Protocol-dependence of middle cerebral artery dilation to modest hypercapnia

There is a need for improved understanding of how different cerebrovascular reactivity (CVR) protocols affect vascular cross-sectional area (CSA) when measures of vascular CSA are not feasible. In human participants, we delivered ~±4mmHg end-tidal partial pressure of CO2 (PETCO2) relative to baseline through controlled delivery, and measured changes in middle cerebral artery (MCA) cross-sectional area (CSA; magnetic resonance imaging (7 Tesla MRI)), blood velocity (transcranial Doppler and Phase contrast MRI), and calculated CVR based on steady-state versus a ramp protocol during two protocols: a 3-minute steady-state (+4mmHg PETCO2) and a ramp (delta of −3 to +4mmHg of PETCO2). We observed that 1) the MCA did not dilate during the ramp protocol, but did dilate during steady-state hypercapnia, and 2) MCA blood velocity CVR was similar between ramp and steady-state hypercapnia protocols, although calculated MCA blood flow CVR was greater during steady-state hypercapnia than during ramp, the discrepancy due to MCA CSA changes during steady-state hypercapnia. Due to the ability to achieve similar levels of MCA blood velocity CVR as steady-state hypercapnia, the lack of change in MCA cross-sectional area, and the minimal expected change in blood pressure, we propose that a ramp model, across a delta of ~−3 to +4mmHg PETCO2, may provide one alternative approach to collecting CVR measures in young adults with TCD when CSA measures are not feasible.


52
Cerebrovascular reactivity (CVR) studies assess changes in cerebral blood flow to a known 53 vasoreactive stimulus (e.g., changes in end-tidal partial pressure of CO2; PETCO2). Measures of 54 CVR are important because attenuated CVR may reflect preclinical vascular pathophysiology and 55 an increased risk of mortality independent from cardiovascular risk factors or stroke incidence (1). 56 The most commonly used technique for CVR measures in humans, transcranial Doppler (TCD) 57 ultrasonography, provides an index of vascular blood flow changes (i.e., blood velocity) because 58 the vascular cross-sectional area (CSA) values required for blood flow calculations (i.e., the 59 product of CSA and blood velocity) are not collected with TCD. Thus, an assumption of an 60 unchanging CSA is typically accepted, raising concern if changes in CSA do occur (2). To 61 circumvent issues related to TCD measures of CVR, some research groups measure all four brain-62 supply (i.e., carotid and vertebral) arteries outside of the brain (3), or use expensive neuroimaging 63 approaches (4,5). 64 An additional concern regarding quantification of CVR is the potential for changes in central 65 hemodynamics during hypercapnia (elevated PETCO2) that could elevate cerebral blood flow due 66 to changes in cardiac output (6) and blood pressure (6,7) and not directly due to cerebrovascular 67 dilation (7). Another complicating factor in quantifying CVR between groups is potential variation 68 in large cerebral artery reactivity, particularly when comparing age differences (8). As an example, 69 our group's previous work showed that compared to younger adults, older adults exhibited 70 attenuated changes in large cerebral artery CSA in response to steady-state hypercapnia (9). 71 However, obtaining cerebral artery CSA data requires access to costly MRI or CT systems. We 72 aim to understand protocol designs that provide accurate CVR estimates when using TCD 73 methods. The "ideal" velocity based CVR protocols conducted using TCD would require: 1) 74 minimal CSA changes by conducting CVR protocols that result in negligible change in CSA, and 75 2) minimal influence of confounding variables such as blood pressure. 76 In the current study, we tested the hypothesis that a ramp (i.e., linear) CVR protocol within the ±5 77 mmHg range of relative changes in PETCO2 would provide minimal changes in CSA while still 78 replicating CVR outcomes from the more standard steady-state hypercapnia CVR protocol. Our 79 rationale for this range of relative changes in PETCO2 comes from the emerging knowledge of a 80 sigmoidal change in MCA CSA, with minimal changes in hypercapnia, within the -5 to +5 mmHg 81 TCD during steady-state hypercapnia (4). Specifically, for a within-subject design, with a Cohen's 110 d of 1.02, alpha level of significance of 0.05, and statistical power of 0.80, we calculated a sample 111 size of 10 and recruited 12 individuals due to our laboratory's expected attrition rate of 10-12% 112 with our neuroimaging studies. Participants were ineligible if they were smokers, pregnant, or had 113 any of the following conditions: Raynaud's disease, respiratory illnesses, diabetes, claustrophobia, 114 history of psychosis, eating disorders, manic or bipolar disorder, major psychiatric conditions, or 115 dependence on alcohol or drugs. 116

