Impact of baseline CO2 on Blood-Oxygenation-Level-Dependent MRI measurements of cerebrovascular reactivity and task-evoked signal activation

Accepted Manuscript Impact of baseline CO2 on Blood-Oxygenation-Level-Dependent MRI measurements of cerebrovascular reactivity and task-evoked signal activation

Christiaan Hendrik Bas van Niftrik, Marco Piccirelli, Oliver Bozinov, Nicolai Maldaner, Catherine Strittmatter, Athina Pangalu, Antonios Valavanis, Luca Regli, Jorn Fierstra PII: DOI: Reference:

S0730-725X(18)30013-4 https://doi.org/10.1016/j.mri.2018.02.002 MRI 8918

To appear in: Received date: Revised date: Accepted date:

4 December 2017 30 January 2018 12 February 2018

Please cite this article as: Christiaan Hendrik Bas van Niftrik, Marco Piccirelli, Oliver Bozinov, Nicolai Maldaner, Catherine Strittmatter, Athina Pangalu, Antonios Valavanis, Luca Regli, Jorn Fierstra , Impact of baseline CO2 on Blood-Oxygenation-LevelDependent MRI measurements of cerebrovascular reactivity and task-evoked signal activation. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Mri(2017), https://doi.org/10.1016/j.mri.2018.02.002

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Impact of baseline CO2 on Blood-Oxygenation-Level-Dependent MRI measurements of cerebrovascular reactivity and task-evoked signal activation

PT

Christiaan Hendrik Bas van Niftrik1,2,*, Marco Piccirelli2,3,*, Oliver Bozinov1,2, Nicolai Maldaner1,2, Catherine Strittmatter1,2, Athina Pangalu2,3, Antonios Valavanis2,3, Luca Regli1,2, Jorn Fierstra1,2

Affiliations:

Department of Neurosurgery, University Hospital Zurich, University of Zurich, Switzerland

2

Clinical Neuroscience Center, University Hospital Zurich

3

Department of Neuroradiology, University Hospital Zurich, University of Zurich, Switzerland

NU

SC

RI

1

MA

* Equal Contributions

Corresponding author:

D

Christiaan Hendrik Bas van Niftrik, MD

Frauenklinikstrasse 10

Phone: +41-44-2551111

AC

Fax: +41-44-2554505

CE

CH-8091 Zurich, Switzerland

PT E

Department of Neurosurgery, University Hospital Zurich

E-mail: [email protected]

Email addresses: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected] Keywords: Cerebrovascular reactivity, Functional Magnetic Resonance Imaging, BOLD, Carbon Dioxide, Humans

ACCEPTED MANUSCRIPT Abstract Neurovascular coupling describes the cascade between neuronal activity and subsequent bloodoxygenation-level-dependent (BOLD) signal increase. Based on this premise, the correlation of this BOLD signal increase with a particular task, such as finger-tapping, is used to map neuronal activation. This signal increase may be dampened in brain areas exhibiting impaired cerebrovascular reactivity (BOLD-CVR), leading to false negative activation.

PT

Blood-oxygen-level-dependent (BOLD) cerebrovascular reactivity (CVR) has also been used to

RI

optimize task evoked BOLD signal changes. To measure BOLD-CVR, controlled BOLD-CVR studies

SC

have commonly been performed using a preset isocapnic carbon dioxide (CO2; ~40mmHg) baseline, independent of subjects’ resting CO2. This arbitrary baseline, however, may influence BOLD-CVR

NU

measurements. We therefore performed BOLD-CVR, as well as BOLD fMRI during a controlled bilateral finger-tapping task in two groups of ten subjects: group A at subject’s resting CO2 and group B

MA

at a preset isocapnic CO2 baseline (40mmHg). Whole brain BOLD-CVR was significantly decreased for group B (group A 0.26 (SD 0.05) vs group B 0.16 (SD 0.05), p<0.001). For the predefined hand area in

D

the precentral cortex, BOLD-CVR and BOLD fMRI signal changes were significantly lower for group

PT E

B (group A 0.20 (SD 0.04) vs group B 0.13 (SD 0.05), p< 0.01; 1.19 (SD 0.31) vs 0.62 (SD 0.37), p< 0.01).CO2 levels significantly influence both BOLD-CVR and BOLD fMRI measurements. Hence, for

CE

an accurate interpretation, baseline CO2 levels and BOLD CVR should be considered complementary to

AC

task evoked BOLD fMRI.

ACCEPTED MANUSCRIPT 1. Introduction In functional magnetic resonance imaging (fMRI), a local increase in cerebral blood flow (CBF) following neuronal activation induces a blood oxygenation-level-dependent (BOLD) signal increase through lowering of the deoxyhemoglobin blood content. The BOLD signal increase is generated by a complex physiological mechanism known as neurovascular coupling.1-3 Fox & Raichle were the first to observe neurovascular coupling by noticing a decrease in oxygen extraction fraction (OEF) subsequent

PT

to neuronal activity.4 We now know that during increased neuronal activity, the increase in regional CBF supersedes the increase in cerebral metabolic rate of oxygen (CMRO2), leading to a physiological

RI

hyperemia.5 The subsequent decrease in deoxyhemoglobin ratio, defined as the ∆ deoxyhemoglobin

SC

before and after a stimulus, during this physiological hyperemia can be mapped out using BOLD fMRI3,

NU

thereby localizing neuronal activity associated with specific tasks, such as finger-tapping. On the other hand, by applying a vasodilatory stimulus, such as carbon dioxide (CO2), BOLD signal

MA

changes can also be induced for the entire brain. From this BOLD signal response to CO2 changes, cerebrovascular reactivity (BOLD-CVR) is measured.6, 7 BOLD-CVR is defined as the change in BOLD

D

signal in the presence of a vasoactive stimulus and is an indicator of the remaining cerebrovascular

PT E

reserve.8 Recent studies in different patient groups have shown that localization of neuronal activity using BOLD fMRI may exhibit false negative activationin brain areas with impaired BOLD-CVR. 9-12.

