cerebral autoregulation dynamics in humans -...

45

Cerebral Autoregulation Dynamics in HumansRune Aaslid, PhD, Kari-Fredrik Lindegaard, MD,

Wilhelm Sorteberg, MD, and Helge Nornes, MD

We studied the response of cerebral blood flow to acute step decreases in arterial blood pressurenoninvasively and nonpharmacologically in 10 normal volunteers during normocapnia, hypo-capnia, and hypercapnia. The step (approximately 20 mm Hg) was induced by rapidly deflatingthigh blood pressure cuffs following a 2-minute inflation above systolic blood pressure.Instantaneous arterial blood pressure was measured by a new servo-cuff method, and cerebralblood flow changes were assessed by transcranial Doppler recording of middle cerebral arteryblood flow velocity. In hypocapnia, full restoration of blood flow to the pretest level was seen asearly as 4.1 seconds after the step decrease in blood pressure, while the response was slower innormocapnia and hypercapnia. The time course of cerebrovascular resistance was calculatedfrom blood pressure and blood flow recordings, and rate of regulation was determined as thenormalized change in cerebrovascular resistance per second during 2.5 seconds just after thestep decrease in blood pressure. The reference for normalization was the calculated change incerebrovascular resistance that would have nullified the effects of the step decrease in arterialblood pressure on cerebral blood flow. The rate of regulation was 0.38, 0.20, and 0.11/sec inhypocapnia, normocapnia, and hypercapnia, respectively. There was a highly significantinverse relation between rate of regulation and PacOj (p<0.001), indicating that the responserate of cerebral autoregulation in awake normal humans is profoundly dependent on vasculartone. (Stroke 1989;20:45-52)

Cerebral autoregulation is a homeostatic mech-anism that minimizes deviations in cerebralblood flow (CBF) when cerebral perfusion

pressure (CPP) changes. Cerebral autoregulationacts through vasomotor effectors that control cere-brovascular resistance (CVR). Previous studies haveconvincingly documented the ability of this physio-logic system to maintain relatively constant CBFwhen CPP is within the range 50-170 mm Hg.1-6

Because this is a very fast-acting homeostaticmechanism,7 the method for measuring autoregu-latory responses should ideally have good timeresolution. Our knowledge about the dynamicresponse of cerebral autoregulation in humans islimited because most indicator methods permit sam-pling of regional CBF at intervals of only minutes.Animal experiments7-10 indicate that the main auto-regulatory response is produced within seconds.Measurement of dynamic response times andresponse rates to characterize the effectiveness ofautoregulation is of interest not only from a physi-

From the Institute of Applied Physiology and Medicine, Seat-tle, Washington (R.A.) and the Department of Neurosurgery,Rikshospitalet, The National Hospital, University of Oslo, Oslo,Norway (K-F.L., W.S., H.N.).

Address for correspondence: Rune Aaslid, PhD, Neurochimr-gie, Inselspital, CH-3010 Bern, Switzerland.

Received January 8, 1988; accepted July 6, 1988.

ologic point of view, but perhaps even more so forits clinical implications.

Autoregulation may or may not be impaired inpatients with significant disease of the cerebralarteries; moreover, autoregulatory capacity may bepartly or completely lost after stroke or subarach-noid hemorrhage. For the diagnosis and manage-ment of such patients, it would seem important toknow and even monitor the effectiveness of cere-bral autoregulation.

To study the dynamics of cerebral autoregu-lation, it is necessary to introduce a step distur-bance (stimulus) in CPP and to record the responsesof CBF and CVR continuously before and after thisstep disturbance. Recent developments in instru-mentation now allow such measurements to bemade by completely noninvasive means. We havealso employed a mechanical noninvasive techniqueto induce a CPP step disturbance without drugs orchanges in the concentration of vasoactive sub-stances in the blood. Our study was designed todetermine the response rate of cerebral autoregu-lation in healthy nonanesthetized humans. Further-more, the influence of hypercapnia and hypocapniaon these response rates was studied.

