Influence of sympathoexcitation at high altitude on cerebrovascular function and ventilatory control in humans

P. N. Ainslie, S. J. E. Lucas, J.-L. Fan, K. N. Thomas, J. D. Cotter, Y. C. Tzeng, Keith R. Burgess


We sought to determine the influence of sympathoexcitation on dynamic cerebral autoregulation (CA), cerebrovascular reactivity, and ventilatory control in humans at high altitude (HA). At sea level (SL) and following 3–10 days at HA (5,050 m), we measured arterial blood gases, ventilation, arterial pressure, and middle cerebral blood velocity (MCAv) before and after combined α- and β-adrenergic blockade. Dynamic CA was quantified using transfer function analysis. Cerebrovascular reactivity was assessed using hypocapnia and hyperoxic hypercapnia. Ventilatory control was assessed from the hypercapnia and during isocapnic hypoxia. Arterial Pco2 and ventilation and its control were unaltered following blockade at both SL and HA. At HA, mean arterial pressure (MAP) was elevated (P < 0.01 vs. SL), but MCAv remained unchanged. Blockade reduced MAP more at HA than at SL (26 vs. 15%, P = 0.048). At HA, gain and coherence in the very-low-frequency (VLF) range (0.02–0.07 Hz) increased, and phase lead was reduced (all P < 0.05 vs. SL). Following blockade at SL, coherence was unchanged, whereas VLF phase lead was reduced (−40 ± 23%; P < 0.01). In contrast, blockade at HA reduced low-frequency coherence (−26 ± 20%; P = 0.01 vs. baseline) and elevated VLF phase lead (by 177 ± 238%; P < 0.01 vs. baseline), fully restoring these parameters back to SL values. Irrespective of this elevation in VLF gain at HA (P < 0.01), blockade increased it comparably at SL and HA (∼43–68%; P < 0.01). Despite elevations in MCAv reactivity to hypercapnia at HA, blockade reduced (P < 0.05) it comparably at SL and HA, effects we attributed to the hypotension and/or abolition of the hypercapnic-induced increase in MAP. With the exception of dynamic CA, we provide evidence of a redundant role of sympathetic nerve activity as a direct mechanism underlying changes in cerebrovascular reactivity and ventilatory control following partial acclimatization to HA. These findings have implications for our understanding of CBF function in the context of pathologies associated with sympathoexcitation and hypoxemia.

  • autoregulation
  • cerebral blood flow
  • reactivity
  • SNA
  • ventilation

the cerebral vasculature adapts rapidly to changes in perfusion pressure [cerebral autoregulation (CA)], autonomic neural activity, and humoral factors (cerebrovascular reactivity). Regulation of cerebral blood flow (CBF) is, therefore, highly controlled and involves a wide spectrum of regulatory mechanisms that, together, work to maintain optimum oxygen and nutrient delivery. Exposure to high altitude (HA) stimulates systemic adrenergic activity, as indicated by increases in muscle sympathetic nerve activity (SNA) (19) and plasma norepinephrine (38). The cerebral vasculature has rich innervation by sympathetic nerves (22), but the influence of adrenergic activity and HA on the cerebral circulation, particularly cerebral hemodynamics, is surprisingly unclear (4). For example, chronic exposure to HA has been reported to elevate cerebrovascular reactivity to both hypocapnia and hypercapnia (12, 42). In contrast, acute sympathetic activation (e.g., handgrip, lower body negative pressure) at sea level (SL) generally leads to maintained cerebrovascular CO2 reactivity (1, 29). Moreover, at SL, sympathetic deactivation (via ganglionic blockade) reduces cerebral reactivity to hypercapnia (37); however, since blood pressure may influence CBF (34), this lowering of reactivity is likely confounded by an attenuated hypercapnia-induced pressure response (26, 37). Thus the effects of chronic HA-induced sympathetic activation appear to manifest differently than acute sympathoexcitation with respect to cerebrovascular reactivity.

The brain can modulate its own blood supply to some degree in the face of changes in arterial pressure (34), a process termed CA, that involves fast (dynamic) and slow (static) regulatory components. These two temporally differentiated components appear to be mediated by different mechanisms, with dynamic CA possessing more effective neural control of CBF (52). For example, removal of autonomic neural activity with ganglion blockade has been shown to increase transfer function gain between beat-to-beat changes in arterial pressure and CBF velocity and reduce the phase lead of CBF velocity to arterial pressure. These changes are indicative of impaired dynamic CA (i.e., a more pressure-passive cerebral circulation), implying that autonomic neural control is tonically active and plays an important role in the human CBF regulation (52).