CE

Furthermore, BOLD CVR has also been extensively investigated to calibrate task evoked fMRI and dissociate the changes in oxygen metabolism from changes in CBF and cerebral blood volume.5, 13

AC

Although different methods exist to measure BOLD-CVR, examinations under controlled CO2 conditions have shown to be more sensitive and reproducible.7, 14, 15 Most studies using controlled CO2 conditions assumed isocapnia around 40 mmHg for all subjects, independently of their resting CO2, and therefore used it as a preset study baseline from which the hypercapnic stimulus (usually a +10 mmHg increase in CO2) is given.16, 17 More recently, however, this concept has been revisited since studies indicated that a gap between subjects’ resting CO2 and a preset CO2 baseline may significantly influence BOLD-CVR readings.18, 19

ACCEPTED MANUSCRIPT This raises the important question whether such a calibrated ‘’isocapnia’’ CO2 baseline in healthy subjects will result in a biased estimation of the subject’s BOLD-CVR, and a bias correction of the fMRI signal activation using BOLD CVR.

To test this hypothesis, we selected ten healthy subjects selected from our existing prospective database that underwent whole brain BOLD-CVR, as well as BOLD fMRI during a controlled bilateral finger-

PT

tapping task at an ‘isocapnic‘ baseline (40mmHg – Group A). We additionally scanned ten other age matched healthy subjects at their own resting CO2 as a control group (Group B).For each subject, we

RI

correlated BOLD-CVR measurements in a predefined hand area of the precentral gyrus with percentage

SC

BOLD fMRI signal change and used BOLD CVR to normalize the task-evoked fMRI signal. We found

MA

NU

that CO2 levels influence both BOLD-CVR as well as BOLD fMRI measurements.

D

2. Materials and methods

PT E

2.1 Data selection and acquisition

All subjects were selected from a healthy subject BOLD-CVR database that we created in 2016. BOLD-

CE

CVR studies from this database were originally performed with a calibrated CO2 baseline of 40mmHg. After February 2017, due to novel studies on this topic18, where the subject’s resting CO2 was favored

AC

as a baseline, we adjusted our protocol accordingly. Therefore for this study, to validate our new protocol, we were able to select 10 healthy subjects scanned at an ‘isocapnic‘ baseline of 40mmHg CO2 and prospectively scanned 10 other age matched healthy subjects at their own resting CO2 as a control group. Exclusion criteria for healthy subjects were smoking, the presence of neurological symptoms, a medical history including any neurological illness affecting the central nervous system or inability to move their hands, as well as any type of traumatic brain or cervical injury. Prior to the study, all participants read and signed an informed consent approved by the Ethics commission of the University Hospital of Zurich and the Kanton of Zurich, Switzerland (Ethics approval: KEK-ZH-Nr. 2012-0427)

ACCEPTED MANUSCRIPT MRI data were acquired on a 3 Tesla Skyra VD13 (Siemens Healthcare, Erlangen, Germany) with a 32channel head coil. BOLD fMRI parameters: axial two dimensional (2D) single-shot EPI sequence planned on the ACPC line plus 20° (on a sagittal image) voxel size 3×3×3 mm3, acquisition matrix 64x64x35 ascending interleaved slice acquisition, slice gap 0.3 mm, GRAPPA factor 2 with 32 reference lines, repetition time (TR)/ echo time (TE) 2000/30 ms, flip angle 85°, bandwidth 2368

PT

Hz/Px, Field of View 192x192 mm2. For every subject, 180 volumes were acquired for the BOLD-CVR study and 135 volumes for the fMRI task based study.

RI

A three dimensional (3D) T1-weighted Magnetization Prepared Rapid Acquisition Gradient Echo

SC

(MPRAGE) image was also acquired with the same orientation as the fMRI scans for overlay purposes

NU

Acquisition parameters were: voxel size of 0.8×0.8×1.0 mm3 with a field of view 230x230x176 mm3 and scan matrix of 288x288x176, TR/TE/TI 2200/5.14/900 ms, with a flip angle 8°. For healthy

MA

subjects with age above 50, we also acquired a Time of Flight MRI scan to exclude large intracranial artery pathologies (i.e. stenosis, occlusion).

PT E

D

2.2 Prospectively controlled CO2 Stimulus The CO2 manipulation was provided by a gas delivery system using a computer controlled gas blender with prospective gas targeting algorithms (RespirActTM, Thornhill Research Institute, Toronto, Canada)

CE

which allows for automated precise targeting of arterial partial pressure of oxygen (PaO2) and carbon

AC

dioxide (PaCO2). Before starting the scan protocol, subjects were positioned in the MRI scanner and were asked to breathe normally. The gas delivery were programmed to be just below the inspiratory volume, causing the subjects to rebreathe for 5-10 minutes to correct for possible hyperventilation (i.e. artificially low resting CO2).

2.3 BOLD-CVR Studies and CO2 protocols For the BOLD-CVR study we preprogrammed the RespirActTM using a standardized CO2 increase of ~10 mmHg above baseline for 80 seconds whereas O2 was maintained at a level of 100 mmHg (i.e. isooxia). There were two groups:

ACCEPTED MANUSCRIPT 

Group A: For these subjects (subject 1-10, Table 1), CO2 baseline was clamped at the subject’s own resting CO2 during 100 seconds. During the BOLD-CVR study, CO2 was then increased by ~10mmHg for 80 seconds and brought back to their resting CO2 value for 100 seconds.



Group B (subjects 11-20, Table 1): Here, the CO2 baseline level of each subject was preset to an isocapnic CO2 baseline of ~40mmHg. The actual scan involved a baseline of 40 mmHg CO2 during 100 seconds, followed by a similar CO2 increase of ~10mmHg for 80 seconds and return

PT

to ~40mmHg baseline for 100 seconds. In both groups, we collected 40 extra BOLD volumes to be able to adjust for potential temporal shift.

RI

(For more information see Van Niftrik, et al. 201720. Therefore, the total number of BOLD volumes in

NU

SC

both groups reached 180.

2.4 fMRI Studies and CO2 protocols

MA

During the fMRI studies, all subjects performed the same motor task, meaning repeated bilateral fingertapping. This consisted of 4 times 30-seconds blocks of resting alternating with 30-seconds blocks of

D

simultaneous self-paced bilateral finger-tapping (fingers to thumb). The switch between blocks was

PT E

timed using a visual queue. The arms were resting along the body on the MRI table. Every subject was instructed before the study and correct execution was controlled for. During the complete finger-tapping

CE

paradigm, the face mask remained connected and CO2 and O2 were kept at the baseline levels maintained during the BOLD-CVR studies done previously (resting CO2 for group A and isocapnic CO2

AC

preset at ~40 mmHg for group B).

