Indexes of Flow and Cross-sectional Area of the Middle Cerebral Artery Using Doppler Ultrasound During Hypoxia and Hypercapnia in Humans
Background and Purpose This study examined changes in cross-sectional area of the middle cerebral artery as assessed by changes in Doppler signal power during hypoxia and hypercapnia. In addition, it examined the degree of consistency among three indexes of cerebral blood flow and velocity: the velocity spectral outline (V̄P), the intensity-weighted mean velocity (V̄IWM), and an index of middle cerebral artery flow (P·V̄IWM). P·V̄IWM was calculated as the product of V̄IWM multiplied by the total power signal. Power is proportional to cross-sectional area of the vessel; this calculation therefore allows for any changes in this variable.
Methods Four protocols were used, each repeated six times for six healthy adults aged 20.8±1.7 years (mean±SD). The first was a control protocol (A) with end-tidal Po2 (etPo2) maintained at 100 mm Hg and etPco2 at 1 to 2 mm Hg above eucapnia throughout. The second was a hypoxic step protocol (B) with etPo2 lowered from control values to 50 mm Hg for 20 minutes. The third was a hypercapnic step protocol (C) with etPco2 elevated from control by 7.5 mm Hg for 20 minutes. The fourth was a combined hypoxic and hypercapnic step protocol (D) lasting 20 minutes. A dynamic end-tidal forcing system was used to control etPco2 and etPo2. Doppler data were collected and stored every 10 milliseconds, and mean values were determined later on a beat-by-beat basis. V̄P, V̄IWM, power, and P·V̄IWM were expressed as a percentage of the average value over a 3-minute period before the step.
Results In protocols A and B, there were no changes in power and there were no differences between V̄P, V̄IWM, and P·V̄IWM. In C, at the relief from hypercapnia, there was a transient nonsignificant increase in power and a transient nonsignificant decrease in both V̄P and V̄IWM compared with P·V̄IWM. In D, during the stimulus period, V̄P was significantly higher than V̄IWM (paired t test, P<.05), but both indexes were not different from P·V̄IWM. In the period that followed relief from hypoxia and hypercapnia, the Doppler power signal was significantly increased by 3.8%. During this period, V̄P and V̄IWM were significantly lower than P·V̄IWM.
Conclusions At the levels of either hypoxia or hypercapnia used in this study, there were no changes in cross-sectional area of the middle cerebral artery, and changes in both V̄P and V̄IWM accurately reflect changes in P·V̄IWM. With combined hypoxia and hypercapnia, however, at the relief from the stimuli when there is a very large and rapid decrease in P·V̄IWM, power is increased, suggesting an increase in the cross-sectional area. During this period, changes in V̄P and V̄IWM underestimate the changes in P·V̄IWM.
Transcranial Doppler ultrasound has been used to measure MCA blood flow velocities, which in turn may be used to provide some sort of index of MCA flow. The index that has most commonly been used has been the mean of the velocities associated with the maximal frequency of the Doppler shift (V̄P), which, if the flow is laminar, will be proportional to the axial flow velocity (equal to the axial flow velocity if the angle of insonation is zero). Under conditions of laminar flow in a rigid tube, axial flow velocity is proportional to true flow. However, if the flow is complex, then the possibility exists that the relationship between V̄P and true flow may not be linear.
One way of circumventing the above problem is to use the entire velocity spectrum and calculate a V̄IWM. Because the intensity of the power spectrum at any velocity should be related directly to the number of ultrasound scatterers (ie, red blood cells) moving at that velocity,1 for flow in a rigid tube V̄IWM should provide a signal proportional to overall flow regardless of the complexity of the flow pattern.
A further problem relates to the fact that the MCA cannot be considered to be a rigid tube, but neither V̄P nor V̄IWM can be considered to remain proportional to blood flow if the area of the vessel is changing. To overcome this, it is possible to use the total power of the reflected Doppler signal as a measure of the total number of ultrasound scatterers. If the sample remains of constant depth and the power and position of the insonating beam are unchanged, then the Doppler power provides an index of cross-sectional area. An index of flow that takes variations in cross-sectional area into account may be calculated as the product of total power and the intensity-weighted mean velocity (P·V̄IWM).