Studies conducted at HA indicate that dynamic CA is impaired on acute exposure to hypoxia or HA (5, 8, 31, 42) and does not improve with partial acclimatization (25). This deterioration in CA is generally manifested by increased coherence and gain and/or reduced phase of the transfer function between blood pressure and middle cerebral blood velocity (MCAv) within the optimal autoregulatory frequency range. A potential explanation of these findings is that, because of the intense sympathetic hyperactivity at HA, the autonomic nervous system may not function appropriately to effectively regulate CBF via CA mechanisms (25), despite favorable changes in hypoxia and hypocapnia that act to improve CA (4). However, the extent to which any changes in CA are under neural control at HA has not been explored.

The main aims of this study were to examine the influence of sympathoexcitation on CBF, cerebrovascular reactivity, and CA at HA. Thus, at SL and HA, we quantified cerebrovascular reactivity to CO2 and the static and dynamic relation between beat-to-beat arterial blood pressure and CBF variations, before and following combined α- and β-adrenergic receptor blockade. Because of evidence for effective neural control of CBF (52), we reasoned that attenuation of sympathoexcitation at HA would normalize the previously reported changes in CA (5, 8, 25, 31, 42). Conversely, at least at SL, acute sympathetic activation does not elevate cerebral reactivity to CO2 (1, 29), and sympathetic deactivation diminishes cerebral reactivity to hypercapnia (37) via its effect on reducing the hypercapnic-induced increase in blood pressure. Therefore, we hypothesized that cerebrovascular reactivity would be reduced by comparable extent following α- and β-adrenergic receptor blockade at both SL and HA. Because of the influence of sympathoexcitation in mediating changes in ventilatory control at both SL (23) and HA (7, 11), our secondary goal was to examine how sympathoexcitation may also influence the control of breathing at HA. We hypothesized that acute SNA blockade would reduce some of the evident changes in peripheral and central chemosensitivity on ascent to HA.



Ten adults (three women) with a mean age of 28 ± 8 yr (mean ± SD), and body mass index of 22 ± 2 kg/m2, participated in this study after giving written, informed consent. Participants were nonsmokers, had no previous history of cardiovascular, cerebrovascular, or respiratory diseases, and were not taking any cardiovascular medications. The study was approved by the Lower South Regional Ethics Committee of Otago and conformed to the standards set by the Declaration of Helsinki.

Experimental Design

After a full familiarization with the experimental procedures outlined below (visit 1), the participants underwent experimental trials at SL and at HA. The ascent profile from SL to the Pyramid laboratory (5,050 m; barometric pressure 413 ± 1 mmHg) for this expedition has been described previously (33). Although some of the experiments performed during this expedition have already been published (12, 13, 15, 33), this blockade experiment was completed at both SL and HA by 9 of the 23 volunteers involved with this research expedition. Therefore, apart from the experimental design, there is no overlap or duplication of data with the present study described herein.

To avoid the confounding influence of altitude illness, as confirmed via the Lake Louise Questionnaire and Environmental Symptoms Questionnaire Cerebral Symptoms, with scores <5 and <0.7, respectively, experimental sessions were carried out between days 3 and 10 after arrival to 5,050 m. Before each experimental session, participants were required to abstain from exercise and alcohol for 24 h, caffeine for 12 h, and a heavy meal for 4 h prior. With the exception of the arterial blood-gas sampling, all experiments were performed with participants in the semirecumbent position. Each experimental testing session comprised: 1) an arterial blood-gas sample; 2) instrumentation; 3) 5- to 10-min resting baseline for assessment of dynamic CA; 4) modified hyperoxic rebreathing and isocapnic hypoxia (see details of methods below); 4) combined α-adrenergic and β-blockade; 5) 60-min rest; and 6) repeat testing of 1–4 (Fig. 1). The order of the modified rebreathing and isocapnic hypoxic rebreathing was randomized, and >5-min recovery was permitted between each trial to restore end-tidal gases to baseline resting values.

Fig. 1.

Schematic of the experimental design. Following equipment setup [ECG, blood pressure (BP), transcranial Doppler], an arterial blood gas (ABG) was obtained. Subjects then rested quietly for ∼20 min [during which time spontaneous BP and middle cerebral blood velocity data were collected for transfer function analysis (TFA)]. In a randomized order, subjects underwent tests to assess cerebrovascular and ventilatory control. Following completion of the tests, subjects received oral propranolol and prozasin. The tests were then repeated ∼60 min later. If required, intravenous saline was administered to maintain mean arterial pressure >70 mmHg. See methods for details.