Subjects and MethodsFive healthy female and five healthy male sub-

jects aged 20-56 (mean 36) years from the hospital

by guest on May 10, 2018

http://stroke.ahajournals.org/D

ownloaded from

http://stroke.ahajournals.org/

46 Stroke Vol 20, No 1, January 1989

staff gave consent to participate in our study afterfull explanation of the research protocol. No sub-ject had any indications of cerebrovascular disease,and their mean blood pressures were <110 mm Hg.The subjects were investigated in the supine posi-tion with the head elevated above the bed bycushions. Step decreases in arterial blood pressure(ABP) were induced by rapid deflation of cuffsaround both thighs after a 2-minute inflation. Ineach subject, six such deflations were performedduring normocapnia, three during voluntary hyper-ventilation, and four while the subject breathed amixture of 5% CO2 in air. (Hypercapnia was inducedin only six subjects.) We recorded instantaneousvalues of end-tidal PCO2 (using an infrared gasanalyzer), thigh cuff pressure, ABP in the rightbrachial artery, and Doppler spectra from the rightmiddle cerebral artery (MCA) sampled with a rateof 25 Hz. After the experiment, the time course ofCVR was calculated from these data. The measure-ments and the calculations used are described inmore detail below.

The right MCA was insonated through the tem-poral bone window using a probe arrangement thatcould be strapped onto the subject's head andlocked in position to permit continuous measure-ments. The MCA was identified as describedpreviously.1112 The prototype transcranial com-puter Doppler instrument1213 incorporated a 64-point fast-Fourier transform to analyze the receivedvelocity spectra. The amplitude outputs of foursuch transforms were averaged over each sampleperiod of 40 msec and stored in memory with ABP,thigh cuff pressure, and end-expiratory PCO2 values.The data were transferred to disks after each indi-vidual experimental run.

Further processing of the Doppler spectra wasperformed off line. The maximum or outline bloodflow velocity, calculated by the standard algorithmof the instrument, corresponds to the blood flowvelocity in the center of the MCA. Because the flowis laminar, the mean cross-sectional velocity will beproportional to the maximum velocity (velocityprofile does not change significantly with variationsin velocity). Previously we have demonstrated alinear correlation between the MCA outline bloodflow velocity and blood flow volume measured inthe internal carotid artery (ICA) by an electromag-netic flowmeter during similar ABP variations.14

The finding of a regression line intercept close to theorigin in the blood velocity-blood flow volumerelation suggested that MCA diameter did not changesignificantly. Therefore, in the present experiments,CBF was assumed to be proportional to MCA bloodvelocity. Absolute measurements of blood flowvolume were not required for the calculation of thedynamic response, which means that the actualslope of the blood velocity-blood flow volume rela-tion did not have to be known.

As a further precaution to avoid erroneous inter-pretation of the velocity data in terms of changes in

CBF, we analyzed the Doppler signal power, cal-culated by squaring each spectral signal amplitudeand adding those 32 components with Dopplershifts toward the probe (MCA direction). Signalpower is proportional to the number of ultrasoundscatterers.15 Because the sample volume of theDoppler instrument is larger than the artery inson-ated, the number of such scatterers is proportionalto the cross-sectional area of the artery. Therefore,any change in vessel cross-sectional area will bereflected proportionally in the power of the receivedDoppler-shifted frequency components. Power wasdetermined over the last second before the stepdecrease in ABP (control) and between seconds 4and 5 after the decrease, when ABP drop wasmaximal.

Instantaneous values of ABP are required for thestudy of autoregulatory dynamics. Conventionalnoninvasive methods employing slow cuff deflationto record systolic and diastolic blood pressure wouldmiss the fast dynamic components. Therefore, weused a servo-cuff method16 that is capable of record-ing the ABP waveform continuously over 1-2 min-utes. A normal size arm cuff was connected to avery fast electropneumatic valve, which produced acuff pressure wave that tracked the ABP wave inreal time. The control signal to the servo wasderived from an 8-MHz Doppler velocity measure-ment in the axillary artery 3 cm proximal to thecuff. The control algorithm was designed to main-tain low blood flow velocity in the artery throughoutthe heart cycle. Then, by the flow restrictionprinciple,16 the artery under the servo-cuff wouldremain partially collapsed, and the servo-cuff pres-sure waveform would equal the intra-arterial bloodpressure waveform. The servo-cuff pressure wave-form was measured by a conventional transducerand filtered by a 10-Hz low-pass filter to eliminatehigh-frequency artifacts due to the Doppler noisepropagating through the servo loop. Such noninva-sive ABP recordings have been compared withintra-arterial blood pressures for step decreases inABP produced by leg cuff deflation (Figure 3 inReference 16). The servo-cuff method was found tomeasure these step decreases very accurately.

