Regulation of brain blood flow and oxygen delivery in elite breath-hold divers

Abstract

The roles of involuntary breathing movements (IBMs) and cerebral oxygen delivery in the tolerance to extreme hypoxemia displayed by elite breath-hold divers are unknown. Cerebral blood flow (CBF), arterial blood gases (ABGs), and cardiorespiratory metrics were measured during maximum dry apneas in elite breath-hold divers (n=17). To isolate the effects of apnea and IBM from the concurrent changes on ABG, end-tidal forcing (‘clamp') was then used to replicate an identical temporal pattern of decreasing arterial PO₂ (PaO₂) and increasing arterial PCO₂ (PaCO₂) while breathing. End-apnea PaO₂ ranged from 23  to 37 mm Hg (30±7 mm Hg). Elevation in mean arterial pressure was greater during apnea than during clamp reaching +54±24% versus 34±26%, respectively; however, CBF increased similarly between apnea and clamp (93.6±28% and 83.4±38%, respectively). This latter observation indicates that during the overall apnea period IBM per se do not augment CBF and that the brain remains sufficiently protected against hypertension. Termination of apnea was not determined by reduced cerebral oxygen delivery; despite 40% to 50% reductions in arterial oxygen content, oxygen delivery was maintained by commensurately increased CBF.

Introduction

The regulation of cerebral blood flow (CBF) with hypoxia is complicated due to the ventilatory response to hypoxia and resultant hypocapnia. Hypoxia per se causes cerebral vasodilation and increases CBF, whereas hypocapnia induces vasoconstriction and results in a reduction in CBF.^{1, 2} Thus, brain perfusion in hypoxia is a balance between the degree of hypoxia, which is influenced by an individual's ventilatory sensitivity to hypoxia, and the cerebrovascular reactivity to hypocapnia.^{3, 4} Indeed, despite profound hypocapnia (12 to 14 mm Hg), CBF increased by ~200% at 7,950 m where oxyhemoglobin saturation (SaO₂) is reduced to ~65%⁵—a physiologic defense of brain tissue oxygenation that appears to prioritize the brainstem.^{6, 7} Data on CBF in humans are generally in the context of either hypobaric hypoxia, i.e., high altitude^{5, 8, 9, 10, 11} or by various means of altering inspired gases to induce acute poikilocapnic or isocapnic hypoxia.^{6, 7, 12}

Elite breath-hold divers experience a similar or greater magnitude of transient hypoxemia as do individuals exposed to very high altitudes, but with concomitant acidosis due to build up of blood and tissue CO₂. Prolonged apnea thus features two factors (hypoxia and hypercapnia) that together generate an extremely potent cerebral vasodilatory stimulus.^{13, 14} Prolonged apneas of >4 minutes may attain values of arterial partial pressures of oxygen (PaO₂) and carbon dioxide (PaCO₂) of ~19 to 25 mm Hg and ~30 to 50 mm Hg, respectively.^{15, 16} The current world record for static apnea is an outstanding 11 minutes 35 seconds. Apnea of such duration likely results in even greater changes to arterial blood gases (ABGs); extremely low SaO₂'s are regularly reported anecdotally (personal correspondence, Croatian National Free-diving Coach), yet the mechanism(s) finally triggering termination of apnea remains unknown. Because of their unique ability to generate and tolerate profound hypoxemia, elite breath-hold divers are an inimitable model for assessing the cerebrovascular regulatory response to extreme changes in ABGs otherwise observed only during pathophysiologic events involving life-threatening hypoxia.

A significant physiologic characteristic of maximum apnea is the occurrence of involuntary breathing movements (IBMs). Triggered by threshold chemoreceptor stimulation—typically occurring halfway through the apnea—these increasingly powerful contractions of the respiratory muscles continue until apnea cessation and result in transient elevations in venous return and cardiac output.¹⁷ The profound changes in ABGs with prolonged apnea cause CBF to increase by >100% (as indexed by velocity in the middle cerebral artery), but IBM in conjunction with progressive increases in sympathetic nervous activity¹⁸ and mean arterial pressure (MAP)¹⁹ may additionally augment cerebral perfusion.¹³ Conversely, cerebral sympathetic nerve activity in other paradigms that induced transient surges in MAP (e.g., rapid eye movement sleep) attenuates surges in CBF,^{20, 21} and chemoreflex augmentation of muscle sympathetic nerve activity is nearly double during apnea than with breathing for the same values of PaO₂ and PaCO₂.²² The specific effects of the apnea-induced changes in blood gases versus the combined blood gas and circulatory effects of IBM on CBF per se remain ambiguous.