Arterial blood gases.

Arterial blood-gas samples from the radial artery were obtained at rest using a 25-gauge needle into a preheparinized syringe. Following standardized calibration, all blood samples were analyzed using an arterial blood-gas analyzing system (NPT 7 series, Radiometer, Copenhagen, Denmark) for pH, partial pressure of arterial O2 (PaO2) and CO2 (PaCO2), bicarbonate concentration ([HCO3]), and arterial O2 saturation.


Ventilation (V̇e) was measured using a heated pneumotachograph (Hans-Rudolph 3813) and expressed in units adjusted to btps. The fractional changes in inspired and expired O2 and CO2 were used to calculate end-tidal Po2 (PetO2) and Pco2 (PetCO2) using fast responding gas analyzers (model CD-3A, Pittsburgh, PA; ML206, ADInstruments, Colorado Springs, CO). The pneumotachograph was calibrated using a 3-liter syringe (Hans-Rudolph 2700, Kansas City, MO) and the gas analyzers were calibrated using known concentrations of O2 and CO2 before each testing session. MCAv (an index of CBF) was measured in the right middle cerebral artery using previously described search and fixation methods (45). In our hands, the day-to-day reproducibility of MCAv has a coefficient of variation of <4.5% (2). Beat-to-beat mean arterial blood pressure (MAP) was monitored using finger photoplethysmography (Finometer, Finapress Medical System). Manual blood pressure measurements by auscultation were also made periodically to check and validate the automated recordings. Cerebrovascular conductance index was subsequently estimated by dividing mean MCAv (MCAvmean) by MAP.

Assessment of dynamic CA using transfer function analysis.

Beat-to-beat MAP and MCAvmean signals recorded during baseline conditions were then cubic spline interpolated and resampled at 4 Hz for spectral and transfer function analysis (TFA), based on the Welch algorithm. Each 5-min recording was first subdivided into five successive windows that overlapped by 50%. The data within each window were linearly detrended, passed through a Hanning window, and subjected to fast Fourier transform analysis. For TFA, the cross-spectrum between MAP and MCAvmean was determined and divided by the MAP autospectrum to derive the transfer function gain, phase, and coherence indexes. Spontaneous MAP and MCAvmean spectral powers, and the mean value of transfer function coherence, gain, and phase were calculated in the very-low- (VLF, 0.02–0.07 Hz) and low-frequency (LF, 0.07–0.20 Hz) ranges, where CA is thought to be operant. To ensure that robust phase and gain estimates within the VLF and LF bands were entered for subsequent analysis, we averaged only those gain and phase values where the corresponding coherence was ≥0.5. In the context of applications of this technique, the absence of dynamic CA would manifest as increases in coherence and gain (17, 51) and reductions in phase (36). Conversely, strong CA would theoretically be associated with reductions in coherence and gain and increases in phase.

Assessment of cerebrovascular reactivity and ventilatory control.


We used the well-established modified rebreathing method for assessing both ventilatory and cerebrovascular CO2 reactivities (3). The details of the modified hyperoxic rebreathing method have been described previously (12, 15). In brief, following 2 min of baseline room-air breathing, participants were instructed to hyperventilate for 5 min to lower and then maintain PetCO2 at 22 ± 2 Torr (SL) and 17 ± 3 Torr (5,050 m). Participants were then switched to the rebreathing bag following an expiration and instructed to take three deep breaths to ensure rapid equalization of Pco2, then breathe normally. The modified rebreathing tests were terminated when either 1) PetCO2 reached 60 Torr; 2) PetO2 could no longer be maintained above 160 Torr; 3) V̇e exceeded 100 l/min; or 4) the participant reached the end of their tolerance. The modified rebreathing data were accumulated on a breath-by-breath basis and analyzed using a specially designed program (Full Fit Rebreathing Programme, version 3.1, Toronto, Canada); this analysis has been detailed in Refs. 12 and 15.


The soda-lime rebreathing technique was used to assess the ventilatory O2 sensitivity as an index of peripheral chemoreflex sensitivity (14). The isocapnic hypoxia was terminated when either 1) peripheral O2 saturation reached 80% at SL and 70% at 5,050 m; 2) PetO2 decreased to 45 Torr at SL and 30 Torr at 5,050 m; 3) the V̇e exceeded 100 l/min; or 4) the participants reached the end of their tolerance. The breath-by-breath V̇e data were plotted against the slope of peripheral O2 saturation.