2.5 Image analysis

2.5.1 Spatial pre-processing Anatomical and functional images were transferred into a free BOLD fMRI preprocessing software called Statistical Parametric Mapping 12 (SPM12, Wellcome Trust Centre for Neuroimaging, Institute of Neurology, University College London, UK). First, slice time correction was applied to all the BOLD volumes. BOLD volumes were orthogonalized in 6 motion estimate directions and realigned to the mean BOLD volume to minimalize the influence of head motion. In case of motion larger as 3mm (1

ACCEPTED MANUSCRIPT voxel size) in any direction, we redeemed the BOLD study as unusable. The high-resolution MPRAGE T1-weighted image was aligned to the mean BOLD image. For construction of normalized volumes, functional and anatomical maps were normalized in Montreal Neurological Institute (MNI) Space of the MNI/ICBM AVG 152 Template using a 12-parameter affine linear transformation and warping of the images using a nonlinear transformation. After removal of all other voxels, in order to decrease the partial volume effects of neighboring CSF voxels, the white and grey matter voxels of the functional

PT

images were spatially smoothed with an isotropic Gaussian kernel of 8 mm full width at half maximum. SPM combined grey matter and white matter maps were set at a threshold of 80% probability and

SC

RI

served as a template for calculating hemispheric BOLD-CVR for each subject.

NU

2.5.2 ROI determination

MA

Similar to Para et al. 2017, we determined the hand region in the left and right hemisphere using the predefined regions described in Yousry et al. 1998.21 We created an individual mask for each subject.

D

Subsequently, a senior staff neuroradiologist (A.P.) checked all maps for anatomical accuracy. These

PT E

masks were used to determine BOLD-CVR and BOLD fMRI signal change during the fMRI fingertap

CE

protocol.

2.5.3. BOLD-CVR calculations

AC

The BOLD-CVR datasets were further processed with custom MATLAB R2013b routines (The MathWorks Inc., Natick, MA, USA). To correct for fMRI signal drift, the BOLD time signal was linearly detrended. Temporal smoothing included applying a low pass filter and a robust Loess smoothing method to the data. The CO2 time course was interpolated and resampled to match the number of BOLD time points. Secondly, the temporal delay on a voxel-wise basis between CO2 and BOLD time course was calculated for improved alignment. BOLD-CVR, defined as percentage BOLD signal change per CO2 change in mmHg (Δ%BOLD/mmHg), was calculated by regressing CO2 against the BOLD signal time course using a

ACCEPTED MANUSCRIPT linear least square fitting and determined for each voxel of the brain, as well as averaged for both hand area masks. Specific details and novelties of this method as compared to other methods to calculate temporal delay and BOLD-CVR have been described in more detail before.22 BOLD-CVR was color coded and presented as an overlay on the high-resolution T1-weighted image.

PT

2.5.4. fMRI Activation Further processing of the fMRI activation data was also done using Matlab2013. For each subject, the

RI

relative change in percent of the BOLD fMRI signal correlating with the finger-tapping task (relative

SC

BOLD fMRI signal change) was determined. After preprocessing, the BOLD signal time course was linear detrended to correct for any signal drift in the data. Then, the block paradigm was convolved with

NU

the hemodynamic response function, derived from SPM, and secondly regressing linearly the convolved

MA

block paradigm time course with the BOLD fMRI signal time course on a voxel-by-voxel basis. Finally, the regression slope was divided by the BOLD signal at baseline to represent relative BOLD fMRI

D

signal change. After applying the anatomical hand area masks, the relative BOLD fMRI signal change

(Figure 1).

CE

2.6 Statistical Analysis

PT E

for the masked area was calculated. For illustrative purposes, we also created a t-value activation map

Statistical analysis was performed using SPSS 23 (IBM, Armonk, NY). The Shapiro-Wilk test was used

AC

to assess normality of distribution. Each variable was redeemed to be normally distributed. Each variable in both groups were compared using an independent sample t-test [p<0.05 was considered significant]. By using BOLD-CVR and relative BOLD fMRI signal change per hand area ROI, we constructed a step-wise linear regression model between the two variables using BOLD-fMRI as in the dependent variable for both groups (Figure 2). BOLD-CVR, resting CO2 levels and baseline CO2 levels were entered as independent variables. The Fisher r-to-z transformation was applied to assess the significance of the difference between both correlation coefficients.

ACCEPTED MANUSCRIPT Last we normalized our regional fMRI signal changes by regional BOLD-CVR to assess the parameter M.5

3. Results 3.1 Baseline Characteristics Baseline characteristics of the both groups and individual characteristics are shown in Table 1,

PT

respectively Table 2. No significant differences were found resting PetCO2 (Group A: PetCO2: 36.4

RI

mmHg (SD 3.4); Group B: PetCO2: 34.1 mmHg (SD 2.3). No subject had a head motion larger than 3

SC

mm and therefore none of the subjects had to be discarded from further analysis.

NU

3.2 BOLD-CVR and fMRI findings

Mean whole brain BOLD-CVR for group A was significantly higher than group B (0.26 (SD 0.05) vs.

MA

0.16 (SD 0.05) BOLD/mmHg, p<0.001), corresponding to a ~40% dampened BOLD-CVR response. A higher BOLD-CVR was also found for each hand area (left hand area, group A vs. group B: 0.20 (SD

D

0.05) vs. 0.14 (SD 0.05), p=0.008; right hand area, group A vs group B: 0.20 (SD 0.03) vs 0.14 (SD

PT E

0.06), p=0.005). The fMRI finger-tapping maps also exhibited a decreased signal change in the bilateral hand areas in group B (relative BOLD fMRI, group A vs. group B, left hand area: 1.22 (SD 0.36) vs

CE

0.61 (SD 0.36), p=0.004; right hand area: 1.16 (SD 0.27) vs 0.63 (SD 0.41), p=0.004), indicating an on

AC

average decrease in relative BOLD fMRI signal change after neuronal activation.