It is often assumed that cerebral ischemia determines the termination of apnea²³ but no study to-date has measured cerebral oxygen delivery. To explicate the relative import of ABGs, cerebral oxygen delivery, and IBM on CBF, we directly measured ABGs and regional CBF throughout maximum dry static apneas in elite breath-hold diving athletes. By directly measuring ABGs during apnea and after the apnea break point, we were able to assess whether a common hypoxic threshold existed among divers. On an individual basis, we then replicated each individual breath-hold time course of ABG ‘profiles' (i.e., declining PaO₂ and increasing PaCO₂) during breathing by end-tidal clamping. We hypothesized that (1) reduced oxygen delivery would determine apnea break point; and (2) the magnitude of CBF increase would be greater during apnea than during clamped breathing due to larger increases in MAP.

Materials and methods

Participants

Seventeen competitive and elite breath-hold divers (4 female; age 29.5±6.5 years; body mass index (BMI) 24±2.5 kg/m²) were selected from the Croatian cities of Split and Zagreb, most of whom were being coached by the Croatian national apnea coach at the time of testing and had been practicing competitive breath hold diving for 2 to 8 years (mean 4.7±2.5 years). There are numerous disciplines of free diving, with many competitors taking part in multiple events. Pool events consist of ‘static' apnea, where the athlete floats face down aiming to maintain apnea for as long as possible; and ‘dynamic' apnea where the diver swims underwater to their maximum distance. In depth events, the goal is to attain the greatest depth, with athletes therefore being regularly exposed to tremendous pressures. The majority of subjects in this study were competitive in pool disciplines of free diving with nine additionally involved in various depth disciplines; divers had not dove to depth within a week preceding the study. Seven of the subjects were world-class free diving competitors, having placed top-ten within the last 3 years in international competition in at least one event. Three subjects had recently set new official world records. Subject characteristics are presented in Table 1. The experimental procedures were approved by the ethical committee of the University of Split School of Medicine, and by the Clinical Research Ethics Board of the University of British Columbia, and conformed to the standards set by the Declaration of Helsinki.

Table 1. Anthropometric and performance characteristics.

	Mean±s.d.	Range
Age (years)	29.9±6.8	19–48
Weight (kg)	77.2±16	53.2–98.4
Height (cm)	180±12	156.8–194
BMI (kg/m2)	24±2.5	19.1–27.1
FVC (L)	6.8±1.4	3.9–8.5
FVC1 (L)	5.4±1.0	3.9–8.5
Body fat index (%, body fat/kg)	10.6±2.5	7.3–17.3
VO2 max (L/min)	4.4±1.1	2.5–5.8
Personal best static apnea (seconds)	377±75	305–558
Personal best dynamic apnea (m)	187±38	125–281
Years breath-hold diving	4.7±2.5	2–8
Breath-hold duration (seconds)	316±60	212–435
End apnea end-tidal PO2 (mm Hg)	37±14	24–77
End apnea end-tidal PCO2 (mm Hg)	45±7	34–60

BMI, body mass index; FVC, forced vital capacity, FVC₁, forced vital capacity in 1 second; body fat index, based on 7-site skin folds; VO₂ max, maximal oxygen uptake (running); static apnea performed floating prone in a pool; dynamic apnea performed in a pool. Values are means±s.d.

Experimental Design

Subjects visited the laboratory on one occasion. The procedures were explained in the subjects' first language and signed informed consent was obtained. Standard pulmonary function measures were first collected (Table 1) before the subject lay supine for the placement of a 20-gauge radial arterial catheter (Arrow, Markham, Ontario, Canada) by ultrasound guidance under local anesthesia (1% lidocaine). The catheter was attached to an in-line waste-less sampling system and a pressure transducer at the level of the right atrium (Edwards Lifesciences VAMP, and TruWave transducer, Irvine, CA, USA). After cannulation, subjects rested for ~30 minutes during the set-up of the monitoring equipment.

Breath-hold divers typically complete a number of preparatory apneas before attempting a maximum apnea. All subjects therefore completed two such preparatory apneas with 2 minutes of resting breathing between. The first apnea was held long enough to elicit seven IBMs, the second apnea was held through ten IBMs. Then, after 7 minutes of rest measurements were obtained while the subject completed a maximum apnea with verbal encouragement from their coach. Arterial samples were drawn every 30 seconds starting 30 to 60 seconds before apnea onset, to 2 minutes after completion of apnea. After a full recovery of at least 30-minutes rest, the end-tidal clamping procedure began. On the basis of the PaO₂ and PaCO₂ profile determined during the apnea, the exact time profile was followed for each individual (see end-tidal forcing below; Figure 1), and verified with arterial samples every 30 seconds.

