Sex comparison of cerebrovascular reactivity to CO₂ at global, lobe, and regional levels in healthy young adults

Abstract

Cerebrovascular and neurologic diseases show sex-specific risks and regional variation, both of which are linked to changes in cerebral blood flow (CBF), suggesting that sex or sex hormones may influence cerebrovascular function. Cerebrovascular reactivity (CVR) to hypercapnia may help predict these outcomes, but the identification of fundamental sex differences in CVR remain equivocal. Most CVR studies use methods that cannot identify regional variation, which may be clinically relevant. This study examined if sex differences exist in CVR, at the whole brain, lobular, and regional levels while controlling for confounding factors like menstrual cycle and hormonal contraception. 87 young, healthy adults (38F, 49M; F:22±4yrs, M:23±4yrs, 18–37yrs) from two similar cohorts were studied using arterial spin labeling (pcASL) in a 3T magnetic resonance imaging (MRI) scanner to derive microvascular CBF at room-air and during hypercapnia (~+8 mmHg end-tidal CO₂). The cohorts had identical inclusion criteria but used different hypercapnia methods. In a subset (n=37), CVR was measured using 4D flow MRI of 11 cerebral arteries. While females displayed greater resting CBF, no sex differences in CVR using pcASL were observed in whole-brain (females=4.1±1.9, males=4.9±3.0%/mmHg), grey matter, white matter, or regions of interest forming the grey matter (all p>0.05). Similarly, 4D flow-derived CVR was not different between sexes at a whole-brain (females=3.5±0.9, males=3.4±1.8%/mmHg) or individual artery level (all p>0.05). These results indicate that when controlling for the confounders of age, menstrual cycle, hormonal contraception, and comorbidities, no sex differences in CVR are detected at global, lobular, arterial, or regional levels.

Graphical Abstract

Background

Cerebrovascular disease afflicts millions of people globally and is associated with altered basal cerebral blood flow (CBF) and the control of the cerebral vasculature (1-4). Specifically, the control of CBF during hypercapnia, termed cerebrovascular reactivity (CVR), is critical in acutely maintaining cerebral acid-base balance and oxygen delivery (5) while providing a window into cerebrovascular health indicative of long-term health outcomes, such as stroke, cognitive decline, or Alzheimer’s disease which often exhibit sex-specific patterns (2, 3). For example, stroke risk is lower in premenopausal women compared to their age-matched male counterparts (6), an effect abolished post-menopause which implicates a potential role of sex or sex hormones as influential in cerebrovascular health. Additionally, these pathologies appear regionally specific in both males and females, such as stroke occurrence being much greater in the parietal lobes or regions supplied by the middle cerebral artery (7, 8). Similarly, regional relationships are observed such that cognitive decline is associated with reduced CVR in some lobes (e.g., temporal) or regions (e.g., hippocampus) related to cognition but not others (e.g., occipital lobe, occipital pole) (9, 10). Consistent with this notion, CVR also appears related to executive function in a regionally heterogeneous manner spanning multiple lobes throughout the brain (11).

Given the associations between CVR and cerebrovascular risk on both global and regional levels, understanding fundamental sex differences in CVR may offer insight into observed sex discrepancies in stroke and neurological disease (6). However, investigations assessing sex-specific CVR responses in young, healthy adults remain equivocal and highly context dependent, with some evidence finding greater CVR in females (12-19), males (20, 21), or neither group (22-26). Even among studies reporting higher CVR in females, such findings appear contingent either on the method of CBF quantification or region being assessed. For example, previous reports using duplex-doppler ultrasound or transcranial doppler ultrasound (TCD) have found greater CVR in females only in the vertebral artery, but not internal carotid artery (18) or only in the middle cerebral artery, but not posterior cerebral artery (19). Interestingly, studies using magnetic resonance imaging (MRI) to assess sex-specific responses report greater CVR in males in the grey matter (GM), but not white matter (WM) (20) while others have found more ubiquitous elevations in CVR across multiple vessels (21). The above evidence points towards regional heterogeneity in CVR between the sexes. Given the evidence of the clinical relevance of regional differences in CVR noted above (9-11), more rigorous investigation of the role of sex in CVR is critical.

The discrepancy in identifying sex differences may arise from methodological differences in quantifying CBF or controlling for critical confounders. Most prior studies used TCD to assess the velocity of intracranial arteries (12, 13, 15-17, 19, 22-26). TCD offers high temporal resolution, but it is now clear that large intracranial arteries like the middle cerebral artery dilate during hypercapnia, resulting in up to ~50% underestimations of CVR (21, 27-30). Alternatively, studies utilizing duplex-doppler ultrasound can derive CBF through arteries for more accurate CVR measures (18), but these measures are limited to extracranial arteries and lack information on potential regional CVR differences. Interestingly, studies using MRI to assess sex-specific responses report greater CVR in males (20, 21). MRI measures have the distinct advantage of providing both global and regional intracranial CBF measures but greater CVR in males in these studies diverge from the bulk of the literature showing either no difference or greater CVR in females.

Further variability in sex comparisons of CVR might stem from study design factors. Specifically, studies have not accounted for menstrual cycle phase (15, 16, 18, 23, 25), allow hormonal contraceptive use (13, 21, 22, 24, 26), or have wide ranges for inclusion of age (17, 23) and overweight/obesity (21, 22). Each of these factors are known to alter vascular function and may influence basal CBF or CVR responses (19, 21, 31-34) and obscure the ability to conclusively test the existence of sex differences in CVR.

The collective literature assessing sex differences in CVR remains inconclusive. Interestingly, studies using TCD (19), Doppler ultrasound (18), and MRI (20) all hint at sex differences in CVR being region-specific. Therefore, the goal of this study was to determine if fundamental sex differences in CVR exist by employing comprehensive MRI on a global, lobular, and regional level while attempting to tightly control potential confounders observed in previous studies. Additionally, we utilized a second MRI method in a subset of participants to assess CVR in an artery-specific manner to further our understanding of regional variations in CVR. Based on previous MRI studies (20, 21), we hypothesized that males would display greater CVR than their age-matched female counterparts on global, lobular, and regional levels as well as by individual intracranial arteries.

Methods

Subjects

In this report, investigations of two separate cohorts from larger studies aimed at mechanistically assessing the regulation of CBF were combined. These similar cohorts were studied at two separate 3T MRI scanners, both conducted at the University of Wisconsin–Madison. Three subjects participated in both cohorts and were assigned to cohort 2 for analyses based on our use of an end-tidal forcing system to administer hypercapnic gas (see Instrumentation). Cohort 1 consisted of a final sample size of 56 subjects (23 females), while cohort 2 consisted of 31 subjects (15 females) after assessing for MRI scan quality, both using identical inclusion/exclusion criteria (see below). Of note, data from cohort 2 comes from the placebo visit of a randomized double-blind, placebo-controlled intervention aimed at mechanistically assessing the control of CBF and CVR, whereas no placebo pill was taken in cohort 1. Written informed consent was obtained before study participation. All procedures were approved by the University of Wisconsin–Madison Health Sciences Institutional Review Board and were in accordance with the Declaration of Helsinki (registered in the ClinicalTrials.gov database (ID: NCT04265053)).

All subjects completed an initial screening visit to determine eligibility with identical inclusion and exclusion criteria. Subjects completed a medical history questionnaire, MRI safety questionnaire, venipuncture, pregnancy test (females only), and anthropometric measurements (height, weight, waist and hip circumference). Brachial artery blood pressure was measured in triplicate using an automated sphygmomanometer (Datex Ohmeda) and the lowest of the 3 measurements was used to assess eligibility. A fasted (≥ 8 hrs) venous blood sample was obtained, and plasma glucose and lipids were measured (CardioChek).

