Differences in the control of breathing between Himalayan and sea-level residents

Abstract

We compared the control of breathing of 12 male Himalayan highlanders with that of 21 male sea-level Caucasian lowlanders using isoxic hyperoxic (= 150 mmHg) and hypoxic (= 50 mmHg) Duffin's rebreathing tests. Highlanders had lower mean ±s.e.m. ventilatory sensitivities to CO₂ than lowlanders at both isoxic tensions (hyperoxic: 2.3 ± 0.3 vs. 4.2 ± 0.3 l min⁻¹ mmHg⁻¹, P= 0.021; hypoxic: 2.8 ± 0.3 vs. 7.1 ± 0.6 l min⁻¹ mmHg⁻¹, P < 0.001), and the usual increase in ventilatory sensitivity to CO₂ induced by hypoxia in lowlanders was absent in highlanders (P= 0.361). Furthermore, the ventilatory recruitment threshold (VRT) CO₂ tensions in highlanders were lower than in lowlanders (hyperoxic: 33.8 ± 0.9 vs. 48.9 ± 0.7 mmHg, P < 0.001; hypoxic: 31.2 ± 1.1 vs. 44.7 ± 0.7 mmHg, P < 0.001). Both groups had reduced ventilatory recruitment thresholds with hypoxia (P < 0.001) and there were no differences in the sub-threshold ventilations (non-chemoreflex drives to breathe) between lowlanders and highlanders at both isoxic tensions (P= 0.982), with a trend for higher basal ventilation during hypoxia (P= 0.052). We conclude that control of breathing in Himalayan highlanders is distinctly different from that of sea-level lowlanders. Specifically, Himalayan highlanders have decreased central and absent peripheral sensitivities to CO₂. Their response to hypoxia was heterogeneous, with the majority decreasing their VRT indicating either a CO₂-independent increase in activity of peripheral chemoreceptor or hypoxia-induced increase in [H⁺] at the central chemoreceptor. In some highlanders, the decrease in VRT was accompanied by an increase in sensitivity to CO₂, while in others VRT remained unchanged and their sub-threshold ventilations increased, although these were not statistically significant.

Introduction

Andean and Tibetan highlanders exhibit different patterns of adaptation to chronic hypoxia of their environment (Leon-Velarde & Richalet, 2006; Beall, 2007; Brutsaert, 2007). Andean highlanders, compared to Tibetans, have lower resting minute ventilations (Beall et al. 1997), higher end-tidal () (Moore, 2000) and blunted hypoxic ventilatory responses (HVRs) (Chiodi, 1957; Severinghaus et al. 1966; Sorensen & Severinghaus, 1968a,b; Velasquez et al. 1968; Lahiri et al. 1969; Cudkowicz et al. 1972; Leon-Velarde et al. 1996; Beall et al. 1997; Gamboa et al. 2003; Brutsaert et al. 2005). Blunting of HVRs in Andeans is thought to be acquired rather than genetic and develops in adolescence (Lahiri et al. 1976; Lahiri, 1980), although recent studies suggest that the acquired blunting may depend on age and on the altitude of exposure (Curran et al. 1995). Tibetans, on the other hand, have HVRs comparable to sea-level residents at sea-level (Hackett et al. 1980; Zhuang et al. 1993; Ge et al. 1994; Beall et al. 1997), so that Tibetan ancestry may actually protect against blunting of the HVR (Curran et al. 1997). The reasons for differences in ventilatory adaptation patterns between these two populations are unclear, but they probably involve differences in the control of breathing (Moore, 2000; Brutsaert et al. 2005).

Control of breathing in humans can be broadly divided into chemoreflex and non-chemoreflex drives to breathe (Fig. 1) (Lloyd & Cunningham, 1963). Non-chemoreflex breathing stimuli include a wakefulness drive (Longobardo et al. 2002), voluntary (cortical) drive (Shea, 1996) and hormonal factors (Jensen et al. 2008), as well as neural and humoral mediating factors that are especially important in the control of breathing during exercise (Bell, 2006; Dempsey, 2006; Haouzi, 2006). The chemoreflex drive to breathe can be further divided into central and peripheral chemoreceptor drives. Both central and peripheral chemoreceptors respond to changes in the hydrogen ion concentration ([H⁺]) in their immediate environments (Torrance, 1996; Nattie & Li, 2009). In contrast to the central chemoreceptors, peripheral chemoreceptors are also sensitive to changes in arterial () via a hypoxia-mediated increase in their sensitivity to [H⁺] (Cunningham, 1987; Torrance, 1996; Kumar & Bin-Jaliah, 2007), and hyperoxia (≥ 150 mmHg) effectively silences this response (Lloyd & Cunningham, 1963; Mohan & Duffin, 1997). Central and peripheral chemoreceptor neural drives are integrated in the medulla to provide the total chemoreflex neural drive (Fink, 1961; Shea, 1996; Mohan & Duffin, 1997; Orem et al. 2002) that, in combination with non-chemoreflex drives, provides ventilatory drive to respiratory muscles. It is important to point out that central and peripheral chemoreceptor drives only affect ventilation if their total sum exceeds a drive threshold so that a ventilatory recruitment threshold (VRT) to [H⁺] or is established (Duffin, 2005). Below VRT, changes in [H⁺] or have no effect on ventilation which is solely dependent on non-chemoreflex drives, such as the ‘wakefulness drive’ (Duffin, 2005).