Figure 1.

Changes in arterial blood gases (ABGs) during maximal apnea. (A) Subject and arterial blood samples before commencing and at near-break point of a maximum apnea. Profound cyanosis in the subject's face is evident; hypoxemic coloration of the arterial blood sample taken at the end of apnea is visible by color. (B) PaO₂ and PaCO₂ before, during, and after apnea (solid lines) and clamp (dashed lines). Clamp was performed by end-tidal forcing—the subject breathed altered inspired fractions of oxygen, carbon dioxide, and nitrogen to replicate their ABG profile measured during a maximal apnea. * indicates significant difference between apnea and clamp (P<0.05).

Measurements

Cerebrovascular measures

Blood velocities in the right middle cerebral artery (MCAv; measured 1 cm from anterior cerebral artery-MCA bifurcation) and left posterior cerebral artery (PCAv; P1 segment) were measured by 2 MHz pulsed transcranial Doppler ultrasound (Spencer Technologies, Seattle, WA, USA). Signal quality was optimized using standardized search techniques as previously described that produce test-retest reliability of ~3% and 2% for MCAv and PCAv, respectively.²⁴

Extracranial blood flow in the right internal carotid (ICA) and left vertebral (VA) arteries was measured with Duplex vascular ultrasound. Right ICA was measured ~2 cm from the carotid bifurcation with a 10-MHz linear array probe (Terason 3000, Teratech, Burlington, MA, USA), and the left VA at the C5-C6 or C5-C4 intravertebral space with a variable frequency high-resolution harmonic probe (Vivid Q, GE, Fairfield, CT, USA). Great care was taken to ensure the same settings and anatomic position of measurement within subjects. With the steering angle set to 60 degrees, the sample volume was placed in the center of the vessel adjusted to cover the vascular lumen. The entire breath hold or clamp protocol was collected as an AVI file with analyses of simultaneous luminal diameter and velocity over ≥10 cardiac cycles every 30 seconds subsequently completed offline at 30 Hz using custom-designed software.²⁵ This semiautomated method of analysis results in better reproducibility of baseline measurements, giving test-retest reliability for ICA and VA flows of ~5% and ~11%, respectively.⁷

Hemodynamics

Manual blood pressure measures were collected before the start of each protocol. Three-lead electrocardiogram and beat-to-beat blood pressures were collected at 1 kHz by both finger photoplethysmography (Finometer, Finipress Medical Systems, Amsterdam, The Netherlands) and direct intraradial arterial pressure. However, due to the frequency of arterial blood sampling, radial artery pressure could only be measured for ~5 to 10 seconds every 30 seconds; MAP is therefore reported from finger photoplethysmograhy calibrated to measured radial arterial pressure. Heart rate (HR) was calculated online from the R-R interval measured by electrocardiogram.

Ventilation and End-Tidal Clamping

Subjects breathed through a mouthpiece for collection of end-tidal gases before and at the break point of apnea, and during the entire end-tidal clamping procedure (see below). Gases were sampled from the mouthpiece and analyzed by a calibrated gas analyzer (ML206 AD Instruments, Colorado Springs, CO, USA) and respiratory flows by pneumotachograph (HR 800 L, Hans Rudolph, Shawnee, KS, USA). Custom software (written in Labview, Austin, TX, USA) determined the breath-by-breath tidal volumes and end-tidal partial pressures of oxygen and carbon dioxide (PETO₂ and PETCO₂). A portable end-tidal forcing system prospectively delivered inspired gases to clamp PETO₂ and PETCO₂ at desired levels. Independently controlled solenoid valves delivered the desired volumes of O₂, CO₂, and N₂ as determined by an error reduction algorithm incorporating PETO₂, PETCO₂, and inspiratory and expiratory tidal volume from the last breath (AirForce, GE Foster, University of British Columbia—Okanagan, Kelowna, Canada).

Blood Gases

Arterial blood samples drawn into preheparinized syringes were analyzed either immediately, or were kept on ice if there was a delay greater than 5 minutes. All samples were analyzed within 30 minutes of collection for pH; PO₂; PCO₂; percent saturation of hemoglobin (SO₂); total hemoglobin (tHb); and, plasma glucose, lactate, and electrolyte concentrations (ABL-90, Radiometer, Copenhagen, Denmark).