All subjects were 18–40 years old (range 18–37) with a systolic blood pressure (SBP) of ≤ 125 mmHg, diastolic blood pressure (DBP) of ≤ 80 mmHg, body mass index (BMI) ≤ 25 kg/m², blood glucose < 100mg/dL, low-density lipoprotein (LDL) < 130 mg/dL, and triglycerides < 150 mg/dL. Exclusion criteria included smoking, taking any cardiovascular, hormonal, or metabolic medication, any overt cardiovascular, neurological, autoimmune, or reproductive disorders and MRI contraindications. Females were not pregnant (confirmed by urine pregnancy test) or lactating, were regularly menstruating and not currently taking hormonal contraceptives.

MRI Study Visit

Pre-visit Procedures

Subjects attended a single MRI visit after fasting for ≥ 8 hours, as well as abstaining from vigorous exercise, caffeine, and non-steroidal anti-inflammatory drugs for ≥ 24 hours. Female subjects were tested in the early follicular phase (days 1–7) of their menstrual cycle, confirmed by the onset of menses. Negative urine pregnancy status was confirmed again on day of study.

Instrumentation

MRI Scanner and Monitoring: Both MRI scanners were General Electric (GE Healthcare, Waukesha, WI, USA) 3 Tesla systems. Cohort 1 used a Discovery MR750, and cohort 2 used a MAGNUS scanner. Head Coil: Cohort 1 used a 48-channel GE head coil while cohort 2 used a 32-channel phased array receive-only head coil (Nova Medical, Wilmington, MA). Breath-by-breath end-tidal carbon dioxide (ETCO₂): Cohort 1 was assessed via capnography (Medrad Veris MR Vital Signs Patient Monitor; Bayer Healthcare, Whippany, NJ) and cohort 2 utilized an end-tidal forcing system (RespirAct, Thornhill Medical, Toronto, Canada). Heart rate and pressure monitoring: In both cohorts, HR was measured with a pulse oximeter and brachial artery blood pressure was measured using an automated sphygmomanometer. HR and ETCO₂ were recorded every minute and averaged across the duration of the T1 and CBF scans (~4 minutes each) whereas SBP, DBP, and mean arterial pressure (MAP) were measured and recorded during the final minute of each scan.

MRI scanning Protocol

Subjects were instructed to lie still and awake with their eyes closed during the MRI scans. A T1-weighted image set and pseudo-continuous Arterial Spin Labeling (pcASL) scans were completed (~4 minutes) during baseline (room air) and pcASL during hypercapnia. Hypercapnia was administered in cohort 1 via inhalation of a 5% CO₂, 21% O₂, N₂ balanced gas mixture through a Hans Rudolph mask (7450 V2 mask and headgear) with a two-way non-rebreathing valve (2700 Large T-Shape) that was tightly fitted over the subject’s nose and mouth. Scans were not performed until ETCO₂ reached a steady state during hypercapnia (typically achieved within ≤3 minutes). In cohort 2, an end-tidal forcing system (RespirAct) was used to passively measure ETCO₂ during the T1 and room air pcASL scans; the system’s preparatory (prep) phase, which necessarily induces a brief period of hypercapnia, was performed after these scans to avoid the influence of this hypercapnic exposure on room air CBF measures. The end-tidal forcing system was then used to actively administer hypercapnia while maintaining end-tidal O₂ at room air levels. After a 1 min ramp towards the target ETCO₂, ETCO₂ plateaued and pcASL scanning began. Both cohorts aimed at increasing ETCO₂ by ~8 mmHg.

To determine if methodological discrepancies are responsible for differences in identifying sex-specific CVR responses, a subset of participants in cohort 1 (24M, 13F) had CBF quantified using four-dimensional (4D) flow scan (4D flow) called phase contrast vastly undersampled isotropic projection reconstruction (PC-VIPR) (33, 34). In these visits, 4D flow was captured at rest prior to resting pcASL, with the second 4D flow scan following immediately after the pcASL scan during steady-state hypercapnia condition using the same 5% CO₂, 21% O₂ gas mixture. Unlike the lobular and regional CBF measures taken by pcASL, 4D flow allows for the quantification of CBF through individual cerebral arteries (see below).

While both methods are built on specific MRI assumptions, they are both validated CBF measures. Using this approach allowed us to compare CVR at the macrovascular and microvascular levels in the same subjects on the same day with the same hypercapnic stimulus, with the aim to bolster confidence in interpreting data toward firm conclusions on sex differences in CVR.

T1 Scan Parameters

Cohort 1:

A T1-weighted MRI scan was completed using an ADNI 3 protocol on a 3T GE 750 scanner. An accelerated sagittal 3D IR-FSPGR (Inversion recovery fast spoiled gradient-echo) sequence was completed then utilized for brain structure analysis. Parameters of the scan were: flip angle = 11; repetition time = 7.2 ms; echo time = 2.9 ms; inversion time = 400 ms; field-of-view = 27 cm; slice thickness = 1 mm; matrix size = 256 x 256; and number of excitations (NEX) = 1. All T1 images were reviewed independently to remove any subjects who demonstrated motion artifact that could reduce validity of CBF measures. In both cohorts, T1 scan quality was determined by assessing the overall weighted image quality (IQR), noise contrast ratio (NCR), and inhomogeneity contrast ratio (ICR) as described by Gaser (10.1093/gigascience/giae049). Briefly, scan quality is graded on a 1–6 scale, with lower numbers being indicative of higher quality or lower motion artifact. Only scans with quality scores of 3.5 or lower in these three metrics (indicative of a “satisfactory” or greater rating) were used for final analyses. Mean ± standard deviation (SD) values for each measure were as follows: IQR 2.58 ± 0.27; NCR 2.78 ± 0.32; ICR 1.16 ± 0.21. Furthermore, all T1 scans were visually assessed for sufficient contrast, low motion artifact, and the absence of overt structural abnormalities. These quantitative and qualitative analyses are indicative of acceptable scan quality, motion artifact, and signal-to-noise ratio.

Cohort 2:

Similarly, T1-weighted MRI scans used an ADNI 3 protocol with an accelerated sagittal 3-D IR-FSPGR sequence. The parameters were as follows: flip angle = 8, repetition time = 9.624 ms, echo time = 4.056 ms, inversion time = 1,060 ms, field of view = 25.6 cm, slice thickness = 1 mm, matrix size = 256 x 256, and number of excitations = 1. As in cohort 1, all T1 images were quality assessed based on motion artifact.

pcASL Labeling Parameters

Cohort 1:

A background-suppressed pcASL was used to assess cerebral perfusion and perform image registration. The pcASL sequence featured a 3D fast spin-echo spiral readout utilizing a stack of variable-density 5 ms readout and eight interleaves. Parameters of the scan were: flip angle = 111; repetition time = 6,031 ms; echo time = 64.8 ms; inversion time = 1,000 ms; field-of-view = 240 mm; slice thickness = 4 mm no gap; matrix size = 128 x 128; number of excitations (NEX) = 1; and labeling radiofrequency amplitude = 0.24 mG. Multi–post-labeling delay (multi-PLD) pcASL acquisition with Hadamard encoding, incorporating three PLDs (1.0, 1.8, and 2.7 s), was used to quantify transit time corrected CBF. During the same imaging sequence / image slab location as the pcASL, we acquired a fluid-suppressed proton density (PD) scan but without radiofrequency labeling preparation. Preprocessing qualitative assessment of all pcASL scans were conducted using the MicroDicom viewer (version 2026.3; https://www.microdicom.com/) to confirm that there was no visually apparent signal loss and that no slices were corrupted or absent for both room air and hypercapnia scans. In total, three participants (1 female, 2 males) in cohort 1 were excluded due to poor scan quality.

