Seventeen participants (11 male and 6 female; mean ±S.D.: aged 31 ± 9 years; height, 1.7 ± 0.1 m; body mass, 69 ± 8 kg) volunteered for this study, which was approved by the Lower South Regional Ethics Committee of Otago and conformed to the standards set by the Declaration of Helsinki. Participants were informed of the experimental procedures and possible risks involved in the study and written informed consent was obtained. All participants underwent full medical screening, including ECG and echo-cardiography assessment. Participants were not taking any medication, all were non-smokers, and none had any history of cardiovascular, cerebrovascular or respiratory disease. In addition, only two participants had previous high-altitude experience, which was >2 years previous to this expedition.

Following full familiarisation with the experimental procedures (outlined below), the participants underwent two or four experimental trials: one at sea level (Dunedin, New Zealand; barometric pressure 755 ± 7 mmHg) and one or three repeated trials while living at high altitude (5050 m; the Ev-K2-CNR Pyramid Laboratory, Khumbu Valley, Nepal, at the base of Mt Everest; barometric pressure 413 ± 1 mmHg). The high-altitude trials were completed upon initial arrival (2–4 days) and then a sub-group of 10 participants were tested twice across a partial acclimatisation period of ∼2 weeks (at days 7–9 and 12–15). Sea-level testing was completed 2 weeks before arriving in Nepal. Participants spent 7 days at Kathmandu (∼1400 m) before flying to Lukla (2860 m). Participants then trekked to the Ev-K2-CNR Pyramid Laboratory over an 8 day period, which included rest days at Namche Bazar (3450 m) and Pheriche (4252 m).

With the exception of arterial blood gas sampling, which was conducted following 10 min supine rest, all experiments were performed with participants in a semi-recumbent position. Each experimental testing session included an arterial blood gas sample, instrumentation, a 5 min baseline data collection period, then 4 min of steady-state hyperoxic hypercapnia (7% CO₂–93% O₂) and 5 min of voluntary hyperventilation to induce steady-state hypocapnia. Hyperoxic hypercapnia was intentionally used in order to eliminate the influence of hypoxic-induced peripheral chemoreceptor activation at high altitude and acutely remove the influence of hypoxia on cerebrovascular tone. For hypocapnia, participants were instructed and given verbal feedback to increase ventilation (via depth and frequency) to lower and then maintain a partial pressure of end-tidal ( ) at 22 ± 2 mmHg (sea level) and 17 ± 2 mmHg (5050 m). The order of each respiratory manipulation was randomised and full recovery was permitted between each trial. These tests were done between 15:00 and 19:00 at sea level and 12:00 and 19:00 at high altitude. The participants were instructed to avoid caffeine, alcohol and exercise within 12 h prior to the experimental testing.

This study was part of a larger research expedition and consequently participants took part in a number of studies conducted during the 2 weeks at the Pyramid Laboratory. However, none of the other studies involved exercise, and recovery time between testing sessions was managed to avoid potential confounding results (e.g. >48 h between all drug intervention studies). In addition, participants were given low-dose acetazolamide (125 mg bd) for the first 7 days of ascent as acute mountain sickness (AMS) prophylaxis, as recommended by others (^{Basnyat et al. 2006}), to minimise the impact that AMS has on cerebrovascular and ventilatory responses (^{Jansen et al. 1999}). Importantly, treatment was discontinued at day 7 to allow sufficient clearance time (i.e. >48 h) before experimental data were collected, since the reported half-life for acetazolamide is ∼10 h (^{Ritschel et al. 1998}) and this low-dose quantity is reported to be 90–100% excreted within 24 h of administration (^{Richalet et al. 2005}). AMS was assessed via the Lake Louise Questionnaire (LLQ; ^{Roach et al. 1993}) and Environmental Symptoms Questionnaire Cerebral Symptoms (ESQ-C; ^{Sampson et al. 1983}) scoring systems. Clinical diagnosis of AMS was determined from a LLQ score of ≥5 in the presence of a headache and an ESQ-C score ≥0.7, as recommended by others (^{Bailey et al. 2006}).