Voluntary hyperventilation was used to assess cerebrovascular reactivity to hypocapnia (48). Steady-state hypocapnic cerebrovascular reactivity was estimated from the slope of the reduction in MCAvmean from the final 2 min of baseline to the voluntary hyperventilation that preceded rebreathing, relative to the reduction in PetCO2 (see below). We did not control for inspired O2 fraction during the voluntary hyperventilation, because the small increases in PetO2 (by 27 ± 6 and 12 ± 6 Torr at SL and 5,050 m, respectively) were considered unlikely to influence either MCAv or its reactivity.

Autonomic blockade.

Sympathetic autonomic blockade was achieved via combined α-adrenergic (prazosin; 1 mg/20 kg body mass) and β-adrenergic blockade (propranolol; 60–80 mg). In some incidences (n = 1 at SL; n = 8 at HA), to prevent excessive hypotension during the combined blockade, we infused between 500 and 1,500 ml of warmed 0.9 saline via rapid (5- to 15-min) infusion before repeat testing. We allowed at least 20 min for blood pressure to stabilize before testing began. If needed, the infusion was continued at a low dose during and following testing.

Statistical Analysis

The effects of HA and SNA blockade on resting variables, cardiorespiratory and cerebrovascular responsiveness to CO2, as well as dynamic CA were assessed using two-way (altitude and blockade) repeated-measures ANOVA with an α-level of 0.05 (SPSS version 17.0, SPSS, Chicago, IL). Pairwise comparisons (Bonferroni corrected) were performed to isolate the effect of HA and blockade on the dependent measures within participants. Data are reported as means ± SD.


All 10 participants completed the full experimental protocol at SL. However, due to time constraints, only nine participants completed the protocol at HA.

Effect of HA and SNA Block on Resting Variables

See Table 1. With the exception of PaO2, blood gases and V̇e were unaltered following sympathetic blockade at both SL and HA (P > 0.18). At HA, MAP was elevated (P < 0.01 vs. SL control), but MCAv remained unchanged (P = 0.29). The blockade tended (P = 0.07) to reduce MCAv [HA: 9 ± 15% (mean ± SD) vs. SL: 4 ± 20%; interaction effect: P = 0.57] and reduced (P < 0.01) MAP, more so at HA (HA: 26 ± 13% vs. SL: 15 ± 12%; interaction effect: P = 0.048). Likewise, heart rate was elevated at HA (10 ± 9%; P < 0.01 vs. SL control) and was reduced by blockade more at HA than at SL (interaction: P < 0.01).

View this table:
Table 1.

Resting cerebrovascular, respiratory, and arterial blood-gas variables before and after autonomic blockade ingestion at sea level and following ascent to 5,050 m

Dynamic CA

See Table 2 and Figs. 2 and 3. Following ascent to HA, VLF MCAv power, coherence, and gain were all elevated, while phase was reduced (all P < 0.05 vs. SL), indicative of CA impairment. While MAP variability was unaltered at altitude (P = 0.16), the blockade reduced total, VLF, and LF MAP variability by a comparable extent (down ∼50–60%; all P < 0.01) at SL and HA (interactions: P > 0.21). Sympathetic blockade reduced MCAv variability in the LF and high-frequency ranges (down ∼30 and 40%; P = 0.01 and <0.01, respectively) to a similar extent at SL and HA (interactions: P = 0.56 and 0.10, respectively). In contrast, while coherence was unchanged (in all frequency ranges) following blockade at SL (P > 0.10), the VLF phase was reduced (40 ± 23%; P < 0.01). Opposite changes occurred following blockade at HA. Specifically, blockade at HA reduced LF coherence (26 ± 20%; P = 0.01 vs. baseline; interaction: P = 0.03) and increased VLF phase lead (by 177 ± 238%; P < 0.01 vs. baseline; interaction: P < 0.01), fully restoring these parameters back to SL values. HA increased the VLF transfer function gain (P < 0.01), and so did blockade, irrespective of altitude (∼43–68%; blockade effect: P < 0.01; interaction: P = 0.61).

View this table:
Table 2.

Summary of baseline spontaneous spectral and transfer function analysis variables before and after autonomic blockade at sea level and high altitude

Fig. 2.

