Indomethacin-induced impairment of regional cerebrovascular reactivity: implications for respiratory control

Abstract

Cerebrovascular reactivity impacts CO₂–[H⁺] washout at the central chemoreceptors and hence has marked influence on the control of ventilation. To date, the integration of cerebral blood flow (CBF) and ventilation has been investigated exclusively with measures of anterior CBF, which has a differential reactivity from the vertebrobasilar system and perfuses the brainstem. We hypothesized that: (1) posterior versus anterior CBF would have a stronger relationship to central chemoreflex magnitude during hypercapnia, and (2) that higher posterior reactivity would lead to a greater hypoxic ventilatory decline (HVD). End-tidal forcing was used to induce steady-state hyperoxic (300 mmHg ) hypercapnia (+3, +6 and +9 mmHg ) and isocapnic hypoxia (45 mmHg ) before and following pharmacological blunting (indomethacin; INDO; 1.45 ± 0.17 mg kg⁻¹) of resting CBF and reactivity. In 22 young healthy volunteers, ventilation, intra-cranial arterial blood velocities and extra-cranial blood flows were measured during these challenges. INDO-induced blunting of cerebrovascular flow responsiveness (CVR) to CO₂ was unrelated to variability in ventilatory sensitivity during hyperoxic hypercapnia. Further results in a sub-group of volunteers (n = 9) revealed that elevations of via end-tidal forcing reduce arterial–jugular venous gradients, attenuating the effect of CBF on chemoreflex responses. During isocapnic hypoxia, vertebral artery CVR was related to the magnitude of HVD (R² = 0.27; P < 0.04; n = 16), suggesting that CO₂–[H⁺] washout from central chemoreceptors modulates hypoxic ventilatory dynamics. No relationships were apparent with anterior CVR. As higher posterior, but not anterior, CVR was linked to HVD, our study highlights the importance of measuring flow in posterior vessels to investigate CBF and ventilatory integration.

Key points

• Anterior and posterior cerebral circulations have differential reactivity to changes in arterial blood gases, but the implications for the chemoreflex control of breathing are unclear.

• Indomethacin-induced blunting of cerebrovascular flow responsiveness did not affect central or peripheral respiratory chemoreflex magnitude using steady-state end-tidal forcing techniques.

• Posterior reactivity was related to hypoxic ventilatory decline, suggesting that CO₂ washout from central chemoreceptors modulates hypoxic ventilatory dynamics.

• Our data indicate that steady-state end-tidal forcing techniques reduce the arterial–venous gradients, attenuating the effect of brain blood flow on ventilatory responses.

• Our study confirms the importance of measuring posterior cerebrovasculature when investigating the link between cerebral blood flow and the chemical control of breathing.

Introduction

The cerebral vasculature is highly sensitive to changes in the partial pressure of arterial carbon dioxide (), dilating during elevations in (i.e. hypercapnia), and constricting during reductions in (i.e. hypocapnia) (Wolff & Lennox, 1930; Kety & Schmidt, 1948; Willie et al. 2012; Coverdale et al. 2014; Verbree et al. 2014). The sensitivity of the cerebral vasculature to changes in is termed cerebrovascular reactivity, where the resulting vessel dilatation causes an increase in cerebral blood flow (CBF) per unit change in (Willie et al. 2012; Skow et al. 2013). It should also be acknowledged that hypercapnic-induced elevations in perfusion pressure may also influence the extent of cerebrovascular reactivity (Willie et al. 2012; Regan et al. 2014); therefore, one cannot fully dissociate between the vasodilatory influence of CO₂ versus that of elevations in perfusion pressure. Therefore, a more parsimonious term is cerebrovascular flow responsiveness (CVR).

Nevertheless, the implications of alterations in CVR are numerous (for review see Ainslie & Duffin, 2009). For example, a blunted CVR to CO₂ alters the tight regulation of CO₂–[H⁺] at the central chemoreceptors (in the medulla), subsequently enhancing the ventilatory response for a given change in (e.g. Xie et al. 2006). A heightened ventilatory response to CO₂ contributes to the development of breathing instability during wakefulness (Fan et al. 2010a) and sleep (Xie et al. 2009), and is important in the pathophysiology of central and mixed sleep apnoea (Javaheri & Dempsey, 2013). Administration of indomethacin (INDO; a non-steroidal and non-selective anti-inflammatory) reduces baseline CBF and CVR to CO₂, contributing to an increased ventilatory sensitivity to CO₂ (Xie et al. 2006, 2009; Fan et al. 2010a). While INDO does not directly affect central chemoreceptor sensitivity per se, it increases ventilatory sensitivity via increased central chemoreceptor stimulation resulting from a blunted CO₂–[H⁺] washout from the brain (i.e. CO₂ accumulation). However, ventilatory sensitivity is unchanged following INDO administration when indexed against changes in the partial pressure of jugular venous CO₂ (; Xie et al. 2006), which is considered a more accurate representation of the central stimulus to breathe than (Fencl, 1986). Therefore, it is evident that the ventilatory response to a given change in is unaltered post INDO as it is the reactivity-induced changes in the − gradient that result in the change of ventilatory sensitivity to CO₂. The findings that ventilatory sensitivity to CO₂ is unaltered following INDO administration using the rebreathing technique in which CO₂ gradients are abolished further supports the theory that it is changes in the − gradients that modulate ventilatory sensitivity (Fan et al. 2010a).

One issue to date, however, is that the aforementioned human studies have limited the assessment of CVR to the anterior circulation (i.e. middle cerebral artery; MCA), despite the fact that the vertebrobasilar (posterior) arterial circuit supplies the brainstem where the central chemoreceptors are located (e.g. Mitchell & Loeschcke, 1963; Ainslie & Duffin, 2009; Guyenet et al. 2010). Moreover, recent studies from our group (Willie et al. 2012; Skow et al. 2013) have demonstrated that anterior and posterior circulations have differential CVR to CO₂, such that reactivity in the MCA is almost twofold greater than in the posterior cerebral artery (PCA) in absolute measures of CVR to CO₂ (i.e. cm s⁻¹ (mmHg )⁻¹).