Calculations

Mean values for MAP, HR, MCAv, and PCAv were averaged over 30 seconds at the time of each arterial blood draw, from which values of pH, PaO₂, PaCO₂, and SaO₂ were obtained. Cerebrovascular conductance was calculated individually for the ICA and the VA as MAP/flow through the respective vessel. Because SaO₂ and PaCO₂ change concomitantly during apnea, cerebrovascular reactivity was calculated within individuals as the slope of the linear regression between flow and the ratio SaO₂/PaCO₂. SaO₂, instead of PaO₂, was used because CBF is linearly related to SaO₂ (whereas the CBF-PaO₂ relationship is exponential) and because SaO₂ provides a clinically meaningful and more easily obtained metric of arterial oxygen content than does PaO₂.

Arterial content of oxygen (CaO₂) was calculated by:

Estimations of Volumetric Cerebral Blood Flows

All but one of the subjects showed IBM that elicited severe contractions of the accessory breathing muscles in the neck and chest. Toward the end of apnea (when IBM became most severe) it was consequently impossible in most individuals to collect accurate ultrasound images and Doppler waveforms of ICA and VA vessel diameter and blood velocity. Likewise at the extremely high rates of ventilation near the end of clamp accurate neck ultrasound was not possible. For each subject, a regression equation was thus calculated between QICA and MCAv, and between QVA and PCAv during each condition (apnea/clamp) and used to estimate the missing flow values. Of the total 768 CBF bins collected across both conditions, 558 were quantified by neck vascular ultrasound, with the remaining 27% imputed from these regression equations. R² values were 0.78±0.17 and were not significantly different between arteries and conditions.

Analysis

That data were normally distributed that was confirmed with a Shapiro-Wilk test. Baseline cardiovascular variables did not differ between male and female subjects and data were therefore pooled. Data are expressed as percentages of total apnea time to compare between subjects, where 100% represent apnea termination, and 110% to 130% represent recovery. To assess the relationship between changes in blood gases and CBF, the slopes of the regressions between the ratio SaO₂/PaCO₂ and total CBF were calculated for each individual duration apnea and clamp. These values were compared by Student's paired t-test. Huynh-Feldt corrected two-way repeated measured ANOVA was used to assess interactions between conditions (apnea versus clamp) and timepoint of apnea/clamp. A priori defined comparisons relative to baseline, and between apnea and clamp, were tested by one-way repeated measures ANOVA (Dunnet's post hoc tests) and paired t-tests, respectively (two-tailed, α=0.05). Analyses were completed with SPSS 16.02 (SPSS Inc., IBM, Chicago, IL, USA) and Prism 5.0 (GraphPad Software Inc., La Jolla, CA, USA). All values in text and tables are mean±s.d.; figures show mean±s.e.m.

Results

Every subject completed the apnea and clamp protocols, however, in two subjects neither ICA nor VA blood flow was successfully measured due to poor quality images. These two subjects were therefore excluded from all further analyses. All data presented are based on n=15 (4 female; age 30±6.8 years; BMI 23.6±2.5 kg/m²). Although females had significantly smaller anthropometric variables and apnea durations, the level of hypoxemia at end-apnea was not different between sexes, and consequently data from the subjects were pooled.

Efficacy of Clamping

The ABG profiles between apnea and clamping were generally very well matched. Due to the nature of end-tidal forcing (and unknown end-tidal to arterial differences), precisely matching the very low PaO₂ values attained during apneas was difficult with the extremely large ventilations and necessarily severely low fraction of inspired oxygen. Nonetheless, PaO₂ was lower by 2.7 mm Hg in clamp only at 80% SaO₂ was lower in clamp at 80% and 100% PaCO₂ was slightly higher (+2.6, +3.2 mm Hg, respectively) during clamp at 80% and 100% and pH was slightly lower in apnea at 40% and 60% (Figure 1; Table 2). These differences were therefore small relative to the magnitude of hypoxic stimulus generated with both apnea and clamping (see Table 2).

Table 2. Cardiovascular variables.