Cohort 2:

pcASL scan parameters for cohort 2 were as follows: flip angle = 111, repetition time = 5,457 ms, echo time = 55.7 ms, inversion time = 1,000 ms, field of view = 240 mm, slice thickness = 4 mm no gap, matrix size = 128 x 128, number of excitations = 1, and labeling radiofrequency amplitude = 0.24 mG. Multi-PLD at 1.0, 1.8, and 2.7s quantified transit time corrected CBF. Fluid-suppressed proton density scan without radiofrequency labeling preparation was acquired during the pcASL imaging sequence. One female was excluded from cohort 2 due to poor scan quality.

PC-VIPR Parameters (4D Flow)

Using 4D flow, blood flow through 11 intracranial conduit arteries is gathered and has been used previously by our lab (35) and others (21). The following scan parameters were used: imaging volume = 22 x 22 x 22 cm³, acquired isotropic spatial resolution = (0.69 mm³), scan time = 5 min 38 s, velocity encoding = 100 cm/s, flip angle = 8°, and repetition time/echo time = 6.7/2.8 ms. After reviewing PC-VIPR scans, 1 female was removed due to low image quality.

Data Processing

pcASL Analysis

T1-weighted and pcASL images were analyzed in MATLAB (MATLAB, The Mathworks, Natick, MA, USA) using a Computational Anatomy Toolbox (CAT12) for SPM12 (statistical Parametric Mapping version 12) as described previously (35). T1-weighted Digital Imaging and Communications in Medicine (DICOM) and pcASL DICOM images were converted into Neuroimaging Informatics Technology Initiative files, then processed in MATLAB. CAT12 was used to resample T1-weighted scans and correct for noise, bias, and global intensities. Arterial transit time estimation was performed by fitting the ASL signal across PLDs using a multi-PLD kinetic model, based on the Buxton framework and its extensions (36) T1-weighted scans were then segmented to obtain grey matter (GM), white matter (WM), and were regionally segmented using the Neuromorphometrics atlas in MNI152NLin2009cAsym space (37). Partial GM masks were co-registered to pcASL scans using an affine transformation and smoothed using a 7 mm full-width at half-maximum Gaussian kernel. Total GM and WM perfusion assessed in native space, while regional perfusion was assessed relative to the Neuromorphometrics atlas (37). All scans were quality assessed independently by two lab members.

PC-VIPR Analysis

PC-VIPR was analyzed as previously reported (35) using validated in-house processing software developed in MATLAB (38). This software allows for processing of segments perpendicular to individual vessels one voxel in width. For each vessel, three to five consecutive cross sections were analyzed and averaged to derive flow, cross-sectional area (CSA), and velocity. This method has been previously used to measure CBF at rest and during hypercapnic and hypoxic gas challenges (21, 35, 39) in healthy adults and a variety of clinical populations (31, 40). Additionally, whole brain CBF was calculated as the sum of left and right internal carotid and left and right vertebral arteries.

Cerebrovascular Reactivity Calculations

pcASL allows the determination of CBF in the whole brain, GM, and WM as well as lobes constituting the GM (i.e., subcortical regions, brainstem and cerebellum, frontal, temporal, occipital, and parietal lobes) and the individual regions within these lobes. Similarly, 4D flow yields blood flow through individual intracranial arteries and allows for the calculation of whole brain CBF from the addition of left and right internal carotid and vertebral arteries. Therefore, CVR was determined on a whole brain, lobular, regional, and individual cerebral artery level using the following calculation: $\text{CVR}(\%∕\text{mmHg})=\%\Delta \text{CBF}∕\Delta {\text{ETCO}}_2$, where $\%\Delta \text{CBF}$ is the relative increase in CBF in the brain area being assessed and ΔETCO₂ is the change in ETCO₂ from room air to hypercapnia (5). As ETCO₂ reached a steady state at both room air and during hypercapnia ( Figure 1), ETCO₂ was averaged over the duration of each pcASL scan to calculate ΔETCO₂. Additionally, changes in cerebrovascular conductance (CBF / MAP) were used to calculate CVR to account for sex differences in MAP (Supplemental Materials).

Figure 1. Demonstration of CVR testing stability.

A) Individual ETCO₂ responses at rest vs hypercapnia for each subject. Cohorts are black (1, solid) or red (2, dashed), with different symbols for sex. While cohort 2 had higher resting ETCO₂ (p < 0.001), the levels were stable for each MRI scan and the rise in ETCO₂ was similar between sexes and cohorts (sex-by-cohort interaction p = 0.624). B) Average ETCO₂ data from panel A. C) 3 subjects completed CVR testing in both cohorts and yielded nearly identical CVR values. A steady state was achieved in both cohorts in males and females, with no change observed across hypercapnia (p = 0.550). No cohort x sex interaction effects were observed (p = 0.624).

Sex Hormone Analyses

In a subset of participants (7M, 13F), sex steroid concentrations were analyzed from fasted venous blood samples taken on the day of the MRI scans. These samples were centrifuged, and serum was drawn and stored at −80°C. Sex steroid concentrations were analyzed at the Wisconsin National Primate Research Center. Concentrations of estradiol, testosterone, and progesterone were determined via liquid chromatography-tandem mass spectrometry. Dehydroepiandrosterone (DHEA-S) and sex hormone-binding globulin (SHBG) were quantified by Roche electrochemiluminescence immunoassay.

Cohort Comparisons

This study combines two cohorts to answer a fundamental question in human cerebrovascular control, as done in previous studies (41-43). Cohorts 1 and 2 were largely similar, with identical inclusion criteria, method of quantifying CBF, and hypercapnic ETCO₂ target and duration. However, these cohorts did differ in hypercapnia delivery (5% CO₂ gas tank versus end-tidal forcing system) and CBF measures that occurred in two different GE 3T MRI scanners. Critical to our conclusions in the current report is determining if these cohorts are comparable. To do so, two-way ANOVA was performed between groups to determine if a cohort-by-sex interaction effect was present in 1) baseline subject characteristics (Supplemental Table 1), 2) CBF and MAP responses to hypercapnia (Figure 2, Supplemental Figure 1, Results text), and 3) the magnitude of rise and subsequent attainment of steady-state ETCO₂ ( Figure 1). Minor differences were observed in baseline characteristics (see Results text) while no sex or sex-by-cohort interactions were observed in CVR, blood pressure responses, or the degree of hypercapnia between cohorts. Therefore, these cohorts appear comparable and are minimally biased by method of hypercapnic delivery or the MRI scanner used. We then chose to combine cohorts to study sex differences in CVR with greater statistical power.

Figure 2. Whole Brain Cerebrovascular Reactivity (CVR) Comparison Between Cohorts.

Whole brain CVR was greater in Cohort 2 (p = 0.019). However, no effect of sex (p = 0.097) or sex-by-cohort interactions were found (p = 0.654), suggesting that the elevated reactivity found in Cohort 2 is not sex-specific. Two-way ANOVA. Flow derived from pcASL MRI.