In addition to changes in CO₂ affecting respiratory and cerebrovascular responses, hypoxia is also both a respiratory stimulant and a potent cerebrovascular dilator (e.g. Kety & Schmidt, 1948; Willie et al. 2012). The initial acute hypoxic ventilatory response (HVR) is mediated by the peripheral chemoreceptors and is rapidly (within minutes) followed by a hypoxic ventilatory decline (HVD) (Robbins, 1995; Powell et al. 1998). The HVD phenomenon persists even when isocapnia is maintained (Sato et al. 1992; Steinback & Poulin, 2007). Although the exact mechanisms causing HVD are uncertain, its occurrence has been speculated to result from, (1) decreased peripheral chemoreceptor sensitivity (Honda et al. 1997), (2) elevations in CBF resulting from hypoxic cerebral vasodilation (Lee & Milhorn, 1975; Neubauer et al. 1985; Poulin & Robbins, 1998), (3) changes in peripheral chemoreflex threshold (Mahamed & Duffin, 2001), and/or (4) disassociation between CNS input and phrenic output (Melton et al. 1992). Although it has been recently reported that there is a greater relative CVR in the posterior arteries during isocapnic hypoxia (Willie et al. 2012; Ogoh et al. 2013), the potential role of CBF in the posterior circulation has not been previously examined during the HVR and subsequent HVD in response to steady-state isocapnic hypoxia.

If differences in regional CVR to hypercapnia and hypoxaemia persist, it follows that studies only employing the CBF measure of MCAv have potentially misinterpreted the relationship between CBF and central chemosensitivity. Furthermore, CBF and ventilatory integration has yet to be assessed with measures of volumetric flow. Therefore, we examined the relationship between regional CBF, CVR, and ventilatory responses to hypercapnia and hypoxia, both before and following INDO. Given the previously reported differential CO₂ (Willie et al. 2012; Skow et al. 2013) and O₂ reactivity (Willie et al. 2012) between anterior and posterior vessels, we hypothesized that that: (1) INDO-induced elevations in ventilatory sensitivity to hypercapnia will be better related to the posterior versus anterior reductions in CVR and (2) greater CVR of the posterior circulation to hypoxia will result in reduced magnitude of HVR and an increased magnitude of HVD. Furthermore, using previously collected data with arterial and jugular venous blood draws, we expected that estimations of in the present study would result in no change in ventilatory sensitivity following INDO administration.

Methods

Subject recruitment

Twenty-two healthy participants (mean age: 22.3 ± 2.4 years; mean BMI: 22.5 ± 2.4 kg m⁻²; 9 females) were recruited from the University of British Columbia Okanagan, pre-screened, familiarized to the experimental protocol and provided written informed consent, in accord with the standards set by the Declaration of Helsinki. Selection criteria for participants were as follows: aged 18–40 years old, non-smoking, non-obese (BMI < 30 kg m⁻²), non-hypertensive, and past/presently free from any known neurological, respiratory and cardiovascular diseases. Participants were not on any medications, except for female participants using oral contraceptives. Participants were asked to refrain from caffeine and alcohol for 12 h, as well as vigorous exercise for 24 h prior to participation in the study. This study received ethical approval from the Clinical Research Ethics Board at the University of British Columbia, and abided by the Canadian Government Tri-Council Policy Statement for Integrity in Research.

Instrumentation and data collection

All data were collected and digitally archived at 200 Hz using an analog-to-digital acquisition system (Powerlab Model 880 ADInstruments, Colorado Springs, CO, USA), interfaced with LabChart Pro software (version 7.2; ADInstruments), and analysed offline. Subjects were instrumented for measures of instantaneous heart rate via standard electrocardiogram (HR, beats min^–1; ADI bioamp ML132) and beat-by-beat blood pressure (Finometer Pro, Finapres Medical Systems, Amster dam, The Netherlands). Finometer blood pressure record ings were calibrated to brachial blood pressure.

The respiratory apparatus incorporated a mouthpiece, nose clip, bacteriological filter, and a spirometer flowhead connected to a custom-designed dynamic end-tidal forcing system (AirForce, GE Foster, Kelowna, BC, Canada). Respiratory flow was measured with a pneumo tachometer (model 800 L, Hans Rudolph, Shawnee, KS, USA). Breath-by-breath %CO₂ and %O₂ were measured using a calibrated gas analyser (ADInstruments; ML206) and corrected for BTPS in millimetres of mercury (mmHg) using the daily atmospheric pressure to calculate the partial pressure of end-tidal CO₂ and O₂ ( and ).

Bilateral transcranial Doppler (TCD) ultrasound was employed using a 2 MHz pulsed Doppler ultrasound system (Spencer Technologies, PMD150B) to insonate, and measure the cerebral blood velocity (CBV) of the MCA and PCA (MCAv and PCAv; cm s⁻¹). The ultrasound probes were secured using a bilateral headpiece (model M600 bilateral head frame, Spencer Technologies). The MCA and PCA were insonated through the anterior temporal window using standardization and location techniques previously described by our group (Willie et al. 2011).

Extra-cranial CBF was measured using a 10 MHz multi frequency linear array vascular ultrasound (Terason T3200, Teratech, Burlington, MA, USA). The internal carotid (ICA) and vertebral arteries (VA) were insonated ipsilateral to the MCA and PCA, respectively. Arterial diameter was measured via B-mode imaging, while peak blood velocity was simultaneously measured with pulse-wave mode. Measurements of ICA diameter and velocity were acquired no less than 2 cm distal to the common carotid bifurcation, with no evidence of turbulent or retrograde flow present during recording. Vertebral artery diameter and velocity were acquired between either the C4–C5 and C5–C6 segments or, where possible, proximal to entry into the vertebral column. The location was determined on an individual basis in an attempt to select the most reproducible measures, with the same location repeated within subjects.

Measurements of the extra-cranial vessels were screen captured and stored as AVI files for offline analysis. Concurrent values for arterial diameter and peak blood velocity were acquired at 30 Hz, using customized edge detection software designed to eliminate observer bias (Black et al. 2008). Internal carotid and vertebral artery flow ( and , respectively; ml min^–1) were subsequently calculated for each vessel using eqn 1 (Evans, 1985; Willie et al. 2012):

1

No flow calculation included less than 12 consecutive and stable cardiac cycles, and where possible included the entire averaging period.

Protocols

Before and 90 min following oral INDO administration (1.45 ± 0.17 mg kg⁻¹), subjects underwent a 10 min baseline period breathing room air while resting supine, followed by two randomly ordered ventilatory tests, each separated by 10 min of rest (Fig. 1):

Figure 1.