Data were averaged over the final minute of baseline, hypocapnia and hypercapnia conditions. The time course of change for all measures was examined using linear mixed modelling (SPSS 17.0, SPSS Inc., IL, USA), which can be employed for unbalanced and mixed (between- and within-subjects factors) repeated measures designs, such as in the present study (^{Cnaan et al. 1997}). Main and interaction effects of time (4 levels) and group (2 levels) were tested, with pair-wise comparisons (Bonferroni corrected) to identify differences between test-time. A planned comparison was also conducted to assess changes from sea-level measures. For all variables tested there was no difference between groups nor was there an interaction effect, therefore all participants were pooled into one group at sea level and upon initial arrival at high altitude. All data are expressed as mean ±S.D. Statistical inferences were based on an α-level of 0.05. The relationship between changes in MCAv and the blood gas balance (i.e. the ratio) at high altitude from sea-level values was tested using linear regression.

Participants' arterial blood gases showed the expected initial respiratory alkalosis at high altitude, which was corrected with partial acclimatisation over the 14 days (Table 1). As anticipated, the lowest and the highest were observed at days 2–4, although the subsequent recovery in was not statistically significant or as pronounced as that for (Table 1). Upon initial arrival at high altitude, MCAv was elevated (up 31 ± 31%, P < 0.01) compared to sea level, but returned to sea-level values by 7–9 days (P= 0.30; Table 1 and Fig. 1). Heart rate was elevated upon initial arrival at high altitude (up 11 ± 13 beats, P < 0.01) and remained elevated over the 14 days (Table 1). No significant changes in blood pressure, cerebral vascular resistance (CVR) and conductance (CVC) were observed across the acclimatisation period compared to sea level (all P > 0.05; Table 1).

The theoretical influence of hypoxia per se on MCAv decreased during the acclimatisation period, as shown by the CO₂-adjusted MCAv data in Table 1. This adaptation over time is consistent with the / ratio (Table 1 and Fig. 1) and the reduced sensitivity to hypoxia (Table 2) across the partial acclimatisation.

Cerebrovascular reactivity to hypercapnia was lower on average at all three testing times at high altitude compared to sea level (Figs 3 and 4B); however, statistical significance was only reached at days 2–4 (down 26 ± 43%, P= 0.02; n= 17) and at days 12–15 (down 24 ± 30%, P= 0.03; n= 10). Cerebrovascular reactivity to hypocapnia was elevated at days 2–4 (up 153 ± 80%, P < 0.01), a response which persisted during the 2 weeks (up 70 ± 47% at days 12–15; Fig. 3). CVR for hypercapnia reactivity tests were greater (1.1 vs. 0.9 mmHg cm s⁻¹, P= 0.03) upon initial arrival compared to sea level, and remained elevated at days 7–9 (1.2 mmHg cm s⁻¹, P < 0.01 vs. sea level) and days 12–15 (1.1 mmHg cm s⁻¹, P= 0.01 vs. sea level). CVR for the hypocapnia tests was unchanged, P= 0.74) across the 2 week period from the sea-level measure (1.7 mmHg cm s⁻¹). Ventilatory response to hypercapnia was elevated upon arrival at high altitude (up 111 ± 127%, P= 0.01; 2.1 ± 1.3 and 4.0 ± 3.0 l min⁻¹ mmHg⁻¹, sea level and days 2–4, respectively) and, on average, remained elevated across the 2 weeks living at high altitude, although not statistically so (P > 0.05 vs. sea level; 3.7 ± 1.9 and 3.2 ± 1.5 l min⁻¹ mmHg⁻¹, days 7–9 and 12–15, respectively; Fig. 4A).

