directional sensitivity of dynamic cerebral autoregulation
TRANSCRIPT
Directional sensitivity of dynamic cerebral autoregulation
1
2
Directional sensitivity of dynamic cerebral autoregulation 3
in squat-stand maneuvers 4
RB Panerai1,2, SC Barnes1, Nath M1,2, N Ball1, TG Robinson1, 2, VJ Haunton1,2 5
1Department of Cardiovascular Sciences, University of Leicester, Leicester, UK. 6
2National institute for Health Research (NIHR) Leicester Biomedical Research Centre, 7
University of Leicester, Leicester, UK. 8
9
10
Running head: Directional sensitivity of dynamic cerebral autoregulation 11
12
13
14
Corresponding author: 15
Ronney B. Panerai 16
Department of Cardiovascular Sciences 17
Room 439, Robert Kilpatrick Clinical Sciences Building 18
University of Leicester 19
PO Box 65 20
Leicester LE2 7LX 21
Email: [email protected] 22
Phone: 0116 252 3130 23 24
25
26
Directional sensitivity of dynamic cerebral autoregulation
Abstract 27
Dynamic cerebral autoregulation (CA), the transient response of cerebral blood flow (CBF) to 28
rapid changes in arterial blood pressure (BP), is usually modelled as a linear mechanism. 29
We tested the hypothesis that dynamic CA can display non-linear behavior resulting from 30
differential efficiency dependent on the direction of BP changes. CBF velocity (CBV, 31
transcranial Doppler), heart rate (HR, 3-lead ECG), continuous BP (Finometer) and end-tidal 32
CO2 (capnograph) were measured in 10 healthy young subjects during 15 squat-stand 33
maneuvers (SSM) with a frequency of 0.05 Hz. The protocol was repeated with a median 34
[IQR] of 44 [35-64] days apart. Dynamic CA was assessed with the autoregulation index 35
(ARI), obtained from CBV step responses estimated with an autoregressive moving-average 36
model. Mean BP, HR, and CBV were different (all p<0.001) between squat and stand, 37
regardless of visits. ARI showed a strong interaction (p<0.001) of SSM with the progression 38
of transients; in general, the mean ARI was higher for the squat phase compared to 39
standing. The changes in ARI were partially explained by concomitant changes in CBV 40
(p=0.023) and pulse pressure (p<0.001), but there was no evidence that ARI differed 41
between visits (p=0.277). These results demonstrate that dynamic CA is dependent on the 42
direction of BP change, but further work is needed to confirm if this finding can be 43
generalised to other physiological conditions, and also to assess its dependency on age, sex 44
and pathology. 45
46
Keywords – cerebral blood flow, transcranial Doppler, arterial blood pressure, posture, non-47
linear behavior 48
49
50
51
Directional sensitivity of dynamic cerebral autoregulation
Introduction 52
Cerebral blood flow (CBF) is tightly controlled by metabolic, myogenic and neurogenic 53
mechanisms, usually assumed to optimise brain perfusion in response to changes in arterial 54
blood pressure (autoregulation), arterial carbon dioxide (CO2 reactivity), and oxygen demand 55
(neurovascular coupling). The speed at which these regulatory mechanisms respond to 56
different stimuli can provide insight into the underlying physiology, and also potentially 57
provide diagnostic and prognostic information in clinical studies. 58
Both CO2 reactivity and neurovascular coupling (NVC) have been shown to be directional 59
mechanisms. In other words, when changing from normocapnia to hypercapnia, CBF 60
increases with a much longer time-constant than in the reverse direction (29). With NVC, 61
different types of sensorimotor and cognitive paradigms have been shown to induce faster 62
increases in CBF at the start of stimulation than that observed when stimulation ceases and 63
flow returns to its baseline level (1, 17, 23, 30). However, in the case of autoregulation, 64
particularly its dynamic response to sudden changes in blood pressure (BP), the presence of 65
directionality remains controversial. Using repeated compression and release of thigh cuffs, 66
Aaslid et al. (2) suggested the existence of directionality based on asymmetric changes in 67
critical closing pressure following increases or decreases in BP. On the other hand, 68
Katsogridakis et al. (16), using a similar protocol, could not identify any differences in 69
measures of dynamic CA when comparing increases with decreases in BP. In a previous 70
study, differences in cerebral blood velocity (CBV) step response could not be detected for 71
maneuvers that induce either a sudden reduction in BP (e.g. single thigh cuff deflation) or 72
increases in BP (e.g. cold stress test)(26). More recently, a study based on the squat-stand 73
maneuver (SSM) concluded that directionality, or ‘hysteresis’, was a feature of BP 74
autoregulation (CA). However, this was ascertained on somewhat ‘static’ estimates of the 75
CA response based on only two values of CBV and BP during either the stand or the 76
squatting phases of the maneuver (8). 77
Directional sensitivity of dynamic cerebral autoregulation
Determining if dynamic CA has a directional, or asymmetrical, behaviour is highly relevant 78
for both improving our understanding of the underlying physiology and potentially optimising 79
the current methods of assessment (13, 37). If dynamic CA is proven to show asymmetry in 80
relation to BP changes, this will impact on the current modelling techniques used to quantify 81
dynamic CA in health and disease, as methods will be required to take into account the 82
directionality of BP changes. On teleological grounds, a greater sensitivity of dynamic CA to 83