Statistical Analysis

Power was determined using G*Power (v3.1.9.7). Of the two previous MRI studies investigating sex differences in CVR (20, 21), the effect sizes observed in GM were very large (Cohen’s d of 1.46). Given the mixed results in the field in identifying CVR, a more conservative effect size of 0.694 was chosen for a priori power analyses based on sex differences observed in WM. A sample size of 34/sex was determined to achieve a power of 0.8 using unpaired samples t tests between sexes. We then chose to combine cohorts in order to exceed this sample size of 34/sex to allow for detection of CVR differences of smaller regions or vessels. Data were assessed for normality and equal variance using the Shapiro-Wilk and Brown-Forsythe tests, respectively. When cohorts were combined and analyzed together, unpaired student’s t tests were used to evaluate sex differences in total, GM, and WM CVR. Non-normal data was tested using the Mann-Whitney U test. Shaffer’s sequential method was used as described previously (35) to correct for statistically comparing multiple regions within a given lobe (e.g., 18 regions comprising the temporal lobe). Finally, ANCOVA was used to determine if cohort or MAP were a significant predictor of CVR responses. Significance was set at p < 0.05. All data are expressed as mean ± SD.

Results

Subject Characteristics and Cohort Comparisons

Subject characteristics are summarized in Table 1. By design, all subjects were normotensive, had a healthy BMI, and glucose and lipid levels associated with low cardiovascular or cerebrovascular disease risk. While all subjects were normotensive, males had significantly higher SBP (p<0.001) and MAP (p = 0.004). Lastly, females had significantly higher concentrations of high-density lipoprotein (HDL; p = 0.002), whereas males had significantly higher triglycerides (p = 0.046). Males had higher serum testosterone and lower SHBG concentrations than females (Table 2). Notably, estradiol and progesterone values in females are consistent with early follicular phase values (44).

Table 1.

Subject characteristics of the combined cohorts. Data are presented as mean ± SD. * Significantly different from males, using unpaired t tests, p < 0.05.

	Males (n = 49)	Females (n = 38)	p-value
Age, yrs	23 ± 4	22 ± 4	0.472
Height, cm	178.5 ± 8.3	165.0 ± 5.8*	<0.001
Weight, kg	72.0 ± 7.3	60.3 ± 6.4*	<0.001
BMI, kg/m2	22.6 ± 1.6	22.1 ± 1.8	0.159
SBP, mmHg	115 ± 7	108 ± 8*	<0.001
DBP, mmHg	68 ± 5	67 ± 6	0.275
MAP, mmHg	84 ± 5	80 ± 6*	0.005
Glucose, mg/dL	81 ± 10	79 ± 8	0.440
Cholesterol, mg/dL
Total	138 ± 28	147 ± 27	0.185
HDL	54 ± 12	62 ± 11*	0.002
LDL	68 ± 23	71 ± 22	0.533
Triglycerides, mg/dL	75 ± 24	64 ±18*	0.020

Table 2.

Serum sex steroid concentrations. Data are presented as mean ± SD. * Significantly different from males, using unpaired t tests, p < 0.05. DHEA-S = dehydroepiandrosterone; SHBG = sex hormone-binding globulin. ^a indicates n = 11.

	Males (n = 7)	Females (n = 13)	p-value
Estradiol, pg/mL	19 ± 12	25 ± 13a	0.340
Testosterone, pg/mL	4035 ± 592	201 ± 68*	<0.001
Progesterone, pg/mL	66 ± 31	72 ± 55	0.768
DHEA-S, μg/mL	313 ± 123	228 ± 113	0.136
SHBG, nmol/L	33 ± 9	61 ± 21*	0.003

To determine if cohorts were comparable, two-way ANOVA was used to determine if the groups differed in subject characteristics or cardiorespiratory responses to different hypercapnic delivery systems. Cohort 2 had greater resting SBP, MAP, and fasting glucose than cohort 1 (all p ≤ 0.011), but no cohort-by-sex interaction effect was determined (p ≥ 0.080; Supplemental Table 1). Similarly, cohorts did not differ in any other recorded baseline measures (all p ≥ 0.130). Two-way ANOVA did reveal that cohort 2 had higher MAP (p = 0.004), ETCO₂ (p < 0.001), and whole brain CVR (p = 0.019) responses to hypercapnia. Importantly, however, no sex or sex-by-cohort interactions were found in MAP (p = 0.191), ETCO₂ (p = 0.624; Figure 1), or CVR (p = 0.654; Figure 2). This indicates that, while CVR and cardiorespiratory variables were greater in the MRI and gas delivery systems used in cohort 2, this effect does not appear to differentially bias the sexes. Therefore, these analyses demonstrate that from technical and demographic perspectives, cohorts are comparable to address the primary aims of this study.

Cardiorespiratory Responses

Cardiorespiratory responses to hypercapnia are presented in Table 3. An overall effect of sex was detected such that males exhibited higher MAP (p = 0.036) and ETCO₂ (p = 0.013) between conditions. HR (p = 0.012) and ETCO₂ (p < 0.001) were significantly increased during hypercapnia. However, the absolute ΔETCO₂ during hypercapnia was not different between sexes (Female = 8.2 ± 2.1 mmHg, Male = 7.3 ± 2.3 mmHg, p = 0.06) or cohorts ( Figure 1). ETCO₂ did not fluctuate during baseline and reached a steady state during hypercapnia (p = 0.461) (Figure 1).

Table 3.

Cardiorespiratory responses to hypercapnia. HR and ETCO₂ values reflect the average over the duration of the pcASL scans. Data are presented as mean ± SD. Analyses performed using two-way ANOVA. * indicates different from males, † different from room air.

		Males	Females	Condition	Sex	Interaction
MAP, mmHg	Room Air	77 ± 8	73 ± 8	0.576	0.036	0.754
	Hypercapnia	77 ± 10	75 ± 9
HR, bpm	Room Air	57 ± 10	58 ± 9	0.012	0.159	0.464
	Hypercapnia	60 ± 12†	63 ± 10†
ETCO2, mmHg	Room Air	38 ± 5	36 ± 3*	<0.001	0.013	0.289
	Hypercapnia	45 ± 4†	44 ± 4†

Whole Brain Cerebrovascular Blood Flow and Reactivity (pcASL)

Similar to our previous reports (31, 35), basal whole brain CBF was greater in females (54 ± 13 mL/100g/min) than males (43 ± 11 mL/100g/min; p < 0.001). We observed no within-sex differences in CVR between cohorts in females (p = 0.067) or males (p = 0.165), shown in Figure 2. Absolute increases in CBF during hypercapnia trended towards being greater in females compared to males for whole brain CBF (16.9 ± 7.6 and 13.8 ± 7.6 ml/100g/min; p = 0.058) while absolute increases in GM flow were significantly higher (20.6 ± 8.9 and 16.4 ± 9.3 ml/100g/min; p = 0.037). However, absolute WM changes were not different between sexes (11.5 ± 5.9 and 9.5 ± 5.1 ml/100g/min; p = 0.10). Sex comparisons of whole brain, GM, and WM CVR are shown in Figure 3. When expressing data as the percent increase in CBF relative to elevations in ETCO₂, sex differences were not observed in whole brain CVR (Female = 4.1 ± 1.9, Male = 4.9 ± 3.0 %/mmHg, p = 0.16), GM CVR (F = 4.2 ± 1.9, M = 4.9 ± 3.0 %/mmHg, p = 0.19), or WM CVR (F = 4.0 ± 2.1, M = 4.6 ± 2.7 %/mmHg, p = 0.25). Because previous reports have found some populations that experience diverging MAP responses to hypercapnia (21), we also expressed CVR relative to changes in cerebrovascular conductance (CBF / MAP); no sex differences in conductance CVR were observed in whole brain, GM, or WM (all p ≥ 0.14; Supplemental Figure 2). To further assess an influence of MAP on CVR responses, ANCOVAs were performed with MAP as a covariate. CVR remained statistically similar at the whole brain, GM, and WM level after controlling for MAP (all p ≥ 0.114). MAP alone was not a significant predictor of CVR in these areas (all p ≥ 0.421) and no sex-by-MAP interaction effects were observed (all p ≥ 0.294). Additionally, no sex differences were found after correcting for cohort as a covariate (all p ≥ 0.561).

