients with ICH from a large tertiary stroke center, it should be noted that their inclusion into the study was nonconsecutive. In particular, patients receiving palliative care were not included for ethical reasons. This may explain the overall low rate of fatal outcome (8%). The exclusion of patients with a presumably high chance of mortality may have biased the study.
All patients received routine neurocritical care, including various combinations of antihypertensive, potentially vasoreactive drugs. Under these conditions, it is difficult to rule out an influence of medication on our autoregulation results. Although none of our patients received nimodipine, 10 patients received Ca2+-channel blocker (usually amlodipine). Amlodipine intake was, however, not associated with a change in autoregulatory parameters in our study. Furthermore, we could not find an influence of other antihypertensive drugs on phase or gain. There is a positive effect of early statin treatment in subarachnoid hemorrhage by improving autoregulation and ameliorating cerebral vasospasms.20 In our study, only 3 patients received statin treatment, and an impact on phase and gain could not be established.
In acute ICH, the rapidity of ipsilateral autoregulatory action (phase) is poorer in patients with low ABP and a large ICH volume. Poorer phase values also associate with poorer clinical status and with a worsened clinical outcome. Dynamic dampening characteristics of autoregulation (gain) are clearly disturbed in patients with ICH from the first day onward, raising the question of whether this is a reason for, or the result of, ICH. The clinical consequences derived from the present study are that low ABP in ICH might impair the rapid autoregulatory regulation. Blood pressure reductions in acute ICH should thus be viewed with caution. Future studies should attempt to measure phase values continuously to establish a causal link among individual blood pressure reductions, autoregulation, and clinical status. Perhaps blood pressure management could also be adapted to phase values, as has been suggested for other more static autoregulatory parameters.
- Received April 25, 2013.
- Revision received July 3, 2013.
- Accepted July 15, 2013.
- © 2013 American Heart Association, Inc.
- Zhang R,
- Zuckerman JH,
- Iwasaki K,
- Wilson TE,
- Crandall CG,
- Levine BD
- Broderick J,
- Connolly S,
- Feldmann E,
- Hanley D,
- Kase C,
- Krieger D,
- et al
- Merino JG,
- Lattimore SU,
- Warach S
- Huttner HB,
- Steiner T,
- Hartmann M,
- Köhrmann M,
- Juettler E,
- Mueller S,
- et al
- Schmidt B,
- Czosnyka M,
- Raabe A,
- Yahya H,
- Schwarze JJ,
- Sackerer D,
- et al
- Cipolla MJ,
- Curry AB
- Brady KM,
- Easley RB,
- Kibler K,
- Kaczka DW,
- Andropoulos D,
- Fraser CD III.,
- et al
- Diedler J,
- Sykora M,
- Rupp A,
- Poli S,
- Karpel-Massler G,
- Sakowitz O,
- et al
- Reinhard M,
- Roth M,
- Guschlbauer B,
- Harloff A,
- Timmer J,
- Czosnyka M,
- et al
- Immink RV,
- van den Born BJ,
- van Montfrans GA,
- Koopmans RP,
- Karemaker JM,
- van Lieshout JJ
- Tseng MY,
- Czosnyka M,
- Richards H,
- Pickard JD,
- Kirkpatrick PJ