Contribution of central hemodynamics and end-tidal CO 2 to cerebrovascular dynamics

Abstract


Introduction
It has been established that aerobic exercise is good for brain health through adaptations of the vascular system [1,2].The recent application of a dynamic cerebrovascular modeling technique can determine the response pathway during aerobic exercise [3] and may be valuable for understanding the underlying mechanisms contributing to cerebrovascular health.This modeling technique is attainable because of utilizing beat-by-beat and breath-bybreath physiological measures during exercise [4].In addition to middle cerebral blood velocity (MCAv), the dynamic modeling can be applied to heart rate, mean arterial pressure, and endtidal CO 2 partial pressure.
Previous investigations have established the MCAv dynamic response during aerobic exercise in healthy adults across the aging spectrum [5][6][7] and in clinical populations, such as stroke [8,9].This technique has revealed that MCAv increases exponentially following the onset of exercise and is a method for identifying unique age-and disease-specific profiles.Evidence suggests MCAv increases from rest during low to moderate-intensity exercise, with the amplitude greatest during moderate-intensity exercise [5].Additionally, MCAv amplitude is negatively associated with age and positively correlated with estimated VO 2 max [5].Within older adults, the time delay is significantly longer (i.e., slower) than in young adults [6], and elevated cerebral resistance in older adults attenuates the MCAv response compared to older adults with normal cerebral resistance [7].Lastly, compared to healthy older adults, individuals post-stroke have a significantly lower MCAv amplitude in both hemispheres [8].
Central hemodynamics and end-tidal CO 2 partial pressure are central to the middle cerebral artery blood velocity (MCAv) dynamic response during acute aerobic exercise.However, there is still a lack of understanding of the central hemodynamic and end-tidal CO 2 partial pressure dynamics and how their responses contribute to MCAv dynamics during aerobic exercise.
. This investigation aimed to characterize and assess the contribution of exercise intensity, central hemodynamics (i.e., heart rate and mean arterial pressure), and end-tidal CO 2 partial pressure dynamics to MCAv dynamics during acute bouts of aerobic exercise in young, healthy adults.We hypothesize that heart rate will significantly contribute to the initiation of exercise, but that mean arterial pressure and end-tidal CO 2 will be significant contributors during the steady-state phase.Additionally, we explored exercise dynamics within males and females.

Study design
The experimental protocol consisted of three separate, 6-min bouts of exercise.The order of the exercise bouts was randomized using a random number generator.All data were collected during a single visit to the laboratory.Data were collected between October 2016 and September 2021, and data were assessed for this investigation in March 2022.

Participant criteria
Participants were screened using our established inclusion/exclusion criteria [3].Inclusion criteria were: 1) the ability to perform repeated bouts of exercise and 2) the ability to travel to the University of Kansas Medical Center for testing.Exclusion criteria were: 1) inability to acquire a middle cerebral artery transcranial Doppler ultrasound (TCD) signal, 2) inability to perform the alternating leg movements on the seated recumbent stepper, 3) diagnosis of any neurological conditions such as multiple sclerosis, 4) pulmonary disease, or 5) diagnosis of myocardial infarction or heart failure.
Participants were asked to abstain from food for two hours [10], caffeine for a minimum of 6 hours [11], and vigorous exercise for 12 hours [6] before testing.Female participants were .verbally questioned regarding menstrual status.We did not directly assess hormone levels, but females exercised during their menstrual cycle's early follicular phase (Days 1-7) [6].
The research was conducted following the principles embodied in the Declaration of Helsinki and following local statutory requirements.The University of Kansas Medical Center Human Subjects Committee approved all experimental procedures (IRB number STUDY00003176).Institutionally approved written informed consent was obtained before participation.

Recumbent stepper familiarization
The Karvonen method was used to determine the appropriate heart rate range for each exercise intensity (HR range = [% exercise intensity (age-predicted HRmax -HRrest)] + HRrest) [3,12].Participants were familiarized with the prescribed rate of 120 steps per minute.The target work rate was determined by setting the initial resistance to 40 watts and then increasing 10 watts every 30 seconds.Once the target heart rate was achieved for the first intensity and maintained for one minute, watts were increased progressively until the target heart rate for the second intensity was achieved and maintained, and then again until the target heart rate for the third intensity was achieved.