Figure 3. Cerebrovascular Reactivity (CVR) of Whole Brain, Grey Matter (GM), and White Matter (WM).

There was no difference between females (n = 38) and males (n = 49) CVR for whole brain (p = 0.16; Mann-Whitney U Statistics), GM (p = 0.19; Mann-Whitney U Statistics), and WM (p = 0.25; Two-tailed unpaired t-test). Flow derived from pcASL MRI.

Lobular and Regional Cerebrovascular Reactivity (pcASL)

CVR values for the six lobes (brainstem/cerebellum, subcortical regions, frontal lobe, temporal lobe, parietal lobe, and occipital lobe) are presented in Figure 4. No sex differences in CVR were observed in the brainstem/cerebellum (F = 5.5 ± 3.8, M = 6.5 ± 5.0 %/mmHg, p = 0.29), subcortical regions (F = 4.1 ± 1.8, M = 4.6 ± 3.3 %/mmHg, p = 0.35), frontal lobe (F = 3.8 ± 1.6, M = 4.4 ± 2.5 %/mmHg, p = 0.23), temporal lobe (F = 3.8 ± 1.6, M = 4.2 ± 3.1 %/mmHg, p = 0.42), parietal lobe (F = 4.3 ± 1.8, M = 4.9 ± 3.3 %/mmHg, p = 0.30), or occipital lobe (F = 5.2 ± 3.4, M = 5.7 ± 4.2 %/mmHg, p = 0.53). Expressing data relative to changes in cerebrovascular conductance yielded no sex differences in any lobe (all p ≥ 0.65; Supplemental Figure 3). Additionally, no sex differences were found after correction for MAP or cohort in any lobe (all p ≥ 0.207). Table 4 provides a complete list of all GM regions of interest within the lobes. After correction for multiple comparisons of regions within a given lobe, no differences were found between the sexes (all p ≥ 0.240). One female was excluded as an outlier from analysis of the cerebellar vernal lobules VIII-X region, with a calculated CVR value > 70 standard deviations away from the mean.

Figure 4. Regional Cerebrovascular Reactivity.

There was no difference between females (n = 38) and males (n = 49) in the brainstem/cerebellum (p = 0.29)^#, subcortical regions (p = 0.35)*, frontal lobe (p = 0.23)*, temporal lobe (p = 0.42)*, parietal lobe (p = 0.30)*, or occipital lobe (p = 0.53)*. * Indicates Mann-Whitney U Statistics and # indicates two-tailed unpaired t-test. Flow derived from pcASL MRI. Brain images created using BioRender.com.

Table 4.

Regional cerebrovascular reactivity. Data presented as mean ± SD. ^a indicates one female outlier excluded in the cerebellar vermal lobules VIII-X (see Results text). Tested by two-tailed unpaired samples t-test, with effect sizes reported as Cohen’s d and corresponding confidence interval (CI). Both the unadjusted and adjusted for multiple comparisons (via Shaffer’s sequential method) p-values are provided, with adjusted p-values exceeding 1.00 being reported as 1.00. No significant differences were found after correction for multiple comparisons.