Since first reported by ^{Severinghaus and colleagues (1966b)}, the increased CBF upon initial exposure to high altitude and subsequent normalisation after a period of acclimatisation has been well described (^{Huang et al. 1987}; ^{Wolff, 2000}; ^{Ainslie & Ogoh, 2010}). Importantly, the changes that we observed using transcranial Doppler (up 31%) were similar to those previously reported using different measuring techniques; e.g. ^{Severinghaus et al. (1966b}), using the Kety–Schmidt method, reported that CBF increased 24% on arrival (6–12 h) at high altitude (3810 m). The mechanisms underlying the regulation of CBF during exposure to hypoxia are influenced by the extent of the hypoxic stimulus itself and on the cerebrovascular sensitivity to hypoxia and CO₂ (^{Ainslie & Ogoh, 2010}), acting in an interactive way (^{Severinghaus, 2001}). The novel information that the current study contributes is the time course of the increase and subsequent normalisation of CBF at high altitude, along with the alterations in blood gas balance across this time course. For example, the low -to- ratio at days 2–4 indicates a greater degree of hypoxic vasodilatation for a given hypocapnic vasoconstriction (Fig. 1) – this balance, prior to ventilatory acclimatisation, accounts for ∼40% of the initial increase in CBF upon arrival at high altitude (Fig. 2). Our data indicate that, within 7–9 days of living at high altitude, normalisation of CBF occurs due to hypoxic-induced elevations in ventilation (Table 1). This influence leads to a higher -to- ratio, although only reaching statistical significance at days 12–15 (P < 0.01 vs. days 2–4), and therefore less hypoxic-induced dilatation and more hypocapnic-induced constriction in the cerebral circulation. Consistent with these findings are the adjusted MCAv data (Table 1) and the reduced hypoxic sensitivity at days 7–9 and 12–15 compared to initial arrival at high altitude (days 2–4, Table 2). These analyses reveal that the hypoxic vasodilatory contribution to CBF, independent of the changes in , is greatest upon initial arrival at high altitude prior to ventilatory adjustments, and that the theoretical influence of hypoxia per se on MCAv is reduced during acclimatisation. Importantly, although the ventilatory acclimatisation relieves some of the hypoxaemia, it is not sufficient to remove the hypoxic vasodilatory effect altogether, since the normalised MCAv is still elevated compared to sea level (Table 1).

At sea level, steady-state cerebrovascular reactivity is more sensitive to increases in (i.e. hypercapnic reactivity) than decreases (i.e. hypocapnic reactivity) (^{Ide et al. 2003}; ^{Xie et al. 2005}; ^{Peebles et al. 2007}). Data from the present study indicate that this relation is reversed upon initial arrival at high altitude, while following partial acclimatisation, hyper- and hypocapnia cerebrovascular reactivity were similar to one another and both significantly different to sea-level responses. Specifically, both steady-state hypercapnic and hypocapnic reactivity were altered (down 26% and up 153%, respectively; both P < 0.05 vs. sea level) upon arrival at 5050 m; responses that remained affected across a 2 week acclimatisation period (at days 12–15; down 24% and up 70%, respectively; both P < 0.05 vs. sea level; Figs 3 and 4). In addition, Fig. 5 shows the typical response across the 2 week acclimatisation for one individual, which clearly shows the reduced hypercapnic reactivity at high altitude compared to the sea-level response.

^{Jansen et al. (1999)} has proposed that the augmented hypocapnic and reduced hypercapnic cerebrovascular reactivity at high altitude could potentially be explained by a resetting of the total cerebral vasomotor reactivity (the sum of the fractional dilatation during hypercapnia and the fractional vasoconstriction during hypocapnia). The altered hypocapnic reactivity observed in the current study is consistent to that observed by ^{Jansen et al. (1999)}, who reported a 71% higher reactivity in healthy newcomers to high altitude (4243 m) compared to a separate group of sea-level controls. Our data confirm this observation and importantly do so using a within-subject study design. Furthermore, the greater augmentation of hypocapnic reactivity and the significant reduction in hypercapnic reactivity observed in the present study indicates that higher altitudes than previously examined may alter cerebrovascular reactivity to a greater extent in healthy newcomers to high altitude. We had previously reported that the MCAv response to hypocapnia (voluntary hyperventilation) was blunted (^{Ainslie et al. 2007b}) or unchanged (^{Ainslie & Burgess, 2008}) at 3840 m, which conflicts with our present findings. One explanation for these differences is that the participants in our earlier study were tested following a 9 day ascent to altitudes >5000 m. Therefore, perhaps the descent (i.e. relative deacclimatisation), and related changes in arterial blood gases, from those at higher altitude are relevant in opposing the hypocapnia reactivity response. Recently, it has been suggested that augmented hypocapnic reactivity at high altitude could be an ‘artefact’ because of the influence of hyperventilation on elevating PaO₂ and thus reducing the hypoxic dilatory stimulus on the cerebrovasculature (^{Moller, 2010}). There are two relevant points to this consideration. First, although PaO₂ within this range will have an effect on CBF dilation, it is unlikely that these changes would be solely responsible for our findings. For example, PaO₂ increases over this two-week time frame (Table 1) due to ventilatory acclimatisation, yet the heightened hypocapnic reactivity persists – a report consistent to that observed at a lower altitude (4243 m; ^{Jansen, 1999}). Second, regardless of the mechanism this augmented hypocapnic sensitivity is the physiological reality in cerebrovascular control: that is, a given drop in PaCO₂ at high altitude results in a greater drop in CBF. The implications of this maybe important in the pathophysiology of a range of cerebrovascular disorders at high altitude (e.g., AMS, CSA, cognitive impairment).