3.3 Correlation of BOLD-CVR vs. fMRI signal change BOLD-CVR data of each subject from both groups are plotted in Figure 1. In subjects with higher BOLD-CVR values, BOLD activation patterns are generally seen in both pre- and postcentral gyri as well as the supplementary motor cortex. As is shown in Figure 1, decreased BOLD-CVR does not only diminish activation in the pre- and postcentral gyrus, but seemingly also limits activation in other areas in the brain, in particular here, a decreased activation in the supplementary motor cortex. Further quantitative statistical analysis was done for the hand areas. Figure 2 shows the correlation between BOLD-CVR and relative BOLD fMRI signal change for these areas. Mean relative BOLD fMRI signal

ACCEPTED MANUSCRIPT change for all subjects demonstrated a very strong and significant correlation to the mean BOLD-CVR (Figure 2, p<0.05, r=0.72). More than half of the variance of relative BOLD fMRI signal change is explained by the mean BOLD-CVR in the hand region (r2=0.53). This correlation was even stronger for group A as for group B (r=0.66 vs r=0.51, respectively), however, no significant difference in correlation was found between group A and group B (z = 0.67, p=0.50). After normalization, when evaluation the maximal BOLD signal to a task, the difference b between both groups decreased but

PT

remained significantly different, (Group A vs. Group B: 6.0±1.1 vs. 4.6±2.6 p=0.03).

RI

4. Discussion

SC

Our findings show that in controlled conditions different CO2 levels can significantly influence BOLD-

NU

BOLD-CVR and fMRI finger-tapping activation in healthy subjects, potentially leading to false-negative measurements. By determining BOLD-CVR and relative BOLD fMRI signal change at two different

MA

controlled CO2 baseline conditions (i.e. subjects’ resting CO2 – group A and a preset ‘‘isocapnic’’ CO2 baseline of ~40mmHg – group B) we demonstrated that subjects from group B on average exhibited a

D

marked overall dampened BOLD-CVR, encompassing the entire brain including the bilateral hand areas.

PT E

A similar effect was seen during task-based fMRI, i.e. a clear overall relative BOLD fMRI signal change reduction within the hand areas in group B. BOLD-CVR was also highly correlated with the percentage

CE

of fMRI signal change (r2= 0.53, p<0.001), implying an association between BOLD-BOLD-CVR and relative BOLD task based fMRI signal change. Interestingly, no significant difference in correlation was

AC

found between group A and group B (z = 0.67, p=0.50). After normalization with BOLD-CVR, the difference between relative BOLD fMRI signal change decreased, however the determined parameter M remained significantly different. This implies a partly coalescence system between both signal changes. However, as the difference remains significant, it also suggest the additional influence of different baseline levels of CO2.

4.1 Neurovascular and neurometabolic effects of hypercapnia In healthy subjects, with cerebral autoregulation expected to be intact, our results show an on average ~40% reduction of BOLD-CVR in a group with a preset CO2 baseline in comparison to a group scanned

ACCEPTED MANUSCRIPT on their own resting CO2. These results can be explained by the interaction between CO2 and BOLD signal. Previously this interaction has been described as sigmoidal, BOLD signal changes can reach a sublinear phase at supramaximal hypercapnic CO2 conditions (as is clearly illustrated in Figures 2 of Sobczyk et al. 2014 and in Bhogal et al. 2014).23, 24 Therefore, BOLD-CVR measurements are deemed to be highly influenced by the CO2 measurement position on this sigmoidal curve. For instance, in patients with severe steno-occlusive disease, we know that cerebral arteries dilate, ipsilateral to the

PT

stenotic or occluded vessel, resulting in a shift of this sigmoidal BOLD/CO2 curve to the left (meaning a higher CO2 measurement position on the BOLD- CO2 curve at a similar CO2 baseline). This results in a

RI

BOLD fMRI signal reduction through a decrease of the maximal vasodilatory capacity. 23 In earlier

SC

studies, ‘isocapnia’ was anticipated to be around 40mmHg CO2. In our study, however, a preset CO2

NU

baseline of 40 mmHg in healthy subjects with a lower resting CO2 resulted in an on average 40% dampened BOLD-CVR. With the premise that ‘isocapnia’ is at the subjects own resting CO2, a preset

MA

baseline of 40mmHg is expected to induce premature vasodilation and shift the BOLD/CO2 levels more towards the sublinear range of the sigmoidal curve. Consequently, using a similar CO2 increase (i.e. 10

D

mmHg), BOLD-CVR would artificially be reduced in an otherwise healthy capillary bed.

PT E

In case of task-based fMRI, neuronal activation causes locally an increase in cerebral blood flow.5 During such a preset ‘isocapnic’ CO2 baseline, which is not expected to impair neuronal activity, the

likewise.

CE

decreased vasodilatory potential (i.e. decreased BOLD-CVR) leads to a decreased fMRI activation,

AC

These findings are in agreement with findings by Cohen et al.25 They found that uncontrolled change in the arterial partial pressure of CO2 induces BOLD signal increase differences; especially hypercapnia induces a dampened relative BOLD fMRI signal increase. Considering CBF, a preset isocapnic baseline may lead to a higher basal CBF. Here, CO2 is thought to decrease the perivascular pH.26, 27 The change in pH causes smooth muscle cells to decrease their resistance and a concomitant CBF increase is seen throughout the brain. As seen with impaired BOLDCVR, during preset isocapnic baseline, a dampened CBF increase is therefore also likely to occur.28 During neuronal activity, an increase in local metabolism is known to cause a functional hyperemia. 5 This causes a relative decrease in deoxyhemoglobin and a BOLD signal increase can be measured.

ACCEPTED MANUSCRIPT Similar to BOLD-CVR, the functional hyperemia is dependent on the potential remaining vasodilatory capacity. Therefore, a direct relationship between BOLD-CVR and BOLD fMRI signal activation has already been presumed in animal and human studies. For instance, Stefanovic et al 29 showed that with increased basal CBF the BOLD fMRI signal increase is dampened. In humans, Para et al.10 found in patients with steno-occlusive disease a similar strong correlation of impaired BOLD-CVR vs. dampened

PT

BOLD fMRI activation as compared to our data (R2=0.65). However, as our data suggests, this is not the only mechanism acting on the BOLD signal in a signal activation study.