increases in BP may confer an evolutionary advantage given the need to protect the 84
microcirculation from surges in BP (21). On the other hand, the argument could be reversed 85
when considering the importance of avoiding syncope following acute hypotension. 86
The SSM is a suitable protocol to investigate the directional sensitivity of CA, given the 87
relatively large changes induced in BP. SSMs provide a better signal-to-noise ratio than 88
corresponding measurements with either spontaneous BP fluctuations or repeated thigh cuff 89
maneuvers (15, 20, 32). On the other hand, SSMs induce considerable changes in PaCO2 90
and other peripheral cardiovascular parameters (5) which could contribute, or even explain, 91
the apparent directionality reported by Brassard et al. (8). To address these concerns, we 92
used a modelling approach more suitable to quantify dynamic CA, to test three inter-related 93
hypotheses: i) that the validated and widely used autoregulation index (ARI), has higher 94
values during squatting as compared to the standing phase of SSMs, ii) that differences in 95
ARI between squatting and standing are reproducible in the same group of subjects, and iii) 96
that differences in ARI between squatting and standing can be explained by corresponding 97
changes in other cerebrovascular or peripheral parameters. 98
99
100
101
Directional sensitivity of dynamic cerebral autoregulation
Methods 102
A subset of data collected for a wider investigation (4) was analysed to test the specific 103
hypotheses of this study. The study was approved by the University of Leicester Ethics 104
Committee (reference 8442-vjh12-cardiovascularsciences). Healthy young subjects (<30 105
years old) provided written informed consent. 106
Physiological measurements 107
Beat-to-beat estimates of BP were obtained by arterial volume-clamping of the digital artery 108
(Finometer, FMS, Amsterdam, Netherlands). Transcranial Doppler CBV was measured in 109
the middle cerebral arteries (MCA) using 2MHz probes (Viasys Companion III) held in place 110
by a custom-built headset. A tilt-sensor was attached to the subject’s right thigh 20cm above 111
the superior border of the patella to measure the angle of the squatting motion. Nasal 112
capnography (Salter labs, ref 4000) was used to measure end-tidal CO2 (EtCO2). Heart rate 113
was measured using three-lead ECG. The right hand was held in position with a sling to 114
minimise movement throughout the recordings and to keep the BP finger cuff at heart height. 115
The servo-reset mechanism of the Finometer was disabled throughout the recordings to 116
allow for a continuous BP trace, but enabled between recordings. Intermittent brachial BP 117
was measured using a validated electrosphygmomanometer (UA 767 BP monitor) to 118
calibrate the Finometer recordings. 119
Continuous analogue recordings were digitised at 500 samples/s by a Physiological Data 120
Acquisition System (PHYSIDAS) designed by the Leicester Medical Physics Department for 121
subsequent analysis. 122
Experimental protocol 123
Experiments were performed in a well-lit, environmentally controlled, laboratory that was free 124
from distraction and maintained at a temperature of 20-24⁰C. Participants were asked to 125
avoid strenuous exercise, caffeine, smoking, large meals and alcohol in the four hours prior 126
to their visit. 127
Directional sensitivity of dynamic cerebral autoregulation
Recordings were performed following a 5-min period of rest standing. For the purposes of 128
this study, a five minute baseline recording was performed with the subject in the standing 129
position, followed by fixed frequency SSM at a frequency of 0.05Hz, corresponding to 15 130
squats and stands, each with a duration of 10 s. A computer program provided visual cues to 131
guide the timing of the squatting motion. A period of instruction and practice preceded the 132
SSM recordings. When performing the SSMs, subjects were instructed to squat down as low 133
as they felt able. They were informed that they would need to perform 15 squats, and to take 134
this into account when choosing their depth. Throughout each recording, subjects were 135
asked to breathe through their nose and to avoid a Valsalva-like maneuver during the SSM. 136
Subjects were invited back into the laboratory at a subsequent date for repeated 137
measurements following the same protocol. 138
Data processing 139
The readings from the Finometer were calibrated to the brachial BP recordings. Data were 140
visually inspected; non-physiological spikes in CBV were removed through linear 141
interpolation. Files that contained any segments of significantly poor TCD signals were 142
excluded from further analyses. Narrow spikes (<100 ms) and artefacts were removed by 143
linear interpolation. Subsequently, all signals were filtered in the forward and reverse 144
direction using an eighth-order Butterworth low-pass filter with a cut-off frequency of 20 Hz. 145
The beginning and the end of each cardiac cycle were detected in the BP signal, and mean 146
values of BP, CBV and heart rate were obtained for each heartbeat. EtCO2 was obtained for 147
each breath, linearly interpolated and resampled to coincide with each cardiac cycle. Beat-148
to-beat parameters were interpolated with a third-order polynomial and resampled at 5 Hz to 149
generate signals with a uniform time base. 150
Under visual inspection, the beginning of each squat and stand phases of the SSM were 151