Figure 1. The control of breathing model (Lloyd & Cunningham, 1963).

The total ventilatory drive is a sum of non-chemoreflex and chemoreflex drives to breathe that are integrated in the respiratory centre. The ventilatory drive exerts its action on the respiratory muscles that affect pulmonary ventilation and result in changes in arterial and . Arterial is ‘sampled’ by the peripheral chemoreceptors located in the carotid bodies, where it determines the peripheral chemoreflex drive. Hypoxia exerts its effect on ventilation via peripheral chemoreceptors where it acts indirectly via increasing the ventilatory sensitivity to CO₂ in most individuals but may also act directly by increasing the overall activity of the receptor. Central chemoreceptors respond to changes in the local [H⁺] environment, which is mediated by brain tissue . Brain tissue is a function of both arterial and cerebral blood flow (CBF), which acts to decrease the brain tissue at higher CBF (Ainslie & Duffin, 2009). The central and peripheral chemoreflex drives add together to form a total chemoreflex drive to breathe.

Control of breathing studies in high altitude natives have attempted to elucidate the observed differences in ventilation between high altitude natives using two methods: the hypoxic ventilatory response (HVR) and the hypercapnic ventilatory response (HCVR) (Moore, 2000; Brutsaert, 2007). As their names imply, the HVR method measures the ventilatory response to a hypoxic stimulus and thereby primarily assesses a peripheral-chemoreceptor-mediated response, while the HCVR method measures ventilatory response to hypercapnia and therefore assesses the combined central and peripheral responses (see Fig. 1 and Duffin, 2007). Several methods of measuring HVR and HCVR have been used to study control of breathing in highlanders and these are summarized in Table 1 and in the Methodological considerations section of the Discussion.

Table 1.

Summary of different methods used to measure hypoxic (HVR) and hypercapnic (HCVR) ventilatory responses in highlanders

		Method	Studies	Advantages	Disadvantages
HVR2	Non-isocapnic	Acute hypoxia or hyperoxia	(Lahiri et al. 1967; Lefrancois et al. 1968; Velasquez et al. 1968; Zhuang et al. 1993; Curran et al. 1995, 1997)	Simple, fast, not affected by HVD	Variable O2 stimulus1 Variable CO2 confounds interpretation of results
		Effect of acute hypoxia/hyperoxia on (non-isocapnic) ventilatory response to CO2	(Severinghaus et al. 1966; Milledge & Lahiri, 1967; Sorensen & Severinghaus, 1968a,b; Lahiri et al. 1969)	Simple, fast	Variable CO2–O2 stimuli1 Confounding effects of CO2 and O2 on ventilation
	Isocapnic	Progressive isocapnic hypoxia	(Forster et al. 1971; Weil et al. 1971; Byrne-Quinn et al. 1972; Hackett et al. 1980; Huang et al. 1984; Zhuang et al. 1993; Curran et al. 1995, 1997; Beall et al. 1997; Gamboa et al. 2003)	vs. graph Allows measurement of HVD Choice of isocapnia level may affect HVR calculation (see Fig. 3)	Longer than acute HVR tests Potentially confounded by HVD Affected by choice of hypercapnea (see Fig. 3)
HCVR	Steady state2	CO2 inhalation	(Forster et al. 1971)	Simple, fast, not affected by HVD	Variable CO2 stimulus1 Variable O2 confounds interpretation of results
		Isocapnic changes in CO2	(Weil et al. 1971; Fatemian et al. 2003)	Constant CO2 stimulus	Variable O2 confounds result interpretation
	Rebreathing	Read's rebreathing	(Byrne-Quinn et al. 1972; Zhuang et al. 1993; Curran et al. 1995, 1997)	Ventilatory sensitivity not affected by changes in CBF2	No separation between central and peripheral chemoreceptors responses No measurement of non-chemoreflex drives to breathe No measurement of VRT

Note that all of the described methods do not measure non-chemoreflex drives to breathe, fail to separate central and peripheral components of the total chemoreflex drive to breathe (except Fatemian et al. 2003), and do not measure ventilatory recruitment threshold (VRT) at which the chemoreflex drive starts to affect ventilation. Detailed discussion of these topics can be found in Duffin (2007). HVD, hypoxic ventilatory decline. ¹These methods rely on delivering fixed inspired concentration of O₂ and CO₂ in attempt to change and/or , respectively. However, the actual change in any given subject will depend on their breathing pattern and ventilatory sensitivity to O₂ and CO₂. ²Measurement of ventilatory response to CO₂–O₂ using steady-state (non-rebreathing) methods is confounded by changes in CBF (see Berkenbosch et al. 1989; Mohan et al. 1999).