Instrumentation
After familiarization, participants were instrumented with (1) a standard 5-lead ECG (Cardiocard, Nasiff Associates, Central Square, NY, USA) with lead II continuously recorded, (2) a finger hemodynamic monitoring system (Finometer, Finapres Medical System, Netherlands), used to measure beat-by-beat arterial pressure, (3) a brachial oscillometric blood pressure cuff (Tango M2 Stress Test Monitor, SunTech Medical, Inc., Morrisville, NC, USA) placed on the contralateral arm to the Finapres and used for comparison to ensure accuracy, and (4) expired CO 2 was continuously measured by a nasal cannula connected to a CO 2 capnograph (BCI Capnocheck Sleep 9004 Smiths Medical, Dublin, OH, USA) [3].
Transcranial Doppler ultrasound (TCD, 2 MHz probes; Multigon Industries Inc., Yonkers, NY, USA) measured MCAv with probes on the right and left trans-temporal windows.A Mueller-Moll probe fixation device held the ultrasound probes in place.The TCD signal was optimized by adjusting the probe angle and depth settings [3].
The baseline recording lasted 90-s, followed by 6-min of continuous exercise within the target HR range.The Borg Scale was used to measure the participants' rating of perceived effort or exertion (RPE) during the exercise bout [15].Participants were asked to rate their perceived exertion on a scale of 6 to 20 immediately following the exercise.To ensure participants returned to a rested state, participants remained seated for a minimum of 10 minutes or until heart rate, blood pressure, and end-tidal CO 2 partial pressure returned to resting values before the next bout started [3].

Data acquisition
All variables were sampled at 500 Hz and interpolated to 2.0 Hz.Data with R-to-R intervals greater than 5 Hz or changes in peak MCAv greater than 10 cm s -1 in a single cardiac .cycle were considered an artifact and censored.Three-second averages were calculated and smoothed with a 9-second sliding window average [3].

Dynamic modeling
Heart rate (HR), mean arterial pressure (MAP), end-tidal CO 2 (P ET CO 2 ), and MCAv dynamics were modeled with a monoexponential curve [16].A custom-written script within R version 4.1.0and the nls function package (R Core Team, Vienna, Austria) was used to model the exercise response [16].R scripts and data related to this manuscript are available from the corresponding author upon reasonable request.
From this modeling, we were able to determine baseline ( BL ), time delay ( TD ), time constant (τ), and steady-state ( SS ) responses (Table 1).BL is the average of 90-s before the onset of exercise.TD is the time delay preceding the exponential increase in the variable of interest.τ is the time constant or time-to-63% of the peak.SS is the absolute change from BL to the 90-s average of steady-state exercise (3 to 4.5 minutes) [3].Table 1.Metrics of the dynamic modeling

Statistical Analysis
Parametric (one-way ANOVA) or non-parametric (Kruskal-Wallis) tests were completed to compare variables following a visual inspection of Q-Q plots and the Shapiro-Wilk test.For .post-hoc comparisons, Bonferroni correction was used for parametric comparisons, and the Dunn test was used for non-parametric comparisons.
For evaluation of the contribution of central hemodynamics and P ET CO 2 to MCAv dynamics, a backward stepwise Akaike Information Criterion (AIC) linear regression model selection was completed.The models included exercise intensity, the preceding dynamic factors, and the equivalent dynamic factors.For example, the model for MCAv TD had exercise intensity, HR BL, MAP BL , P ET CO 2BL , and MCAv BL , along with HR TD , MAP TD , and P ET CO 2TD .The same modeling technique was used within sex as well.
Data analyses were performed with R version 4.1.0(R Core Team, Vienna, Austria).Data are presented as mean ± standard deviation unless otherwise noted, and statistical significance was evaluated at α < 0.05.
Acquired heart rates during the steady-state phase were all within the estimated HR range (Table 2).Post-hoc comparisons between exercise bouts for HR SS were all significantly different within the total sample (all comparisons p<0.001).Within males, Low HR SS was significantly lower than Mod2 (p<0.001), and Mod1 was significantly lower than Mod2 (p=0.01).There was not a significant difference between Low and Mod1 (p=0.06).Within females, comparisons between exercise bouts for HR SS were all different, with HR significantly increasing between exercise bouts (Low vs. Mod1, p<0.001;Low vs. Mod2, p<0.001;Mod1 vs. Mod2, p<0.01).Watts for Low was significantly lower than Mod2 (p<0.001) for the total sample, but Mod1 was not significantly different than Low (p=0.19) or Mod2 (p=0.25).Within males, Low was significantly lower than Mod2 (p=0.02), but Mod1 was not significantly different than Low (p=0.51) or Mod2 (p=0.35).Similarly, within females, Low was significantly lower than Mod2 (p<0.001), but Mod1 was not significantly different than Low (p=0.18) or Mod2 (p=0.08).