marked and the resulting transient changes in BP, CBV, heart rate, EtCO2, and pulse 152
pressure (PP) were extracted for further analyses. This corresponded to 15 squat and 15 153
Directional sensitivity of dynamic cerebral autoregulation
standing transients for each subject. Although each phase lasted 10 s, the extracted 154
transient data were extended to 12 s to allow for the transition phase when moving from 155
squat to stand and vice-versa. The same time interval was adopted to average 156
corresponding values of BP, HR, EtCO2, CBV and pulse pressure (PP). The dynamic 157
relationship between BP and CBV for each transient was modelled using an autoregressive-158
moving average (ARMA) structure as previously described (14, 27). The CBV step response 159
to a sudden change in BP was compared with 10 template curves proposed by Tiecks et al. 160
(35) and the best fit curve corresponded to the ARI. Values of ARI = 0 indicate absence of 161
CA, whilst ARI = 9 corresponds to the most efficient CA that can be observed (35). This 162
procedure generated a value of ARI for each of the 15 squat and 15 standing transients in 163
each subject. The quality of the CBV step response fitting to Tiecks et al. model was 164
expressed by the normalised mean square error (NMSE). As previously described, NMSE 165
values ≤ 0.3 guarantee CBV step responses that are physiologically plausible (28). 166
167
Statistical analysis 168
For each transient, ARI values were rejected for NMSE>0.3 and were replaced by the mean 169
value of the preceding and subsequent transients. We modelled the data on ARI along the 170
15 transients of a subject during SSMs for both visits by a repeated measures model. We 171
developed the model in two stages. First, we identified the correlation structure between the 172
repeated measures within the same subject; we observed that the second order 173
autoregressive process along with the first order moving average represented the 174
appropriate correlation structure for the data. In the second step, we developed a multi-175
variable linear model incorporating categorical variables: maneuvers (two levels: squat and 176
standing), transients (15 levels) and visit (two levels) as predictors. The model also included 177
a two-way interaction effect of maneuvers and transients. In addition, we explored the 178
association of other covariates with ARI, namely MAP, HR, EtCO2, CBFV and PP. Using a 179
stepwise model selection based approach, we only retained the covariates that were 180
Directional sensitivity of dynamic cerebral autoregulation
statistically significant (p<0.05). The final model included CBV and PP as covariates. We 181
used Akaike's information criterion (AIC) and Bayesian information criterion (BIC) to perform 182
the model selection in both stages. For other cerebral autoregulation (CA) covariates, 183
namely MAP, HR, EtCO2, CBV and PP, we investigated the effect of visit and maneuver 184
using similar repeated measures models accounting for the correlation between 185
measurements within the same subject. All statistical tests were two-sided with type 1 error 186
rate (p-value) of 0.05 to determine statistical significance. The fitting of a repeated measures 187
model was carried out using the R package nlme in R software environment (version 3.4). 188
189
190
191
Directional sensitivity of dynamic cerebral autoregulation
Results 192
Measurements on two separate occasions were performed in 11 subjects (2 females) with a 193
median [IQR] interval of 44 [35-64] days between visits. One male subject had poor 194
recordings of BP with the Finometer in his second visit to the laboratory and was removed 195
from the study. Good quality recordings with bilateral values of CBV were obtained in the 196
other 10 subjects, aged 22 ± 1 years. Standing at rest (‘baseline’) did not show any 197
differences between visits for the parameters in Table 1. With the exception of EtCO2, all 198
other parameters showed highly significant differences between the squat and stand phases 199
of the maneuver (Table 1). 200
The NMSE was >0.3 in 34 (5.7%) transients for the right MCA and 25 (4.1%) transients from 201
the left MCA. An example of a transient with NMSE>0.30 is shown in Fig. 4 (transient #10). 202
These outliers were interpolated as described above; they represented a small fraction of the 203
600 transients analysed for each hemisphere and were randomly distributed across visits, 204
subjects and maneuvers. Mean ± SD values of NMSE (<0.30) for the right and left MCAs for 205
both visits were 0.094 ± 0.050 and 0.095 ± 0.048, respectively. 206
No significant differences were found between values of CBV and ARI for the right and left 207
MCAs either at baseline or for the squatting and standing phases of SSM; mean values for 208
the two hemispheres were used in all subsequent analyses. 209
The CBV transients during both squatting and standing showed temporal patterns consistent 210
with an active CA response to the corresponding BP rise for the 15 squatting (Figs. 1 & 2) 211
and BP drop for the 15 standing (Figs. 3 & 4) responses from a single subject. Similar 212
behaviour was observed in all the other subjects. 213
Mean BP, PP and HR, but not EtCO2, showed marked differences between squatting and 214
standing regardless of visits (Fig. 5). Similar behavior was observed for CBV and ARI as 215
shown in Fig. 6. Most parameters showed a different behavior at the beginning of SSM, as 216
compared to steady-state values (Figs. 5 & 6), but this pattern was more accentuated for the 217