	Male CVR (%/mmHg)	Female CVR (%/mmHg)	p-value	Adjusted p-value	Cohen’s d [95% CI]
Brain stem/cerebellum
Brain stem	5.4 ± 3.7	5.3 ± 4.1	0.947	0.947	−0.14 [−0.44, 0.41]
Cerebellum exterior	5.4 ± 4.6	5.0 ± 3.5	0.666	1.00	−0.09 [−0.52, 0.33]
Cerebellum white matter	5.3 ± 5.2	4.8 ± 3.4	0.583	1.00	−0.12 [−0.54, 0.31]
Cerebellar vermal lobules I–V	9.6 ± 8.0	6.7 ± 4.8	0.048	0.240	−0.43 [−0.86, 0.01]
Cerebellar vermal lobules VI–VII	8.7 ± 7.7	6.5 ± 5.4	0.148	0.740	−0.32 [−0.80, 0.06]
Cerebellar vermal lobules VIII–Xa	4.8 ± 6.6	5.4 ± 5.2	0.642	1.00	0.10 [−0.33, 0.53]
Subcortical regions
Amygdala	4.8 ± 4.6	4.1 ± 2.7	0.387	1.00	−0.19 [−0.61, 0.24]
Thalamus	7.2 ± 5.6	6.1 ± 3.5	0.301	1.00	−0.23 [−0.65, 0.20]
Ventral diencephalon	5.0 ± 3.9	5.0 ± 3.6	0.953	1.00	−0.01 [−0.41. 0.44]
Optic chiasm	2.4 ± 4.6	2.7 ± 4.9	0.775	1.00	0.06 [−0.36, 0.49]
Anterior cingulate gyrus	4.3 ± 3.0	3.6 ± 2.4	0.205	1.00	−0.28 [−0.70, 0.15]
Middle cingulate gyrus	4.9 ± 3.6	4.0 ± 1.8	0.187	1.00	−0.29 [−0.71, 0.14]
Posterior cingulate gyrus	5.9 ± 4.0	4.5 ± 2.3	0.046	0.506	−0.44 [−0.86, −0.07]
Subcallosal area	3.8 ± 4.3	3.8 ± 1.8	0.998	0.998	−0.01 [−0.42, 0.42]
Accumbens	5.3 ± 5.4	4.5 ± 1.9	0.397	1.00	−0.18 [−0.61, 0.24]
Caudate	4.9 ± 4.3	4.3 ± 2.6	0.505	1.00	−0.15 [−0.57, 0.28]
Pallidum	4.5 ± 4.4	4.1 ± 2.5	0.593	1.00	−0.12 [−0.54, 0.31]
Putamen	4.6 ± 4.5	3.9 ± 2.3	0.382	1.00	−0.19 [−0.61, 0.24]
Frontal lobe
Basal forebrain	4.3 ± 3.9	3.9 ± 2.9	0.587	1.00	−0.12 [−0.54, 0.31]
Anterior orbital gyrus	4.9 ± 3.0	3.9 ± 2.4	0.097	1.00	−0.36 [−0.79, 0.07]
Frontal operculum	4.8 ± 4.3	3.6 ± 3.4	0.169	1.00	−0.30 [−0.73, 0.13]
Frontal pole	4.7 ± 2.8	3.4 ± 3.3	0.055	1.00	−0.42 [−0.85, 0.01]
Gyrus rectus	3.9 ± 3.9	3.2 ± 2.0	0.288	1.00	−0.23 [−0.66, 0.20]
Lateral orbital gyrus	4.6 ± 3.5	3.7 ± 2.3	0.199	1.00	−0.28 [−0.71, 0.15]
Medial frontal cortex	4.2 ± 3.5	3.6 ± 1.8	0.344	1.00	−0.21 [−0.63, 0.22]
Middle frontal gyrus	4.6 ± 2.9	4.0 ± 2.1	0.206	1.00	−0.25 [−0.67, 0.18]
Medial orbital gyrus	4.5 ± 2.9	3.9 ± 2.0	0.241	1.00	−0.26 [−0.68, 0.17]
Precentral gyrus medial segment	5.7 ± 5.5	4.6 ± 2.3	0.224	1.00	−0.27 [−0.69, 0.16]
Superior frontal gyrus medial segment	4.3 ± 2.8	3.6 ± 1.9	0.188	1.00	−0.29 [−0.71, 0.14]
Opercular part of the inferior frontal gyrus	4.6 ± 3.5	3.7 ± 1.9	0.147	1.00	−0.32 [−0.74, 0.10]
Orbital part of the inferior frontal gyrus	4.3 ± 2.9	3.8 ± 2.0	0.345	1.00	−0.21 [−0.63, 0.22]
Posterior orbital gyrus	3.8 ± 3.1	3.6 ± 1.9	0.636	0.636	−0.10 [−0.53, 0.32]
Precentral gyrus	5.3 ± 3.7	4.3 ± 1.9	0.156	1.00	−0.31 [−0.73, 0.12]
Superior frontal gyrus	4.2 ± 3.3	3.9 ± 2.0	0.555	1.00	−0.13 [−0.55, 0.30]
Supplementary motor cortex	4.9 ± 3.8	4.0 ± 2.1	0.187	1.00	−0.29 [−0.71, 0.14]
Triangular part of the inferior frontal gyrus	4.4 ± 2.9	3.7 ± 1.8	0.211	1.00	−0.27 [−0.70, 0.15]
Temporal lobe
Hippocampus	4.5 ± 3.9	4.0 ± 2.3	0.487	1.00	−0.15 [−0.58, 0.27]
Anterior insula	4.9 ± 4.7	4.0 ± 2.6	0.263	1.00	−0.24 [−0.67, 0.18]
Entorhinal area	4.2 ± 4.0	3.9 ± 2.4	0.722	0.722	−0.08 [−0.50, 0.35]
Fusiform gyrus	5.4 ± 4.0	4.7 ± 3.1	0.323	1.00	−0.22 [−0.64, 0.21]
Inferior temporal gyrus	4.9 ± 3.4	4.4 ± 2.3	0.403	1.00	−0.18 [−0.61, 0.24]
Middle temporal gyrus	4.4 ± 3.1	3.7 ± 2.1	0.239	1.00	−0.26 [−0.68, 0.17]
Parahippocampal gyrus	4.5 ± 3.9	4.2 ± 2.6	0.704	1.00	−0.08 [−0.51, 0.34]
Posterior insula	5.3 ± 5.4	4.5 ± 2.8	0.415	1.00	−0.18 [−0.60, 0.25]
Planum polare	3.7 ± 3.8	3.1 ± 1.9	0.328	1.00	−0.21 [−0.64, 0.21]
Planum temporale	3.9 ± 3.3	3.3 ± 2.1	0.318	1.00	−0.22 [−0.64, 0.21]
Superior temporal gyrus	3.9 ± 3.1	3.2 ± 1.8	0.191	1.00	−0.29 [−0.71, 0.14]
Temporal pole	3.4 ± 3.5	3.9 ± 2.4	0.425	1.00	0.17 [−0.25, 0.60]
Transverse temporal gyrus	4.0 ± 4.6	3.0 ± 2.1	0.254	1.00	−0.25 [−0.67, 0.18]
Parietal lobe
Angular gyrus	4.6 ± 2.9	3.9 ± 2.4	0.233	1.00	−0.26 [−0.68, 0.17]
Central operculum	4.9 ± 3.8	4.0 ± 2.2	0.201	0.603	−0.28 [−0.70, 0.15]
Postcentral gyrus medial segment	5.9 ± 6.4	5.4 ± 2.6	0.669	1.00	−0.09 [−0.51, 0.33]
Precuneus	5.6 ± 4.3	4.6 ± 2.3	0.179	0.716	−0.29 [−0.72, 0.13]
Parietal operculum	4.3 ± 3.4	3.7 ± 2.3	0.311	1.00	−0.22 [−0.65, 0.21]
Postcentral gyrus	5.4 ± 4.0	4.7 ± 2.6	0.390	1.00	−0.19 [−0.61, 0.24]
Supramarginal gyrus	5.1 ± 3.8	3.9 ± 2.1	0.096	1.00	−0.36 [−0.79, 0.06]
Superior parietal lobule	5.2 ± 4.7	4.9 ± 2.5	0.664	1.00	−0.09 [−0.52, 0.33]
Occipital lobe
Calcarine cortex	5.9 ± 4.6	4.8 ± 3.9	0.252	1.00	−0.25 [−0.67, 0.18]
Cuneus	6.1 ± 4.5	4.9 ± 3.4	0.171	1.00	−0.30 [−0.72, 0.13]
Inferior occipital gyrus	5.8 ± 5.3	5.3 ± 4.1	0.653	1.00	−0.10 [−0.52, 0.33]
Lingual gyrus	6.6 ± 5.0	5.0 ± 3.9	0.105	0.735	−0.36 [−0.78, 0.07]
Middle occipital gyrus	4.7 ± 3.5	4.5 ± 2.6	0.761	1.00	−0.07 [−0.49, 0.36]
Occipital pole	6.8 ± 7.0	6.5 ± 5.6	0.814	0.814	−0.05 [−0.48, 0.37]
Occipital fusiform gyrus	6.3 ± 5.7	6.5 ± 4.6	0.882	1.00	0.03 [−0.39, 0.46]
Superior occipital gyrus	5.8 ± 4.7	5.4 ± 3.4	0.729	1.00	−0.08 [−0.50, 0.35]

Cerebral Artery Reactivity (4D flow)

In a subset of cohort 1, 4D flow was used to derive CVR of individual intracranial vessels and whole brain CVR. No sex differences in CVR were found in whole brain CBF (F = 3.5 ± 0.9, M = 3.4 ± 1.8 %/mmHg) or the 11 arteries assessed (Figure 5). These conclusions remain unchanged when calculating CVR relative to changes in cerebrovascular conductance (Supplemental Figure 4). Table 5 provides sex-specific data in a table of flow, CSA, and velocity of each large cerebral artery during room air and hypercapnia.

Figure 5. Cerebrovascular Reactivity of Individual Arteries.

No sex differences were observed in CVR at the whole brain level as well as in the basilar artery, or the left and right internal carotid (ICA), anterior cerebral (ACA), vertebral (VA), posterior cerebral (PCA), or middle cerebral (MCA) arteries (all p > 0.05; unpaired t tests). Flow through these arteries derived from 4D flow MRI.

Table 5.

Individual vessel flow, cross-sectional area (CSA), and blood flow velocity at rest and during hypercapnia by sex using 4D flow MRI (24M, 13F). * indicates significantly different from males, † indicates significantly different from room air; 1 male (^a) or female (^b) subject excluded for some vessels due to poor image quality. Tested using two-way ANOVA. ICA = internal carotid artery; MCA = middle cerebral artery; ACA = anterior cerebral artery; PCA = posterior cerebral artery; VA = vertebral artery.