Changes in very-low-frequency (VLF; 0.02–0.07 Hz) phase (A), gain (B), and coherence (C) sea level and high altitude (5,050 m) before (solid bars) and following (shaded bars) sympathetic blockade. aU, arbitrary units. Values are means ± SD. Different from *sea level, †baseline, and ††sea level blockade effect (P < 0.05).

Fig. 3.

Changes in low-frequency (LF; 0.07–0.2 Hz) phase (A), gain (B), and coherence (C) sea level and high altitude (5,050 m) before (solid bars) and following (shaded bars) sympathetic blockade. Values are means ± SD. ††Different from sea level blockade effect (P < 0.05).

Effects of HA and SNA Block on Cerebrovascular CO2 Reactivity

See Table 3. At 5,050 m, the MCAv-CO2 reactivity was elevated by 39 ± 55% in hypercapnia (P = 0.03 vs. SL) and by 109 ± 54% in hypocapnia (P = 0.02). While the blockade had no effect on hypocapnic reactivity at either altitude (P = 0.74; interaction: P = 0.30), hypercapnic reactivity was reduced following blockade to a similar extent at SL and at HA (P = 0.03; interaction: P = 0.70).

View this table:
Table 3.

Cerebrovascular and cardiovascular sensitivity to CO2 and ventilatory control before and after autonomic blockade at sea level and following ascent to 5,050 m

Effects of HA and SNA Block on Ventilatory and Cardiovascular Sensitivities

See Table 3. HA augmented V̇e sensitivity to both CO2 (up 131 ± 132%; P = 0.03) and O2 (up 160 ± 113%; P = 0.02), whereas no effects of sympathetic blockade were evident, irrespective of altitude (P = 0.59 and 0.26, respectively). At HA, both the MAP-CO2 and heart rate-CO2 reactivity were elevated (P < 0.05). Of note, the elevation in MAP-CO2 reactivity was correlated with the increase in MCAv-CO2 reactivity (R2 = 0.41; P < 0.01). Although blockade reduced the MAP-CO2 reactivity, by a similar extent at both SL and HA (P < 0.01; interaction: P = 0.71), these reductions were unrelated to the reduction in MCAv-CO2 reactivity at both SL and HA (R2 = 0.04 and 0.03, respectively).


Our main novel findings are as follows: 1) a lack of influence of SNA on resting CBF at HA (during semirecumbency); 2) a role of SNA at HA in regulating dynamic CA; 3) despite elevations in CBF reactivity to hypercapnia at HA, blockade reduced it comparably at SL and HA; both effects we attributed to the hypotension and/or abolition of the hypercapnic-induced increase in MAP; and 4) a redundant role of autonomic activity as a mechanism underlying changes in ventilatory control following partial acclimatization to HA. Before brief discussion in each of these related topics, pertinent aspects related to our experimental methodology are considered.

Methodological Considerations

Transcranial Doppler ultrasound.

Measurements of MCAv were used as indexes of global CBF, CA, and reactivity. Because transcranial Doppler ultrasound methodology assumes that vessel diameter does not change with HA and that the angle of insonation remains constant, caution should be used when interpreting absolute CBF values (reviewed in Refs. 3, 45). Nevertheless, our laboratory has recently reported that absolute changes in MCAv on initial arrival at HA mirrors the reported increases in CBF, as determined by the direct Fick method (reviewed in Ref. 33). These coincidental findings are supported by a recent report of a preserved MCA diameter at 5,300 m compared with SL (46). Irrespective of changes in MCA diameter, since determinations of dynamic CA and cerebrovascular reactivity are based on stimulus-response principles, absolute CBF values are not as important as reliable and repeatable recordings with short (beat-to-beat) time resolution. For these reasons, transcranial Doppler ultrasound is a well-suited technique for the questions addressed here, especially in regards to field-based research.

Autonomic blockade.