Comparison of dynamic factors across exercise bouts
Within the total sample, resting HR (HR BL ) was not significantly different between exercise bouts.The time to an exponential increase in HR after the initiation of exercise (HR TD ) for Low was significantly earlier than Mod2 (p=0.047), but no difference between Low and Mod1 (p=0.93) or Mod1 and Mod2 (p=0.07;Table 3).The time to 63%-of-peak (HRτ) for Low was significantly slower than Mod1 and Mod2 (p=0.01,p=0.01), but there was no difference in time between Mod1 and Mod2 (p=0.82).The absolute change in response from baseline to steadystate (HR SS ) for Low was significantly less than Mod1 and Mod2, and Mod1 was significantly lower than Mod2 (all post-hoc comparisons, p<0.0001).MAP SS for Low was significantly less than Mod1 (p=0.01) and Mod2 (p<0.0001), but no difference in absolute change between Mod1 and Mod2 (p=0.06).For P ET CO 2 and MCAv, dynamic factors were not significantly different across exercise intensities.  .Within males, HR BL , HR TD , and HRτ were not significantly different across exercise bouts.However, HR SS for Low was significantly lower than Mod1 and Mod2 (p<0.01 and p<0.001), but Mod1 was not significantly lower than Mod2 (p=0.18).Additionally, MAP SS for Low was significantly lower than Mod2 (p<0.01), but no difference in absolute change between Low and Mod1 (p=0.08) or Mod1 and Mod2 (p=0.42).For P ET CO 2 and MCAv, dynamic factors were not significantly different across exercise bouts.Similarly, within females, HR BL , HR TD , and HRτ were not significantly different across exercise bouts.However, HR SS for Low was significantly lower than Mod1 and Mod2 (p<0.01 and p<0.001), and Mod1 was significantly lower than Mod2 (p<0.01).Additionally, MAP SS for Low was significantly lower than Mod1 and Mod2 (p=0.04 and p<0.001), and Mod1 was significantly lower than Mod2 (pp=0.04).For P ET CO 2 and MCAv, dynamic factors were not significantly different across exercise bouts.