117
Testing was completed between 10am -2pm. Participants refrained from exercise, alcohol, drugs, 118 and caffeine within 12 hours prior to testing. We used TCD and MRI to assess the cerebral 119 vasoreactivity in response to steady-state (three minutes) bouts of hypercapnia (HC), and a ramp 120 protocol from hypocapnia to hypercapnia (four minutes) (Fig. 1). 121 Steady-state and ramp protocols were each conducted twice, once for each of the TCD and MRI 122 portions of the testing sessions and the order of TCD and MRI trials was randomized across 123 participants. Unfortunately, we were unable to randomize the order of CVR protocols as it was 124 difficult to stop and restart the RespirAct™ (Thornhill Research Inc., Toronto, Ontario, Canada) without doing extensive recalibration. The desired ventilatory rate was set to 12 breaths/min using 126 a visual metronome for each session and was projected on a screen during the MRI scan. Our goal 127 was to have the protocols fall within the ±5 mmHg from baseline PETCO2 range. Following a 128 familiarization period of four minutes, the order and duration of protocols occurred as follows: 1) 129 baseline (1 minute), 2) steady-state hypercapnia (target was +5 mmHg, although only reached 130 ~+4mmHg; three minutes), 3) recovery (2 minutes) 4) baseline (1 minute), 5) -5 mmHg PETCO2 131 hypocapnia (brief hyperventilation; target was +5mmHg, although only reached ~ -3mmHg) 30 132 seconds), 6) incremental increase (ramp) from ~ -3 mmHg hypocapnia to ~ +4 mmHg relative 133 PETCO2 hypercapnia (four minutes), 7) recovery (two minutes). 134

Manipulating target PETCO2 stimulus 135
Prior to the MRI scan, participants were fitted with a facemask attached to the RespirAct™ system, 136 a modified sequential gas delivery breathing circuit(13) was used to clamp PETCO2 levels at the 137 desired +5 or -5 mmHg (depending on protocol). Breathing rate and tidal volumes were calibrated 138 prior to starting the breathing sequence. 139

MCA blood velocity and systemic blood pressure 140
While supine, continuous beat-to-beat arterial blood pressure was monitored using a Finapres® 141 Finometer system, where a finger cuff was placed on the middle phalange of the third finger, and 142 the finger blood pressure was calibrated with an upper arm cuff (Finapres® Medical Systems, 143 Amsterdam, Netherlands). The MCA was insonated with a 2 MHz ultrasound probe placed at the 144 temporal window and the peak blood flow velocity envelope was collected using the Neurovision 145 TCD System (Multigon Industries Inc., NY, USA). All analog data were sampled at 1000 Hz using 146 the PowerLab data acquisition system (ADInstruments, Dunedin, Otago, New Zealand). 147 MR Innovations Inc., Detroit, MI, USA). Care was taken to avoid any peripheral voxels within the 185 MCA lumen. The magnitude PC-MRI data was used to locate the MCA lumen. The peak velocities 186 were calculated for each subject for baseline and hypercapnic states. 187

Data extraction 188
The LabChart text files, the TCD and MRI session RespirAct™ breath-by-breath PETCO2 values, 189 and the measured CSA were aligned using RStudio (v. 2020) (14) for data extraction from specific 190 epochs as indicated by the event comments in each file. The "print" function in the "magicfor"