Despite numerous HVR studies, interpretation of their findings is problematic as HVR per se is not a measure of individual components of the control of breathing model (Fig. 1), but is rather an indicator of the integrated response from several components (Duffin, 2007). Furthermore, methodological differences between HVR studies (Moore, 2000), as well as their inability to separately study the individual components of the ventilatory control model (Moore, 2000; Duffin, 2007), limit the usability of HVR in comparing control of breathing between high altitude populations. With reference to a control of breathing model (Lloyd & Cunningham, 1963) (Fig. 1), fundamental limitations of the HVR methods include their inability to: (1) separate the contribution of chemoreflex from non-chemoreflex drives to breathe, (2) delineate the contribution of central and peripheral chemoreceptors, (3) identify VRT at which chemoreceptor drives start to produce an increase in pulmonary ventilation, and (4) prevent attenuation of measured ventilatory sensitivities due to cerebral blood flow (CBF)-induced washout of the [H⁺] from the central chemoreceptor (for steady-state methods only) (Berkenbosch et al. 1989; Mohan et al. 1999).

In the present study, we used Duffin's rebreathing method (Duffin et al. 2000) to fill in the missing information about control of breathing in highlanders by comparing ventilatory control in Himalayan highlanders to that of lowlanders at sea-level.

Methods

Subjects

The study was approved by the Ethics Review Board of the University Health Network, Toronto, Canada; Queen's University Ethics Review Board, Kingston, Canada; and Ladakh Institute of Prevention for the Study of Environmental, Occupational, Life Style and High Altitude Related Diseases, Ladakh, India. All studies were performed in accordance with the Declaration of Helsinki of the World Medical Association (2004). For lowlander studies, 21 healthy Caucasian male subjects were recruited from the Queen's University student population. Their mean ±s.d. age, height and body mass were 21.7 ± 2.5 years, 180.2 ± 5.4 cm and 80.4 ± 10.4 kg, respectively. For highlander studies, 12 healthy male long-term residents of Korzok village, Ladakh, India (altitude, 4550 m) were recruited. Their mean ±s.d. age, height and body mass were 40.8 ± 9.8 years, 159.9 ± 5.3 cm and 60.4 ± 10.6 kg, respectively. Each subject gave informed written consent before commencing the study.

General protocol

Each lowlander subject visited the laboratory at Queen's University (altitude, 90 m; barometric pressure, 750 mmHg; ambient air temperature, 20–25°C) for an initial (1.5 h) familiarization test on all the rebreathing apparatus and upon return to the laboratory they each performed either two (10 subjects) or four (11 subjects) pairs of hypoxic and hyperoxic rebreathing tests in random order, with each test separated by a 20 min rest period. At high altitude, the experiments were conducted in a heated research tent outside the Korzok village (altitude, 4550 m; barometric pressure, 442 mmHg; air temperature inside the tent, 18–25°C). On the first day of experiments, highlander subjects were familiarized with the equipment and the test by first watching one of the research scientists perform both hypoxic and hyperoxic rebreathing tests on themselves, and then breathing on the rebreathing apparatus and performing several trial rebreathing runs. At the same time, the nature of the experiments and specific instructions were provided in the highlander's native language through an interpreter, and all questions were answered. Highlanders returned to the tent on the following days and completed at least two pairs of hypoxic and hyperoxic tests in random order separated by at least 20 min of rest time between the tests.

Chemoreflex assessment

Testing method

Duffin's rebreathing tests (Duffin & McAvoy, 1988; Mohan & Duffin, 1997) were used to assess the chemoreflexes. Their interpretation is given in Duffin et al. (2000), and they have been extensively discussed previously (Duffin et al. 2000; Mahamed & Duffin, 2001; Mateika et al. 2004). In this modification of Read's original methods (Read, 1967), the rebreathing stage is preceded by 5 min of hyperventilation and isoxia is maintained throughout the test (Fig. 2).

Figure 2. Differences between Read's (Read, 1967) and Duffin's (Duffin et al. 2000) rebreathing methods (Mohan et al. 1999).

The top two graphs for each method illustrate the change in ventilation and with time during both tests, while the bottom graphs summarize the ventilatory response to CO₂. In Read's method, rebreathing is initiated above a eucapnic level. As a result, the ventilatory response to CO₂ is measured in the hypercapnic range only. In Duffin's method, rebreathing is preceded by 5 min of hyperventilation, so the rebreathing starts at a hypocapnic below the ventilatory recruitment threshold (VRT). As the bottom right graph illustrates, following the start of rebreathing, ventilation remains unchanged below the VRT despite the rising . Duffin's method allows measurement of sub-VRT ventilation, which corresponds to the non-chemoreflex drive to breathe, and the VRT. Since the test is repeated at two isoxic tensions, one hypoxic ( of 50 mmHg) and one hyperoxic ( of 150 mmHg), the test allows us to measure separately the contribution of central and peripheral chemoreceptors (since hyperoxia effectively silences the peripheral chemoreflex).