The purpose of the present study was to examine changes in Doppler signal power (as an index of cross-sectional area of the MCA), as well as the degree of consistency among the three indexes for flow, namely, V̄P, V̄IWM, and P·V̄IWM. Data are taken from a previous study of the dynamics of the response of the cerebral circulation to step changes in etPco2 and etPo2,2 which provide the necessary variations in cerebral blood flow.
Subjects and Methods
This study involved six young adults. The study requirements were fully explained to all participants, with each giving informed consent before participation in the study. The research was approved by the Central Oxford Research Ethics Committee. Participants were not taking any medication, and none of the participants had a history of cardiovascular, cerebrovascular, or respiratory disease.
Each participant visited the laboratory on six or seven occasions, each lasting 4 to 5 hours. On each visit, one repetition of each of four protocols was performed in a randomly assigned order, with each repetition lasting approximately 40 minutes. Before the experiments were started, room air measurements of resting Doppler signals and etPco2 were collected. The subject's natural etPco2 was measured using a nasal catheter, which disturbs quiet breathing less than a mouthpiece and nose clip arrangement.
The four protocols are shown in Fig 1⇓ with the actual data obtained. For the control protocol (Fig 1A⇓), etPo2 was maintained at 100 mm Hg and etPco2 at 1 to 2 mm Hg above the subject's normal value for 30 minutes. Each of the test protocols started with a short 6- to 7-minute period during which etPo2 was maintained at 100 mm Hg and etPco2 at 1 to 2 mm Hg above the subject's natural value as determined that day. Then etPo2 and/or etPco2 were altered rapidly (over one or two breaths) to a new set of desired levels, according to each protocol described below, and maintained constant for 20 minutes. Finally, etPo2 and/or etPco2 were returned (again within one or two breaths) to their initial euoxic and near-eucapnic values and maintained constant for a further 10 minutes. For the hypoxic protocol (Fig 1B⇓), etPo2 was lowered to 50 mm Hg while etPco2 was continually maintained at 1 to 2 mm Hg above the subject's normal value. For the hypercapnic protocol (Fig 1C⇓), etPo2 was continued at 100 mm Hg while etPco2 was elevated by 7.5 mm Hg (ie, between 8.5 and 9.5 mm Hg above the subject's normal value). Finally, for the protocol combining hypoxia with hypercapnia (Fig 1D⇓), etPo2 was lowered to 50 mm Hg and etPco2 was elevated by 7.5 mm Hg.
This study used the technique of dynamic end-tidal forcing to regulate the end-tidal gas tensions accurately and to generate the desired rapid steps in end-tidal composition as required.3 The apparatus and technique have been described previously.2 4 5
Apparatus and Technique for Measurement of Cerebral Blood Flow
A 2-MHz pulsed Doppler ultrasound system (PCDop 842, SciMed) was used to measure back-scattered Doppler signals from the right MCA. The Doppler system was adapted to make the Doppler signals (maximum and intensity-weighted mean Doppler frequency shifts and total power) available as analogue signals. These were updated each time a new spectrum was calculated (every 10 milliseconds). The signals were sampled every 10 milliseconds using a data acquisition package (DAQWare, National Instruments) running on another computer. These signals were saved for later analysis.
The MCA was identified by an insonation pathway through the right temporal window just above the zygomatic arch using standard search techniques that have been described previously.6 7 Ultrasound gel was applied to the subject's skin and hair at the temporal window and to the probe before the experimenter proceeded to locate and identify the main segment of the MCA. Optimization of the Doppler signals from the MCA was performed by varying the sample volume depth in incremental steps and at each depth varying the angle of insonation to obtain the best quality signals for the Doppler frequency shifts that corresponded to the maximum power signal. Ultrasound gel was then reapplied sparingly to both the insonation site and the probe before the probe was securely positioned in a headband device (Müller and Moll Fixation, Nicolet Instruments Ltd) to ensure optimal insonation position and angle for the duration of the experiment.
Averaging Within Experimental Sessions
The data for MCA blood flow comprise one observation each for V̄P, V̄IWM, and power every 10 milliseconds. P·V̄IWM was calculated every 10 milliseconds as the product of V̄IWM and power, thus allowing for any systematic change in diameter throughout the cardiac cycle. The data for V̄P, V̄IWM, power, and P·V̄IWM were averaged over each heartbeat to give beat-by-beat values. The beat-by-beat data were further averaged to give one value every 15 seconds. Finally, for the statistical analysis, the 15-second data were averaged to give three 3-minute values for baseline (−3 to 0 minutes), stimulus (+17 to +20 minutes), and recovery (+27 to +30 minutes) periods.