We attempted to characterize sympathetic control of CBF, dynamic CA, and reactivity using SNA blockade. As in previous studies (52), because hypotension was produced, our stimulus may be inappropriate, given that cerebral SNA may counterregulate against rises, rather than falls, in cerebral perfusion pressure (10, 44). However, at least with TFA, it has been shown that autonomic modulation of CA still persists following ganglionic blockade, with and without hypotension (52). In support of these findings, we found no correlation between the degree of hypotension at SL or HA with changes in TFA metrics. Finally, indirect and direct evidence indicates that regional differences in SNA control exist (6, 10, 44). For example, pharmacologically induced hypotension and hypertension in humans (39) is accompanied by increases and decreases in muscle SNA, respectively. In contrast, cerebral SNA in lambs has been shown to increase with transient hypertension, but not with hypotension (10). Elevations in SNA would be expected if cerebral sympathetic activity simply paralleled outflow to other vasculature beds (e.g., muscle); however, based on the animal study mentioned above, this may not be the case, since muscle SNA activity may not be reflective of brain SNA during acute hypotension. While speculative, the lack of cerebral sympathetic excitation during hypotension could be of teleological advantage to mediate cerebral vasodilatation, whereas sympathetic excitation and cerebral constriction might be an adverse physiological response. Nevertheless, although we are confident in the effectiveness of the blockade (as reflected in significant arterial hypotension), as in other studies (26, 37, 52), we do not know if the blockade equally influences α- and β-receptors in the brain, nor can we separate peripheral from central influences (see below).

Assessment of CA.

We assessed pressure-flow relationships via TFA. It should be noted that there is no “gold standard” or normative data for the assessment of dynamic CA. We also acknowledge that the physiological and clinical implications for changes in dynamic CA and related TFA metrics are still unclear (45). For example, dynamic CA incorporates both active CA as well as windkessel (i.e., compliance) components; as such, we have interpreted our findings cautiously.

Experimental design.

Because each experimental protocol took 5–6 h per person, our studies were conducted during a time period spanning 3–12 days at 5,050 m. Thus partial acclimatization may influence and add to the variability in the physiological responses to SNA blockade. Although we acknowledge this as an unavoidable cofounder, we did not observe any correlations between the dependent outcome variables with the day of testing at 5,050 m.

Autonomic Nervous Activity, CBF, and HA

Large cerebral arteries are innervated by adrenergic fibers originating from the ipsilateral superior cervical ganglion (24); however, smaller arteries or arterioles seem to have less adrenergic innervations (24). Because of differences in distribution of α- and β-receptors, β-receptor-mediated vasodilatation occurs mainly in small cerebral vessels, whereas α-receptor-mediated vasoconstriction seems to preferentially affect large cerebral arteries (16). Thus sympathetic stimulation causes vasoconstriction in large vessels, at least in experimental animals. However, CBF is well maintained as long as arterial pressure remains in the autoregulatory range. This maintenance of CBF is mediated via a concomitant decrease in pial vessel resistance (9), caused by either autoregulatory or β-adrenergic vasodilatation. We have recently documented that the balance of arterial blood gases accounts for a large part of the observed variability (∼40%), leading to changes in CBF at HA (33). Although HA causes marked elevations in adrenergic activity (19), and despite some evidence of SNA in regulating CBF (21, 35), our findings show that acute sympathetic deactivation via α- and β-blockade does not influence resting CBF (when semirecumbent), irrespective of HA sympathoexcitation (at least systemically) and related acute sympathoinhibition.

Influence of Hypoxia on Dynamic CA

The absence of CA has previously been thought to manifest as reductions in phase (36) and increases in coherence and gain (17, 51). In agreement with previous studies at HA (5, 8, 25, 31, 42), we observed these changes following ascent to HA (Table 2). These findings, collectively, indicate that HA exposure gives rise to a cerebral circulation that is more passive to VLF blood pressure fluctuations, as indicated by an increase in coherence and a reduction in VLF phase lead. Although the gain metric is a less reliable index of CA, the increase in VLF gain is also supportive of impaired CA. However, because such changes in CA parameters do not improve following partial acclimatization (25) and may not be involved in the etiology of altitude illness (8, 42), it seems reasonable to speculate that such changes are either adaptive in nature or reflective of hypoxia on TFA parameters.

SNA Control of Dynamic CA at SL and HA

Previous studies have shown that CA is, at least partly, under a neural influence (52). The present study is the first to examine the extent to which any changes in CA are under neural control at HA. Our findings at SL are also consistent with a recent report that dynamic CA is likely influenced by sympathetic rather than cholinergic mechanisms (18). Of note, our findings show that blockade at SL results in comparable changes in dynamic CA to those observed at HA with sympathoexcitation (i.e., without blockade). Although these findings are seemingly at odds with a neural control of CA (52), the related elevations in arterial pressure at HA may lead to changes in steady-state cerebrovascular resistance and/or vascular compliance that can modulate the dynamic pressure-flow relationships (50).