Schematic experimental outline for CO₂ and O₂ perturbations pre-and post INDO using end-tidal forcing

Test 1, hyperoxic hypercapnia involved a 1 min baseline stage and four subsequent stages (0, +3, +6, +9 Torr ). Dotted lines separate transition periods from steady-state end-tidal clamping. Test 2, isocapnic hypoxia had a 1 min baseline stage followed by 10 min of steady-state isocapnic hypoxia (+1 , 45 Torr ). HVR, HVD, and time to peak were calculated as shown in this figure. Subjects were then orally administered INDO, and the tests were repeated following 90 min.

Hyperoxic hypercapnic test

Using an end-tidal forcing system as described previously (Querido et al. 2013; Foster et al. 2014), the steady-state hyperoxic CO₂ test comprised three hyperoxic (300 mmHg ) hypercapnic steps above individual hyperoxic baseline eupnoeic CO₂ levels (+3, +6, and +9 mmHg ). The duration of each stage was 3 min once steady-state end tidal gases were achieved (∼4 min total each).

Isocapnic hypoxia

Using an end-tidal forcing system, isocapnic (eupnoeic +1 mmHg) hypoxia (45 mmHg ) was maintained for 10 min after achieving steady-state end-tidal gas clamping. Peripheral oxyhaemoglobin saturation was monitored for safety throughout.

Data analysis

Baseline

Breath-by-breath and , minute ventilation (; l min⁻¹), CBV (MCAv, PCAv), CBF (, ), mean arterial pressure (MAP; mmHg) and HR (min⁻¹) were quantified during baseline while breathing room air, both before and following INDO administration. All measures represent a 1 min average prior to the beginning of the experimental protocols (see Table1).

Table 1.

Baseline respiratory and haemodynamic values pre- and post INDO for both the hypercapnic and hypoxic chemoreflex tests, as well as vessel reactivity during the respective tests

		Hypercapnia				Hypoxia
	Baseline values	Pre	Post	Delta	Baseline values	Pre	Post	Delta
	(mmHg)	38.6 ± 2.9	38.6 ± 2.4	−0.0 ± 1.9	(mmHg)	38.5 ± 2.8	39.3 ± 2.3	0.8 ± 2.6
	(mmHg)	95.8 ± 6.0	98.8 ± 4.0	−3.0 ± 7.2	(mmHg)	99.8 ± 6.7	96.9 ± 7.2	−2.8 ± 7.0
HR	(beats min–1)	72.0 ± 16.1	63.0 ± 12.1*	−9.0 ± 7.4	(beats min–1)	68.6 ± 15.1	63.9 ± 11.8	−4.7 ± 9.8
MAP	(mmHg)	89.0 ± 9.0	91.9 ± 9.5	2.9 ± 8.9	(mmHg)	89.3 ± 12.5	91.5 ± 13.7	2.2 ± 20.1
	(l min−1)	13.8 ± 2.8	14.7 ± 4.8	0.9 ± 2.1	(l min−1)	13.6 ± 3.0	15.5 ± 4.2	1.9 ± 3.8
MCAv	(cm s−1)	70.0 ± 16.8	49.1 ± 13.9*	−20.9 ± 15.8	(cm s−1)	68.3 ± 15.2	48.7 ± 14.6*	−19.6
PCAv	(cm s−1)	46.4 ± 12.8	35.7 ± 11.3*	−10.7 ± 6.6	(cm s−1)	44.7 ± 8.6	37.5 ± 15.7*	−7.2 ± 9.8
	(ml min−1)	225.6 ± 61.4	147.5 ± 24.0*	−78.1 ± 48.2	(ml min−1)	233.2 ± 56.7	156.7 ± 33.2*	−76.5 ± 58.0
	(ml min−1)	90.1 ± 38.4	60.4 ± 27.8*	−29.7 ± 20.6	(ml min−1)	97.1 ± 45.5	67.0 ± 33.6*	−30.1 ± 13.7
	Absolute reactivity				Absolute reactivity
MCAv		3.0 ± 1.0	1.6 ± 0.7*	−1.4 ± 1.1		1.4 ± 0.4	1.0 ± 0.4*	−0.3 ± 0.5
	(cm s–1 (mmHg)−1)				(cm s–1 (–%)−1)
PCAv		2.0 ± 0.7	1.0 ± 0.6*	−1.0 ± 0.8		0.9 ± 0.3	0.8 ± 0.3	−0.1 ± 0.3
		18.0 ± 4.1	9.2 ± 4.2*	−8.6 ± 4.4		5.9 ± 3.5	4.3 ± 2.8	−1.7 ± 2.2
	(ml min–1  (mmHg)−1)				(ml s–1 (–%)−1)
		7.3 ± 4.2	2.9 ± 1.3*	−3.8 ± 4.0		3.1 ± 1.8	1.6 ± 0.9*	−1.1 ± 1.4
	(l min–1  (mmHg )−1)	2.3 ± 0.8	2.5 ± 1.0	0.2 ± 0.6	(l min–1 (–%)−1)	1.3 ± 0.7	1.6 ± 0.9	0.3 ± 0.7
	Relative reactivity				Relative reactivity
MCAv		4.4 ± 1.2	3.2 ± 1.2*	−1.3 ± 1.9		2.1 ± 0.5	2.2 ± 0.8	0.2 ± 0.9
PCAv		4.4 ± 1.5	3.0 ± 1.3*	−1.3 ± 2.3		1.9 ± 0.6	2.1 ± 0.9	0.2 ± 0.8
	(% (mmHg )–1)	8.4 ± 1.7	5.6 ± 2.1*	−2.6 ± 2.0	(% (–%)−1)	2.7 ± 1.9	2.8 ± 2.0	−0.0 ± 1.4
		7.6 ± 2.5	5.5 ± 2.8*	−2.0 ± 3.7		3.4 ± 1.3	3.2 ± 1.5	0.5 ± 1.9
		16.7 ± 6.6	15.9 ± 7.6	−0.7 ± 4.7		10.9 ± 4.8	10.6 ± 4.8	−0.3 ± 4.8

^*Denotes a significant change from baseline post INDO administration (P < 0.05).

End-tidal forcing tests

During the steady-state hypercapnic test, the last 30 s of each stage was averaged for , CBV, CBF, MAP and HR before and after INDO administration. Relative responses to CO₂ or O₂ perturbations were normalized to baseline for each individual subject (see Table1). For the isocapnic hypoxia test, continuous 15 s averages were calculated for all respiratory and cerebrovascular data, except and , which were averaged over 30 s. Note that image quality for some measures was not of a high enough standard for analysis and use within the study, and as such, the sample size for volumetric flow measures is lower than all other variables. Specific sample sizes are noted for all statistical comparisons involving volumetric blood flow in the results section.