The servo-cuff was placed on the right arm so thatany nerve stimulation from cuff pulsations would bedirected to the hemisphere contralateral to the siteof CBF measurement. The hydrostatic zero pointwas assumed to be at the level of the circle of Willis,and the fluid column between this level and theservo-cuff level (approximately 12 mm Hg) wassubtracted to give an ABP reading approximatingCPP. This assumption was based on the hydrostaticeffects of the position used. Because the right atrialblood pressure is lower than the hydrostatic columnbetween the heart and the head, venous sinuspressure and intracranial pressure (ICP) will beclose to 0 in a normal subject in the supine positionwith the head elevated.



ownloaded from


Aaslid et al Cerebral Autoregulation Dynamics 47

100-

\

8

FIGURE 1. Recordings of arterial blood pressure (ABP)and thigh cuff pressure (upper), middle cerebral artery(MCA) velocity (middle), and end-expiratory carbon diox-ide partial pressure pCO: (lower) from one experiment.Thigh cuffs were released at time 0 (arrow), causingtransient drop in ABP and MCA blood velocity. Autoreg-ulatory action is seen from the blood velocity returning topretest values after only 4 seconds. Absolute units ofmeasure, all pressures in mm Hg and velocity in cmlsec.

Conventional thigh cuffs were modified with largertubings (5 mm i.d.) and connected with a large-boreY-piece. The thigh cuffs were wrapped around boththighs and inflated above the systolic blood pres-sure of the subject. Each subject was maintained inthis state for 120 seconds to produce leg hyperemia.Thigh cuff pressure was measured by a transducerconnected to the computer system. The thigh cuffswere deflated abruptly by pulling the tube off theY-piece. Thigh cuff pressure fell below diastolicblood pressure in <200 msec (Figure 1).

Recordings from each subject in each PacO2 statewere averaged with a common time reference. Time0 for this averaging algorithm was the abrupt fall inthigh cuff pressure. To determine the overall meanresponse, all individual recordings were averagedfor each PacOj state (Figure 2). MCA velocity andABP tracings were both smoothed through a digitallow-pass filter (-3 dB at 1 Hz and -17 dB at 2 Hz)to dampen harmonics of the pulse wave. Controlvalues of CBF (proportional to velocity) and ABPwere defined by their means during the 4 secondsbefore thigh cuff release. Changes in arterial bloodpressure (AABP) and cerebral blood flow (ACBF)were determined relative to these control values.By this approach, the subject was his own refer-ence. Control values were defined for each PaC02state.

The time course of CVR was determined bydividing ABP by CBF for each time point. Thecomputer then plotted curves as shown in Figure 2,

in which all variables are expressed relative to theirrespective controls.

In the present experiments, ABP was decreasedabruptly and remained low for a limited time, approx-imately 6-7 seconds; then reflex cardiovascularaction gradually restored ABP almost to the controlvalue. Before the onset of the reflex effect, it waspossible to equate AABP with a step stimulus. Themagnitude of the step was calculated by subtractingcontrol ABP from mean ABP during the intervalfrom 1 to 3.5 seconds. This value was then dividedby control ABP to obtain the relative step, AABP.

During the interval from 1 to 3.5 seconds, CVRchanged with time (T) in an approximately linearfashion. A regression line could be fitted to thesedata (Figure 2, lower). The slope of this regressionline (ACVR/AT) defines the rate at which CVRchanges. This rate depends on AABP. Full restora-tion of CBF would theoretically occur if AC VR wasequal to AABP. The rate of regulation (RoR) wastherefore defined as RoR=(ACVR/AT)-AABP.According to this definition, RoR of 0.2/sec (as inFigure 2, middle) implies a per-second adjustmentof 20% of the full CVR change necessary to fullycompensate for AABP.

ResultsRecordings from one representative experiment

are displayed in Figure 1. The upper panel showssuperimposed tracings of ABP and thigh cuff pres-sure. The rapid release of the thigh cuffs elicited anabrupt decrease in ABP approximately 200 mseclater. This decrease is also seen in the MCA veloc-ity tracing (Figure 1, middle). While ABP remainedlow for the four or five heart beats, MCA velocityreturned to control. After this phase, there was anovershoot in MCA velocity while ABP returned tocontrol.

The averaged and filtered tracings from the entireseries are presented in Figure 2 for each Pacc>2 stateas listed in Table 1. AABP was practically identicalacross states as seen in Figure 2. A AABP ofapproximately 20%, within the normal physiologicrange of blood pressure disturbances, was achieved.Moreover, this AABP was sufficient to evoke amarked response of autoregulation (CBF response)as seen in Figure 2.