RI

So, an alternative hypothesis can also be postulated: The BOLD signal is not only influenced by CBF,

SC

but also by other factors, such as CBV and CMRO2.5 In the beginning, inducing a short period of

NU

vasodilatation with iso-oxic hypercapnia was considered iso-metabolic.30, 31 Later studies also showed no differences in CMRO2 to hypercapnia.32, 33 In electrophysiological studies, the amplitude of the

MA

signal as well as the waves do not change, further supporting the theory of iso-metabolism.34 Nevertheless, studies using longer duration of hypercapnia found opposite results. Changes in CMRO2

D

were reported in resting state fMRI during a prolonged hypercapnic baseline. For instance, Xu et al35

PT E

found a decrease of 13% in CRMO2, indicating a decrease in metabolism during hypercapnia. Animal studies have elaborated on these findings. In particular, Zappe et al. showed a diminished neuronal

CE

activity to a visual stimulus with hypercapnia(10% CO2).36 A decrease in CMRO2 during prolonged hypercapnia means an overall increase in oxyhemoglobin and

AC

therefore a subsequent change in the ratio oxy/deoxyhemoglobin as compared to normocapnia. Consequently, this would decrease the BOLD signal change induced by a neuronal or vasodilatory stimulus. In stroke patients, it has been shown that neurons in areas with decreased BOLD-CVR ‘hibernate’ (i.e. eliciting a significant decrease in metabolism in areas with impaired BOLD-CVR).37 This may also explain results reported by Blicher et al.11 in similar patients, who did not elicit a BOLD fMRI increase, despite CBF and CBV increases. This underscores the importance of a complementary BOLD-CVR examination. Interestingly, this ‘hibernating’ state could be reversed after revascularization, indicating that it is a potential safety measure of neurons. However, if this can also occur in healthy subjects exposed to prolonged hypercapnia and to what precise extent this might affects

ACCEPTED MANUSCRIPT the BOLD signal in our and other studies is still unclear. Future projects shall focus on these open questions. The difference between parameter M (calibrated fMRI) between both groups can also be explained by either one of those two hypothesis. The BOLD response to CO2 is usually seen as supramaximal, contrary to the task evoked fMRI signal response. As only some subjects in group B would reach their dilatory maximum during the task-evoked fMRI, the average between the two groups became more

PT

equal but was still significantly different. Otherwise, the decrease of cerebral metabolism during prolonged hypercapnia could explain the difference with a preserved dilatory reserve capacity at

SC

RI

40mmHg.

NU

4.2 Significance of BOLD-CVR findings

MA

Recently, MRI studies investigating BOLD-CVR in combination with CO2 has been extensively investigated and validated.38 For instance, BOLD-CVR impairment has been found in subjects with steno-occlusive disease28, 39, brain tumors9, 40, 41 , traumatic brain injury42, leukoaraiosis43, and

PT E

D

neurocognitive and psychological deficits44, often using a healthy reference cohort45. In initial studies, CO2 changes were mostly induced using breath holding or CO2 inhalation.7 However, with more elaborate methods available (i.e. in a controlled manner during iso-oxia) the question rose about an

CE

optimal resting baseline to perform these studies. Most studies with controlled CO2 conditions used a

AC

uniform normocapnic CO2 baseline of 40 mmHg on which all the subjects were tested independently of their resting CO2 at the time of scanning.16, 22, 46 Later studies reconsidered this premise and used the individual resting CO2 as a baseline condition.18 In our study, we have found a significant difference between BOLD-CVR measured at resting CO2 and at a set CO2 of ~40 mmHg, in line with findings by others.19, 47, 48. Halani et al19 assessed BOLD-CVR during different calibrated CO2 baselines. (i.e. hypocapnia, normocapnia and hypercapnia). They found an increased BOLD-CVR response to a vasodilatory stimulus from normocapnia as compared to the other two states. These findings were reproduced in a follow up study done by Golestani et al.48 However, we should mention that in both

ACCEPTED MANUSCRIPT studies a uniform CO2 value was chosen (i.e. 40 mmHg for normocapnia). Given our current work, we now show BOLD-CVR to be more impaired when not performed at individual resting CO2. Interestingly, Battisti 201149, as well as Tancredi and Hoge 201350 stated that BOLD-CVR studies should be conducted within the linear range of the sigmoidal curve, which according to them lie between 30 and 45 mmHg. Although all our subjects had their resting CO2 baselines between these limits, we deem it unlikely that in subjects with low resting CO2 values (i.e. resting CO2 of 32 mmHg)

PT

these limits still stand. In these subjects, we hypothesize a physiological shift in the sigmoidal curve to

SC

RI

the left.

4.3 Significance of fMRI findings

NU

In the current clinical practice, fMRI is mainly used to locate eloquent areas in the brain, for example in

MA

patients with brain tumors. However, it is known that in these subjects, tumors are prone to elicit neurovascular uncoupling, especially in larger tumor volumes.51 For instance, Zaca et al 2014 found that a reduction in peritumoral BOLD-CVR corresponded to a reduction in fMRI activation.9 Moreover,

PT E

D

patients with diffuse gliomas can exhibit whole brain BOLD-CVR impairment.52 This increases the likelihood of neurovascular uncoupling and heightens the risk for a surgeon to damage an eloquent areas. Given this observation and in agreement with our findings, task-based fMRI in these patients

CE

should not be performed without a corroborated BOLD-CVR study to identify subjects at risk for

AC

neurovascular uncoupling.

Interestingly, Para et al found a lower limit of BOLD-CVR of 0.05 below which fMRI activation was almost completely eradicated. We did not find such a limit for BOLD-CVR in our subjects, as, none of our healthy subject had an average BOLD-CVR below that threshold.

4.4 Breathing pattern change during finger-tapping Interestingly, we observed a change in breathing pattern during the active finger-tapping (either hypoor hyperventilation). This is an important observation as such alterations in breathing pattern can result in significant CO2 fluctuations. Such CO2 fluctuations during the finger-tapping paradigm may obscure

ACCEPTED MANUSCRIPT detection of relevant fMRI signal changes. This potential confounder is eradicated with our calibrated fMRI finger-tapping protocol as this paradigm maintains CO2 levels independent of breathing frequency.10 Therefore, we believe it is imperative to use calibrated (i.e. breathing independent as well as CO2 and O2 controlled) fMRI protocol in which BOLD-CVR and fMRI are investigated simultaneously.