Calculations

For the hyperoxic hypercapnic trials, linear regression was performed on the ventilatory () and cerebrovascular data (MCAv, PCAv, and ) for all four vessels against to calculate ventilatory sensitivity and CVR, respectively. Using unpublished data we calculated a regression equation to predict from . Subsequently, respiratory and cerebrovascular reactivity were recalculated as a function of in the same manner in which they were calculated for . The resulting regression equation was:

2

This calculation was performed in an attempt to index cerebrovascular reactivity against a more representative value for the stimulus to breathe (Fencl, 1986), as has been done in previous investigations (Xie et al. 2006; Peebles et al. 2007).

The hypoxic ventilatory response (HVR) and CVR were calculated as both the absolute and relative change from baseline to the zenith of their respective responses, for measures of , MCAv, PCAv, , and . The HVR was calculated as a function of the concurrent reduction in arterial oxyhaemoglobin saturation (; i.e. l min^–1 ( –%)⁻¹). Using a previously established equation (Severinghaus, 1979), was calculated from to avoid the issue of time delay inherent to pulse oximetry when calculating cerebrovascular and ventilatory reactivity. As HVR (peak ) is a variable response in magnitude and time of occurrence, the time at which the zenith occurred varied between individuals. Within individuals, absolute hypoxic ventilatory decline (HVD) was calculated as the reduction in from the zenith to nadir, while relative HVD was calculated as a percentage of the initial HVR using eqn 3:

3

Statistics

Statistical analysis was performed using SPSS (IBM Statistics, Version 21, 2012). Student's paired t tests were used to compare baseline variables before and following INDO administration within-subjects. Paired t tests were also used to compare differences between anterior and posterior cerebrovascular reactivity and ventilatory sensitivity before and after INDO administration. Pearson R correlations were used to assess the relationship between cerebrovascular responsiveness and central and peripheral respiratory chemosensitivity, within subjects. Significance was assumed at P < 0.05.

Results

Baseline effects of indomethacin (Table1)

Following INDO administration, baseline MCAv and PCAv were reduced by 28 ± 17% and 23 ± 12%, respectively. Similarly, blood flow through the ICA (n = 7) and VA (n = 12) was reduced by 32 ± 13% and 32 ± 18%, respectively (Fig. 2). The relative decrease in CBV and CBF due to INDO did not differ between the anterior and posterior circulations. Post INDO, MAP, end-tidal gases and were unaltered while HR was lower prior to the hypercapnic (72 vs. 63 beats min^–1; P < 0.001), but not the hypoxic test. See Table1 for all baseline parameters.

Figure 2.

Absolute cerebrovascular flow responsiveness to CO₂ during steady-state hyperoxic hypercapnia pre- and post INDO

○, individual values; ▪, mean values. Mean cerebrovascular data plotted against steady-state hypercapnia steps with end-tidal forcing (0, +3, +6 and +9 ). A, absolute MCAv responses pre- and post INDO (cm s^–1 (mmHg )⁻¹; n = 13); B, absolute PCAv responses pre- and post INDO (cm s^–1 (mmHg )⁻¹); C, absolute ICA responses pre- and post INDO (ml min^–1 (mmHg )⁻¹); D, absolute VA responses pre- and post INDO (ml min^–1 (mmHg )⁻¹). *Significant change from baseline post INDO administration (P < 0.05).

Regional CBF reactivity (Table1)

During the pre-INDO trial CVR in the MCA was higher than the PCA in absolute (3.0 ± 1.0 vs. 2.0 ± 0.7 cm s^–1 (mmHg )⁻¹; P = 0.01), but not relative (4.4 ± 1.2 vs. 4.4 ± 1.5% (mmHg )⁻¹; P = 0.91) terms. Post INDO, regional differences (MCA vs. PCA) persisted in absolute (1.7 ± 0.7 vs. 1.1 ± 0.6 cm s^–1 (mmHg )⁻¹; P = 0.02) but not relative (3.2 ± 1.2 vs. 3.0 ± 1.3% (mmHg )⁻¹; P = 0.78) reactivity.

Absolute CVR of the ICA was significantly higher than that of the VA (18.1 ± 4.1 vs. 7.3 ± 4.2 ml min^–1 (mmHg )⁻¹; n = 7; P = 0.003) while there was no difference in relative CVR (8.4 ± 1.7 vs. 7.6 ± 2.5% (mmHg )⁻¹; n = 7; P = 0.43) pre-INDO. Absolute CVR in the ICA remained higher than that of the VA post INDO (9.2 ± 4.2 vs. 2.9 ± 1.3 ml min^–1 (mmHg )⁻¹; n = 6; P = 0.03); however, relative CVR of the VA was higher than that of the ICA post INDO (6.0 ± 2.1 vs. 8.1 ± 2.1% (mmHg )⁻¹; n = 6; P = 0.02).

Hypoxic CVR during the pre-INDO trial was greater in the MCA than PCA in absolute (1.4 ± 0.4 vs. 0.8 ± 0.3 cm s^–1 (–)⁻¹; n = 21; P < 0.001) but not in relative (2.0 ± 0.5 vs. 1.9 ± 0.6 cm s^–1 (–)⁻¹; n = 21; P = 0.52) terms. Absolute CVR of the MCA remained higher than that of the PCA post INDO (n = 13; P = 0.04) while there was no difference in relative CVR (n = 13; P = 0.9).

Absolute CVR of the ICA to hypoxia was higher than that of the VA (6.8 ± 3.3 vs. 3.3 ± 1.5 ml min^–1 (–)⁻¹; n = 11; P < 0.001) while relative CVR was also the same between vessels (3.0 ± 1.7 vs. 3.2 ± 1.2 ml min^–1 (–)⁻¹; n = 11; P = 0.40). Following INDO, absolute CVR of the ICA remained higher than that of the VA (P = 0.02) with no difference in relative CVR (P = 0.29).

Cerebrovascular CO₂ reactivity and ventilatory sensitivity to hypercapnia

Following INDO there was a 42 ± 28% (P = 0.001) reduction in absolute CVR of the MCA while relative CVR was reduced by 20 ± 43% (P = 0.03). The PCAv showed near identical reductions as compared to the MCA in both absolute and relative CVR of 42 ± 38% (P = 0.001) and 17 ± 55% (P = 0.04), respectively. Absolute CVR of the ICA was reduced by 48 ± 19% (n = 7; P = 0.002) while relative CVR was reduced by 32 ± 22% (n = 7; P = 0.01). The VA decreased 45 ± 32% (n = 12; P = 0.007) in absolute CVR, and there was an 18 ± 48% tendency for a reduction in relative CVR (n = 12; P = 0.08). Note that the smaller reduction in relative CVR of all vessels compared to absolute reactivity is likely to be result of reduced baseline flow consequently inflating the post-INDO percentage changes from baseline. Despite these marked reductions of CVR in both TCD and volumetric flow measures of the anterior and posterior vasculature, ventilatory sensitivity to elevations in was unaltered following INDO administration (Fig. 3).