The CBF response depended on the Pacc>2 state.In hypocapnia (Figure 2, left), autoregulatory actionwas very rapid; after only 1.9 seconds ACBF wasreduced to half. After 4.2 seconds CBF was evengreater than control. This finding indicated an oscil-latory autoregulatory response that was damped at8 seconds. The autoregulatory response was slowerin normocapnia (Figure 2, middle); the time forhalf-maximal response was 3.4 seconds, and over-shoot was seen only when ABP was increasinggradually (8-12 seconds). This type of overshootwas probably not due to an oscillatory autoregu-latory response but was more likely to be due to thedelay in autoregulatory action compensating for the



ownloaded from



-5 10 sec 10 sec -5 10 sec

pC0 2 = 22 .2 mmHg pC0 2 = 37.1 mmHg pCOo = 46.9 mmHg

FIGURE 2. Averaged findings of all experiments during hypocapnia Qeft), normocapnia (middle), and hypercapnia(right). Normalized arterial blood pressure (ABP) (upper) and cerebral blood flow (CBF) (middle) from transcranialDoppler recordings; calculated cerebrovascular resistance (CVR) (lower). Thigh cuffs were released at time 0, causingstep decrease in ABP 250 msec later; ABP decrease reached maximum in another 250 msec. Straight lines through CVRcurves were determined by regression analysis of data in interval from 1 to 3.5 seconds after thigh cuff release and wereused for calculating rate of regulation. All tracings are shown in normalized units relative to control prerelease valuesduring time from —4 to 0 seconds.

rising ABP. Practically no overshoot and a much-delayed half-maximal response (5.2 seconds) wasfound in hypercapnia (Figure 2, right).

RoR was calculated for each subject for eachPacc>2 state. The results are shown in Figure 3, withlines connecting the findings in each subject. Theopen squares (not connected) represent the curvesshown in Figure 2, in which all responses at eachtime were averaged before determination of RoR.These averages came very close to the regressionline determined by analysis of all the individualresponses. There was a highly significant correla-

tion between RoR and Pacc^ (r=-0.72, /?<0.001).RoR almost doubled relative to normocapnia, from0.20 to 0.38/sec, during hyperventilation. Breathing5% CO2 in air caused a 45% reduction in RoRrelative to normocapnia, to 0.11/sec. As seen inFigure 3, nine subjects had comparable RoRs, whileone subject had a much more rapid response, withan RoR as high as 0.8/sec during hyperventilation.In this subject only 2 seconds were necessary tocompensate fully for the step decrease in ABP. Thissubject had borderline hypertension during the inves-tigation (mean ABP of 109 mm Hg), which was 24

TABLE 1. Parameters Measured in Determining Cerebral Autoregulation Dynamics in Humans

ParameterPco^ (mm Hg)Control VMCA (cm/sec)

Control ABP (mm Hg)A ABP(%)RoR (sec"1)A Power (%)

Hypocapnia (n=10)(22.2 mm Hg)

22.2±0.646.1±3.184.5±4.121.9+1.00.38±0.04

-0.4±1.0

Pacxh state

Normocapnia (n= 10)(37.1 mmHg)

37.1±0.867.4±5.981.6+4.224.1±1.1O.2O±O.3O

-2.55±1.2

Hypercapnia (n=6)(46.9 mm Hg)

46.9±0.589.3±3.488.2±7.221.5±2.20.11 ±0.02

Data are mean±SEM. Pcoj, partial pressure of carbon dioxide in expired air; VMCA, blood velocity in middle cerebral artery; ABP,mean arterial blood pressure; A ABP, percentage decrease in arterial blood pressure; RoR, rate of regulation; A Power, percentagechange in power of reflected Doppler signal 4-5 seconds after blood pressure step decrease.



ownloaded from



RoR

0.7 - -

0.6 - -

0.5 - -

0.4 - -

0 .3 - -

0 . 2 - -

0.1 - -

0 - -15 20 25 30 35

Pco2

H h40 45 50

[ ramHg 3

FIGURE 3. Relation between end-expiratory carbon diox-ide partial pressure pCO2 and rate of regulation (RoR) in10 healthy subjects. Observations in each subject (•) areconnected by lines. Dotted line was determined by regres-sion analysis of all data (y=0.621 -0.011 x; r= -0.717,p<0.001). D, averages shown in Figure 2, when allindividual responses in each pCO2 state were averaged.

mm Hg higher than the mean of the group and 16mm Hg higher than the next highest ABP measured.This observation suggested that there could be apossible correlation between ABP and RoR. Regres-sion analysis of the relation between RoR and ABPin normocapnia was therefore performed, and analmost-significant (r=0.4, p=0.055) correlation wasfound. This result, however, should be viewed fromthe perspective that our study was not initiallydesigned to correlate autoregulatory dynamics withchanges in control ABP and that these findings werenot the outcome of a systematic experimentalapproach.