In addition to the absolute values, normalized beat-by-beat values were calculated for V̄P, V̄IWM, power, and P·V̄IWM. The beat-by-beat data during the 3-minute period immediately before the onset of the stimulus were averaged, and this value was used as a baseline. The 3-minute baseline period was normalized to 100% and used to calculate the percent change over time in these variables (on a beat-by-beat basis) over the rest of the experimental record. The beat-by-beat normalized data were then further averaged to give one value every 15 seconds. Again, the 15-second data were averaged to give three 3-minute values for baseline, stimulus, and recovery periods.
To examine changes in Doppler signal power within each of the experimental protocols, the percent changes in Doppler power signal values obtained during the 3-minute stimulus and recovery periods were compared with the 3-minute baseline value (ie, 100%). Furthermore, to examine the degree of consistency among three indexes of cerebral blood flow, comparisons were made among the values obtained for V̄P, V̄IWM, and P·V̄IWM during the stimulus and recovery periods. Thus, paired t tests were used to (1) compare changes in Doppler power, (2) compare V̄P with V̄IWM, and (3) compare V̄P and V̄IWM with P·V̄IWM. Because several pairwise comparisons were performed within each protocol, a Bonferroni correction was applied to make allowance for multiple comparisons. The overall level of statistical significance was taken as P<.05, which, with the Bonferroni correction, equates to a level of significance of P<.0083 for each of the individual comparisons when six comparisons are made.
The average age of the six subjects was 20.8±1.7 years (mean±SD), average height was 183.3±5.0 cm, and average weight was 74.2±10.0 kg. None had a history of cardiovascular or respiratory disease, and all had normal systolic (116.3±6.3 mm Hg), diastolic (79.0±4.0 mm Hg), and mean arterial (91.4±3.7 mm Hg) blood pressure. The average insonation depth (the distance from the probe to the start of the Doppler sample volume for detecting signals from the MCA) was 5.02±0.03 cm. Small variations in depth between subjects are attributed to differences in skull size.7 The air-breathing data for each subject are listed in Table 1⇓.
Doppler Power Signal During Hypoxia and Hypercapnia
Ensemble averages for the group means for the normalized Doppler power signal are shown in Fig 2⇓. The data in Fig 2⇓ were obtained by first ensemble averaging all repetitions of each protocol in a single subject over the 15-second periods and then ensemble averaging the responses of all six subjects. The group means for the 3-minute averages for the normalized Doppler power are shown in Table 2⇓. In the control and hypoxic protocols, no changes were apparent in the Doppler power signal from the baseline value of 100% (Fig 2A and 2B⇓⇓). This was confirmed on the three averages by t tests. In hypercapnia, at the relief from the stimulus, there appeared to be a small increase in the power signal (Fig 2C⇓), but this was not significant as assessed by t test. In combined hypoxia and hypercapnia, during the recovery period from the stimuli, the Doppler power signal increased by 3.8% (Fig 2D⇓), and this increase was significant (P<.001, t test).
Comparison of V̄P With V̄IWM
Ensemble averages for the group means for the normalized velocities are shown in Fig 2⇑. The data in Fig 2⇑ were obtained by first ensemble averaging all repetitions of each protocol in a single subject over 15-second periods and then ensemble averaging the responses among the six subjects. The group means for the 3-minute averages are presented as absolute values for the 3-minute means of V̄P and V̄IWM (ie, in centimeters per second) in Table 2⇑ and as normalized responses in Table 3⇓. In general, the group responses show that the temporal profiles for V̄P and V̄IWM appear to be very similar. One exception is noted in combined hypoxia and hypercapnia: V̄P and V̄IWM appeared to be different during the stimulus period (Fig 2D⇑). When this difference was assessed statistically (paired t test), it was significant (P<.001).