	Condition N=15	Baseline	20%	40%	60%	80%	100%	Recovery	2-way ANOVA
									Condition	Time	Interaction
pH	Apnea	7.48±0.06a	7.47±0.05	7.42±0.04a,b	7.39±0.03a,b	7.37±0.03b	7.35±0.03b	7.45±0.04b	0.507	<0.001	0.001
	Clamp	7.45±0.06	7.48±0.06	7.45±0.05	7.41±0.03b	7.36±0.04b	7.35±0.04b	7.44±0.05
PaCO2 (mm Hg)	Apnea	34.2±7.1a	32.8±6.0	38.2±5.3b	44.6±4.9a,b	49.0±5.7a,b	51.0±6.7a,b	35.1±5.8	0.005	<0.001	0.011
	Clamp	37.1±7.4	34.3±6.7	37.8±6.2	43.0±4.8b	51.8±6.9b	54.3±7.0b	36.6±7.5
PaO2 (mm Hg)	Apnea	104±19.2	109±11.4	81.9±11.0b	55.7±7.6b	40.0±6.6a,b	29.5±6.5b	103±10.2	0.169	<0.001	0.892
	Clamp	103±14.7	111±20.9	77.4±16.9b	54.9±9.0b	37.0±6.6b	26.9±8.9b	102±22.2
SaO2 (%)	Apnea	98.3±2.0	98.9±0.5	96.7±1.7	89.3±4.6b	75.4±7.2a,b	56.7±11.3a,b	98.6±0.8	0.015	<0.001	0.037
	Clamp	98.4±1.1	98.8±1.1	95.9±3.2	89.5±4.8b	70.7±8.5b	48.3±16.8b	97.7±3.3
HR (b/min)	Apnea	77±14	83±22	78 ±19	69±16	61±16a,b	67±13a	76±18	0.045	0.098	<0.001
	Clamp	74±12	76±14	73±14	75±14	84±11b	90±11b	77±16
MAP (mm Hg)	Apnea	89.3±9.1	96.8±10a	104±13a,b	115±14a,b	130±14a,b	135±19a,b	96.8±13a	<0.001	<0.001	<0.001
	Clamp	83.7±13	82.9±13	85.4±3.0	92.3±6.3b	108±8.0b	111±8.5b	81.8±0.7

HR, heart rate; MAP, mean arterial pressure; PaCO₂ and PaCO₂, partial pressures arterial carbon dioxide and oxygen.

^aP<0.05 versus Clamp.

^bP<0.05 versus baseline.

Cardiovascular and Respiratory Effects of Apnea and Clamping

Table 2 displays the cardiovascular variables during apnea and clamp. All subjects hyperventilated to some extent before apnea; PaCO₂ before apnea was 34.2±7.1 versus 37.1±7.4 mm Hg before clamp. All other baseline variables shown in Table 2 were statistically similar between apnea and clamp. Maximum apnea times ranged from 212 to 435 seconds (mean 316±60 seconds) and yielded end-apnea PaO₂ ranging from 23 to 36.9 mm Hg (mean 29.5±6.5 mm Hg). Single-breath PETO₂ and PETCO₂ were 37±14  and 45±7 mm Hg, respectively (Table 1), overestimating PaO₂ and underestimating PaCO₂—the mean ET-aPO₂ and ET-aPCO₂ differences were 7.5±12.9 mm Hg and −5.1±6.1 mm Hg, respectively. All but one subject had IBMs that began at 50±11% of the total apnea time.

With onset of apnea all subjects displayed a transient decrease in MAP (−11.9±12%), which remained below baseline at 10% apnea duration in all but three subjects. In all subjects, MAP increased throughout the apnea to a maximum of 54±24% above baseline values, a significantly greater increase than during clamp where MAP increased 34±26% in fact, the increase in MAP was greater in apnea at all times from 20% through 120% (i.e., recovery) of apnea/clamp; Figure 2). At 80% and 100% apnea, HR decreased to below baseline values at which points it was significantly less than HR during clamp (Table 2).

Figure 2.

Percent changes in mean arterial pressure (MAP; top), cerebral blood flow (CBF; middle) and cerebral oxygen delivery (DO₂; bottom). The top plot shows changes in MAP during apnea (solid line) and clamp (dashed line). Despite MAP being greater during apnea than during clamp at every time point, CBF is only greater 30% to 50% into the apnea, suggesting a cerebral autoregulatory attenuation of maximum CBF. The middle plot depicts changes in total CBF during apnea (solid line) and clamp (dashed lines) against the percent duration of apnea. Global cerebral delivery of oxygen never falls below baseline during apnea indicating critical reduction of DO₂ is not responsible for apnea breakpoint.

Cerebrovascular Effects of Apnea and Clamping

Global CBF (gCBF) began to increase immediately after the decrease in MAP with apnea onset (Figure 3). During clamp gCBF did not increase significantly until 60% of clamp duration. Conversely, during apnea, gCBF was significantly increased from baseline by 40% apnea duration (Figure 2). Until 60% duration, apnea caused greater increases in gCBF than did clamp, at which point the CBF change became similar between interventions, despite the consistently greater MAP in apnea than in clamp. Flow in the ICA and VA, as well as MCAv and PCAv showed similar trends for both apnea and clamp (Table 3). Cerebral oxygen delivery was never attenuated during either apnea or clamp—i.e., the progressively increasing CBF compensated for the decrease in CaO₂ (Figure 2). Indeed, cerebral O₂ delivery significantly increased from baseline at 60% and 80% duration of apnea, and at 80% duration of clamp (Figure 2). The slopes of SaO₂/PaCO₂ versus total ΔCBF (%) were not different between apnea and clamp (apnea 79.9±21 mm Hg⁻¹; clamp 73.3±24 mm Hg⁻¹). SaO₂, PaCO₂, and ΔMAP during both apnea (R²=0.54, 0.62, 0.57) and clamp (R²=0.57, 0.37, 0.61) were positively related to the increase in CBF (P<0.05; Figure 4).