Analysis of the Doppler-shifted power spectrumduring normocapnia demonstrated that the signalpower was reduced by 2.55±1.2% (mean±SEM) at4-5 seconds compared with control. This reductionwas not significant (p>0.05) and could possibly beexplained by the slight reduction in MCA bloodvelocity at the time of power calculation. Duringhypocapnia, when velocity had returned to control at4 seconds, a reduction in signal power was not seen(0.4±1.0%). If the MCA had reacted to the stepdecrease in ABP with compensatory vasodilation, a

marked increase in power would have been expected.Such an effect was not seen in any subject, and thelargest increase recorded was only 2.3%.

DiscussionCerebral autoregulation is an extremely sensitive

mechanism, and some attempts to investigate itsaction have stranded on the experimental design,which induced diminished responsiveness by inju-ries and lengthy preparations.17 Investigations inawake, normal humans carry special weight becausethey are closer to the normal physiologic state thananimal experiments. Furthermore, it is necessary toensure that the interventions used to induce changesin ABP do not have a direct effect upon cerebralvascular smooth muscle.3 A recent review on thecerebral circulation18 states: "Perhaps the mostphysiological responses to alterations in perfusionpressure are those induced by changes in posture inunanesthetized humans." Our study was designedto minimize interventions and to study the normalphysiologic response to acute reductions in ABPand CPP. Our model might be representative ofmany situations in daily activity involving posturechanges, etc., in which a rapid homeostatic action isnecessary to preserve CBF. Because we employednew techniques, we will first discuss the validity ofour measurements and the approach we used toevoke the autoregulatory response.

Blood Pressure StepThe ideal stimulus to cerebral autoregulation is an

abrupt change in CPP that is maintained during theentire interval in which autoregulatory adjustmentstake place. In this respect, our method was notentirely ideal in the sense that ABP was lowered ina stepwise fashion for only 5-7 seconds beforereflexes started to restore ABP. (The response ofthese reflexes might be an interesting study initself.) In normocapnia, and even more so in hypo-capnia, this time was sufficient for autoregulation tobe observed with almost full restoration of CBF.Therefore, we limited our analysis to a time duringwhich ABP decrease was very close to an idealstep. The first second of the step was also excludedfrom analysis because CBF during this period mightpossibly reflect blood flow to and fro in the compli-ant arterial system. Exclusion of this first secondalso eliminated transients because of the arterialpulse transmission delays between ABP measuredin the arm and CBF measured in the cerebralarteries.

Because our study was performed in normalhealthy subjects, we could assume a low ICP thatwas not influenced significantly by either ABPchanges or the cerebral vasodilation necessary toadjust for the decrease in CPP. If the latter effecthad been present to any significant degree (such asin subjects with increased ICP and a less compliantintracranial space), this would have meant a sec-ondary decrease in CPP caused by vasodilation,



ownloaded from



and compensatory CBF restoration would havebeen delayed.

Another possible source of error in our experi-mental design could be the CO2-rich and hypoxicblood from the legs returning to the circulation andinfluencing cerebral vascular tone. The transporttime from the legs to the cerebral vascular bedthrough the right heart, the pulmonary circulation,the left heart, the aorta, and the carotid arteries isapproximately 15 seconds. The mixed blood fromthe legs would therefore arrive in the brain longafter the data for analysis were collected. Theinfluence of any such PacO2 increase even after thisperiod was probably small because no secondaryincrease in CBF was seen even after 15 seconds.

Cerebral Blood Flow MeasurementsThe validity of our assumption that CBF (in the

MCA territory) is proportional to blood velocity inthe MCA depends on the two premises that thecross-sectional area of the MCA does not changesignificantly and that blood flow in the MCA trunkis not circumvented by collateral networks. Thelatter premise is almost self-evident in our experi-ments. Because the measurement site was distal tothe circle of Willis, the only collaterals possiblewere the leptomeningeal anastomoses over the con-vexities of the cerebrum. This source of collateralblood supply has a high flow resistance. Moreover,perfusion pressure in the anterior and posteriorcerebral artery territories changed equally with thatin the MCA. Therefore, there could have been noperfusion pressure gradient to drive the bloodthrough peripheral collateral channels connectingthese territories.

The first premise was investigated by recordingthe Doppler signal spectral power before and afterthe step decrease in ABP. The late period of thestep was chosen because blood flow velocity wascomparable to that before the step so that reflectedpower could be compared at similar spectral distri-butions. The transcranial approach offers an idealsetting for using signal power to investigate changesin the cross-sectional area of arteries. The samplevolume of the 2-MHz instrument is slightly largerthan the artery studied.19 The MCA was insonatedat an angle of approximately 20° in most subjects.Also, the probe was fixed to the subject's head sothat movement artifacts were avoided during therelatively short period of recording. The basal cere-bral arteries are anchored so that displacementduring the maneuvers is prevented.