Comparison of V̄P and V̄IWM With P·V̄IWM
Ensemble averages for the group means for the normalized flow index (P·V̄IWM) are shown in Fig 2⇑. The data for P·V̄IWM were obtained by first ensemble averaging all repetitions of each protocol in a single subject over 15-second periods and then ensemble averaging the responses among the six subjects. The group means for the 3-minute averages are presented in Table 3⇑. In general, the group responses show that the temporal profiles for V̄P and V̄IWM appear to be very similar to those for P·V̄IWM. Exceptions are noted in hypercapnia and in combined hypoxia and hypercapnia: during the periods of recovery, V̄P and V̄IWM appear to underestimate P·V̄IWM (Fig 2C and 2D⇑⇑). These differences correspond to the changes in power that have been noted above. In hypercapnia, these differences among V̄P, V̄IWM, and P·V̄IWM were not significant as assessed from the 3-minute means (Table 3⇑). However, in combined hypoxia and hypercapnia, the 3-minute means for V̄P and V̄IWM were significantly lower than for P·V̄IWM (Table 3⇑).
To highlight further the effects of changes in Doppler power on the Doppler velocities, we calculated the ratios between the normalized velocities and P·V̄IWM (ie, V̄P/P·V̄IWM and V̄IWM/P·V̄IWM). If the changes in the various indexes of MCA flow match, then the value of this index should be one; this index provides a useful graphic illustration of variations among the indexes throughout the experimental periods, including the transients. Ensemble averages for the group means for these ratios are shown in Fig 2⇑. In the control and hypoxic protocols, there were no apparent changes in V̄P/P·V̄IWM and V̄IWM/P·V̄IWM (Fig 2A and 2B⇑⇑). In hypercapnia, at the relief from the stimulus, transient decreases in V̄P/P·V̄IWM and V̄IWM/P·V̄IWM are observed (Fig 2C⇑). In combined hypoxia and hypercapnia, in the recovery period when Doppler power was significantly increased, V̄P/P·V̄IWM and V̄IWM/P·V̄IWM are lower than unity (Fig 2D⇑).
This study used the technique of transcranial Doppler ultrasound to detect changes in cross-sectional area of the MCA, as assessed by changes in Doppler power, during hypoxia and hypercapnia in humans. Additionally, this study assessed the degree of consistency among three indexes of cerebral blood flow obtained with transcranial Doppler ultrasound during hypoxia and hypercapnia. Two new findings are reported. First, at the levels of either hypoxia or hypercapnia alone used in this study, there was very little change in MCA cross-sectional area. However, in the combined hypoxic and hypercapnic protocol, at the relief of the stimuli when there was a very large and rapid decrease in cerebral blood flow, Doppler signal power was increased by 3.8%, suggesting an increase in the cross-sectional area of the MCA. Second, at the levels of stimulation associated with either hypoxia or hypercapnia alone, changes in V̄P and V̄IWM (and P·V̄IWM) were all very similar. However, in the combined hypoxic and hypercapnic protocol, the change in V̄P was significantly greater than the change in V̄IWM. Additionally, at the relief of the stimuli when Doppler signal power was increased, the changes in V̄P and V̄IWM significantly underestimated the change in P·V̄IWM.
Comparison With Other Measures of Cross-sectional Area Change
Evidence is now available from several studies in humans to suggest that various interventions that result in steady-state changes in cerebral hemodynamics are associated with only small changes in the caliber of the larger cerebral vessels.
Studies using transcranial Doppler ultrasound have generally taken one of two approaches to address the issue of whether, with various interventions, there are changes in caliber of large cerebral vessels such as the MCA. First, studies have correlated changes in V̄P with more direct indexes of changes in cerebral blood flow. Results from these studies show good agreement between changes in V̄P in the MCA and the change in V̄P in the internal carotid artery (combined with B-mode measurements of the diameter of the internal carotid artery)8 and between changes in V̄P in the MCA and changes in cerebral blood flow seen using techniques such as xenon washout,9 single-photon emission CT,10 and electromagnetic flowmeters11 during hypercapnia, blood pressure variations, or acetazolamide injections.8 11 12 Studies using the second approach have assessed changes in cross-sectional area as reflected by changes in Doppler power. Small changes in Doppler power have been reported in response to hypercapnia and hypocapnia,13 step decreases in arterial blood pressure,14 15 orthostasis,16 and anesthetic agents.17
The present study is the first to provide continuous beat-by-beat measurements of Doppler power during periods of sustained hypoxia and hypercapnia. Our results show no changes in Doppler power during “pure” hypoxia and hypercapnia and only small changes in Doppler power during combined hypoxia and hypercapnia. Thus, our findings are consistent with the suggestion that the caliber of the larger cerebral vessels changes very little when cerebral blood flow is moderately increased. Our result of an unchanged Doppler power during hypercapnia is different from that of Müller and Casty,13 who reported a 20% increase in Doppler power during hypercapnia (using a rebreathing technique). However, Müller and Casty also reported a large standard deviation in Doppler power (±18% versus ±6% in the present study; Table 2⇑), and it is unclear whether any attempts were made to deal with some of the potential difficulties associated with the use of Doppler power (ie, the need for several repetitions to ensure a high signal-to-noise ratio).