Directional sensitivity of dynamic cerebral autoregulation
ARI (Figs. 6.C & 6.D). The repeated measures model of ARI showed a strong interaction 218
effect (p<0.0001) of SSM with the progression of transients, indicating that the differences 219
between squat and stand were also dependent on the progression of the maneuvers, as 220
observed in Fig. 6. The fraction of the total variance explained by these variables (maneuver, 221
transients and visits) was expressed by R2=0.254 (p<0.0001). The ARI differences were also 222
associated with changes in CBV, which increased R2 to 0.356 (p=0.023) and PP, which 223
increased R2 to 0.414 (p=0.0002). HR was only borderline significant, increasing R2 to 0.441 224
(p=0.073), but there was no evidence that ARI differences were associated with EtCO2 225
(p=0.50) or BP (p=0.86). The estimates of slopes (mean ± SE) of CBV, PP and HR with 226
transients were -0.0254 ± 0.0111 s.cm-1, -0.0347 ± 0.0094 mmHg-1 and 0.0156 ± 0.0084 227
bpm-1, respectively. There was no evidence that mean ARI differed between visits (p=0.277). 228
229
230
Discussion 231
Using the autoregulation index (ARI), a well-established method to quantify the strength of 232
dynamic CA (32), we have confirmed the recent results of Brassard et al. (8), suggesting that 233
the squatting phase of SSM has a more efficient CA response than the corresponding 234
standing phase. However, by uniquely analysing the time-series of ARI and other cerebral 235
and systemic parameters (Figs. 5 & 6), we have also shown that the initial maneuvers, of a 236
total sequence of 15 squats and stands, induced cerebro- and cardiovascular changes that 237
were unstable, in comparison with the steady-state behavior observed with the subsequent 238
transients (Figs. 5 & 6). Of considerable relevance, we have observed that the changes in 239
ARI and other parameters were highly reproducible at a second visit to the laboratory, 240
approximately six weeks apart, thus suggesting that directional sensitivity is a consistent 241
phenomenon, not related to the cognitive reaction to a new procedure. When considering the 242
potential contribution of other parameters to help explain the differences in ARI between the 243
Directional sensitivity of dynamic cerebral autoregulation
squatting and standing phases of the maneuvers, we found that both CBV and PP had 244
significant effects, contrary to BP, EtCO2 and HR. From these results, we have been able to 245
confirm the three main hypotheses of the study, as listed in the Introduction. These findings 246
stimulate a debate about the underlying mechanisms controlling CBF in humans and 247
strongly suggest the need to reassess current techniques for modelling dynamic CA. 248
249
Physiological considerations 250
Further studies are needed to confirm our findings and to establish if the directionality of 251
dynamic CA is age dependent. CBV and PP changes had a significant effect on explaining 252
the variance of the ARI differences between squatting and standing (ΔARISQ-ST). When the 253
borderline contribution of HR was included, there was a 73.6 % increase in R2 (from 0.254 to 254
0.441). However, variance estimates do not reflect the sign of the slopes (β coefficients) of 255
each co-variate. In Figs. 5 & 6, it is possible to estimate that, from the 7th to the 15th 256
maneuver, when most variables showed a more steady behaviour, changes in ARI, CBV, PP 257
and HR, from stand to squat, were approximately 2 (ARI units), 20 cm/s, -10 mmHg and -15 258
bpm, respectively. Based on the slopes of the CBV and PP given above, the CBV effects 259
would correspond to attenuation of the ARI change of approximately -0.5 (ARI units), with a 260
corresponding increase of 0.35 (ARI units) for the PP contribution, and – 0.23 (ARI units) 261
due to HR, if that is also included. Therefore, neither of these effects would explain the 262
overall changes of the order of 2 ARI units, and, moreover, they show opposite influences 263
that tend to cancel each other. Therefore, despite their significant influence on ΔARISQ-ST, 264
differences, CBV and PP, and HR, if also considered, have a minor effect in explaining the 265
more steady changes in ARI from stand to squat observed from the 7th to the 15th transient. 266
Associations derived from the repeated-measures linear model do not provide reliable 267
evidence of causality, and it is possible that other underlying variables could be mediating 268
these statistical associations. Several reasons point towards sympathetic neural activity 269
(SNA) as a strong candidate for this role (9). 270
Directional sensitivity of dynamic cerebral autoregulation
Previous studies that could not detect asymmetry of dynamic CA (16, 26), involved BP 271
changes that were much smaller than in our case (Table 1, Fig. 5) and other studies where 272
asymmetry was reported (8, 36). Although Aaslid et al. (2), concluded in favor of asymmetry 273
in CA, despite relatively smaller changes in BP, this was only found in severe head injury 274
patients, and not in their healthy volunteers. Directionality of CA in head injury patients was 275
also reported using the Mx index (31). In these two studies, it is possible that directionality 276
was caused by the underlying pathology, with different mechanisms compared to healthy 277
subjects, including exaggerated SNA. On the other hand, the lack of association between 278
ΔARISQ-ST and ΔBPSQ-ST suggests that the amplitude of the BP change might not be the 279