Our findings reveal that blockade at HA reduced LF coherence and elevated phase lead, fully restoring these metrics back to SL blockade values. Although these findings are supportive of a neural influence on explaining changes in coherence and phase at HA, blockade also resulted in further increases in VLF transfer function gain. These findings appear contradictory, given that elevation in gain is commonly interpreted as indicating impairment in CA, while reductions in coherence and elevations in phase are interpreted as an improvement in CA. We considered two possible explanations for these novel findings. First, it is possible that the CA impairment at HA, as indicated by reductions in VLF phase lead and elevations in coherence and gain, is normalized to SL values by removal of sympathoexcitation. However, if this were the case, then VLF coherence should decrease, given that CA is inherently nonlinear and would give rise to nonlinear pressure-flow relationship. Furthermore, this explanation would not account for the discordant rise in gain following blockade, which indicates that CA further worsens following blockade at altitudes. Thus we considered a more parsimonious explanation based on recent findings implicating the role of steady-state vascular resistance and compliance as determinants of dynamic cerebral-pressure flow relations. As Zhang et al. (50) reported, it is possible for an increase in vascular compliance to give rise to concurrent increases in both gain and phase, without necessarily implicating any change in CA per se. Thus at HA, prazosin administration may have enhanced steady-state cerebrovascular compliance, which, coupled with the rise in MAP spectral power at altitude, enhances capacitive blood flow (43) to give rise to concurrent increases in VLF gain, as well as phase, without altering coherence. While speculative, other unmeasured factors, such as a greater reliance on angiotensin II and vasopressin to maintain arterial pressure and/or small elevations in intracranial pressure (47) at HA, may have influenced cerebrovascular compliance and hence the gain of the arterial pressure transfer function independent of the other CA responses following sympathetic blockade. We should acknowledge, however, that our study was not designed to explain the differential changes in CA metrics. Although multiple measures of CA may be used to add confidence to our study conclusions (41), the physiological interpretation of using more than one metric is currently unclear.

Differential Influence of SNA Blockade on Blood Pressure

Despite comparable doses of α-adrenergic and β-blockade blockades, there was a greater degree of arterial hypotension (−26 ± 13% HA vs. −15 ± 11% SL; Table 1) and bradycardia at HA than at SL. Moreover, because we needed to infuse 0.5–1.5 liters of saline at HA to maintain blood pressure in a safe range, the impact of hypotension is likely to be an underestimation of the influence of the blockade. The mechanisms for these differences in the effectiveness α-adrenergic and β-blockade are not clear, but likely reflect the influence of marked tonic sympathoexcitation, potential dehydration, and related mild hypertension at HA.

SNA Control of Cerebrovascular Reactivity

The role of SNA in affecting cerebrovascular reactivity is unclear, with acute sympathetic activation (e.g., handgrip, lower body negative pressure) leading to a maintained (1, 29) or diminished cerebrovascular CO2 reactivity (49). The former finding is especially hard to reconcile, since cerebrovascular CO2 reactivity was reported to be reduced during acute sympathetic deactivation (via ganglionic blockade), a phenomenon interpreted to suggest a moderate direct effect of the SNA on the cerebral vasculature (26). Conversely, in a well-controlled experiment by Przybyrowski and coworkers (37), it was reported that apnea-induced elevations in MCAv were attenuated by more than one-third when related elevations in MAP were prevented using ganglionic blockade. Moreover, they found the blockade attenuated cerebrovascular CO2 reactivity to hypercapnia, but not hypocapnia. Our finding that blockade selectively reduced the hypercapnic-induced pressure effect (Table 3) at both SL and HA is entirely consistent with this report.