The findings from the above studies generally agree with those of more invasive approaches that have been used to assess changes in cross-sectional area of the large cerebral vessels. Giller et al18 measured the diameters of human cerebral arteries during craniotomies under moderate changes in mean blood pressure and etPco2 and reported small diameter changes (3% to 4%) in the large cerebral arteries (including the main branch of the MCA). In another study, Huber and Handa19 used contrast-agent angiography and found a 3.8% increase in vessel diameter in response to hypercapnia.
Thus, the general consensus from the studies so far published is that although changes in the vessel caliber of large cerebral vessels have been reported, the changes appear to be quite small. Our results are in agreement with this and therefore provide some reassurance that studies reporting the changes in velocities using transcranial Doppler ultrasound provide indexes of changes in underlying flow that are not distorted through changes in cross-sectional area of the vessel.
V̄P, V̄IWM, and P·V̄IWM as Measures of Cerebral Blood Flow
The results from this study indicate that there is very little difference among these indexes when assessing cerebral blood flow responses to modest stimuli. Given this, the use of either V̄P or V̄IWM has considerable advantages over the use of P·V̄IWM. First, an absolute value can be ascribed to the velocities, whereas the reflected power depends entirely on the power and other properties of the insonating beam. The only caveat to this statement is that the values for velocities will always be somewhat lower than the true values because the insonating beam will not be completely axial. However, the effect of this is likely to be small (eg, an angle of insonation of 30° difference gives an error of <13%, and in general the true value for this angle is likely to be substantially less than this). Second, the position and fixation of the probe are far less critical for successful measurement of velocities than for power.
There are some additional disadvantages associated with the use of P·V̄IWM as a flow index. First, small movements of the probe (and therefore sample volume) can cause some changes in the reflected power signal, which would then be wrongly interpreted as changes in flow. In the present study, this potential problem was avoided by always taking extreme care to identify accurately the center of the insonated vessel by maximizing the Doppler power signal and to secure firmly the probe and headpiece to minimize any noise in the power signal due to probe movements. Second, each protocol was repeated several times, and the results for each protocol were averaged, thereby minimizing any effect due to noise artifact, as well as the random error associated with the measurement of blood flow using Doppler ultrasound.20 Third, the use of P·V̄IWM requires the need to optimize and maximize both the reflected Doppler power signal and V̄P (or V̄IWM), and this can add a significant amount of time to assessments of cerebral hemodynamics with transcranial Doppler ultrasound. Despite these difficulties in measuring power, it can provide a useful check on whether there are changes in cross-sectional area in situations in which this might be suspected. In such conditions, the use of P·V̄IWM is preferable because this index should result in a constant estimation of flow.14
Turning to the two measurements for velocity, V̄P has been used much more widely than V̄IWM. The resting absolute values for V̄P in this study (Table 2⇑) are very similar to those published elsewhere,6 21 22 but for V̄IWM we have not managed to find published data for comparison. Most of the studies using transcranial Doppler ultrasound have elected to use V̄P instead of V̄IWM or P·V̄IWM as an index for flow because it has been suggested that V̄P is easier to measure accurately than V̄IWM.9 Indeed, to our knowledge, this is the first study to assess the relationship between V̄P and V̄IWM when cerebral blood flow is increased moderately during “pure” hypoxia or hypercapnia. Our results show that under such conditions both indexes accurately reflect changes in P·V̄IWM. However, in combined hypoxia and hypercapnia, the change in V̄P was significantly larger than the change in V̄IWM.