Figure 3.

Pre- and post INDO central chemoreflex responsiveness to steady-state hypercapnia

A, mean ventilatory data (n = 14) plotted against steady-state hypercapnia steps with end-tidal forcing (0, +3, +6 and +9 ). Mean central chemoreflex responsiveness to elevations in both (▪ and □) and (• and ○). Error bars represent one standard deviation. B, individual (□ and ○) and mean (▪ and •) ventilatory reactivity to elevations in (▪ and □) and (• and ○).

Relationship between cerebrovascular CO₂ reactivity and ventilatory sensitivity to hypercapnia

Ventilatory to hypercapnia during the pre-INDO trials was positively correlated with both relative (R² = 0.55; n = 8; P = 0.035) and absolute (R² = 0.50; n = 8; P = 0.05) CVR to CO₂ in the ICA. This relationship with ventilatory sensitivity was not present in the MCA, or in the posterior vessels (PCA and VA). Following INDO, there were no relationships between the absolute and relative reductions in CVR of any of the cerebral vessels to variability in ventilatory sensitivity (Fig.4).

Figure 4.

Relationship between the INDO-induced reduction in cerebrovascular CO₂ reactivity and variability in central chemoreflex CO₂ sensitivity pre- and post INDO

Delta (Δ) values were calculated within-subject from absolute cerebrovascular and respiratory reactivity slopes (see Figs 2 and 3). A, ΔMCAv CO₂ reactivity plotted against Δ sensitivity, within-subject (n = 13). B, ΔPCAv CO₂ reactivity plotted against Δ sensitivity, within-subject (n = 14). C, ΔICA CO₂ reactivity plotted against ΔCCR reactivity, within-subject (n = 7). D, ΔVA CO₂ reactivity plotted against Δ sensitivity, within-subject (n = 12).

Relationship between jugular venous CO₂ reactivity and ventilation

Upon estimating as a function of , there were no correlations present between the variability in ventilatory sensitivity and changes in any vessels’ CVR post INDO, both absolute and relative. Similar to the pre-INDO data, however, relative CVR of the ICA to was positively correlated to both absolute (R² = 0.49; n = 8; P = 0.05) and relative (R² = 0.55; n = 8; P = 0.04) ventilatory sensitivity (l min^–1 (mmHg)⁻¹). The reduction in CVR persisted in all vessels after estimation of from (P < 0.01).

Cerebrovascular O₂ reactivity and HVD

Following administration of INDO, absolute hypoxic CVR of the VA, but not PCA, was reduced (3.09 ± 1.8 vs. 1.62 ± 0.9 ml min^–1 (–%)⁻¹; n = 7; P = 0.04; Fig. 5) while the magnitude of HVD was elevated (−14.2 ± 11.4 vs. −16.6 ± 10.9 l min⁻¹; P = 0.02). These concurrent changes were correlated (R² = 0.56; n = 7; P = 0.05). A reduction in CVR of the MCA (1.4 ± 0.4 vs. 1.0 ± 0.4 cm s^–1 (–%)⁻¹; P = 0.02), but not ICA, was observed post INDO (Fig. 5); however, the change in CVR of the MCA was unrelated to the change in HVD (R² = 0.04; P = 0.52). Note that the magnitude of HVR was not different pre- and post INDO.

Figure 5.

Absolute cerebrovascular flow responsiveness to hypoxia during steady-state isocapnic hypoxia pre- and post INDO

○, individual values; ▪, mean values. A, absolute VA responses pre- and post INDO (ml min^–1 (–%)⁻¹; n = 7); B, absolute ICA responses pre- and post INDO (ml min^–1 (–%)⁻¹; n = 7). C, absolute PCAv responses pre- and post INDO (cm s^–1 (–%)⁻¹; n = 14); D, absolute MCAv responses pre- and post INDO (cm s^–1 (–%)⁻¹; n = 13); *Significant change from baseline post INDO administration (P < 0.05).

Relationship between O₂ reactivity and HVD

The magnitude of HVD was not related (all: R² < 0.02) to the peak increase MCAv, PCAv or ICA flow (or the related reactivity of these vessels; see Fig. 6) pre- and post INDO. In contrast, the magnitude of HVD was correlated with both the relative increase in VA flow (R² = 0.32; n = 16; P = 0.02) and relative CVR to hypoxia of the VA (R² = 0.27; n = 16; P = 0.04) upon exposure to hypoxia pre-INDO. Moreover, this relationship persisted post INDO for both the relative increase in VA flow (R² = 0.81; n = 10; P < 0.001) and relative CVR of the VA (R² = 0.58; n = 10; P = 0.01). Upon pooling the pre/post-INDO data, a similar relationship is found (R² = 0.37; n = 26; P<0.001; Fig. 6A). The time to peak was negatively correlated with absolute CVR of the VA following INDO administration (R² = 0.50; n = 10; P = 0.02).

Figure 6.

Relationship between hypoxic ventilatory decline (HVD) and relative cerebrovascular flow responsiveness to hypoxia during steady-state isocapnic hypoxia

A, in the VA; B, in the ICA; C, in the PCA; D, in the MCA. The relationship between HVD (calculated as the zenith minus the nadir of ventilation during hypoxia) and relative reactivity of all vessels is plotted with the regression line and R² value representing the relationship of the pooled pre- (•) and post-INDO (○) data.

Discussion

The principle findings of our study were as follows: (1) despite INDO blunting the CVR to CO₂ in all cerebral vessels, ventilatory CO₂ sensitivity was unaltered when assessed with end-tidal forcing, (2) although there were comparable INDO-induced reductions in regional CBF ( and ), a reduction in CVR to hypoxia was evident in the posterior circulation (), and (3) the degree of HVD magnitude during isocapnic hypoxia was correlated to CVR of the VA, which represents the upstream perfusate of the brainstem and thus the central chemoreceptors.