The hyperventilation ensures that the rebreathing starts at a CO₂ level below VRT, and enables both the VRT and the sub-threshold (basal) ventilation to be measured (Mohan et al. 1999). The latter measures the contribution of the non-chemoreflex drives to breathe (Shea, 1996), such as the wakefulness drive (Fink, 1961). Rebreathing is repeated at two different isoxic tensions: one hyperoxic (= 150 mmHg) to silence peripheral chemoreceptor and allow isolated measurement of the central chemoreceptor response (Lloyd & Cunningham, 1963), and one hypoxic (= 50 mmHg) to allow measurement of the combined central and peripheral chemoreceptor responses (Fig. 3). The difference between hypoxic and hyperoxic responses represents the contribution of the peripheral chemoreflex and can be used to calculate HVR at any given isocapnic (Fig. 3).

Figure 3. Hypercapnic ventilatory response.

The graph displays two isoxic responses: hyperoxic (= 150 mmHg) representing central chemoreflex response, and hypoxic (= 50 mmHg) representing the addition of central and peripheral chemoreflexes responses. The slope of each isoxic response represents sensitivity of the chemoreflex to CO₂. The inflection point at which ventilation starts to increase in response to increasing is the ventilatory recruitment threshold (VRT), where the chemoreflex neural drive to breathe exceeds a drive threshold and starts to produce an increase in pulmonary ventilation. Ventilation below VRT represents non-chemoreflex drives to breathe and is known as the basal ventilation. The differences in ventilation between isoxic rebreathing lines at any given isocapnic can be used to calculate the hypoxic ventilatory response (indicated by vertical arrows). Note that the choice of isocapnic affects the magnitude of the measured HVR even within the same subject (Duffin, 2007), with higher HVRs measured at higher isocapnic values in the illustrated example. Note also that the magnitude of HVR provides little information about the characteristics of the control of breathing model, as HVR magnitude is dependent on the combination of central and peripheral chemoreflex responses.

In contrast to previously used HVR and HCVR methods, Duffin's rebreathing technique allows direct measurement of the individual components of the control of breathing model (Fig. 1) (Lloyd & Cunningham, 1963) including: (1) measurement of non-chemoreflex drives to breathe, (2) separate measurement of central and peripheral chemoreceptor drives, and (3) identification of VRTs for both central and peripheral chemoreceptors. Furthermore, Duffin's rebreathing ensures equilibration between arterial, venous and tissue (Read, 1967; Read & Leigh, 1967) and thereby avoids attenuation of the measured ventilatory response due to CBF-induced washout of the and consequent lowering of [H⁺] at the central chemoreceptors that occurs with steady-state HVR techniques (Berkenbosch et al. 1989; Mohan et al. 1999; Ainslie & Duffin, 2009).

Testing protocol

The test order was randomly assigned. All subjects were seated comfortably upright during the rebreathing tests and wore a finger pulse oximetry probe. They breathed via a mouthpiece, with nose clips, through a bacterial filter (Allegiance, Healthcare Corp., McGaw Park, IL, USA) connected to one side of a Series 2870 three-way sliding valve (Hans Rudolph Inc., Kansas City, MO, USA) that provided either ambient air or gas from the rebreathing bag. The rebreathing bag had a volume of 5 l and an inlet for oxygen. It was primed with a CO₂–O₂ mixture to ensure appropriate equilibration at the beginning of a rebreathing test (Mohan & Duffin, 1997). Inspired and expired partial pressures of CO₂ and O₂ were sampled at the mouth and monitored throughout the test using a respiratory mass spectrometer (Perkin Elmer MGA 1100) at the Queen's University laboratory and infra-red gas analyser (RespirAct, Thornhill Research Inc., Toronto, Canada) at Ladakh. Ventilation was measured using bi-directional volume turbines (VMM-2A; Alpha Technologies, Laguna Niguel, CA, USA at Queens University and Universal Ventilation Meter, VacuMed, Ventura, CA, USA at Ladakh). Continuous data were input to a computer via an analog-to-digital converter (DAQCard-6024E, National Instruments, Austin, TX, USA). A specially written program (LabVIEW, National Instruments, Austin, TX, USA; source code available on request) analysed the data to provide a file of breath-by-breath , end-tidal () and ventilation. In addition, the program operated a solenoid valve controlling the flow of oxygen to the rebreathing bag to maintain at 150 mmHg (hyperoxic) or at 50 mmHg (hypoxic) during rebreathing. The and analysers were calibrated using gas from cylinders of analysed medical grade compressed gases, and ventilation was calibrated using a 3 l calibration syringe (Model R5530B, Vacumed, Ventura, CA, USA). Portable pulse oximeters (OXI; Radiometer Copenhagen, Copenhagen, Denmark in lowlanders, and Autocorr Plus Vital Signs Monitor, BCI international, USA in highlanders) were used to measure arterial oxygen saturation and heart rate.

Each rebreathing test began with a 5 min hyperventilation of ambient air, with subjects coached to maintain their between 19 and 25 mmHg for lowlanders and between 10 and 15 mmHg for highlanders. The target range was selected based on the resting mixed venous in each population. Since resting in highlanders was on average 10 mmHg less than that in lowlanders and assuming similar mixed venous to end-tidal gradients in both populations, the hyperventilation target for highlanders was set at 10 mmHg lower than that for lowlanders.