MCAv dynamics
. For MCAv TD , all baseline and P ET CO 2TD factors dropped from the model (Table 4).Exercise intensity remained in the model, but its contribution was not significant.HR TD and MAP TD significantly contributed to MCAv TD , such that the longer HR TD and MAP TD were to their exponential increase after initiation of exercise, the longer the exponential increase in MCAv TD .Exercise intensity, HR TD , and MAP TD accounted for 17% of the adjusted shared variation.
Within males, exercise intensity and MAP BL , HR TD , and P ET CO 2TD dropped from the model.Only MCAv BL and MAP TD were significant contributors, with HRBL and P ET CO 2BL contributing non-significantly.Of the significant contributing factors, a higher MCAv at baseline (MCAv BL ) and the longer MAP TD was to the exponential increase after initiation of exercise, the longer the exponential increase in MCAv TD .HR BL , P ET CO 2BL , MCAv BL, and MAP TD accounted for 27% of the adjusted shared variation.
Within females, MAP BL , MCAv BL , and P ET CO 2TD dropped from the model.HR TD and MAP TD significantly contributed to MCAv TD , such that the longer HR TD and MAP TD were to their exponential increase after initiation of exercise, the longer the exponential increase in MCAv TD .Exercise intensity, HR BL , P ET CO 2BL , HR TD, and MAP TD accounted for 37% of the adjusted shared variation.
. Table 6 For each variable, the primary number is the β-weight or standardized regression coefficient.A dash (-) indicates that the variable was dropped during the backward AIC model selection.95% confidence intervals are reported in brackets.BL , baseline.HR, heart rate.MAP, mean arterial pressure.MCAv, middle cerebral artery blood velocity.P ET CO 2 , end-tidal CO 2 .TD , time delay.*, p-value < 0.05.**, p-value <0.01.***, p-value < 0.001.For MCAvτ, only TD dynamic factors were included in the model because HRτ, MAPτ, and P ET CO 2 τ all occurred after MCAvτ (Table 7).Exercise intensity and MAP TD dropped from the model.HR TD remained in the model, but its contribution was not significant.P ET CO 2TD and MCAv TD significantly contributed to MCAvτ, such that the longer P ET CO 2TD was to the exponential increase after initiation of exercise, the longer the time to 63-of-peak (MCAvτ).Additionally, the longer MCAv TD was to the exponential increase after the initiation of exercise, the less time to MCAvτ.HR TD , P ET CO 2TD , and MCAv TD accounted for 21% of the adjusted shared variation.
For males, exercise intensity, P ET CO 2TD , and MCAv TD dropped from the model.HR TD remained in the model, but its contribution was not significant.MAP TD significantly contributed to MCAvτ, such that the longer MAP TD was to the exponential increase after initiation of exercise, the less time to MCAvτ.HR TD and MAP TD accounted for 48% of the adjusted shared variation.
For females, exercise intensity and P ET CO 2TD dropped from the model.HR TD remained in the model, but its contribution was not significant.MAP TD significantly contributed to MCAvτ, such that the longer MAP TD was to the exponential increase after initiation of exercise, the longer the time to MCAvτ.Additionally, the longer MCAv TD was to the exponential increase after the initiation of exercise, the less time to MCAvτ.HR TD , P ET CO 2TD , and MCAv TD accounted for 21% of the adjusted shared variation.HR TD , MAP TD , and MCAv TD accounted for 61% of the adjusted shared variation.For each variable, the primary number is the β-weight or standardized regression coefficient.A dash (-) indicates that the variable was dropped during the backward AIC model selection.95% confidence intervals are reported in brackets.HR, heart rate.MAP, mean arterial pressure.MCAv, middle cerebral artery blood velocity.P ET CO 2 , end-tidal CO 2 .τ, tau.TD , time delay.*, pvalue < 0.05.**, p-value <0.01.***, p-value < 0.001.
For MCAv SS , exercise intensity, all τ dynamic factors (except MCAvτ), and HR SS dropped from the model.All remaining dynamic factors significantly contributed to MCAv SS .The longer the time to 63-of-peak (MCAvτ), the greater the absolute change in MAP SS and PETCO2 SS ; the larger the absolute change in MCAv SS .MCAvτ, MAP SS , and PETCO2 SS accounted for 60% of the adjusted shared variation.For males, exercise intensity, all τ dynamic factors, and HR SS dropped from the model.All remaining dynamic factors significantly contributed to MCAv SS , such that the greater the absolute change in MAP SS and P ET CO 2SS , the larger the absolute change in MCAv SS .MAP SS and P ET CO 2SS accounted for 43% of the adjusted shared variation.
For females, HRτ, MAPτ, and HR SS dropped from the model.Exercise intensity and P ET CO 2SS remained in the model but did not significantly contribute.Like the total sample, MCAvτ, MAP SS , and P ET CO 2SS were all significant contributors.Thus, the longer the time to 63of-peak (MCAvτ), the greater the absolute change in MAP SS and P ET CO 2SS ; the larger the absolute change in MCAv SS .MCAvτ, MAP SS , and P ET CO 2SS accounted for 78% of the adjusted shared variation.