Differential cerebrovascular reactivity to hypercapnia and hypoxia

In agreement with previous studies (Willie et al. 2012; Skow et al. 2013) we observed greater absolute but not relative CVR to CO₂ in the anterior compared to posterior circulation under control conditions. However, following INDO administration, the relative CVR to CO₂ of the VA was higher than that of the ICA, despite no difference between the MCA and PCA. As it is the vertebrobasilar system that perfuses brainstem and respiratory centres, we provide evidence that it may be inappropriate to relate anterior and/or velocity measures of CVR to ventilatory sensitivity as we (Fan et al. 2010a,b) and others (Xie et al. 2005, 2006, 2009) have previously done. To the best of our knowledge, it is not known whether there are regional differences in prostanoid-mediated cerebrovascular dilation between the ICA and VA.

During isocapnic hypoxia, we found no statistical differences in relative CVR between the anterior and posterior circulations. While this finding is in contrast to an earlier study (Willie et al. 2012; Lewis et al. 2014) that reported higher relative CVR in the VA compared to ICA (based on n = 11), we report no difference in CVR (n = 11). This disparity is likely to be due to differing characteristics of both the hypoxic exposure and calculation of reactivity.

Cerebrovascular reactivity to hypercapnia and chemoreflex sensitivity

During hypercapnia, increases in CBF act to attenuate decreases in medullary pH (Neubauer et al. 1985) via the washout of CO₂ and hence reduction in brain tissue (partial pressure of brain tissue carbon dioxide ()). Irrespective of the type or magnitude of hypercapnic challenge, will always be greater than due to metabolic CO₂ production, resulting in a positive − gradient. Increases in CBF act to increase central pH by reducing this gradient, as greater flow through the arterial system will draw closer to equilibrium with (see Fig. 7). Previous data have exemplified this affect of cerebrovascular reactivity on central /[H+] in animals (Chapman et al. 1979; Neubauer et al. 1985).

Figure 7.

A theoretical schematic representing the relationship between two fluid compartments separated by a semi permeable membrane

When two fluid compartments separated by a semipermeable membrane are adjacent to one another, a square wave change in the concentration of a substance in one fluid compartment (C1) will be 95% equilibrated with the opposite compartment (C2) over three time constants. A time constant (τ) represents the time required for a volume of fluid equal to that of C1 to flow through C2 (e.g. if C1 is 500 ml, 1.5 l would have to flow through C2 to reach 95% equilibration). This model can be applied to the cerebral vasculature and brain tissue compartment to understand the effect of cerebrovascular CO₂ reactivity in producing varying stimuli to breathe despite the same value for (or ), as is seen in our placebo/INDO intervention. The major limitation of this model is that if we use C2 to represent arterial blood vessels proximal to the central chemoreceptors, and C1 to represent the tissue compartment where the central chemoreceptors reside, C1 is, in contrast to a steady-state, constantly producing CO₂ via metabolism. Therefore, a simple square wave change cannot be assumed between blood and tissue CO₂, but for simplicity we can assume that the relationship between arterial and tissue CO₂ resides somewhere on the dashed line, based upon the magnitude of flow and metabolic production of CO₂ at any given time. Assuming constant CO₂ production, higher flow would result in a rightward shift down the line, and an overall reduced blood to tissue gradient.

Human studies (Xie et al. 2006; Fan et al. 2010b) utilizing steady-state elevations in fractional inspired CO₂ (+2, 4, and 6%, Xie et al. 2006; 7%, Fan et al. 2010b) showed that an INDO-induced reduction in CVR to CO₂ results in a concurrent increase in ventilatory sensitivity to hypercapnia. However, no correlative data between the concurrent ∼70% reduction in CVR and ∼40–60% increase in ventilatory gain were reported (Xie et al. 2006). To the best of our knowledge, we are the first to report the magnitude to which these two variables co-vary during pharmacological reductions in CVR to CO₂ (Fig.4). Thus, we provide insight into the magnitude by which CVR to CO₂ may influence ventilatory sensitivity within subjects. Furthermore, we are the first to integrate CVR to CO₂ and ventilatory sensitivity using volumetric CBF measurements ( and ). Past conclusions (Xie et al. 2005, 2006, 2009; Fan et al. 2010a) are based on TCD recordings of the MCA alone, the validity of which is now unclear, especially during changes in CO₂ (Giller, 2003; Willie et al. 2012; Ainslie & Hoiland, 2014; Coverdale et al. 2014; Verbree et al. 2014) and hypoxia (Wilson et al. 2011; Willie et al. 2012, 2013). Moreover, these recordings were of anterior, and not posterior flow, the potential implications of which we discussed above (see section headed ‘Differential cerebrovascular reactivity to hypercapnia and hypoxia’).

Brainstem blood flow and HVD

During isocapnic hypoxia we demonstrate for the first time that a higher posterior, but not anterior, CVR to hypoxia () is selectively linked to a larger HVD. There was no relationship between CVR and HVR, which is perhaps not surprising as HVR is mediated by the peripheral chemoreceptors and peak preceded peak VA flow by 150 ± 138 s, highlighting that there could be no time order effect of posterior flow on HVR. Previously, Poulin & Robbins (1998) reported that CVR to hypoxia was not related to variability in HVD; however, their study used MCAv as a surrogate for brainstem flow and recent data indicate that relative posterior CVR to hypoxia () is greater than that of anterior circulation (; Willie et al. 2012). In addition, the possibility that the MCA probably dilates in this degree of hypoxaemia should be considered (Wilson et al. 2011; Willie et al. 2012, 2013). As represents flow through the vertebrobasilar system which perfuses the downstream respiratory centres, the current and previous data (Willie et al. 2012; Lewis et al. 2014) highlight the need to utilize both posterior and volumetric blood flow measures when investigating the relationship between CVR and ventilation. The mechanism by which increased reactivity increases the magnitude of HVD in humans is somewhat speculative. However, animal data indicate that an increase in CO₂–[H⁺] washout at the central chemoreceptors via elevations in brainstem flow results in a reduction of the ‘central’ drive to breathe and hence the presence of HVD (Neubauer et al. 1985; Fig. 7). This is consistent with a previous investigation by Suzuki et al. (1989) in which they report a reduction in during isocapnic hypoxia (Suzuki et al. 1989). Assuming constant cerebral metabolism (Ainslie et al. 2014), the concomitant percentage change in flow can be estimated using the Fick equation as follows:

4

The estimated percentage change calculated for Suzuki et al. (1989) is slightly greater compared to ours (∼78% vs. ∼60%), probably due to the index of global, not posterior flow changes. As CBF calculated via the Fick method tends to be greater than that of ultrasound, this further supports the influence of VA reactivity on CO₂–[H⁺] washout and HVD (Suzuki et al. 1989). Furthermore, greater CO₂–[H⁺] washout is likely responsible for reducing ventilatory drive, therefore, reaching and initiating HVD earlier as time to peak was inversely correlated to absolute CVR of the VA to hypoxia. While our findings do support CO₂–[H⁺] washout modulating HVD, our study design cannot provide insight into the contributions from other factors such as decreased peripheral chemoreceptor sensitivity (Honda et al. 1997), changes in peripheral chemoreceptor threshold (Mahamed & Duffin, 2001), and/or disassociation between CNS input and phrenic output (Melton et al. 1992).