	Flow (mL/min)		CSA (mm2)		Mean Velocity (cm/s)
	Baseline	Hypercapnia	Baseline	Hypercapnia	Baseline	Hypercapnia
L ICA
Male	243 ± 49	299 ± 55†	15.5 ± 3.5	16.8 ± 4.0†	26.6 ± 5.0	30.3 ± 5.7†
Female	272 ± 48	334 ± 56†	14.3 ± 3.7	16.3 ± 3.6†	32.6 ± 5.0	34.6 ± 3.9*†
R ICA
Male	245 ± 41	305 ± 55†	16.2 ± 2.6	17.2 ± 2.4†	25.6 ± 4.8	29.7 ± 5.1†
Female	218 ± 34*	268 ± 47*†	12.6 ± 1.9*	14.2 ± 2.1*†	29.2 ± 4.2*	31.8 ± 4.2†
L MCAb
Male	158 ± 27	177 ± 33†	7.7 ± 1.1	8.0 ± 1.2	34.3 ± 4.3	36.9 ± 4.0†
Female	148 ± 23	162 ± 39†	6.8 ± 1.0*	7.5 ± 1.1†	36.2 ± 4.6	36.0 ± 5.9
R MCAb
Male	153 ± 27	169 ± 32†	7.5 ± 1.3	7.8 ± 1.1	34.3 ± 4.7	36.3 ± 4.9†
Female	143 ± 24	154 ± 34†	6.5 ± 0.8 *	7.1 ± 0.9†	36.5 ± 3.3	36.2 ± 6.2
L ACAa
Male	80 ± 23	98 ± 24†	4.8 ± 0.9	5.0 ± 0.9†	27.4 ± 5.4	32.5 ± 5.0†
Female	96 ± 34	114 ± 43†	4.7 ± 1.4	5.0 ± 1.5†	34.0 ± 3.6*	36.7 ± 5.3*†
R ACA
Male	93 ± 24	115 ± 29†	5.4 ± 1.0	5.7 ± 1.0†	28.4 ± 4.3	33.5 ± 4.1†
Female	68 ± 27*	82 ± 31*†	3.8 ± 1.0 *	4.3 ± 1.2*†	29.0 ± 5.7	31.1 ± 4.8†
L PCAa,b
Male	64 ± 15	79 ± 21†	5.5 ± 0.8	5.8 ± 0.8	19.6 ± 3.7	22.6 ± 4.7†
Female	72 ± 20	87 ± 26†	5.2 ± 1.3	5.4 ± 0.7	24.1 ± 5.6*	26.8 ± 5.5*†
R PCA
Male	65 ± 16	80 ± 18†	5.3 ± 0.9	5.6 ± 0.9†	20.5 ± 4.0	23.7 ± 3.8†
Female	73 ± 27	81 ± 35†	4.7 ± 1.1	5.2 ± 0.9†	25.4 ± 6.6*	28.6 ± 8.3*†
L VA
Male	80 ± 28	100 ± 40†	5.9 ± 1.1	6.1 ± 1.3†	22.2 ± 5.0	26.4 ± 6.4†
Female	87 ± 24	108 ± 33†	5.7 ± 0.8	6.4 ± 1.1†	25.2 ± 5.1	27.8 ± 5.7†
R VAa
Male	91 ± 23	111 ± 29†	6.4 ± 1.2	6.7 ± 1.3†	23.5 ± 3.5	27.4 ± 4.6†
Female	90 ± 36	115 ± 45†	5.8 ± 1.5	6.5 ± 1.6†	24.6 ± 6.8	28.2 ± 7.7†
Basilar
Male	148 ± 33	185 ± 41†	8.8 ± 1.6	9.4 ± 1.9†	28.2 ± 4.8	33.0 ± 5.7†
Female	162 ± 44	199 ± 54†	8.4 ± 1.0	9.2 ± 1.3†	32.1 ± 6.7*	35.8 ± 6.9†

Discussion

To our knowledge, this is the largest MRI-based assessment to date assessing sex differences in CVR in young, healthy adults at both global and regional levels. In this report, we combined two cohorts with rigid inclusion criteria that controlled for several factors that may influence CVR. Contrary to our hypothesis, we observed no sex differences in CVR in the whole brain, GM, and WM. Particularly novel to this report is the first assessment of lobular and region of interest comparisons in CVR using pcASL, which also lacked differences between the sexes. This finding was confirmed using 4D flow MRI to assess reactivity of individual intracranial arteries in a subset of 37 subjects. Taken together, these data support the concept of a lack of fundamental sex differences in CVR and serve as a robust reference set for expanding sex-related CVR studies in a variety of clinically and physiologically relevant scenarios including hormonal contraception, menstrual cycle phases, aging, and comorbidities.

The major novelty and strength of the current report is the simultaneous assessment of CVR both in the whole brain as well as individual lobes, regions, and arteries which adds substantial new data at high spatial resolution. Cerebrovascular pathology displays regional heterogeneity (7, 8) and local reductions in CVR are related to cognitive task performance (9). Therefore, the testing of CVR at varying levels of spatial fidelity is critical in understanding fundamental sex differences in cerebrovascular function that are linked to long-term brain health. To our knowledge, we provide the first data to compare sex-specific microvascular CVR responses at a lobular and regional level using pcASL. Importantly, we used 4D flow to subsequently verify these findings at the macrovascular level. While previous reports using Doppler ultrasound (18) or TCD (19) have been used to assess vessel-specific heterogeneity, these methods lack the regional specificity that MRI techniques provide. Of the previous MRI studies assessing the role of sex in CVR, GM, and WM have been studied using blood-oxygen level dependent MRI (20) while 4D flow has been used to measure CVR in the basilar, internal carotid, and middle cerebral arteries (21).

Our rigorous inclusion criteria, notably greater samples sizes, and inclusion of two validated MRI methodologies support the null hypothesis and provide evidence for a lack of fundamental sex differences in CVR in young, healthy adults. Given this finding, it is worth drawing attention to the striking range of CVR responses, evidenced by the presence of near-zero or negative CVR values for some individuals (Figs. 2-5). This is despite our selection of young, healthy adults and normalization of data as relative increases in CBF and ETCO₂. High interindividual variability is observed in many previous studies of healthy populations using TCD or MRI. Specifically, multiple reports have shown individuals displaying ~0 %/mmHg or negative CVR measured both at regional and whole brain levels via TCD and MRI (12, 19, 47-49). This is surprising, given the consistent observation that, on average, CVR results in increases in CBF in this population. The presence of extremely low or negative CVR found in the current report and others suggests that there may be unknown factors that influence CVR beyond sex, sex hormones, hormonal contraception, or traditional cardiometabolic measures and controlled for in this study. Alternatively, this may highlight variability inherent to the method of CBF quantification or hypercapnic administration that adds to the heterogeneity observed in CVR responses. Unfortunately, the present and prior work are unable to offer insight into these sources of intersubject variance.

Our data suggest there are not fundamental sex differences in CVR at multiple perfusion levels, consistent with some previous reports in single cerebral arteries (22-26). Interestingly, we observed a non-significant 8–19% higher CVR in males using pcASL (Figures 3, 4), depending on which perfusion level of whole brain, GM, WM, or lobe is tested, which aligns with two prior MRI reports demonstrating greater CVR in males (20, 21). However, post hoc power analyses indicate sample sizes of 183–1386 participants per sex would be required to detect statistically discernable sex differences. These large sample sizes required may explain, in part, the equivocal findings between studies in humans that commonly have 15 or less participants per sex (12, 14, 15, 19-22, 25). The present data suggest that the primary drivers of cerebrovascular responses to hypercapnia are largely independent of biological sex (e.g., the direct effects of H⁺ or bicarbonate) (5, 50, 51). The clinical relevance of these non-significant differences between sexes remains unclear, as studies in older adults indicate CVR decrements on the order of 20–25% to distinguish pathological risk (e.g. stroke, mortality, Alzheimer’s Disease; (52, 53)). In the context of those findings, and our stated 8–19% greater CVR (in males) are not statistically significant, the clinical ramifications of these trends in our data remain speculative at this stage.

However, the relationship between CVR and cerebrovascular pathologies (2, 3), which exhibit sex-specific risk (6), has evoked great interest in understanding the role of biological sex on CVR. This is complemented by animal data suggesting a role for sex hormones, such as testosterone and estrogen, in modifying enzyme expression (e.g., the cyclooxygenase pathway) that are mechanistically implicated in CVR (54-56). However, the lack of sex differences observed by our group in this study and a previous TCD report (24) despite differences in some sex steroid concentrations (e.g., testosterone, SHBG; Table 2) indicates that this mechanistic animal work does not translate to notable differences in CVR in humans. Alternatively, it may be that the beneficial effects of individual sex hormones are masked by effects of other hormones, such as a potential antagonistic role of follicle stimulating hormone and progesterone on vascular function observed in the periphery (57, 58). Investigations on the roles of these sex hormones on cerebrovascular function across the lifespan warrant further investigation.