Subjects then exhaled completely and were switched to the rebreathing bag where they took three deep breaths to facilitate rapid equilibration of in the bag, lungs and arterial blood to that of mixed venous blood. This equilibration was verified by observing a plateau in the end-tidal , and was a prerequisite for continuing the test. The rebreathing test ended when ventilation exceeded 100 l min⁻¹, or exceeded 60 mmHg, or if discomfort occurred. The testing environment was quiet, with minimal distractions. Subjects were instructed to relax and close their eyes during the rebreathing experiments.

Rebreathing test analysis

Rebreathing test data were analysed using a specially written program (LabVIEW). After eliminating the initial three equilibration breaths, as well as sighs, swallows and breaths incorrectly detected by the acquisition software, breath-by-breath values were plotted against time and fitted with a least squares regression line. This line provided a predicted value of as a function of time, thereby minimizing inter-breath variability, and tidal volume (ml BTPS), respiratory rate (breaths min⁻¹) and ventilation (l min⁻¹ BTPS) were plotted against the predicted (mmHg).

These plots (see Fig. 4 for an example) were fitted by dividing them into two segments separated by a breakpoint corresponding to the Vrt, defined as the where ventilation starts to increase in response to rising (Duffin et al. 2000). The first segment was fitted with either an exponential decline to a final value, or a mean, and measured sub-VRT ventilation () that represents non-chemoreflex ventilation drive. The second segment was fitted with a straight line whose slope measured the sensitivity (). Model fitting was based on minimizing the sum of least squares for non-linear regression using the LabVIEW software (Levenberg-Marquardt algorithm). The furthest outlying points were automatically discarded until an r² value of > 0.95 was achieved.

Figure 4.

Breath-by-breath ventilation vs. end-tidal during an isoxic hypoxic Duffin's rebreathing test for one highlander illustrating the measures derived.

Statistical analyses

All results are reported as mean ±s.e.m. except where noted, with significance set at P < 0.05. The measures obtained from the repeated tests on lowlanders were averaged. A two-way, repeated measures ANOVA with factors population (highlanders vs. lowlanders) and isoxia (hypoxic vs. hyperoxic) and post hoc Tukey's tests were used to detect differences in sub-threshold ventilations, ventilatory recruitment thresholds and chemoreflex sensitivities to CO₂.

Results

Each subject completed the rebreathing protocol. One highlander subject (039) was unable to complete the hyperoxic protocol and his data were withdrawn from analysis. Table 2 shows values of , VRT and for all subjects.

Table 2.

Individual subject values for non-chemoreflex drives to breathe (), ventilatory recruitment thresholds (VRT) and sensitivities to CO₂ () for all rebreathing tests

Highlanders	Isoxic hyperoxic tests = 150 mmHg)			Isoxic hypoxic tests = 50 mmHg)
	(l min−1)	VRT (mmHg)	(l min−1 mmHg−1)	(l min−1)	VRT (mmHg)	(l min−1 mmHg−1)
001	11.5	30.5	3.5	23.5	27.0	2.8
011	23.5	40.0	2.8	18.0	38.0	2.1
014	7.2	34.0	5.3	24.7	33.0	5.4
019	6.1	33.0	1.1	4.9	30.5	2.2
020	14.0	35.0	2.1	13.2	33.0	3.8
039	n/a	n/a	n/a	5.3	23.5	3.1
051	3.5	37.0	1.9	9.8	37.0	1.7
061	11.8	36.0	1.4	9.4	30.7	1.9
062	19.4	33.0	2.0	21.9	34.0	2.2
065	12.3	30.0	2.0	17.8	30.0	3.2
066	12.7	31.0	2.1	10.6	28.7	2.9
068	14.7	32.5	1.6	19.7	29.0	2.7
Mean (s.e.m.)	12.4 (1.6)	33.8 (0.9)*†	2.3 (0.3)†	14.9 (1.9)	31.2 (1.1)*†	2.8 (0.3)†
Lowlanders	Isoxic hyperoxic tests = 150 mmHg)			Isoxic hypoxic tests = 50 mmHg)
	(l min−1)	VRT (mmHg)	(l min−1 mmHg−1)	(l min−1)	VRT (mmHg)	(l min−1 mmHg−1)
BY	18.2	53.2	4.2	19.2	45.8	8.3
EV	16.6	54.7	4.4	17.0	50.7	4.9
GM	14.3	49.0	4.6	10.4	45.5	5.0
ID	14.6	42.3	5.3	16.9	38.6	12.1
KM	11.6	46.0	2.6	7.8	41.0	5.0
KMM	21.4	52.3	2.9	10.2	43.5	7.3
MD	15.9	47.4	4.2	20.2	44.4	9.5
PR	12.1	53.0	8.2	13.8	47.0	13.5
RG	18.8	52.3	4.0	14.3	47.9	4.7
SC	9.6	52.6	4.3	11.1	49.8	9.1
SM	18.8	47.0	3.9	28.6	48.7	10.4
BS	10.3	47.0	4.4	10.1	43.7	7.6
DG	11.2	47.1	6.1	15.4	43.1	7.1
DJ	7.1	48.7	5.6	7.7	46.5	8.8
GM	9.4	49.2	1.9	9.0	44.8	2.9
JS	9.1	45.2	3.8	7.2	40.4	5.8
KD	14.2	48.1	2.5	21.5	41.9	5.8
MF	6.9	48.1	4.3	7.0	43.5	5.8
MH	6.7	46.5	1.9	7.9	42.7	3.5
MO	9.3	46.3	2.3	10.7	43.1	2.9
PJ	14.2	51.4	6.3	17.9	47.0	8.8
Mean (s.e.m.)	12.9 (0.9)	48.9 (0.7)*†	4.2 (0.3)*†	13.5 (1.2)	44.7 (0.7)*†	7.1 (0.6)*†