Consistent with others, we also find that, despite chronic SNA elevations, CBF reactivity is elevated at HA (33, 42). The CBF reactivity to CO2 is increased at altitude primarily because of a decrease in extracellular HCO3 that accompanies the chronic reduction in PaCO2 and possibly increased tissue lactate at altitude. At SL, because CBF depends on extracellular pH, the vascular response to CO2 is inversely related to extracellular HCO3 (27, 28, 30). Thus, at HA, because of the logarithmic relation between PaCO2 and pH [i.e., change (Δ) in pH = Δlog PaCO2], any change in PaCO2 at HA will result in a greater pH change compared with the equivalent SL PaCO2, and thus the lower PaCO2 at HA may overestimate the hypercapnia responses (40). As an additional mechanism to explain why CBF reactivity to CO2 is increased at altitude, we found that the elevation in MAP-CO2 reactivity was correlated with the increase in MCAv-CO2 reactivity (R2 = 0.41; P < 0.01) at HA. In contrast, the comparable reductions in MCAv-CO2 reactivity following SNA blockade at both SL and HA were not related to reductions MAP-CO2 reactivity. We speculate that elevations in MCAv reactivity to hypercapnia are mediated, in part, via the hypercapnic-induced systemic pressure following chronic sympathoexcitation, and that reduction in MCAv reactivity via acute sympathoinhibition is mediated by some other mechanism(s). In support of the latter, it was reported that hypotensive animals had a lack of change in cerebral vessel diameter in response to increases or decreases in PaCO2 (i.e., CBF reactivity is abolished), due to the vessels being already maximally dilated (20). Thus a compromise in the capacity of the cerebral vessels to dilate when the PaCO2 is elevated could indicate that, in hypotension, the maintenance of cerebral perfusion takes precedence over the maintenance of a normal tissue Pco2. Finally, it would be remiss not to consider a technical alternative to explain our findings. Although we mentioned above about the caution needed in using transcranial Doppler to reflect absolute flow, since adrenergic innervation occurs primarily in large cerebral arteries, blockade could result in MCA diameter increases and smaller changes in velocity for the same change in Pco2. Although this effect could account for the decrease in MCA velocity reactivity at SL and at altitude, consistent with our laboratory's previous report of the influence of MAP on MCAv in the absence of changes in PaCO2 (34), we observed comparable correlations between the degree of hypotension with the related decline in MCAv following blockade at both SL (R2 = 0.48) and at HA (R2 = 0.46). Thus we feel it unlikely that changes in diameter, if they did occur, would explain the differential influence of HA on our findings.

Role of the Autonomic Nervous System in Ventilatory Control at HA

Sympathetic nervous supply to the carotid body increases the frequency of sinus nerve chemosensory discharges (11). Moreover, Asano et al. (7) reported that, using circulating and urinary noradrenaline concentrations or spillover as indirect measures of SNA, the progressive increase in sympathoexcitation in humans during exposure to HA was correlated with the increase in resting V̇e. Thus it seems reasonable that hypoxic-induced elevations in SNA would account for some of the changes in ventilatory control at HA. Into the context of chronic hypoxia, our findings extend an earlier report that an altered autonomic activity following 8 h of hypoxia does not underlie the acclimatization observed in ventilatory control (32). Accordingly, alterations in SNA following acute and chronic exposure to hypoxia do not appear to play a role in ventilatory control at HA.

In conclusion, with the exception of dynamic CA, we provide evidence of a redundant role of SNA as a mechanism underlying changes in cerebrovascular reactivity and ventilatory control following partial acclimatization to HA. These findings may have implications for our understanding of CBF function in the context of other pathologies associated with sympathoexcitation and/or hypoxemia.


This study was supported by the Otago Medical Research Foundation, SPARC New Zealand, the Peninsula Health Care and Air Liquide. This study was carried out within the framework of the Ev-K2-CNR Project in collaboration with the Nepal Academy of Science and Technology, as foreseen in the Memorandum of Understanding between Nepal and Italy, and thanks to a contribution from the Italian National Research Council.


No conflicts of interest, financial or otherwise, are declared by the author(s).


Author contributions: P.N.A., S.J.E.L., K.N.T., J.D.C., Y.-C.T., and K.R.B. conception and design of research; P.N.A., S.J.E.L., M.F., K.N.T., J.D.C., and K.R.B. performed experiments; P.N.A., S.J.E.L., M.F., K.N.T., Y.-C.T., and K.R.B. analyzed data; P.N.A., S.J.E.L., M.F., J.D.C., Y.-C.T., and K.R.B. interpreted results of experiments; P.N.A. and K.N.T. prepared figures; P.N.A., K.N.T., J.D.C., and Y.-C.T. drafted manuscript; P.N.A., S.J.E.L., M.F., K.N.T., J.D.C., Y.-C.T., and K.R.B. edited and revised manuscript; P.N.A., M.F., K.N.T., J.D.C., Y.-C.T., and K.R.B. approved final version of manuscript.


The authors are thankful to J. Duffin, who kindly providing technical assistance and the rebreathing analysis program. We thank Drs. R. I. A. Lucas, K. C. Peebles, R. Basnyat, and Joseph Donnelly for technical assistance during the experimental testing. We extend our thanks to ADInstruments and Compumedics Ltd. for the use of laboratory equipment.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
View Abstract