Ramp protocol 202
The target PETCO2 and ramp protocol schematic is shown in Fig. 1, panel a. The target ventilation 203 rate was 12 breaths/minute and participants were coached using a visual metronome. In order to 204 equalize the spacing on the time axis when plotting the achieved PETCO2 data, the PETCO2 and 205 corresponding time vectors were resampled to a fixed 12 breaths/minute sampling rate using the 206 base R "approx" function in RStudio. Thus, two time vectors were created: 1) a target time vector 207 that is based off of a 12 breath/minute ventilation rate and 2) a "fixed" time vector that is based 208 off of resampling each participants data to meet the target time vector. The inter-individual 209 differences for the "fixed" time vector are indicated by horizontal error bars (mean ± S.D.;

MCA blood velocity reactivity calculations 239
The ramp and steady-state hypercapnia MCA blood velocity cerebrovascular reactivity (CVR) 240 measures are shown in Fig. 5 panel a. For the ramp protocol, the MCA blood velocity slope (Fig. 3 panel c) and the PETCO2 (for TCD session; Fig. 3  For the steady-state hypercapnia protocol, the difference between the average baseline before start 245 of hypercapnia and the average of the last minute of hypercapnia (i.e., the difference between the 246 SSHC and B conditions in Fig. 4)  (2) 249

MCA blood flow reactivity calculations 250
The ramp and steady-state hypercapnia MCA blood flow CVR measures are shown in Fig. 5  calculated during the steady-state hypercapnia protocol, the percentage change from baseline for 261 each of the MCA blood velocity or blood flow were calculated (Fig. 5, panel c). 262

263
Inter-rater variability was assessed using Bland-Altman analysis for 45 randomly selected images. 264 Pearson's correlation coefficient was used to test the correlation for inter-modality (TCD vs. PC-265 MRI; Fig. 2) MCA blood velocity measures for both baseline and steady-state hypercapnia states 266 (P < 0.05 was considered to be statistically significant using a two-tailed test). In addition, linear 267 slope analysis was conducted to assess changes in PETCO2, MAP, MCA CSA, and MCA blood velocity variables during the ramp protocol. A probability level of P<0.05 indicates a non-zero 269 slope (the linear fits and p-values are labelled on Fig. 3 panels a-e). One-tailed paired t-tests were 270 conducted to compare variable responses during the steady-state hypercapnia condition versus 271 baseline (Fig. 4). Finally, to compare the MCA blood velocity CVR values and the MCA blood 272 flow CVR values calculated during the steady-state hypercapnia protocol, the percentage change 273 from baseline for each of the MCA blood velocity or blood flow was calculated (Fig. 5). 274

275
Our CVR protocols with target PETCO2 are illustrated in Fig. 1  Slope analysis during the ramp protocol indicated significant slopes (P<0.05; Fig. 3

Figure 5 -Middle cerebral artery (MCA) blood velocity and flow reactivity across protocols
Top row: Calculated MCA blood velocity cerebrovascular reactivity (CVR) was not different between ramp and steady-state hypercapnia (SSHC; panel a), while MCA blood flow CVR was higher in SSHC compared to the ramp protocol (panel b). The percent change in calculated MCA blood flow from baseline was higher than the percent change in MCA blood velocity from baseline (panel c) with SSHC. Where statistical significance occurs, p-values are indicated above data (ns = not significant; paired t-test; α level significance 0.05). N=12 for all variables with data presented as mean±S.D. Bottom row: Individual data points (n=12) comparing ramp to steadystate hypercapnia (SSHC) responses (panels a-b) for the same variable in that column as top row or MCA blood velocity to calculated MCA blood flow (panel c). The diagonal line (identity line) indicates where data would fall if there was no measurable effect of protocol on CVR measures (panels a-b) or effect of accounting for MCA CSA in flow calculations (compared to using MCAv alone as an index of flow) when assessing %change of flow or MCAv during SSHC (panel c). Data above the identity line indicates an increase in variable measure with SSHC (compared to ramp; panels a-b) or higher %change in flow value for a given %change in MCAv.