main determinant of directionality. Alternatively, it is possible to speculate that the rate of 280
change of BP, usually termed rate-sensitivity, could be the dominant factor responsible for 281
the asymmetry in dynamic CA. One key study in this case would be to perform SSM with a 282
modified protocol where the change from squat-to-stand, and vice-versa, would be 283
performed gradually, instead of the rapid change in posture traditionally used in SSM (5, 8, 284
12, 32, 34, 36). Two previous studies though suggest that directionality might not be the 285
result of a rapid rate of change in BP; as it was also observed following relatively slower 286
changes in BP induced by phenylephrine and nitroprusside in healthy subjects (36), and it 287
was again noted in a review of 40 previous studies of static CA which concluded that CA 288
was more efficient for increases (26 studies) than decreases (23 studies) in mean BP (24). 289
As suggested by these investigators, directional sensitivity could result from SNA activation 290
as described by Casaglia et al. (11), showing increases in SNA before surges in BP, but not 291
before drops in BP. In fact, it is possible to see a clear distinction in SNA temporal patterns 292
in their Fig. 3 (11), showing much faster rates of rise in SNA than the rate of fall in the return 293
to baseline. Although the influence of sympathetic activation could be manifested in heart 294
rate, the lack of association between ΔARISQ-ST and ΔHRSQ-ST (p=0.07) cannot be used to 295
dismiss the involvement of SNA as the root cause of directionality for two main reasons. 296
First, as indicated in Fig. 5, it is more likely that HR changes were dominated by the 297
baroreflex and, second, sympathetic cardiac activity cannot be accepted as representative of 298
Directional sensitivity of dynamic cerebral autoregulation
arteriolar sympathetic control due to the well-known regional differences in SNA. Gradual 299
changes in SNA during the 15 distinct SSM could also help to explain the non-stationary 300
behavior of CBV and ARI in (Fig. 6), where differences between values at squat and stand 301
were not noticeable at the beginning, but became more established with the progression of 302
the repeated maneuvers. 303
The recent finding that systolic BP values during SSM reflect a more efficient dynamic CA 304
than corresponding diastolic values could also help to explain the directional sensitivity of the 305
ARI observed in our study (34). As suggested by the authors (34), greater efficiency of 306
dynamic CA might protect the brain from the larger values of systolic pressure that could 307
lead to hyperperfusion and haemorrhage. Considering that systolic BP values are much 308
larger during squatting compared to standing, this greater efficiency of the cerebral 309
circulation to counteract changes in systolic pressure could explain our main findings, 310
including the association of PP with the ΔARISQ-ST. 311
In summary, further research is needed to explain the physiological mechanism responsible 312
for the directional sensitivity of the CA dynamic response. One interesting possibility would 313
be to investigate if directionality is already present in the myogenic response of isolated 314
cerebral arteries. 315
316
Clinical implications 317
Previous applications of the repeated SSM in physiological and clinical studies have 318
extracted parameters from the entire recording (4, 5, 6, 12, 18, 32, 33, 34).The more detailed 319
information provided by this study suggests that more robust results might be obtained by 320
removing the first 2 or 3 maneuvers from the analysis (Fig. 6). Further work is needed to 321
determine if an even smaller number of maneuvers, say from #3 to #9 (Figs. 5 & 6) would 322
suffice to provide reliable data. 323
Directional sensitivity of dynamic cerebral autoregulation
Translation of current knowledge of the pathophysiology of dynamic CA to clinical practice 324
has been hindered by concerns about the reliability of existing parameters and protocols 325
adopted in its assessment (9, 23, 25). The main motivation for adopting SSM for assessment 326
of dynamic CA is the improved quality of estimates, as indicated by values of coherence 327
approaching 1.0 and the excellent reproducibility achieved within several days apart (4, 32). 328
As discussed below, the presence of directionality in CBF control requires some re-329
evaluation of current methodology for assessment of dynamic CA, but it also provides 330
potential opportunities to improve the sensitivity and specificity of diagnostic/prognostic 331
techniques, as suggested by studies that identified asymmetry of CA in patients with head 332
injury, despite relatively low amplitude of BP changes (2, 31). Given the amplitude of the ARI 333
differences observed in volunteers (Fig. 6), it is tempting to speculate that having two 334
measures of CA in each patient, from a single SSM, might be more pathognomonic in 335
different clinical conditions. Clearly, much more work is needed to test these different 336
hypotheses. 337
338
Assessment of dynamic autoregulation 339
Current techniques for assessment of dynamic CA can be classified as directional or non-340
directional. The former is represented by methods such as thigh cuff deflation, that induces 341
only a sudden drop in BP and the latter, by the widely used approach based on spontaneous 342
fluctuations in BP, where transient increases and reductions in BP are both present and 343