Mean ±s.e.m. values for each condition and subject populations are compared with (*) values indicting differences between hypoxic and hyperoxic tests within populations and (†) values indicating difference between populations determined by repeated measures ANOVA. Note that highlander subject 039 was unable to complete hyperoxic rebreathing tests and his data were withdrawn from analysis.

There was no difference in basal ventilation between highlanders and lowlanders during both isoxic tensions (hyperoxic: 12.9 ± 0.9 vs. 12.4 ± 1.6 l min⁻¹; hypoxic: 13.5 ± 1.2 vs. 14.9 ± 1.9 l min⁻¹, P= 0.982), with a trend for higher basal ventilation during hypoxia (P= 0.052). Highlanders had lower ventilatory sensitivities to CO₂ compared to lowlanders at both isoxic tensions (hyperoxic: 2.3 ± 0.3 vs. 4.2 ± 0.3 l min⁻¹ mmHg⁻¹, P= 0.021; hypoxic: 2.8 ± 0.3 vs. 7.1 ± 0.6 l min⁻¹ mmHg⁻¹, P < 0.001). Interestingly, the hypoxia-induced increase in ventilatory sensitivity to CO₂ observed in lowlanders (P < 0.001) was absent in highlanders (P= 0.361). Furthermore, ventilatory recruitment thresholds were lower in highlanders compared to lowlanders (hyperoxic: 33.8 ± 0.9 vs. 48.9 ± 0.7 mmHg, P < 0.001; hypoxic: 31.2 ± 1.1 vs. 44.7 ± 0.7 mmHg, P < 0.001), although the ventilatory recruitment thresholds decreased with hypoxia in both groups (P < 0.001). Figure 5 summarizes these findings for both populations.

Figure 5. Mean ±s.e.m. breath-by-breath ventilation vs. during isoxic hyperoxic and hypoxic Duffin's rebreathing tests for all highlander and lowlander subjects.

In both populations, the continuous lines represent hyperoxic (= 150 mmHg) and dashed lines represent hypoxic (= 50 mmHg) rebreathing tests. Dotted lines for each test represent s.e.m. responses. Note that highlanders had lower ventilatory recruitment thresholds (VRT) than lowlanders at both isoxic tensions probably due to differences in acid–base status between the two populations. Additionally, highlanders had lower central chemoreflex sensitivities to CO₂ compared to lowlanders and responded to acute hypoxia by decreasing their VRT rather than increasing the ventilatory CO₂ sensitivity like lowlanders.

Discussion

This study is the first to independently examine all three components of the ‘Oxford’ control of breathing model depicted in Fig. 1 (Lloyd & Cunningham, 1963) in highlanders, namely the central and peripheral chemoreflex drives to breathe and the non-chemoreflex drive to breathe, and compare them to those of lowlanders residing at sea-level. As Fig. 5 illustrates, the control of breathing in highlanders differs from that in lowlanders in several aspects that are discussed below.

Non-chemoreflex drives to breathe

There was no difference in the non-chemoreflex drives to breathe between highlanders and lowlanders, as indicated by similar basal (below VRT) ventilations in the two populations (Table 2). According to the control of breathing model (Lloyd & Cunningham, 1963) (Fig. 1), any differences in the control of breathing between Himalayan highlanders and lowlanders are therefore solely due to differences in their chemoreflexes. Note that the basal ventilations in our subjects fall at the higher end of the range of those reported previously (2.5–22 l min⁻¹, see Table 3). Since basal ventilation is a measure of the non-chemoreflex drive to breathe, this finding may suggest that the non-chemoreflex drive to breathe was increased in our subjects, possibly due to increased experimental anxiety secondary to unfamiliarity with the respiratory experiments. It therefore follows that familiarizing the subjects with the experimental protocol by repeating the experiments multiple times would reduce experimental anxiety and basal ventilations. However, Jensen et al. (2010), whose study agrees well with our data, recently showed that basal ventilation in a given subject was not affected by multiple repetitions of the rebreathing experiments within and between days, suggesting that a significant familiarization effect does not exist (Jensen et al. 2010). Furthermore, even if significant experimental anxiety was present in the current study, our lowlander and highlander subjects were equally naïve to the nature of respiratory experiments and were therefore probably affected to the same degree, with the familiarization protocols during their first visit to the research tent aimed at reducing this anxiety. Lastly, other methods of studying the control of breathing, including classic HVR tests, are affected by experimental anxiety to the same extent as current experiments. As a result, the reported basal ventilations are probably the best representation of the non-chemoreflex drives to breathe in the studied populations during the described testing conditions.