An additional explanation contributing to a lack of sex differences may be the effect of baseline ETCO₂. Females displayed greater resting CBF and lower resting ETCO₂, both of which are common observations in the literature (12, 17, 24, 35). There is evidence that lower resting ETCO₂ may yield greater CVR (59), presumably as a result of a greater vasodilatory reserve. It may be that the lack of sex differences observed in the current report is a result of the lower baseline ETCO₂ in females allowing for greater absolute increases in CBF, resulting in similar relative increases in CBF and therefore CVR between sexes. This is supported by the absolute change in GM CBF being greater in females than males (20.6 ± 8.9 vs 16.4 ± 9.3 mL/100g/min; p = 0.037) whereas relative changes in GM CBF were similar between sexes (34.2 ± 17.7 vs 36.78 ± 27.1%; p = 0.61). Similar relationships are seen when assessing whole brain CVC, where absolute increases were greater in females (p = 0.031) while relative increases were similar between groups (p = 0.441), suggesting that differences in baseline MAP does not modify this relationship.

Most investigations of sex-specific CVR responses have used TCD, finding that females either display greater (12-16, 19) or equivalent (22-26) CVR to males. Thus, we used 4D flow to derive CVR in 11 arteries constituting the cerebral circulation to assess vessel-specific differences. While no sex differences were observed in CVR in any vessel (Figure 5), we do provide a robust data table of flow, CSA, and velocity of these vessels at rest and during hypercapnia (Table 5) that may prove useful in future TCD studies to adjust for changes in CSA that result in substantial underestimations of CVR (27). Notably, we observed dilation in 8 arteries in males and 10 arteries in females, further challenging the notion of an assumed constant vessel diameter inherent in TCD use (27-29). This may be particularly impactful in vessels such as the middle cerebral artery (MCA) which display sex-specific responses in CSA responses to hypercapnia in the current report, with both the left and right MCA significantly dilating in females but not males (Table 5). This, in turn, may bias underestimations of CVR in females in TCD studies.

Limitations

While the analysis of combined cohorts with rigorous experimental control strengthens statistical power, subtle differences in the MRI scanners, differences in hypercapnia administration, or unmeasured characteristic differences between cohorts may obscure our analysis of sex differences in CVR. These unmeasured characteristics include not controlling for the time of day tested or characteristics of sleep, as there is evidence of diurnal variation in CVR in both sexes that may be driven by sleep duration, quality, and efficiency or nocturnal reductions in blood pressure (60). Similarly, acute fluctuations in testosterone in both sexes are noted (61, 62), potentially contributing to these diurnal variations. Additionally, cohort 2 was involved in a study that utilized a placebo pill not used in cohort 1. However, given the lack of differences between CVR values at both study sites as well as no cohort differences in relevant subject characteristics indicates that small variations between groups likely hold minimal impact on conclusions drawn from this study. This is further supported by our statistical analyses yielding no sex or sex-by-cohort effect in CVR, suggesting that our conclusion of a lack of sex differences is not altered when accounting for differences between cohorts (ANOVA) or when using MAP or cohort as covariates (ANCOVA). Another limitation of this study is the lack of testing across the menstrual cycle. As our primary research question was in studying fundamental sex differences in CVR, testing female participants in the early follicular phase was most appropriate to minimize any short-term effects of fluctuating sex hormones on CVR to focus on the effect of sex rather than the role of sex hormones (Table 2). This, alongside our exclusion of hormonal contraceptive use, limits the generalizability of our results to other hormonal conditions, such as in other phases of the menstrual cycle or in females using various methods of hormonal contraceptives. We also did not collect pregnancy history or parity data, which may introduce variability between the female participants. Furthermore, our age range spanned 19 years, and age can impact CVR, but no correlations between CVR and age were significant (p all ≥ 0.166). Finally, graded hypercapnia protocols (19) may provide a more nuanced insight intro cerebrovascular control in contrast to our single target of +8 mmHg ETCO₂. However, hypercapnia responses appear largely linear at the target ETCO₂ typically used in most human CVR studies and are therefore reflective of reactivity in these ranges (5).

Conclusion

There does not appear to be a fundamental sex difference in CVR as measured at both macrovascular and microvascular levels in young, healthy adults when rigorously controlling for potential confounders. The lack of sex differences appears consistent when assessing global CVR as well as individual GM lobes, regions of interest within GM, WM, and individual cerebral arteries. In the context of prior work, our data are consistent with the concept that males may demonstrate insignificantly higher CVR when controlling for aforementioned limitations, but that this difference is minor due, in part, by high heterogeneity in CVR responses in both sexes. Importantly, these new data establish robust global and regional microvascular and arterial macrovascular CVR reference data to build on the CVR-by-sex conceptual framework for future studies to investigate a wide variety of physiologic scenarios. Follow-up studies are needed to assess the mechanistic governors of CVR outside of traditional cardiometabolic health markers defining the healthy young adult to understand the heterogeneity inherent to CVR. Further, more rigorous testing of the influence of the menstrual cycle and hormonal contraceptives are needed to expand the translatability of the current report.

Supplementary Material

https://doi.org/10.6084/m9.figshare.31876381

Supplemental Table 1. Subject characteristics between cohorts. Data are presented as mean ± SD. Analyses performed using two-way ANOVA. * Significantly different from males within the same cohort, p < 0.05.

Supplemental Figure 1. Demonstration of MAP stability. A) Individual MAP at rest vs hypercapnia for each subject. Cohorts are black (1, solid) or red (2, dashed), with different symbols for sex. No sex-by-cohort interaction effects were observed (p = 0.191). B) Average MAP data from panel A.

Supplemental Figure 2. Cerebrovascular Reactivity Using Changes in Conductance. Cerebrovascular conductance was calculated as CBF/MAP. When reporting reactivity relative to changes in conductance, no sex differences are observed in the whole brain, grey matter, or white matter. No sex differences were observed, assessed using unpaired samples t-tests. Flow derived from pcASL MRI. No cohort x sex interaction effects were observed at any perfusion level (p ≤ 0.299).

Supplemental Figure 3. Regional Cerebrovascular Reactivity Using Changes in Conductance. Cerebrovascular conductance was calculated as CBF/MAP. When reporting reactivity relative to changes in conductance, no sex differences are observed in brainstem/cerebellum, subcortical regions, frontal lobe, temporal lobe, parietal lobe, or occipital lobe. Flow derived from pcASL MRI. Brain images created using BioRender.com.

Supplemental Figure 4. 4D-Flow Cerebrovascular Reactivity Using Changes in Conductance by 4D Flow MRI approach Cerebrovascular conductance was calculated as CBF/MAP. When reporting reactivity relative to changes in conductance, no sex differences are observed in the whole brain, the basilar artery, or the left and right internal carotid (ICA), anterior cerebral (ACA), vertebral (VA), posterior cerebral (PCA), or middle cerebral (MCA) arteries. Flow derived from 4D flow MRI.

New & Noteworthy:

There were no significant sex differences in cerebrovascular reactivity to CO2 detected at global, lobular, arterial, or regional levels in young healthy adults. This study rigorously controlled for confounders of age, menstrual cycle, hormonal contraception, and comorbidities.