cannot be easily separated. Even with protocols that can induce clearly identifiable increases 344
and reductions in BP, such as the SSM, and the related sit-to-stand approach, modelling has 345
usually been performed with transfer function analysis (TFA) that does not distinguish 346
positive from negative changes in BP (13). If the asymmetry of dynamic CA is confirmed by 347
further studies, it will be important to move towards modelling techniques that could reflect 348
the directional sensitivity we, and others, have observed. As demonstrated in this study, 349
SSM combined with ARMA modelling would be a strong possibility, but alternatives are also 350
Directional sensitivity of dynamic cerebral autoregulation
needed to allow investigation of dynamic CA in patients that cannot perform the SSM, or in 351
physiological studies where the protocol would not allow SSMs simultaneously with other 352
physiological challenges. 353
Compared to previous studies where ARMA modelling was adopted to estimate dynamic CA 354
parameters (3, 19), our study also showed innovation in the use of segments of data only 12 355
s long, as compared to 60 s or more used for analysis of time-varying estimates of ARI (14, 356
27). Although these relatively short segments would be expected to produce unreliable 357
estimates, surprisingly, the NMSE of CBV step responses from 1200 transients led to 358
rejection of only 59 estimates of ARI, that had to be replaced by interpolated values (28). 359
Undoubtedly, these exceptional results were possible due to the large changes in BP caused 360
by SSM and the corresponding high signal-to-noise ratios of the BP and CBV transients 361
(Figs. 1 & 3). 362
Previous application of TFA to SSM data, yielded estimates of ARI that were significantly 363
lower during SSM as compared to baseline standing (4). In the present study though, it is not 364
possible to directly compare values obtained for the squat or standing phases of SSM with 365
the baseline period, given the different modelling approaches used in each case, to suit the 366
much longer data segments available during baseline and the much reduced amplitude of 367
BP spontaneous fluctuations at rest. For this reason, it might be more prudent if future 368
studies report on multiple estimates of ARI, or other indices of dynamic CA, such as the 369
phase difference between CBV and BP (13), to provide a more comprehensive picture of the 370
effects of physiological changes, or disease, on parameters at rest as well as during the 371
separate phases of squatting and standing. Similar considerations would apply to the use of 372
the sit-to-stand maneuver. 373
Limitations of the study 374
Changes in CBV can reflect corresponding changes in CBF if the diameter of the insonated 375
artery (MCA) remains constant. Changes in MCA diameter, due to either squatting or 376
Directional sensitivity of dynamic cerebral autoregulation
standing, could distort values of CBV changes, but, it would be very unlikely that the change 377
in MCA diameter would present a temporal pattern that would nullify the differences in ARI 378
values observed between the squatting and standing phases of the maneuver. 379
Our study was not designed to address potential effects of sex on the directionality of 380
dynamic CA, but this important aspect of phenotype deserves further investigation. In our 381
population, only two subjects were females and their results were not dissimilar from those 382
of the majority of male subjects. 383
Finally, our conclusions were derived from CBV measurements in the MCA and cannot be 384
generalised to other large intra-cerebral arteries like the PCA or ACA. 385
386
Perspectives and Significance 387
Better efficiency of dynamic CA, following relatively large changes in BP resulting from the 388
squat maneuver, in comparison with drops in BP resulting from standing up, could be 389
partially explained by corresponding differences CBV and PP, but not by absolute BP levels 390
or other parameters, such as HR, or EtCO2, but were highly reproducible in repeated 391
measurements taken several weeks apart. The underlying physiological mechanisms 392
responsible for the directional sensitivity of dynamic CA need further investigation, including 393
the potential involvement of sympathetic nervous system activation during increases in BP. 394
Clinical applications of dynamic CA assessment might benefit from the additional information 395
that can be derived with new analytical methods that are able to discriminate between the 396
autoregulatory response to increases or reductions in BP. Further work is also needed to 397
investigate the influence of age, sex and pathology on the directional sensitivity of dynamic 398
CA. 399
400
401
Directional sensitivity of dynamic cerebral autoregulation
ACKNOWLEDGEMENTS 402
We would like to thank the volunteers for the time dedicated to the study. TGR is an NIHR 403
Senior Investigator. 404
405
GRANTS 406
None to declare 407
408
DISCLOSURES 409
No conflicts of interest, financial or otherwise, are declared by the author(s). 410
411
AUTHOR CONTRIBUTIONS 412
V.J.H., T.G.R. and R.B.P. designed and planned study; S.C.B., N.B. and V.J.H. set up 413
experiment and protocol; S.C.B. and N.B. performed data collection; R.B.P. wrote dedicated 414
software; S.C.B., R.B.P. and M.N. performed data analysis; R.B.P. drafted manuscript; N.B., 415
S.C.B., V.J.H., T.G.R, R.B.P. and M.N. revised and approved final version of the manuscript. 416
417
418
419