According to theories of ultrasonic scatteringfrom blood, any change in artery cross-sectionalarea would cause a proportional increase in signalpower.15 Even if this relation were not perfectbecause of inhomogeneities in the ultrasonic beamshape over the artery, we would at least expect thata significant change in area would cause a signifi-cant change in power. The change in signal powermeasured during normocapnia was an order of

magnitude lower than the changes in ABP and CBF;there was no change during hypocapnia. Theseresults strongly suggest that the human MCA cross-sectional area does not change significantly during a20-mm Hg step decrease in ABP. This conclusion isin agreement with findings of very stiff walls ofhuman cerebral arteries20 and with studies compar-ing ICA blood flow and MCA blood velocity whenABP changed.14 Experimental studies21-23 haveshown that the main autoregulatory action inresponse to ABP decreases is located in smallvessels (diameter of <40 pirn) in the brain paren-chyma. Arterioles with diameters of 322 /xmincreased their calibers only minimally when ABPwas lowered from 90 to 70 mm Hg.21 If this findingis extrapolated to the human MCA, which is largerby an order of magnitude, no significant activeregulation of MCA diameter would be expected.In conclusion, the assumption of proportionalitybetween MCA blood velocity and flow in ourexperimental protocol is supported by the avail-able evidence.

Mechanism of AutoregulationThe response of autoregulation to step increases

in ABP was studied in baboons by Symon et al.7

They found a very rapid CVR adjustment, with aninitial peak at 0.5 seconds and a secondary peak at3-4 seconds after the step. Since CBF was mea-sured on the outflow side in Labb6s vein, it ispossible that the initial peak was caused by thedelay in the transmission of the flow wave througharterioles, capillaries, and small veins. This delaywas clearly seen in one of their other recordings(Figure 9 of Reference 7) when no autoregulationwas present due to cerebral ischemia. Therefore,the 3-4-second peak was probably the real autoreg-ulatory response, and it corresponds to our findingsof an overshoot at approximately 4.5 seconds (Figure2, left). It should be noted that step increases wereused in the baboon experiments,7 whereas we usedstep decreases. Our data indicate an increase inRoR and overshoot with increasing vasoconstric-tion, and this could explain the somewhat faster andlarger overshoot observed by Symon et al.7

Studies on the influence of Paco2 on the staticCPP-CBF characteristics show that the curvebecomes more parallel to the CPP axis duringhypocapnia, whereas hypercapnia shifts the curvetoward a passive-resistive CPP-CBF relation.1-418

Interpreted in terms of a feedback system,18 anerror signal (imbalance in blood flow) is necessaryto activate changes in vasomotor tone. If the open-loop gain is high, only a small disturbance in CBF issufficient to cause a large reaction in CVR. Thismeans that hypocapnia would increase feedbackgain, whereas hypercapnia would decrease and evenabolish the gain.

The changes in RoR could have been mediatedthrough such changes in feedback gain. When gainincreases, RoR becomes both more rapid and more



ownloaded from



oscillatory. This is a well-known experience inman-made control systems with time constants anddelays in the feedback loop. Hypothetically, assum-ing a potent vasodilator regulator substance involvedin a metabolic feedback loop,24-23 the overshoot incerebral autoregulatory response could be explainedas follows: Initially there is a step decrease in ABPand an immediate reduction of CBF. Slight hypoxiadevelops in the neurons and their environment, and(through some unknown pathway) regulator sub-stance is released locally.25 With some delay, thetone of vascular smooth muscle is decreased by thevasodilatory properties of the regulatory substance.Because of these time constants/delays, it couldwell be that more regulator substance than is requiredfor restoration of normal CBF is released before thevascular smooth muscle has responded to the actionof the regulator substance. This situation wouldcause an overshoot in CBF when the effect of theregulator substance on smooth muscle was fullydeveloped. The CBF overshoot would, in turn,create a slightly hyperoxic environment, reduceproduction of the regulator substance, and causesmooth muscle to contract again to achieve thedesired CBF and metabolic environment. Reason-ing along these lines, it is perfectly conceivable thata metabolic feedback system with a high open-loopgain could explain the overshoot of the cerebralautoregulation response seen with a large stepincrease in ABF and/or with hypocapnia (this study).