Figure 3.

Cerebral blood flows over time during maximal apnea (left) and end-tidal forcing (right). Apnea times ranged from 212 to 435 seconds. During clamp the identical duration and changes in arterial blood gases were replicated while the subjects breathed.

Table 3. Cerebrovascular variables.

	Condition (sample size)	Baseline	20%	40%	60%	80%	100%	Recovery	2-way ANOVA
									Condition	Time	Interaction
aQICA (mL/min)	Apnea (12)	189±60a	181±60a	265±153b	329±172b	380±167b	396±176b	245±128	0.224	<0.001	0.256
	Clamp (11)	244±91.6	218±75.1	246±88.0	328±120b	444±125b	446±153b	273±126
DO2 (mL O2/min)	Apnea (12)	38.0±11.0a	35.3±12.4	50.1±27.0	59.6±28.9b	57.4±22.8b	43.6±17.8	50.3±25.8	0.485	<0.001	0.196
	Clamp (11)	48.0±16.4	43.0±13.63	47.6±13.7	58.1±19.5	61.8±15.2b	44.3±16.0	53.7±25.8
CVC (mL/min/mm Hg)	Apnea (12)	2.1±0.6a	1.8±0.5a	2.5±1.1	2.8±1.2a,b	2.9±1.1a,b	2.9±1.3a,b	2.6±1.5a	<0.001	<0.001	<0.163
	Clamp (11)	2.9±0.8	2.7±0.8	2.9±0.9	3.6±0.9b	4.1 ±0.8b	4.0 ±1.1b	3.4±1.5
aQVA (ml/min)	Apnea (13)	81.1±32a	80.1±29b	98.3±32b	124±44b	145±43b	149±40b	103±28	0.421	<0.001	0.23
	Clamp (15)	92.2±25	84.7±21	95.6±27	122±47b	149±53b	154±52b	116±45
DO2 (mL O2/min)	Apnea (13)	16.3±6.4	15.6±5.9	19.0±6.0	22.3±6.9b	22.2±6.2b	16.9±5.5	21.2±7.1b	0.797	<0.001	0.415
	Clamp (15)	18.2±5.3	16.7±4.6	18.2±5.3	21.7±9.3	21.2±8.6	14.9±6.6	22.4±9.5
CVC (mL/min/mm Hg)	Apnea (13)	0.9±0.3a	0.8±0.2a	0.9±0.2a	1.1±0.3a,b	1.1±0.3a,b	1.1±0.3a,b	1.1±0.4a,b	<0.001	<0.001	0.323
	Clamp (15)	1.1±0.3	1.0±0.3	1.1±0.4	1.3±0.5	1.4±0.5b	1.4±0.4b	1.4±0.4b
MCAv (cm/s)	Apnea (15)	50.9±8.6a	56.1±12.3b	70.4±15a,b	85.4±13.2a,b	99.2±13b	101±12b	60.2±14b	0.042	<0.001	<0.001
	Clamp (15)	58.0±12	52.5±9.8	60.0±12	78.4±15.4b	98.6±12b	98.7 ±13b	59.8±13
PCAv (cm/s)	Apnea (15)	33.3±9.3a	35.0±9.6b	44.4±14.2a,b	54.6±15.6b	64.2±12.7b	66.9±10.8b	41.7±12.6b	0.704	<0.001	0.013
	Clamp (15)	38.3±9.7	34.8 ±6.5	38.9±8.0	50.4±13b	65.3±13b	68.3±12b	41.2±12

CVC, cerebrovascular conductance (flow or velocity/MAP); DO₂, oxygen delivery through vessel; MAP, mean arterial pressure; MCAv, middle cerebral artery blood velocity; PCAv, posterior cerebral artery blood velocity; QICA, internal carotid artery blood flow; QVA, vertebral artery blood flow. Values are quantified flow through the respective vessel except in most subjects at 80% and 100% where flows were estimated from the regression equation derived between QICA–MCAv and QVA–PCAv; see Materials and Methods. Sample sizes vary between measured vessels due to inadequate image quality for VA and ICA measures.

^aP<0.05 versus Clamp.