PT

4.5 Limitations

RI

In this study, we scanned subjects at either resting CO2 or a set CO2 baseline. Although our data

SC

strongly suggest that increasing the CO2 level above the subject’s resting state can both cause BOLDCVR impairment and less fMRI task evoked signal activation, we did not scan the same subjects with

NU

both protocols. Therefore, we are unable to make such statements for the individual subject. A more

MA

definitive answer can only be given in a future paper using the same protocol in the same subjects. Despite this limitation we saw a consistent response over all the subjects scanned at a high then normal

D

CO2 baseline. We only investigated the hand area in the motor cortex. We have opted for the hand area,

PT E

as it is an area consistently activated within a known and small region in the brain. Para et al already adopted this functionality of our concept successfully.10 However, other known regions involved in finger-tapping, like supplementary motor area, frontal operculum, basal ganglia and cerebellum, have

CE

not been investigated.

AC

Only subjects without a known neurological medical history, neurological symptoms and use of medicine were included in our study. It is possible that in subjects, especially the older subjects, a dormant-asymptomatic neurovascular pathology could be present. Moreover, BOLD-CVR in the eldest subject in group B (subject 2) might be partially damped by age and vascular stiffening, and therefore show reduced BOLD-CVR 53. However, group B consists further mostly of younger subjects and no significant differences were found between the two groups (see Table1). Moreover, low BOLD-CVR and fMRI values were also present in our younger subjects. As was mentioned before, BOLD-CVR is not only influenced by cerebral blood flow (CBF), but might also be influenced by OEF or CMRO2. Therefore, we wish to state that our results are not directly

ACCEPTED MANUSCRIPT applicable to CBF changes. CBF has shown to be linear over a larger range of CO2 BOLD-CVR, however non-linear at the sublinear phase of the BOLD-CO2 curve.19, 54 This could be an explanation for the weaker correlation seen in group B. Especially in regard to the ROI analysis another explanation has to be stated. Besides, the CBF and OEF coupling, the BOLD responses in both investigations can be influenced by the amount of baseline deoxyhemoglobin within a voxel. (known as parameter M).5 As differences can be seen between

PT

subjects as well as between regions in the brain, this could have induced more pronounced differences.

RI

We have checked for this difference by normalizing task-evoked fMRI.

SC

5. Conclusions

NU

Our main finding is that BOLD-CVR and fMRI activation are significantly lower when measured at a set ‘‘normocapnic’’ baseline of 40mmHg then at the subjects’ physiological resting CO2.

MA

For an accurate interpretation, baseline CO2 levels and BOLD CVR should be considered

PT E

D

complementary to task evoked BOLD fMRI.

6. Funding

CE

This research project was funded by the “Forschungskredit: Postdoc 2016 (FK-16-040) ” and “Swiss

AC

Cancer League 2017 (KFS-3975-08-2016-R)”.

7. Declaration of conflicting interest J. Fierstra was as an external consultant at Thornhill Research Inc., Toronto, Canada (TRI), a spin-off company from the University Health Network Toronto. The company developed the RespirAct TM, which is currently dispensed as a non-commercial research tool around the world. He has worked from June 2009 until April 2011 while working on PhD related research projects involving the RespirActTM system. The other authors report no conflicting interests regarding the research, authorship and/or publication of this article.

ACCEPTED MANUSCRIPT 8. Authors contributions CHBvN and MP equally contributed to the manuscript. Conceived and designed the experiments: CHBvN, MP and JF. Performed the experiments: CHBvN, MP, NM, CS and JF. ROI analysis: CHBvN, CS, AP. Analyzed the data: CHBvN. Interpreted results of experiments: CHBvN, MP, OB, AP, LR, JF. Drafted the manuscript: CHBvN, MP, JF. Revised the manuscript: CHBvN, MP, OB, CS, NM, AP, AV, LR, JF. Approved final version of manuscript: CHBvN, MP, OB, AP, AV, LR, JF.

PT

9. References: Uncategorized References

AC

CE

PT E

D

MA

NU

SC

RI

1. Bandettini PA, Wong EC, Hinks RS, Tikofsky RS, Hyde JS. Time course EPI of human brain function during task activation. Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine 1992;25:390-397. 2. Kwong KK, Belliveau JW, Chesler DA, et al. Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci U S A 1992;89:5675-5679. 3. Ogawa S, Tank DW, Menon R, et al. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci U S A 1992;89:5951-5955. 4. Fox PT, Raichle ME. Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects. Proc Natl Acad Sci U S A 1986;83:11401144. 5. Davis TL, Kwong KK, Weisskoff RM, Rosen BR. Calibrated functional MRI: mapping the dynamics of oxidative metabolism. Proc Natl Acad Sci U S A 1998;95:1834-1839. 6. Ito S, Mardimae A, Han J, et al. Non-invasive prospective targeting of arterial P(CO2) in subjects at rest. The Journal of physiology 2008;586:3675-3682. 7. Fierstra J, Sobczyk O, Battisti-Charbonney A, et al. Measuring cerebrovascular reactivity: what stimulus to use? The Journal of physiology 2013;591:5809-5821. 8. Mutch WA, Mandell DM, Fisher JA, et al. Approaches to brain stress testing: BOLD magnetic resonance imaging with computer-controlled delivery of carbon dioxide. PloS one 2012;7:e47443. 9. Zaca D, Jovicich J, Nadar SR, Voyvodic JT, Pillai JJ. Cerebrovascular reactivity mapping in patients with low grade gliomas undergoing presurgical sensorimotor mapping with BOLD fMRI. Journal of magnetic resonance imaging : JMRI 2014;40:383-390. 10. Para AE, Sam K, Poublanc J, Fisher JA, Crawley AP, Mikulis DJ. Invalidation of fMRI experiments secondary to neurovascular uncoupling in patients with cerebrovascular disease. Journal of magnetic resonance imaging : JMRI 2017. 11. Blicher JU, Stagg CJ, O'Shea J, et al. Visualization of altered neurovascular coupling in chronic stroke patients using multimodal functional MRI. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2012;32:2044-2054. 12. Siero JC, Hartkamp NS, Donahue MJ, et al. Neuronal activation induced BOLD and CBF responses upon acetazolamide administration in patients with steno-occlusive artery disease. NeuroImage 2015;105:276-285. 13. Bright MG, Croal PL, Blockley NP, Bulte DP. Multiparametric measurement of cerebral physiology using calibrated fMRI. NeuroImage 2017. 14. Sobczyk O, Crawley AP, Poublanc J, et al. Identifying Significant Changes in Cerebrovascular Reactivity to Carbon Dioxide. AJNR American journal of neuroradiology 2016;37:818-824. 15. Kassner A, Winter JD, Poublanc J, Mikulis DJ, Crawley AP. Blood-oxygen level dependent MRI measures of cerebrovascular reactivity using a controlled respiratory challenge: reproducibility and gender differences. Journal of magnetic resonance imaging : JMRI 2010;31:298-304. 16. Sam K, Conklin J, Holmes KR, et al. Impaired dynamic cerebrovascular response to hypercapnia predicts development of white matter hyperintensities. NeuroImage Clinical 2016;11:796-801.