The hypothesis7 that the rapid autoregulatoryresponse is produced by a myogenic mechanism isnot supported when our results are compared withthose from an earlier study in which purely func-tional stimuli (light on the retina) were used toevoke CBF responses.26 The comparable RoR fromthe visual evoked CBF study was 0.22/sec for thelights-off response and 0.32/sec for the lights-onresponse. The half-time responses were also slightlyfaster than in the present experiments. ABP washigher in the visual evoked CBF study becausethere was no step decrease in ABP, and this mightaccount for the slightly faster autoregulatoryresponses. Studies on isolated posterior cerebralarteries of rats27 or pial arteries of monkeys28 showedthat the myogenic response developed over a periodof 1-10 minutes. This is more than an order ofmagnitude slower than the autoregulatory responsewe observed.

Because the time response of CBF to stepdecreases in ABP was practically identical to that offunctional stimuli in normal humans, it seems rea-sonable to assume that the same feedback loop isinvolved in both. This is in agreement with studiesusing venous blood pressure increases to discrimi-nate between the metabolic and the myogenichypotheses of cerebral autoregulation. This simulusresulted in arteriolar vasodilation29 and constantCBF,30 whereas vasoconstriction and significantlyreduced CBF would have been the result expectedfrom a myogenic mechanism.

In summary, our study lends support to thehypothesis6 that the rapid response of the cerebralautoregulation in normal human physiology is medi-ated by a metabolic mechanism. Furthermore, wehave shown that the cerebral autoregulatory actionsare significantly faster than those of the barorecep-tor reflex regulating ABP.

Clinical ImplicationsInvestigations on physiologic parameters in awake

normal humans carry special weight because theymore closely represent the normal physiologic statethan can be achieved in animal experiments. Ourclinically most significant finding is the strong depen-dence of RoR on cerebral vasomotor tone. Measur-ing RoR with the present technique is completelynoninvasive and nonpharmacologic. Therefore, weenvision that it will be possible to obtain clinicallyrelevant information on autoregulatory gain and thecerebral vasomotor state from such noninvasiveresponse studies. This new level of insight may beof value in the day-to-day management of patientsunder neurointensive care, for diagnostic purposes,and to obtain improved insight into the pathophys-iology of many cerebrovascular disorders.

References1. Harper AM: Autoregulation of cerebral blood flow: Influ-

ence of the arterial blood pressure on the blood flow throughthe cerebral cortex. J Neurol Neurosurg Psychiatry 1966;29:398-403

2. Rape I a CE, Green HD: Autoregulation of canine cerebralblood flow. Circ Res 1964;14(suppl I):I-205-I-212

3. Lassen NA: Autoregulation of cerebral blood flow. Circ Res1964;14(suppl I): 1-201-1-204

4. Ekstrom-Jodal B, Haggendal E, Under L-E, Nilsson NJ:Cerebral blood flow autoregulation at high arterial pressuresand different levels of carbon dioxide tension in dogs. EurNeurol 1971 ;6:6-10

5. Strandgaard S, MacKenzie ET, Sengupta D, Rowan JO,Lassen NA, Harper AM: Upper limit of autoregulation ofcerebral blood flow in the baboon. Circ Res 1974;34:435-440

6. Berne RM, Winn HR, Rubio R: The local regulation ofcerebral blood flow. Prog Cardiovasc Dis 1981;24:243-260

7. Symon L, Held K, Dorsch NWC: A study of regionalautoregulation in the cerebral circulation to increased perfu-sion pressure in normocapnia and hypercapnia. Stroke 1973;4:139-147

8. Lubbers DW, Leiniger-Follert E: Capillary flow in the braincortex during changes in oxygen supply and state of activa-tion, in Cerebral Vascular Smooth Muscle and its Control.Ciba Foundation Symposium 56. Amsterdam/New York,Elsevier/Excerpta Medica/North-Holland, 1978, pp 21-43

9. Silver IA: Cellular microenvironment in relation to localblood flow, in Cerebral Vascular Smooth Muscle and itsControl. Ciba Foundation Symposium 56. Amsterdam/NewYork, Elsevier/Excerpta Medica/North-Holland, 1978, pp49-61

10. Nilsson B, Rehncrona S, Siesjd BK: Coupling of cerebralmetabolism and blood flow in epileptic seizures, hypoxia andhypoglycaemia, in Cerebral Vascular Smooth Muscle and itsControl. Ciba Foundation Symposium 56. Amsterdam/NewYork, Elsevier/Excerpta Medica/North-Holland, 1978, pp199-214