Table 3.

Summary of basal ventilation measurements with Duffin's rebreathing method in lowlanders residing at sea-level

Study	Subjects	Basal ventilation (mean ±s.d., l min−1)
		Hyperoxic (150 mmHg)	Hypoxic (50 mmHg)
Mohan & Duffin, 1997	7 males	Basal ventilations were measured at different isoxic values (100, 80, 60 and 40 mmHg) and varied between subjects from 3.1 to 14.7 l min−1. Change in had no effect on basal ventilation (correlation coefficient of −0.0248)
Mahamed & Duffin, 2001	5 males and 2 females	6.1 ± 0.8	8.5 ± 1.0
Mahamed et al. 2003	6 males and 3 females	3.9 ± 0.6	6.0 ± 0.8
Somogyi et al. 2005	6 males	5.0 ± 1.03	5.9 ± 0.9
Jensen et al. 2005	14 males and 14 females	16.2 ± 2.1 in males 9.9 ± 1.1 in females	17.7 ± 2.1 in males 10.5 ± 1.3 in females
Fan et al. 2010	11 males and 6 females	10.5 ± 7.4 (sea level) 11.1 ± 6.3 (altitude)	Not available
Jensen et al. 2010	12 males and 8 females	Mean, ∼9 (range, 2.5 to 19)	Mean, ∼9 (range, 1.5 to 22)
Present study	21 male lowlanders	12.9 ± 0.9 (range, 3.5 to 23.5)	13.5 ± 1.2 (range, 4.9 to 24.7)
	12 male highlanders	12.4 ± 1.6 (range, 6.7 to 21.4)	14.9 ± 1.9 (range, 7.0 to 28.6)

Note that more recent studies had higher averages and wider ranges of basal ventilations. Also note that the level of hypoxic rebreathing has no effect on basal ventilation (Mohan & Duffin, 1997).

Ventilatory recruitment thresholds

The highlanders had decreased VRTs compared to lowlanders during both hypoxic and hyperoxic rebreathing tests. The leftward shift of the VRTs in highlanders suggests that a lower was required to exceed the VRT in highlanders compared to lowlanders. Since both central and peripheral chemoreceptors are actually [H⁺] sensors, interpretation of this result should consider the acid–base status in both populations. According to the Henderson–Hasselbach equation (Nunn, 1993), the relationship between [H⁺] and can be described as follows:

where [HCO₃⁻] is the bicarbonate ion concentration. In a hypothetical sea-level resident at sea-level, [H⁺] is approximately 40 nm l⁻¹, is 40 mmHg and [HCO₃⁻] is 24 mm l⁻¹. At altitude, hypoxia-induced hyperventilation results in a reduction of that leads to a reduction in [H⁺] and respiratory alkalosis according to the above equation. Highlanders compensate for respiratory alkalosis by presumably reducing their [HCO₃⁻] through increased renal excretion, thereby restoring the /[HCO₃⁻] ratio to sea-level values and normalizing [H⁺]. Although the /[HCO₃⁻] ratio in adapted highlanders is comparable to that of lowlanders at sea-level, the lower [HCO₃⁻] in highlanders alters the relationship between [H⁺] and in this population such that a lower is required to attain the [H⁺] of 40 nm l⁻¹ in highlanders compared to lowlanders (Duffin, 2005). For example, if hypoxia-induced hyperventilation reduced CO₂ from 40 to 30 mmHg, and HCO₃⁻ fell from 24 to 18 mm l⁻¹, then the overall ratio of CO₂/HCO₃⁻ would be maintained at 5/3 as in sea-level lowlanders, but normal [H⁺] of 40 nm l⁻¹ would be achieved at a lower of 30 mmHg rather than 40 mmHg, as at sea-level. Considering that the highlanders in our study have an adapted acid–base status (Santolaya 1989), the observed difference in VRTs can be explained by the altered [H⁺]– relationship in highlanders compared to lowlanders, with the assumption that the chemoreceptor thresholds [H⁺] are similar (Duffin, 2005). However, since we did not perform acid–base measurements in the present study, the above explanation requires further experimental validation.