^bP<0.05 versus baseline.

Figure 4.

Relationship between ΔCBF and ΔMAP during apnea (left) and clamp (right). Cerebral blood flow (CBF) percentage change from baseline was positively related to the percentage change from baseline of mean arterial pressure (MAP) (P<0.05) both for apnea (left) and for clamp (right).

Discussion

This is the first study to combine measures of regional CBF and arterial blood sampling during maximal dry apneas in elite breath-hold divers, and also to precisely simulate a maximal apnea using dynamic end-tidal forcing. Our principal findings were that (1) apnea breakpoint is not induced by a critical attenuation of cerebral oxygen delivery. Cerebral oxygen delivery was never attenuated during apnea being instead maintained by CBF increases of up to 100% of baseline values; (2) changes in CBF were relatively homogenous between regions perfused by the ICA and VA; and (3) despite significantly greater increases in MAP during apnea relative to clamped (breathing) conditions, the increase in CBF was similar indicating that mechanisms responsible for buffering surges in CBF remain effective during maximal dry apnea.

Apnea Break Point

As early as 1908, Hill and Flack²⁶ observed that—in healthy volunteers—the break point of volitional apnea was not solely a function of chemoreflex stress as identical PETO₂ and PETCO₂ values could be sustained for longer and with greater ease during rebreathing. Indeed, our volunteers consistently noted they believed they could continue breathing at the end of clamp—despite their blood gas levels being at the same level (or, in a number of trials, even more hypoxemic) as their apnea break point. We could also not identify a threshold PaO₂, PaCO₂, or ratio of SaO₂/PaCO₂ common among individuals at end apnea, and although most of the subjects reached PaO₂ values in the mid-twenties, cerebral oxygen delivery was never attenuated. The factors that allow some individuals to maintain apnea longer than others are their greater oxygen stores (i.e., lung volume, blood, and tissue), lower oxygen consumption and psychologic factors.¹⁵ Therefore, while a threshold PaO₂, CaO₂, or oxygen delivery per se may not stimulate apnea break point, oxygen metabolism clearly has a crucial role in apnea duration in elite breath-hold divers.

Regulation of Cerebral Blood Flow During Apnea

A number of studies have assessed the relative sensitivity of ICA and VA to changes in ABGs, but their heterogeneous methodologies make definitive elucidation of regional cerebrovascular responses to hypoxia and PCO₂ flux difficult. At high altitude three studies have found no difference between ICA and VA flow increase with ascent to altitudes ranging from 4,300 to 5,260 m.^{10, 11, 27} Conversely, sea level sensitivity to isocapnic or poikilocapnic hypoxia is greater in the VA.^{7, 12} Euoxic hypocapnia also elicits greater decreases in VA flow through a broad range of PaCO₂ (~15 to 40 mm Hg);⁷ however, this is not a universal finding through smaller changes in PCO₂.²⁸ It thus seems that during simultaneous changes in PaO₂ and PaCO₂, ICA and VA reactivities are similar, as was the case in the present study where ICA and VA showed similar profiles during clamp and apnea. That both vessels also had similar profiles during apnea shows their additive response to altered perfusion pressure. The IBM of prolonged apnea is associated with oscillations in intrathoracic pressure that augment left ventricular stroke volume,^{13, 19, 29} suggesting that CBF (as indexed by MCAv) is augmented by IBM during the latter half of a prolonged apnea. The present study does not support this conclusion. We were able to dissociate the effects of blood gases and IBM during apnea: CBF increased more with apnea only during the first 50% of apnea/clamp duration (preceding onset of IBM) due to the greater apnea-induced increase in MAP. Initiation of apnea decreases right heart filling^{19, 29} producing a transient MAP decrease that in turn facilitates baroreflex mediated increases in sympathetic activity.¹⁷ Inspiration inhibits sympathetic outflow, but in the absence of ventilatory restraint, it increases unchecked until the end of apnea^{17, 22} thus driving peripheral vasoconstriction and sustained increases in MAP. The augmentation of CBF by MAP could be interpreted to serve maintenance of cerebral oxygenation, but the present data show this has no effect on cerebral oxygen delivery at apnea termination when presumably CBF is most critical. Thus, it seems IBM are not necessary for the overall prolonged apnea—indeed, the one subject who never experienced an IBM maintained apnea for 366 seconds reaching PaO₂=27.7 mm Hg. Involuntary breathing movements may therefore instead be a consequence of chemoreceptor stimulation and consequent efferent respiratory motor output.³⁰ Given that the delivery of oxygen never fell below baseline values in either vessel, the present data indicate that even during radical hypoxemia commensurate increases in CBF sate the entire brain's requirement for oxygen and, by extension, that global oxygen delivery cannot be the mechanism stimulating apnea break point.