11. Aaslid R, Markwalder T-M, Nornes H: Noninvasive trans-crania] Doppler ultrasound recording of flow velocity inbasal cerebral arteries. J Neurosurg 1982;57:769-774



ownloaded from



12. Aaslid R: Transcranial Doppler examination techniques, inAaslid R: (ed): Transcranial Doppler Sonography. Vienna/New York, Springer-Veriag, 1986, pp 39-59

13. Aaslid R: The Doppler principle applied to measurement ofblood flow velocity in cerebral arteries, in Aaslid R: (ed):Transcranial Doppler Sonography. Vienna/New York,Springer-Verlag, 1986, pp 22-38

14. Lindegaard K-F, Lundar T, Wiberg J, Sj6berg D, Aaslid R,Nornes H: Variations in middle cerebral artery blood flowinvestigated with noninvasive transcranial blood velocitymeasurements. Stroke 1987;18:1025-1030

15. Arts MGJ, Roevros JMJG: On the instantaneous measure-ment of blood flow by ultrasonic means. MedBiol Eng 1972;10:23-30

16. Aaslid R, Brubakk AO: Accuracy of an ultrasound Dopplerservo method for noninvasive determination of instanta-neous and mean arterial blood pressure. Circulation 1981;64:753-759

17. Sagawa K, Guyton AC: Pressure-flow relationships in iso-lated canine cerebral circulation. Am J Physiol 1961;200:711-714

18. Heistad DD, Kontos HA: Cerebral circulation, in HandbookPhysiology, sect 2: The Cardiovascular System. Bethesda,Md, American Physiological Society, 1983, vol 3, pp 137—182

19. Grolimund P: Transmission of ultrasound through the tem-poral bone, in Aaslid R (ed): Transcranial Doppler Sonog-raphy. Vienna/New York, Springer-Veriag, 1986, pp 10-21

20. Hayashi K, Handa H, Nagasawa S, Okumura A, MoritakeK: Stiffness and elastic behavior of human intracranial andextracranial arteries. Biomed Eng Comput 1980;13:175—183

21. Kontos HA, Wei EP, Navari RM, Levasseur JE, RosenblumWI, Patterson JL: Responses of cerebral arteries and arte-

rioles to acute hypotension and hypertension. Am J Physiol1978;234:H371-H383

22. Stromberg DD, Fox JR: Pressures in the pial arterial micro-circulation of the cat during changes in systemic arterialblood pressure. Circ Res 1972^1:229-239

23. Baumbach GL, Heistad DD: Effects of sympathetic stimu-lation and changes in arterial pressure on segmental resis-tance of cerebral vessels in rabbits and cats. Circ Res 1983;52:527-533

24. Winn HR, Rubio R, Beme RM: Brain adenosine productionduring 60 seconds of ischemia. Circ Res 1979;45:486-492

25. Winn HR, Rubio R, Berne RM: The role of adenosine in theregulation of cerebral blood flow. J Cereb Blood Flow Metab1981;l:239-244

26. Aaslid R: Visually evoked dynamic blood flow response ofthe human cerebral circulation. Stroke 1987;18:771-775

27. Halpern W, Osol G: Influence of transmural pressure onmyogenic responses of isolated cerebral arteries of the rat.Ann Biomed Eng 1985; 13:287-294

28. Bevan JA, Hwa JJ: Myogenic tone and cerebral vascularautoregulation: The role of a stretch-dependent mechanism.Ann Biomed Eng 1985;13:281-286

29. Wei EP, Kontos HA: Responses of cerebral arterioles toincreased venous pressure. Am J Physiol 1982;243:H442-H447

30. Wagner E, Traystman RJ: Cerebral venous outflow andarterial microsphere flow with elevated venous pressure. AmJ Physiol 1983;244:H505-H512

KEY WORDSblood flow •

• autoregulationultrasonics

cerebral arteries • cerebral



ownloaded from


R Aaslid, K F Lindegaard, W Sorteberg and H NornesCerebral autoregulation dynamics in humans.

Print ISSN: 0039-2499. Online ISSN: 1524-4628 Copyright © 1989 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Stroke doi: 10.1161/01.STR.20.1.45

1989;20:45-52Stroke.

http://stroke.ahajournals.org/content/20/1/45World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://stroke.ahajournals.org//subscriptions/

is online at: Stroke Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer available in the

Permissions in the middle column of the Web page under Services. Further information about this process isOnce the online version of the published article for which permission is being requested is located, click Request

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Stroke Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:



ownloaded from

http://stroke.ahajournals.org/content/20/1/45

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://stroke.ahajournals.org//subscriptions/


cerebral autoregulation dynamics in humans -...

Documents