Discussion
This investigation suggests that HR, MAP, and P ET CO 2 contribution varies throughout the MCAv dynamic response pathways during aerobic exercise, and contribution differs within males and females.Time to the exponential increase in HR and MAP are significant contributing factors during the initial response to exercise, but 83% of the variation is still unexplained.Continuing into the τ phase, MAP is no longer a contributor, but the time to the exponential increase in P ET CO 2 and MCAv is, but 79% of the variation is still unexplained.It is not until the steady-state phase that the MCAv time-to-63% of peak (τ) and the absolute changes in MAP and P ET CO2 contribute most to MCAv SS phase, explaining 60% of the variation.This evidence suggests that at the initiation of exercise, other factors beyond these central physiological components play significant roles across the dynamic pathway that explain the dynamic response of MCAv.
Cardiac regulation, directly and indirectly, affects cerebral vasculature [17][18][19].Previous research has reported that changes in cardiac output influence cerebral blood velocity during exercise in healthy young adults [19].At the onset of dynamic exercise, oxygen consumption increases and continues to increase over the first minute of steady-state exercise and then plateaus as the oxygen uptake and transport match the demand of the tissues.The increase in cardiac output is due to an initial increase in stroke volume and heart rate, with both variables plateauing within two minutes of steady-state exercise.Concerning our findings, cardiac output, specifically stroke volume, may significantly contribute to MCAv during exercise initiation [13].Once cardiac output plateaus in the steady-state phase of the exercise, its contribution may .lessen, and other factors such as MAP and P ET CO 2TD may contribute more, as seen in our findings of 60% for MCAv SS .
Our findings suggest that sexes may differ in central hemodynamic adjustments and adaptations to dynamic exercise [20].Regarding the sex-related difference in hemodynamic response during exercise, it has been found that males have higher systolic blood pressure during exercise than females, probably due to females' blunted sympathetic response and higher vasodilatory state of females [21].One potentially confounding factor is the fluctuation in estrogens throughout the menstrual cycle, which can impact blood volume, systemic vascular resistance, and ventricular functions [22].As all the females were in the early follicular phase (Days 1-7), we do not believe hormonal factors played a significant role in the variability within females, but it is unclear if this resulted in differences between sexes.The potential role played by autonomic activity and hormone fluctuations should be investigated.Furthermore, the potential gender-related hemodynamic difference is the lower maximum level of stroke volume reached during dynamic exercise by women compared to men.The smaller cardiac size due to smaller body may be responsible for the reduced stroke volume and cardiac output often reported in females [23,24].Central command has been almost completely overlooked by scientists within the cerebrovascular function field, and such considerations of their effects on MCAv should included in future exercise investigations and may explain some of the variation.
Aerobic exercise is commonly used to stress the cardiovascular system, which allows for the quantification and evaluation of cardiovascular disease severity.However, this modeling technique could assess cerebrovascular function and disease during aerobic exercise.Declines in cerebral blood velocity are often observed with advancing age and disease.The decline in cerebral blood velocity may be observed at rest but is often enhanced in response to various challenges.They are thought to reflect a deterioration in cerebrovascular reserve, or "the ability of cerebral blood vessels to respond to increased metabolic demand and chemical, mechanical, or neural stimuli."[25] Understanding the dynamic response of associated physiological .variables within young adults will allow us to understand responses within aging populations.Because aging and pathology may affect the hemodynamic and CO 2 reactions to exercise, this may also be reflected in the association with the MCAv response.Furthermore, we will be able to investigate how exercise interventions affect the dynamic response of hemodynamics and P ET CO 2 and how their relationship to MCAv changes as a result of training.
We recognize limitations to the present investigation.Although TCD can directly measure cerebral blood velocity, it cannot measure the diameter of a vessel.The current evidence of whether exercise induces changes in MCA diameter is conflicting [26,27].Our methodology assumes a constant MCA diameter during the exercise transition.However, if the diameter does change, a change is likely to be very slight in larger basal arteries, such as the MCA, and it would not affect blood velocity to a great extent [28].Within our protocol, we increase the watts systematically over the first 30-s of the exercise.Exercise intensity was a non-significant factor at the initiation of exercise, and it may be explained by how we set watts.Despite this finding, we believe this methodology is essential and should be implemented to reduce the participant from eliciting a Valsalva maneuver and altering the hemodynamic responses at the initiation of exercise.Lastly, the sample was small, especially within sex groups, and limited to young, healthy adults, and future study comparisons should consider this.
These findings provide characterization and novel insight into central hemodynamics and PETCO2's contribution to MCAv dynamics during aerobic exercise.Heart rate, mean arterial pressure and end-tidal CO 2 partial pressure contribute little at the initiation of exercise but mean arterial pressure and end-tidal CO 2 partial pressure contribute to most of the MCAv response during steady-state exercise.Further work is necessary to support these findings by investigating other factors, such as cardiac output and sex differences, that help explain the variation in the MCAv response.Furthermore, as we have characterized the contribution of hemodynamics and end-tidal CO 2 partial pressure to MCAv dynamics, an investigation into exercise intervention effects on dynamics is underway.

Fig 1 .
Fig 1.Average dynamic response during each exercise bout.Mean dynamic heart rate

Table 4 .
Exercise dynamic characteristics within males (n = 9)The p-value is the main effect of exercise intensity on the dependent variables.Bold p-values indicate significance evaluated at α < 0.05, and symbols represent significance between exercise bouts.The Bonferroni correction method was used for the posthoc multiple comparisons.*, Low vs. Mod1 significantly different.†, Low vs. Mod2 significantly different.‡, Mod1 vs. Mod2 significantly different.MCAv, middle cerebral artery blood velocity.

Table 5 .
Exercise dynamic characteristics within femalesThe p-value is the main effect of exercise intensity on the dependent variables.Bold p-values indicate significance evaluated at α < 0.05, and symbols represent significance between exercise bouts.The Bonferroni correction method was used for the posthoc multiple comparisons.*, Low vs. Mod1 significantly different.†, Low vs. Mod2 significantly different.‡, Mod1 vs. Mod2 significantly different.MCAv, middle cerebral artery blood velocity.

.
Regression results for time delay (TD)