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

17. Fierstra J, Spieth S, Tran L, et al. Severely impaired cerebrovascular reserve in patients with cerebral proliferative angiopathy. Journal of neurosurgery Pediatrics 2011;8:310-315. 18. Fisher JA, Sobczyk O, Crawley A, et al. Assessing cerebrovascular reactivity by the pattern of response to progressive hypercapnia. Human brain mapping 2017. 19. Halani S, Kwinta JB, Golestani AM, Khatamian YB, Chen JJ. Comparing cerebrovascular reactivity measured using BOLD and cerebral blood flow MRI: The effect of basal vascular tension on vasodilatory and vasoconstrictive reactivity. NeuroImage 2015;110:110-123. 20. van Niftrik CHB, Piccirelli M, Bozinov O, et al. Iterative analysis of cerebrovascular reactivity dynamic response by temporal decomposition. Brain and behavior 2017:e00705. 21. Yousry TA, Schmid UD, Alkadhi H, et al. Localization of the motor hand area to a knob on the precentral gyrus. A new landmark. Brain : a journal of neurology 1997;120 ( Pt 1):141-157. 22. Sommer WH, Bollwein C, Thierfelder KM, et al. Crossed cerebellar diaschisis in patients with acute middle cerebral artery infarction: Occurrence and perfusion characteristics. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2016;36:743-754. 23. Sobczyk O, Battisti-Charbonney A, Fierstra J, et al. A conceptual model for CO(2)-induced redistribution of cerebral blood flow with experimental confirmation using BOLD MRI. NeuroImage 2014;92:56-68. 24. Bhogal AA, Siero JC, Fisher JA, et al. Investigating the non-linearity of the BOLD cerebrovascular reactivity response to targeted hypo/hypercapnia at 7T. NeuroImage 2014;98:296-305. 25. Cohen ER, Ugurbil K, Kim SG. Effect of basal conditions on the magnitude and dynamics of the blood oxygenation level-dependent fMRI response. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2002;22:1042-1053. 26. Kety SS, Schmidt CF. THE EFFECTS OF ALTERED ARTERIAL TENSIONS OF CARBON DIOXIDE AND OXYGEN ON CEREBRAL BLOOD FLOW AND CEREBRAL OXYGEN CONSUMPTION OF NORMAL YOUNG MEN. The Journal of clinical investigation 1948;27:484-492. 27. Grubb RL, Jr., Raichle ME, Eichling JO, Ter-Pogossian MM. The effects of changes in PaCO2 on cerebral blood volume, blood flow, and vascular mean transit time. Stroke; a journal of cerebral circulation 1974;5:630-639. 28. Mandell DM, Han JS, Poublanc J, et al. Mapping cerebrovascular reactivity using blood oxygen level-dependent MRI in Patients with arterial steno-occlusive disease: comparison with arterial spin labeling MRI. Stroke; a journal of cerebral circulation 2008;39:2021-2028. 29. Stefanovic B, Warnking JM, Rylander KM, Pike GB. The effect of global cerebral vasodilation on focal activation hemodynamics. NeuroImage 2006;30:726-734. 30. Yezhuvath US, Lewis-Amezcua K, Varghese R, Xiao G, Lu H. On the assessment of cerebrovascular reactivity using hypercapnia BOLD MRI. NMR in biomedicine 2009;22:779-786. 31. Yablonskiy DA. Cerebral metabolic rate in hypercapnia: controversy continues. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2011;31:1502-1503. 32. Chen JJ, Pike GB. Global cerebral oxidative metabolism during hypercapnia and hypocapnia in humans: implications for BOLD fMRI. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2010;30:1094-1099. 33. Jain V, Langham MC, Floyd TF, Jain G, Magland JF, Wehrli FW. Rapid magnetic resonance measurement of global cerebral metabolic rate of oxygen consumption in humans during rest and hypercapnia. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2011;31:1504-1512. 34. Hall EL, Driver ID, Croal PL, et al. The effect of hypercapnia on resting and stimulus induced MEG signals. NeuroImage 2011;58:1034-1043. 35. Xu F, Uh J, Brier MR, et al. The influence of carbon dioxide on brain activity and metabolism in conscious humans. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2011;31:58-67. 36. Zappe AC, Uludag K, Oeltermann A, Ugurbil K, Logothetis NK. The influence of moderate hypercapnia on neural activity in the anesthetized nonhuman primate. Cerebral cortex 2008;18:2666-2673. 37. Jussen D, Zdunczyk A, Schmidt S, et al. Motor plasticity after extra-intracranial bypass surgery in occlusive cerebrovascular disease. Neurology 2016;87:27-35.