Chemoreflex stimulus – methodological considerations

Previous data have indicated that CVR to CO₂ indexed against , but not and , is correlated to ventilatory sensitivity to CO₂ (Peebles et al. 2007). It is important to consider that does not directly act upon the central medullary chemoreceptors. Instead, accumulation of [H⁺] from increases in stimulates breathing downstream from changes in . This provides a reasonable explanation for the lack of correlation found between each vessel's reactivity and ventilatory sensitivity to elevations in both pre- and post INDO. As is accepted as a more appropriate index of the local CO₂–[H⁺] stimulus at the medullary central chemoreceptors (Fencl, 1986; Xie et al. 2006), these results may simply be due to the inherent methodological limitations of using as a surrogate for /[H⁺] in brain tissue. As such, we derived a regression equation using unpublished data from our lab that included several of the same subjects as the present study to estimate from during hypercapnia at sea level using the same end-tidal forcing system. The estimation of CVR as a function of is shown in Fig. 3. No change in ventilatory sensitivity pre- vs. post INDO was observed when expressed against , despite a simultaneous reduction in CVR of all vessels similar to that of the data. We interpret these findings to suggest that, as ventilatory sensitivity expressed with either or were unaffected by reductions in CVR to hypercapnia, the end-tidal forcing approach we used reduced any impact of gradients on central chemosensitivity (Fig. 8C).

Figure 8.

Arterial to jugular venous carbon dioxide gradients and ventilatory sensitivity during varying chemoreflex tests

The change in pre- and post INDO (continuous and dotted lines, respectively) during hypercapnia as a function of steady-state fractional inspired CO₂ (Xie et al. 2006), rebreathing (Fan et al. 2010a), and end-tidal forcing methods of controlling (current study). The bottom panel represents the changes in − gradient expected using each method. A, for the steady-state method, the pre-INDO gradient data points (•) are calculated from Peebles et al. 2007, while the post-INDO data points (○) are theoretical. As INDO-induced reductions in cerebrovascular CO₂ reactivity should reduce CO₂ washout, it is expected that is not reduced to as great an extent post INDO, keeping the − gradient larger than in the pre-INDO scenario and resulting in increased ventilatory sensitivity. B, during rebreathing, the gradient between and is theoretically eliminated (Read & Leigh, 1967) and is independent of reactivity and thus should be the same pre- (•) and post INDO (○). This latter notion is supported by no changes in ventilatory sensitivity pre- and post INDO using the rebreathing method (Fan et al. 2010a). C, similar to the rebreathing method, we found no change in ventilatory sensitivity pre- and post INDO with end-tidal forcing. The pre-INDO (•) data points represent the − gradient recorded from previously collected data in our laboratory, while post-INDO (○) points represent the likelihood that − gradient magnitude was unchanged, explaining the similar ventilatory sensitivity pre- and post INDO.

Clinical implications

Blunted CVR is associated with increased breathing instability at both rest (Fan et al. 2010a) and during sleep (Xie et al. 2009). Moreover, INDO-induced reductions in CVR is reported to worsen obstructive sleep apnoea at sea level (Burgess et al. 2014b) and central sleep apnoea at high altitude (Burgess et al. 2014a). Reductions in CVR are also associated with the occurrence of central sleep apnoea in congestive heart failure patients (e.g. Xie et al. 2005), and these patients have been shown to have an increased central chemoreflex magnitude (Xie et al. 1995; Solin et al. 2000; Topor et al. 2001). Our study indicates that due to differential CVR between the anterior and posterior cerebrovasculature, it may be more appropriate to investigate ventilatory and cerebrovascular integration with posterior vessel recordings in clinical populations. Furthermore, hypoxia-induced cerebral dilatation and CO₂–[H⁺] washout from central chemoreceptors may exacerbate the effects of high peripheral chemoreflex magnitude on breathing instability by augmenting the resulting brain tissue hypocapnia and withdrawal of central chemoreceptor stimulus (Dempsey & Strakud, 1986; Dempsey et al. 2004, 2010).

Methodological considerations

Pharmacological manipulation of CBF

The utility of INDO in assessing the effects of a reduction in CVR to CO₂ is exemplary due to its ability to elicit a large response in the absence of unwanted side-effects. For example, INDO is capable of reducing CVR to CO₂ by ∼40–60% (this study), while not affecting resting ventilation (e.g. Fan et al. 2010a), cerebral metabolism (e.g. Kraaier et al. 1992), or plasma catecholamines (e.g. Staessen et al. 1984; Wennmalm et al. 1984; Green et al. 1987). Furthermore, any possible affect of INDO on the peripheral chemoreceptors is unlikely. For example, Xie et al. (2006) reported similar ventilatory sensitivity in both hyperoxic and normoxic hypercapnia, highlighting that INDO drives changes in ventilatory sensitivity through modulation of central chemoreceptor stimulation (Xie et al. 2006).

Comparing methods utilized for gas challenges

During incremental hypercapnia (from ∼40 to 55 mmHg), recent data from our group show that during end-tidal forcing (authors’ unpublished data) the − gradient is much smaller (∼1–2 mmHg) as opposed to (∼4–6 mmHg) during increases in fractional inspired CO₂ (; Peebles et al. 2007). The utility of steady-state , rebreathing, and end-tidal forcing methods for the assessment of cerebrovascular and ventilatory integration has been extensively reviewed (Ainslie & Duffin, 2009). While clearly depending on the experimental question, there is little consensus on the most appropriate method to investigate ventilatory and cerebrovascular integration and consideration of the aforementioned differences in − gradients between methods is of importance. To date, studies implicating CVR to CO₂ in the control of breathing have exclusively employed alterations in as the breathing stimulus, with our use of end-tidal forcing combined with regional CBF being the first in this experimental context. In light of our study, it is now apparent that reductions in CVR elicit an increase in ventilatory sensitivity for steady-state tests (Xie et al. 2006), but not rebreathing (Fan et al. 2010a) or end-tidal forcing (current data; Fig. 8). Therefore, end-tidal forcing is similar to rebreathing in that variability in ventilatory sensitivity to hypercapnia is independent of CVR (Fan et al. 2010a). One additional consideration is that, during a steady-state test, and subsequently are influenced by ventilation, and as such so is CVR. In contrast, end-tidal forcing results in a fixed change in and subsequently independent of ventilation, and therefore, CVR during such a test can be considered ‘ventilatory independent’ (Ainslie & Duffin, 2009).