Directional sensitivity of dynamic cerebral autoregulation
REFERENCES 420
1. Aaslid R. Visually evoked dynamic blood flow response of the human cerebral circulation. 421
Stroke 1987;18:771-5. 422
423
2. Aaslid R, Blaha M, Sviri G, Douville C, Newell DW. Asymmetric dynamic cerebral 424
autoregulatory response to cyclic stimuli. Stroke 2007;38:1465-9. 425
426
4. Barnes SC, Ball N, Haunton VJ, Panerai RB, Robinson TG. Randon squat-stand maneuvers: a 427
novel approach for assessment of dynamic cerebral autoregulation? J Applied Physiol 428
2017;123:558-566. 429
430
5. Barnes SC, Ball N, Haunton VJ, Robinson TG, Panerai RB. The cerebro-cardiovascular 431
response to periodic squat-stand maneuvers in healthy subjects: a time-domain analysis. Am 432
J Physiol Heart Circ Physiol 2017;313:H1240-H1248. 433
434
6. Birch AA, Dirnhuber MJ, Hartley-Davies R, Iannotti F, Neil-Dwyer G. Assessment of 435
autoregulation by means of periodic changes in blood pressure. Stroke 1995;26:834-7. 436
437
7. Bland JM, Altman DG. Statistical methods to assess agreement between two methods of 438
clinical measurement. The Lancet 1986;1:307-310. 439
440
8. Brassard P, Ferland-Dutil H, Smirl JD, Paquette M, Le Blanc O, Malefant S, Ainslie PN. 441
Evidence for hysteresis in the cerebral pressure-flow relationship in healthy men. Am J 442
Physiol Heart Circ Physiol 2017;312:H701-H704. 443
444
9. Brassard P, Tymko MM, Ainslie PN. Sympathetic control of the brain circulation: 445
appreciating the complexities to better understand the controversy. Autonon Neurosci 446
2017;207:37-47. 447
448
10. Brodie FG, Atkins ER, Robinson TG, Panerai RB. Reliability of dynamic cerebral 449
autoregulation measurements using spontaneous fluctuations in blood pressure. Clin Sci 450
2009;116:513-520. 451
452
Directional sensitivity of dynamic cerebral autoregulation
11. Cassaglia PA, Griffiths RI, Walker AM. Cerebral sympathetic nerve activity has a major 453
regulatory role in the cerebral circulation in REM sleep. J Applied Physiol 2009;106:1050-454
1056. 455
12. Claassen JAHR, Levine BD, Zhang R. Dynamic cerebral autoregulation during repeated squat-456
stand maneuvers. J Applied Physiol 2009; 106:153-160. 457
458
13. Claassen JAHR, Meel-van den Abeelen ASS, Simpson DM, Panerai RB. Transfer function 459
analysis of dynamic cerebral autoregulation: a white paper from the International 460
Autoregulation Research Network (CARNet). J Cereb Blood Flow Metab 2016;36:665-680. 461
462
14. Dineen NE, Brodie FG, Robinson TG, Panerai RB. Continuous estimates of dynamic cerebral 463
autoregulation during transient hypocapnia and hypercapnia. J Applied Physiol 464
2010;108:604-613. 465
466
15. Katogridakis E, Bush G, Fan L, Birch AA, Simpson DM, Allen R, Potter JF, Panerai RB. 467
Detection of impaired cerebral autoregulation improves by increasing arterial blood 468
pressure variability. J Cereb Blood Flow Metab 2013;33:519-23. 469
470
16. Katsogridakis E, Simpson DM, Bush G, Fan L, Allen R, Potter JF, Panerai RB. Coherent 471
averaging of pseudorandom binary stimuli: is the dynamic cerebral autoregulatory response 472
symmetrical? Physiol Meas 2017; 38:2164-75. 473
474
17. Kennerly AJ, Harris S, Bruyns-Haylett M, Boorman L, Zheng Y, Jones M, Berwick J. Early and 475
late stimulus-evoked cortical hemodynamic responses provide insight into the neurogenic 476
nature of neurovascular coupling. J Cereb Blood Flow Metab 2012;32:468-80. 477
478
18. Labrecque L, Rahimaly K, Imhoff S, Paquette M, Le Blanc O, Malefant S, Lucas SJE, Bailey 479
DM, Smirl JD, Brassard P. Diminished dynamic cerebral autoregulatory capacity with forced 480
oscillations in mean arterial pressure with elevated cardiorespiratory fitness. Physiol Rep 481
2017;5(2) pii:e13486. 482
483
19. Liu Y, Birch AA, Allen R. Dynamic cerebral autoregulation assessment using an ARX model: 484
comparative study using step response and phase shift analysis. Med Eng Phys 2003;25:647-485
653. 486
Directional sensitivity of dynamic cerebral autoregulation
20. Liu J, Simpson DM, Allen R. High spontaneous fluctuations in arterial blood pressure 487
improves the assessment of cerebral autoregulation. Physiol Meas 2005;26:725-41. 488
489
21 Lucas SJE, Cotter JD, Brassard P, Bailey DM. High-intensity interval exercise and 490
cerebrovascular health: curiosity, cause, and consequence. J Cereb Blood Flow Metab 491
2015;35:902-911. 492
493
22. Mahony P, Panerai RB, Deverson ST, Hayes PD, Evans DH. Assessment of the thigh cuff 494
technique for measurement of dynamic cerebral autoregulation. Stroke 2000;31:476-480. 495
496
23. Moody M, Panerai RB, Eames PJ, Potter JF. Cerebral and systemic hemodynamic changes 497
during cognitive and motor activation paradigms. Am J Physiol Regul Integr Comp Physiol 498
2005; 288:R1581-R1588. 499
500
24. Numan T, Bain AR, Hoiland RL, Smirl JD, Lewis NC, Ainslie PN. Static autoregulation in 501
humans: a review and reanalysis. Med Eng Phys 2014;36:1487-1495. 502
503
25. Ortega-Guttierrez S, Petersen N, Masurkar A, Reccius A, Huang A, Li M, Choi JH, Marshall 504
RS. Reliability, asymmetry, and age influence on dynamic cerebral autoregulation measured 505
by spontaneous fluctuations of blood pressure and cerebral blood flow velocities in healthy 506
individuals. J Neuroimaging 2014;24:379-386. 507
508
26. Panerai RB, Dawson SL, Eames PJ, Potter JF. Cerebral blood flow velocity response to 509
induced and spontaneous sudden changes in arterial blood pressure. Am J Physiol Heart Circ 510
Physiol 2001;280:H2162-H2174. 511
512