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

38. Herzig R, Hlustik P, Skoloudik D, et al. Assessment of the cerebral vasomotor reactivity in internal carotid artery occlusion using a transcranial Doppler sonography and functional MRI. Journal of neuroimaging : official journal of the American Society of Neuroimaging 2008;18:38-45. 39. van Niftrik CH, Piccirelli M, Bozinov O, et al. Fine tuning breath-hold-based cerebrovascular reactivity analysis models. Brain and behavior 2016;6:e00426. 40. Fierstra J, van Niftrik B, Piccirelli M, et al. Altered intraoperative cerebrovascular reactivity in brain areas of high-grade glioma recurrence. Magnetic resonance imaging 2016;34:803-808. 41. Strother MK, Buckingham C, Faraco CC, et al. Crossed cerebellar diaschisis after stroke identified noninvasively with cerebral blood flow-weighted arterial spin labeling MRI. Eur J Radiol 2016;85:136-142. 42. da Costa L, van Niftrik CB, Crane D, Fierstra J, Bethune A. Temporal Profile of Cerebrovascular Reactivity Impairment, Gray Matter Volumes, and Persistent Symptoms after Mild Traumatic Head Injury. Front Neurol 2016;7:70. 43. Sam K, Crawley AP, Poublanc J, et al. Vascular Dysfunction in Leukoaraiosis. AJNR American journal of neuroradiology 2016;37:2258-2264. 44. Marshall RS, Festa JR, Cheung YK, et al. Cerebral hemodynamics and cognitive impairment: baseline data from the RECON trial. Neurology 2012;78:250-255. 45. Sobczyk O, Battisti-Charbonney A, Poublanc J, et al. Assessing cerebrovascular reactivity abnormality by comparison to a reference atlas. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2015;35:213-220. 46. Esposito G, Fierstra J, Kronenburg A, Regli L. A comment on "Contralateral cerebral hemodynamic changes after unilateral direct revascularization in patients with moyamoya disease". Neurosurgical review 2012;35:141-143; author reply 143. 47. Carrera E, Kim DJ, Castellani G, et al. Effect of hyper- and hypocapnia on cerebral arterial compliance in normal subjects. Journal of neuroimaging : official journal of the American Society of Neuroimaging 2011;21:121-125. 48. Golestani AM, Kwinta JB, Strother SC, Khatamian YB, Chen JJ. The association between cerebrovascular reactivity and resting-state fMRI functional connectivity in healthy adults: The influence of basal carbon dioxide. NeuroImage 2016;132:301-313. 49. Battisti-Charbonney A, Fisher J, Duffin J. The cerebrovascular response to carbon dioxide in humans. The Journal of physiology 2011;589:3039-3048. 50. Tancredi FB, Hoge RD. Comparison of cerebral vascular reactivity measures obtained using breathholding and CO2 inhalation. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 2013;33:1066-1074. 51. Southwell DG, Hervey-Jumper SL, Perry DW, Berger MS. Intraoperative mapping during repeat awake craniotomy reveals the functional plasticity of adult cortex. J Neurosurg 2016;124:1460-1469. 52. Fierstra J, van Niftrik C, Piccirelli M, et al. Diffuse gliomas exhibit whole brain impaired cerebrovascular reactivity. Magnetic resonance imaging 2017;45:78-83. 53. Leoni RF, Oliveira IA, Pontes-Neto OM, Santos AC, Leite JP. Cerebral blood flow and vasoreactivity in aging: an arterial spin labeling study. Brazilian journal of medical and biological research = Revista brasileira de pesquisas medicas e biologicas 2017;50:e5670. 54. Fierstra J, van Niftrik C, Warnock G, et al. Staging Hemodynamic Failure With Blood OxygenLevel–Dependent Functional Magnetic Resonance Imaging Cerebrovascular Reactivity. Stroke; a journal of cerebral circulation 2018:STROKEAHA.117.020010. 55. Poublanc J, Crawley A, Sobzcyk O, et al. Measuring cerebrovascular reactivity: the dynamic response to a step hypercapnic stimulus. 2015.

ACCEPTED MANUSCRIPT Figure captions: Figure 1: Representative slice of the BOLD-CVR and task-based fMRI maps of all subjects. BOLDCVR was presented between -0.6 and 0.6% BOLD signal change/mmHg CO2 change. The fMRI task based maps were shown for t values between 5 (Family Wise Error Rate of 0.05) and 8 to obtain an optimal visual comparison. Subjects in group A – scanned at resting CO2 – clearly demonstrate an overall heightened BOLD-CVR as well as fMRI activation as opposed to group B subjects – scanned at isocapnic baseline of 40mmHg. As can be seen, not all subjects in group B experience a decrease in

PT

fMRI activation, e.g. subject 1 group B, who had a resting CO2 baseline of 38. BOLD: Blood-Oxygenation-Level-Dependent, CO2: carbon dioxide, BOLD-CVR: cerebrovascular

RI

reactivity, fMRI: functional magnetic resonance imaging

SC

Figure 2: Correlation between mean task-based BOLD fMRI signal activation and BOLD CVR within the hand area ROI for all subjects. Each point represents either the left or the right hand areas Panel A

NU

shows the overall correlation included all the subjects of both groups. Panel B and C show the two respective groups separately. The squared correlation coefficients (R2) are shown in the top right corner

MA

of each Panel. More than half of the variance of the relative change in fMRI is explained by the BOLDCVR considering the two groups together – Panel A. No significant difference was found between correlation B and C (p=0.5)

D

BOLD: Blood-Oxygenation-Level-Dependent, BOLD-CVR: cerebrovascular reactivity, fMRI:

AC

CE

PT E

functional magnetic resonance imaging,

ACCEPTED MANUSCRIPT Table 1: Baseline Characteristics (mean ± SD)

BOLD-CVR fMRI

Group A

Group B

Age Gender

35.1 ± 11.5 6 F and 4 M

33.2 ± 12.3 7 F and 3 M

Mean resting PetCO2 Mean PetCO2 baseline

36.4 ± 3.4 37.3 ± 2.5

34.1 ± 2.3 41.3 ± 2.6*

Mean PetCO2 hypercapnia

46.9 ± 2.9

51.0 ± 2.5*

Mean PetCO2

37.2 ± 3.2

41.0 ± 2.4*

AC

CE

PT E

D

MA

NU

SC

RI

PT

* Significant (p<0.05) difference between group A and B Abbreviations: PetCO2 = partial pressure of end tidal carbon dioxide, CVR = blood oxygen level dependent derived cerebrovascular reactivity, fMRI = blood oxygen level dependent task based magnetic resonance imaging SD = Standard Deviation

Figure 1

Figure 2