As a smaller − gradient results in less of an effect of CVR on the central CO₂–[H⁺]-mediated stimulus to breath (Read & Leigh, 1967), it would seem plausible that the − gradients in the current study were reduced to an extent that did not allow for appreciable changes in ventilatory sensitivity due to CO₂ washout, and downstream medullary pH changes. However, as seen in Fig. 8, the initial gradients at rest between a steady-state test and end-tidal forcing are quite similar. Interestingly, if the slope change in the gradient ( − , in mmHg) versus the increase in is calculated, there is less of a change in gradients during end-tidal forcing than with the steady-state method (−0.27 mmHg −  (mmHg) vs. −0.33 mmHg −  (mmHg); see fig. 8). As CBF is linearly related to above eupnoea, this reflects that flow has less of an effect on gradient narrowing during end-tidal forcing compared to steady-state tests. As such, when using end-tidal forcing the smaller gradient change prior to INDO administration results in a reduced − . Therefore, the impact of INDO-induced reductions in CVR will be inherently smaller, reducing the consequent influence on ventilatory sensitivity. Whilst speculative, there may be a threshold change in gradient magnitude necessary to affect ventilatory sensitivity, which we have not achieved in our study as gradients narrow to a lesser extent during end-tidal forcing compared with steady-state changes in . This provides a possible explanation for our observation of no change in ventilatory sensitivity after the large INDO-induced reduction in CVR. The implications of the inherent ventilatory dependency of different chemoreflex tests in the assessment of ventilatory and cerebrovascular reactivity are unclear and should be considered in future research of this type (Fierstra et al. 2013).

Conclusions

Our study is the first to use end-tidal forcing to explore the relationship between CVR in anterior intra- and extra-cranial vessels and ventilatory chemoreflexes. The main conclusions of this study are that despite large reductions in CVR of posterior intra- and extra-cranial arteries in the vertebrobasilar circuit using INDO, ventilatory sensitivity remained unchanged with end-tidal forcing. Previous studies showing increased central chemoreflex magnitude with INDO used steady-state increases in . Thus, we speculate that end-tidal forcing is a somewhat intermediate method between steady-state changes in (where − gradients are maintained) and rebreathing (where − gradients are reduced or eliminated). Our data also reinforce the need to measure posterior vasculature when exploring the effects of cerebrovascular regulation and the chemoreflex control of breathing. Lastly, we conclude that in humans, posterior cerebral blood flow mediates CO₂–[H⁺] washout at the central chemoreceptors during steady-state hypoxia and, in part, mediates HVD. Future studies are now needed to better explore the role of brainstem blood flow in the regulation of breathing during physiological (e.g. sleep, exercise, ageing) and pathological conditions (e.g. congestive heart failure, central sleep apnoea).

Translational perspective.

The present study highlights two very important methodological factors pertaining to the assessment of cerebrovascular flow responsiveness (CVR) and ventilatory integration in humans: (1) measurements of posterior blood flow (an index of blood flow in the brainstem) are essential to adequately determine the effect of CVR on stimuli to breathe and (2) end tidal forcing – a widely used and sophisticated approach – reduces the effect of CVR on the arterial to brain tissue carbon dioxide gradient to an extent that does not allow for appreciable effects on ventilatory sensitivity after Indomethacin administration. Therefore, depending on the research question, in both clinical and experimental research it may be more appropriate to utilize steady state changes in fractional inspired carbon dioxide. With this approach reactivity induced changes in gradient magnitude are apparent (see Fig.7), thereby permitting the interaction between CVR and ventilatory control to be assessed. As all previous studies have exclusively assessed anterior CVR and its integration with ventilatory sensitivity, the influence of CVR on ventilation may have been partially underestimated or misinterpreted. For example, impairments in CVR are known to augment breathy instability in healthy humans (Xie et al. 2006), and exacerbate central sleep apnea at high altitude (Burgess et al. 2014a), whilst the occurrence of central sleep apnea in congestive heart failure patients is also associated with impairments in CVR (Xie et al. 2005). Thus, the implication of CVR in clinical ventilatory dysregulation is apparent (Morgan et al. 2010) and future research should further quantify posterior CVR in these populations.

Glossary

Abbreviations

CBF

cerebral blood flow

CBV

cerebral blood velocity

CO₂

carbon dioxide

fractional inspired CO₂

HR

heart rate

HVD

hypoxic ventilatory decline

HVR

hypoxic ventilatory response

ICA

internal carotid artery

INDO

indomethacin

MAP

mean arterial pressure

MCA

middle cerebral artery

MCAv

middle cerebral artery blood velocity

partial pressure of arterial oxygen

partial pressure of arterial carbon dioxide

partial pressure of brain tissue carbon dioxide

PCA

posterior cerebral artery

PCAv

posterior cerebral artery blood velocity

partial pressure of end-tidal carbon dioxide

partial pressure of end-tidal oxygen

partial pressure of jugular venous carbon dioxide

internal carotid artery blood flow

vertebral artery blood flow

arterial oxyhaemoglobin saturation

TCD

transcranial Doppler ultrasound

ventilation

VA

vertebral artery

Additional information

Competing interests

No author declares a conflict of interest.

Author contributions

R.H., study organization, data collection and analysis, data interpretation, manuscript first draft; P.N.A., study design, laboratory space, funding, ethics, manuscript writing, critical revisions; K.W., data collection and analysis, data interpretation; K.J.S., study organization, data collection and analysis, critical revisions; A.B., data collection and analysis, critical revisions; C.K.W., ethics, data collection and analysis, critical revisions; B.M., medical supervision; T.A.D., study design and organization, data collection, analysis and interpretation, manuscript writing, critical revisions. All authors gave feedback on and approved the final manuscript.

Funding

P.N.A. and the work conducted in this project were supported by a Canada Research Chair in Cerebrovascular Physiology and an NSERC Discovery Grant. T.A.D. was supported via an MRU sabbatical.