27. Panerai RB, Eames PJ, Potter JF. Variability of time-domain indices of dynamic cerebral 513
autoregulation. Physiol Meas 2003;24:367-381. 514
515
28. Panerai RB, Haunton VJ, Salinet ASM, Robinson TG. Statistical criteria for estimation of the 516
cerebral autoregulation index (ARI) at rest. Physiol Meas 2016;37:661-680. 517
518
29. Poulin MJ, Liang PJ, Robbins PA. Dynamics of the cerebral blood flow response to step 519
changes in end-tidal PCO2 and PO2 in humans. J Applied Physiol 1996;81:1084-95. 520
Directional sensitivity of dynamic cerebral autoregulation
521
30. Salinet ASM, Robinson TG, Panerai RB. Active, passive, and motor imagery paradigms: 522
component analysis to assess neurovascular coupling. J Applied Physiol 2013;114:1406-12. 523
524
31. Schmidt B, Klingelhofer J, Perkes I, Czosnyka M. Cerebral autoregulatory response depends 525
on the direction of change in perfusion pressure. J Neurotrauma 2009;26:651-656. 526
527
32. Smirl JD, Hoffman K, Tzeng YC, Hansen AE, Ainslie PN. Methodological comparison of 528
active- and passive-driven oscillations in blood pressure: implications for the assessment of 529
cerebral pressure-flow relationships. J Applied Physiol 2015;119:487-501. 530
531
33. Smirl JD, Lucas SJE, Lewis NCS, duManior GR, Smith KS, Bakker A, Basnyat AS, Ainslie PN. 532
Cerebral pressure-flow relationship in lowlanders and natives at high altitude. J Cereb Blood 533
Flow Metab 2014;34:248-257. 534
535
34. Smirl JD, Wright AD, Ainslie PN, Tzeng YC, van Donkelaar P. Differential systolic and 536
diastolic regulation of the cerebral pressure-flow relationship during squat-stand 537
manoeuvres. Acta Neurochir Suppl 2018;126:263-268. 538
539
35. Tiecks FP, Lam AM, Aaslid R, Newell DW. Comparison of static and dynamic cerebral 540
autoregulation measurements. Stroke 1995;26:1014-1019. 541
542
36. Tzeng YC, Willie CK, Atkinson G, Lucas SJE, Wong A, Ainslie PN. Cerebrovascular regulation 543
during transient hypotension and hypertension in humans. Hypertension 2010;56:268-273. 544
545
37. Xiong L, Liu X, Shang T, Smieleswski P, Donnelly J, Guo Z, YangY, Leung T, Czosnyka M, 546
Zhang R, Liu J, Wong KS. Impaired cerebral autoregulation: measurement and application to 547
stroke. J Neurol Neurosurg Psych 2017;88:520-31. 548
549
550
551
552
553
Directional sensitivity of dynamic cerebral autoregulation
Tables 554
555
556
Table 1. Population mean ± SD for systemic and cerebral hemodynamic parameters. 557
Parameter Visit 1 Visit 2
Baseline Squat Stand Baseline Squat Stand p-value
Visit p-value
SSM
Mean BP (mmHg) 96.5 ± 6.4 111.2 ± 18.9 89.7 ± 14.4 90.0 ± 17.2 109.0 ± 9.5 88.0 ± 4.7 0.71 <0.0001
Heart rate (bpm) 87.2 ± 7.2 87.4 ± 9.7 103.8 ± 11.7 93.7 ± 14.1 92.8 ± 10.6 107.9 ± 12.6 0.35 <0.0001
EtCO2 (mmHg) 36.4 ± 2.1 41.3 ± 3.7 40.9 ± 3.4 37.3 ± 2.6 41.5 ± 3.8 41.3 ± 3.3 0.35 <0.0001
PP (mmHg) 44.8 ± 6.8 65.2 ± 14.8 72.9 ± 15.4 32.0 ± 13.6 59.1 ± 12.5 68.5 ± 15.0 0.51 0.0087
CBV (cm.s-1
) 55.8 ±10.1 71.6 ± 14.3 53.2 ± 8.2 53.6 ± 8.0 67.4 ± 13.1 48.4 ± 7.2 0.15 <0.0001
558
BP, mean arterial blood pressure; EtCO2, end-tidal CO2; PP, pulse pressure; CBV, cerebral blood velocity. 559
Directional sensitivity of dynamic cerebral autoregulation
Figure legends 560
561
Figure 1 – Representative arterial blood pressure (continuous line, mmHg) and cerebral 562
blood velocity (dashed line, cm.s-1) for 15 squat maneuvers performed by a 21 year-old 563
female subject. Baseline values at t=0 s were removed for both signals. 564
Figure 2 – Cerebral blood velocity step response (continuous line) estimated with ARMA 565
model for the arterial blood pressure and cerebral blood velocity transients in Fig. 1 and best 566
fit Tiecks model (dashed line) for 15 squat maneuvers. Responses are in cm.mmHg-1.s-1. 567
Corresponding values of ARI ranged from 4.9 (transient 1) to 7.5 (transient 3) with median 568
[IQR] of 6.1 [5.9-6.6]. 569
Figure 3 – Corresponding changes in arterial blood pressure (continuous line, mmHg) and 570
cerebral blood velocity (dashed line, cm.s-1) for 15 standing maneuvers performed for the 571
same 21 year-old female subject with data presented in Fig. 1. Baseline values at t=0 were 572
removed from both signals. 573
Figure 4 – Cerebral blood velocity (CBV) step response (continuous line) estimated with 574
ARMA model for the arterial blood pressure and CBV transients in Fig. 3 and best fit Tiecks 575
model (dashed line) for 15 standing maneuvers. Responses are in cm.mmHg-1.s-1. 576
Corresponding values of ARI ranged from 0 (transient 10) to 7.0 (transient 2). The 577
normalised mean square error for fitting Tiecks model to the CBV step response for transient 578
#10 was above the threshold of 0.30 (see Methods) and the corresponding value of ARI was 579
then replaced by the average ARI of transients #9 and #11. The median [IQR] for the 15 580
transients was 5.2 [3.1-6.0]. 581
Figure 5 – Population averages for each of the 15 transient responses during squat (circles, 582
continuous line) and standing (squares, dashed line) maneuvers performed at 0.05 Hz. Error 583
bars represent ±1 SE.( A,B) Mean arterial blood pressure; (C,D) Heart rate; (E,F) End-tidal 584
CO2; (G,H) Pulse pressure. Visit 1 (A,C,E,G); visit 2 (B,D,F,H). 585
Figure 6 – Population averages for each of the 15 transient responses during squat (circles, 586
continuous line) and standing (squares, dashed line) maneuvers performed at 0.05 Hz. Error 587
bars represent ±1 SE.( A,B) Cerebral blood velocity; (C,D) Autoregulation index (ARI). Visit 588
1 (A,C); visit 2 (B,D). 589
590
591
592