Despite the widespread use of V̄P, V̄IWM does involve fewer assumptions about the nature of the flow; furthermore, differences between the two indexes were detected experimentally in the combined stimulus protocol. Given the complex nature of the flow, it is difficult to say exactly why V̄P should change more than V̄IWM, but theoretically the change in V̄IWM is more likely to represent the true changes in flow. For established steady laminar flow in a rigid cylindrical tube, the ratio of V̄IWM to V̄P (V̄IWM/V̄P) is 0.5. Our results in combined hypoxia and hypercapnia give a value for this ratio of 0.63. Assuming an MCA diameter in the region of 0.3 cm,23 we can calculate Reynolds numbers in the region of 291 to 420 and 451 to 665 for V̄IWM and V̄P, respectively. Thus, the deviation from a ratio of 0.5 is likely to be due to bending of the flow at junctions, elastic walls, and nonsteady flow rather than any turbulence induced through a high Reynolds number (critical Reynolds number for turbulent flow is approximately 2000).
Results from the present study suggest that the caliber of the MCA does not change significantly under conditions of moderate hypoxia or hypercapnia alone; therefore, V̄P and V̄IWM appear to be acceptable as indexes of changes in cerebral blood flow. Further investigations are needed to evaluate the relationships between V̄P, V̄IWM, and P·V̄IWM to determine whether these results apply to other interventions. The use of the Doppler power signal will be useful in future investigations to assess conditions in which there are alterations in the caliber of the MCA.
Selected Abbreviations and Acronyms
|MCA||=||middle cerebral artery|
|P·V̄IWM||=||flow index (product of total power and intensity-weighted mean velocity)|
|V̄IWM||=||intensity-weighted mean flow velocity|
|V̄P||=||flow velocity spectral outline|
This study was approved by the Central Oxford Research Ethics Committee. It was supported by the Wellcome Trust and by an MRC (Canada) postdoctoral research fellowship (Dr Poulin). We wish to acknowledge the skilled technical assistance from David O'Connor and the volunteers for their participation in the study.
- Received June 5, 1996.
- Revision received August 13, 1996.
- Accepted August 13, 1996.
- Copyright © 1996 by American Heart Association
Poulin MJ, Liang P-J, Robbins PA. Dynamics of the cerebral blood flow response to step changes in end-tidal PCO2 and PO2 in humans. J Appl Physiol. 1996;81:1084-1095.
Swanson GD, Bellville JW. Step changes in end tidal CO2: methods and implications. J Appl Physiol. 1975;39:377-385.
Howson MG, Khamnei S, McIntyre ME, O'Connor DF, Robbins PA. A rapid computer controlled binary gas mixing system for studies in respiratory control. J Physiol (Lond). 1987;403:103P. Abstract.
Robbins PA, Swanson GD, Howson MG. A prediction correction scheme for forcing alveolar gases along certain time courses. J Appl Physiol. 1982;52:1353-1357.
Kleiser B, Scholl D, Widder B. Doppler CO2 and diamox test: decreased reliability by changes of the vessel diameter? Cerebrovasc Dis. 1995;5:397-402.
Dahl A, Lindegaard KF, Russell D, Nyberg-Hansen R, Rootwelt K, Sorteberg W, Nornes H. A comparison of transcranial Doppler and cerebral blood flow studies to assess cerebral vasoreactivity. Stroke. 1992;23:15-19.
Lindegaard KF, Lundar T, Wiberg J, Sojberg D, Aaslid R, Nornes H. Variations in middle cerebral artery blood flow investigated with non-invasive transcranial blood velocity measurements. Stroke. 1987;18:1025-1030.
Bishop CC, Powell S, Rutt D, Browse NL. Transcranial Doppler measurement of middle cerebral artery blood flow velocity: a validation study. Stroke. 1986;17:913-915.
Müller HR, Casty M. CO2 reactivity of middle cerebral artery truncal caliber. J Ultrasound Med. 1991;10:S47. Abstract.
Aaslid R, Newell DW, Stooss R, Sorteberg W, Lindegaard KF. Assessment of cerebral autoregulation dynamics from simultaneous arterial and venous transcranial Doppler recordings in humans. Stroke. 1991;22:1148-1154.
Aaslid R, Lindegaard KF, Sorteberg W, Nornes H. Cerebral autoregulation dynamics in humans. Stroke. 1989;20:45-52.
Müller HR, Casty M, Moll R, Zehnder R. Response of middle cerebral artery volume flow to orthostasis. Cerebrovasc Dis. 1991;1:82-89.