Cerebral Autoregulation During Apnea

Cerebral autoregulation is the term given to the buffering of changes in CBF from changes in perfusion pressure principally facilitated by myogenic³¹ and autonomic mechanisms (reviewed in Willie et al⁴). Quantifying the relationship between blood pressure and CBF during either dynamic changes in blood pressure or during resting steady state conventionally assesses cerebral autoregulation. Of these numerous metrics used to quantify cerebral autoregulation few can be used interchangeably.³² In the present study, we indexed cerebral autoregulation by comparing the CBF response during apnea versus that during clamp, when MAP was much higher during the former. Despite a greater surge in MAP, and the very similar changes in PaO₂ and PaCO₂, the CBF increase during apnea was similar to that during clamp. A mechanism buffering further increases in CBF was thus necessarily present.

Sympathetic activity restrains augmented perfusion during surges in MAP,^{20, 21, 33, 34, 35} and is increased by ~20% more during apnea compared with breathing at similar arterial blood gas tensions as indexed by muscle sympathetic nerve activity.²² It therefore seems likely that the observed similarly in CBF increase between apnea and clamp, despite greater MAP during the former, is a function of increased sympathetic nervous activity during apnea. Previous CBF estimations by transcranial Doppler ultrasound may be interpreted to conflict with the present findings. For example, Kjeld et al³⁶ showed enhanced CBF during maximum breath-hold with concomitant facial immersion in cold water based on transcranial Doppler ultrasound estimation of CBF. This apparent CBF increase during facial immersion manifest in spite of lower cardiac output, stroke volume, HR, and MAP during facial immersion. As shown in the present study, and numerous others (e.g., ref. 7, 37, 38), CBF increases approximately 0.4% to 0.6% per mm Hg increase in MAP. The finding by Kjeld et al³⁶ that CBF increases more when these variables are lower is therefore somewhat surprising. It is possible that the MCA constricts during increased sympathetic activity, yielding an artifactually augmented estimate of CBF. Indeed, it has been recently shown that the diameter of the MCA is not static.^{5, 7, 11, 39} Cross et al⁴⁰ interpreted increased phase synchronization during apnea to indicate impaired cerebral autoregulation, but various metrics that ostensibly quantify the efficacy of cerebral autoregulation have been shown to yield disparate results.³² However, that CBF did not increase further than during clamp despite a much greater increase in MAP is evidence of cerebral autoregulatory efficacy without the confounding influence of more quantitatively complex metrics of autoregulation and the potentially invalid assumptions of transcranial Doppler ultrasound. Nevertheless, a worthwhile future study is to determine whether volumetric CBF is further augmented during maximum apneas with facial immersion, as the similar CBF increase during apnea and clamp in the present study suggest a threshold for CBF increase.

Arterial Blood Gases

A number of studies have previously estimated changes in ABGs based on sampling of end-tidal gases on the first breath after apnea. We observed a 7.5±12.9 mm Hg overestimation of PaO₂ by end-tidal sampling. Our PETO₂ values (37±14 mm Hg) were much higher than those reported in similar studies with similar breath-hold durations (e.g., 26.9±7.5 mm Hg, 284±25 seconds;¹⁵ 26±4.5 mm Hg, 309±38 seconds¹⁶) suggesting that either the subjects in these studies experienced a greater degree of hypoxemia than our subjects, or a methodological problem with our single breath end-tidal tension method. With respect to the latter, effort was made to ensure each subject expired fully but some individuals experienced an impaired level of consciousness at end-apnea making it difficult to ensure this was the case. Given that neither of these previous studies reported oxygen saturation it is difficult to further assess their degree of hypoxemia relative to that reported here; however, certainly it can be observed in the present data that some individuals desaturated more quickly than others, so it is possible differences between cohorts account for these disparate PETO₂ findings. Regardless, PaO₂ is a far more robust measure of hypoxemia than PETO₂ as it reflects any unquantifiable degree of shunt or ventilation-perfusion mismatch that may be manifest during apnea.

Conclusions

The present data indicate that the profound hypoxemia of extreme breath holding does not produce marked global brain hypoxia. On the contrary, cerebral oxygen delivery is never attenuated through the duration of breath hold, suggesting an alternative mechanism triggering apnea breakpoint. Despite an approximate 50% increase in MAP at end-apnea, CBF is buffered from increasing more than ~100% of baseline values indicating cerebral autoregulation is not impaired during prolonged apnea in elite breath-hold divers.

The authors declare no conflict of interest.