One potential explanation for the reduced dilatory capacity may be because of increased muscle sympathetic nervous activity (MSNA), consistently shown to occur at high altitude using a variety of direct and indirect MSNA measurement techniques (^{Saito et al. 1988}; ^{Duplain et al. 1999}; ^{Calbet, 2003}; ^{Hansen & Sander, 2003}). However, an elevated MCAv response to hypercapnia during rebreathing protocols that we (^{Fan et al. 2010a}) and others (^{Subudhi et al. 2010}) have observed during hypobaric hypoxia (natural and simulated), along with the likelihood of an additional elevation in MSNA during such hypercapnic testing (^{Claassen et al. 2007}), indicates that the mechanism for a reduced hypercapnia response during steady-state hypercapnia is more probably linked to changes in ventilation sensitivity. Such findings are consistent with the view that cerebral SNA is differentially controlled from MSNA (^{Ainslie & Tzeng, 2010}).

In a parallel study to the present study, we observed an enhanced hypercapnic reactivity response during a modified rebreathing protocol (^{Fan et al. 2010a}), whereas these steady-state data show an attenuated response. The critical difference between the steady-state and rebreathing methods for the assessment of hypercapnic cerebral reactivity is that the rebreathing protocol isolates the cerebrovascular reactivity response from any change in ventilatory sensitivity (i.e. as a consequence of living at high altitude), while the steady-state protocol includes this ventilatory response to (^{Ainslie & Duffin, 2009}). The influence of changes in ventilation sensitivity at high altitude is illustrated in these steady-state cerebrovascular reactivity data. For example, one participant showed a negative change in MCAv with increased (Fig. 4B), which was accompanied by the greatest increase in ventilatory sensitivity at the same time points (Fig. 4A). Our findings of a reduction in cerebrovascular reactivity to CO₂ and reciprocal elevations in ventilatory sensitivity are broadly consistent with the findings of ^{Xie et al. (2006)} who, using indomethacin to impair hypercapnic cerebrovascular reactivity in healthy subjects at sea level, observed an augmented hypercapnia ventilatory response. The findings of Xie et al. highlighted the important role of cerebrovascular responsiveness in ventilatory control in the awake human. Thus, in addition to the powerful contribution of the peripheral chemoreceptors, the extent to which changes in cerebrovascular reactivity may also mediate the development of ventilatory acclimatisation and related breathing instability during sleep at high altitude warrants future consideration.

An enhanced CBF responsiveness to hypocapnia may be a contributing factor to the commonly, but not universally (^{Thomas et al. 2010}), observed reductions in orthostatic tolerance during hypoxic conditions (^{Blaber et al. 2003}). Importantly, the enhanced hypocapnic cerebrovascular response would increase an individual's risk of cerebral ischaemia during transient reductions in and potentially may lead to impaired cognitive function during exposure to high altitude (^{Hornbein et al. 1989}) and facilitate neurological injury (^{Wilson et al. 2009}). The mechanisms underlying these reactivity responses as well as resting CBF changes warrant further study.

Tight control of the cerebrovascular CO₂ reactivity provides an important regulatory mechanism to minimise changes in brain [H⁺] at the level of the central chemoreceptor, and thereby stabilise breathing during fluctuating levels of chemical stimuli (^{Ainslie & Duffin, 2009}). Consequently, a reduction in CBF hypercapnic responsiveness to CO₂ and/or an augmentation in the hypocapnic range, as observed in the current study, may account in part for the development of periodic breathing commonly observed in newcomers to high altitude (^{Ainslie et al. 2007b}). The causative link between altered cerebrovascular function and the occurrence of breathing instability at high altitude is incompletely understood and warrants further investigation.

The balance of arterial blood gases accounted for a large part of the observed variability leading to changes in CBF at high altitude. High altitude exposure alters the cerebrovascular reactivity to hypercapnia and hypocapnia differentially and they remained distorted during partial acclimatisation. These alterations in cerebrovascular reactivity to CO₂ may, in part, also affect ventilatory sensitivity and breathing stability.