321
This is the first study to provide MCA CSA measures across ramp and steady-state hypercapnia 322 protocols, enabled by the ability to obtain MCA CSA measurements every 14 seconds with 323 prospective targeting of PETCO2. This approach enabled direct comparisons of the steady-state 324 versus a ramp protocols to establish valid CVR calculations using only measures of flow velocity. 325 The noteworthy findings of this study are that: 1) the MCA CSA did not change during the ramp 326 protocol with delta ~ -3 to +4 mmHg of PETCO2, but did increase with steady-state hypercapnia ~ +4 mmHg of PETCO2), 2) blood pressure remained stable throughout duration of both ramp and 328 steady-state protocols, and 3) CVR measures based on MCA blood velocity cerebrovascular 329 reactivity was not different between ramp and steady-state protocols, but 4) MCA blood flow-330 based CVR was greater during steady-state compared to the ramp protocol. Taken

MRI -MCA CSA & CVR measures 345
In the current study, PETCO2 values at the ends of both ramp and steady-state hypercapnia 346 protocols, and the calculated MCA blood velocity CVR measures, were not different between the 347 two protocols. Yet, the PETCO2 changes during the ramp protocol did not affect MCA CSA in 348 same the way that the steady-state hypercapnia protocol did. This is the first study to use a sequential gas delivery circuit (via the RespirAct™) when assessing 378 MCA CSA responses during hypercapnia, thereby more closely aligning partial pressure of arterial 379 and end-tidal CO2 levels (20). Interestingly, we achieved similar MCA blood flow CVR during 380 steady-state hypercapnia as previous work (2), and our MCA blood velocity CVR for both the 381 steady-state hypercapnia (2,21) and ramp protocol (7) were in agreement with previous studies. 382 As recommended by Regan et al. (7), when using a limited range of PETCO2 values during a 383 hypercapnic protocol, a linear approach to CVR analysis is appropriate. 384 Everything considered, a CVR measure based solely on TCD-acquired MCA blood velocity 385 measures during a ramp hypercapnia protocol seems to elicit a similar CVR outcome as the 386 commonly used steady-state protocol, but without the limitation of potential changes in MCA CSA. CVR outcome measures involve calculating MCA blood velocity changes for a given change 388 in PETCO2. Similarly, our study illustrates that even modest values of change in PETCO2 achieve 389 a comparable value of CVR at higher hypercapnic doses. 390

391
Our overall target range for manipulating PETCO2 during the cerebrovascular reactivity was 392 between the -∆5 to +∆5 from baseline PETCO2 (10). While our findings support a lack of MCA 393 CSA changes during ramp protocols of -∆3 to +∆4 mmHg in PETCO2, we cannot make conclusive 394 remarks on MCA CSA changes during a ramp protocol of ±∆5mmHg in PETCO2. However, as 395 mentioned above, our values of ∆PETCO2 fell within our target ±∆5mmHg from baseline PETCO2 396 range, the CVR measures are consistent with previous studies where higher levels of PETCO2 397 were used, and we found negligible increases in MAP. Therefore, the modest level of PETCO2 398 achieved here appear to have achieved the major objectives. The ramp protocol designed for the 399 current study also achieved an optimal balance between stimulus and central hemodynamics. We 400 acknowledge however, that we did not account for the impact of the ~2 mmHg rise in MAP during 401 the ramp protocol, which we expect to be negligible in impacting MCA CSA. 402 Although we were unable to measure continuous blood pressure and MCA blood velocity during 403 the MRI trial, we measured blood velocity data during the MRI session via the PC-MRI sequence. 404 Testing of the TCD and MRI segments of the study were collected consecutively within a 2-hour 405 window with the order of tests varied across participants. The absolute values for MCA blood 406 velocities measured by PC-MRI were lower than our TCD measures, although we suspect this had 407 to do with our pre-set Venc value choice of 100 cm/s which may have cut off some of the higher 408 velocities. Our rationale for not choosing a higher Venc than 100 cm/s was the possibility of cutting 409 off lower velocity values during baseline. Regardless, the increase in MCA blood velocity during 410 hypercapnia (from baseline) were in agreement between the two techniques (TCD: ~∆18 cm/s and 411 PC-MRI: ~∆20 cm/s). Thus, we assume that the blood pressure responses during the TCD and 412 MRI sessions were similar as well. 413 Finally, the current results are delimited to young healthy adults Thus, additional studies are 414 needed to understand the effects of age, a group that demonstrates greater MAP responses to 415 hypercapnia and variable responses in CSA changes (9), or other differentiating conditions.