Central chemoreflex sensitivity to CO₂

The sensitivity of the central chemoreceptor to CO_2, as indicated by the ventilatory sensitivity during hyperoxic rebreathing, was lower in highlanders compared to lowlanders (2.5 ± 0.4 vs. 4.2 ± 0.3 l min⁻¹ mmHg⁻¹, P= 0.011). Hyperoxia effectively silences the peripheral chemoreceptor (Lloyd & Cunningham, 1963; Mohan & Duffin, 1997), and therefore the ventilatory sensitivity measured during hyperoxic rebreathing can be taken as a measure of central chemoreceptor sensitivity (Duffin, 2007). The difference in central chemoreceptor sensitivity to CO₂ between highlanders and lowlanders may represent an intrinsic difference between central chemoreceptors in two populations, or as shown previously, may be related to the difference in anthropomorphic characteristics between highlanders and lowlanders. Highlanders were older (40.8 ± 2.6 vs. 21.7 ± 0.5 years), lighter (60.4 ± 2.9 vs. 80.4 ± 2.2 kg) and shorter (159.9 ± 1.5 vs. 180.2 ± 1.2 cm) than lowlanders (P < 0.001 for all comparisons). Since the ventilatory response to hypercapnia decreases with age (Nishimura et al. 1991; Jones et al. 1993; Poulin et al. 1993; McGurk et al. 1995) and increases with weight (Marcus et al. 1994), the observed lower central CO₂ chemosensivity in our highlander subjects may be the result of their older age and smaller body size. Repeating the experiments with age-matched controls may clarify the mechanism responsible for this observation.

Previous studies of the ventilatory response to CO₂ reported similar ventilatory sensitivities to CO₂ between highlanders and lowlanders (Weil et al. 1971; Zhuang et al. 1993; Curran et al. 1995), but these studies measured the total ventilatory sensitivity to CO₂ and failed to separate contributions from the central and peripheral chemoreceptors. Fatemian et al. (2003) employed a multifrequency binary sequence in (Pedersen et al. 1999) using an end-tidal forcing technique in an attempt to study the central and peripheral components of the chemoreflex response to CO₂ separately. They found that during euoxia ( of 100 mmHg), the total and peripheral chemoreceptor sensitivities to CO₂ in highlanders were double those of the sea-level residents at sea-level, with hypoxia ( of 50 mmHg) abolishing this difference through proportionally greater hypoxia-induced increases in total and peripheral chemoreceptor sensitivities in highlanders. Our results contradict the findings of Fatemian et al. (2003) in that we found the central chemoreceptor drive to breathe to be depressed in highlanders compared to lowlanders. These differences could potentially be attributed to the different high altitude populations studied (Himalayans in our study vs. Andeans in the study of Fatemian et al. 2003), to the different levels of hyperoxia used ( 150 mmHg in ours vs. 100 mmHg in their study) and to the potential confounding effect of CBF on the measured ventilatory sensitivity of the central chemoreceptor (Berkenbosch et al. 1989; Mohan et al. 1999). Fatemian et al. (2003) did not measure CBF in their study, but if their highlanders had an impaired cerebrovascular response, as shown in other studies (Norcliffe et al. 2005), then the washout of and subsequent decrease in [H⁺] at the central chemoreceptors would be decreased, leading to higher measured central chemoreflex sensitivity.

Ventilatory response to hypoxia

Ventilatory response to hypoxia in highlanders was markedly different from that in lowlanders. Unlike lowlanders, who responded to hypoxia by increasing the sensitivity of their ventilatory response to CO₂ (Mohan & Duffin, 1997), the highlanders seemed to decrease their VRT in response to hypoxia with no change in the sensitivity to CO₂. It should be noted, however, that highlanders displayed marked heterogeneity in their ventilatory response to hypoxia, from which we identified three distinct patterns (Fig. 6). While most highlanders seemed to decrease their VRT in response to hypoxia (Pattern B – the only change that was statistically significant), some also increased their basal ventilation without a change in sensitivity or VRT (Pattern A), while others increased their sensitivity in addition to a change in VRT (Pattern C). Our limited sample size prevents us from commenting on whether these observations represent certain subsets within the highlander population, but it demonstrates that the ventilatory response to hypoxia in highlanders may be heterogeneous, which could further confound interpretation of ventilatory control studies. More studies with larger sample sizes are required to further elucidate the nature of this heterogeneity.

Figure 6. Heterogeneity of the observed responses to hypoxia in highlanders.

Three patterns of ventilatory response to hypoxia were observed in the highlander population and these are represented by the top panel of graphs. The bottom panel represents individual (dotted) and mean (continuous) responses to hypoxia in basal ventilation (), ventilatory recruitment threshold (VRT) and ventilatory sensitivity to CO₂ () among all highlanders, with (*) indicating statistically significant differences (P < 0.05). Pattern A was characterized by an increase in basal ventilation during hypoxic rebreathing, with no change in ventilatory recruitment threshold (VRT) or sensitivity to CO₂ (). Pattern B was characterized by a decrease in VRT with hypoxia, and no change in or . Pattern C was characterized by increase in with hypoxia, and no change in or VRT. Note that only the decrease in VRT with hypoxia (pattern B) was statistically significant across all subjects.

The ventilatory response to hypoxia in sea-level residents is mediated by peripheral chemoreceptors via an increase in the peripheral chemoreceptor sensitivity to [H⁺] (Cunningham, 1987; Torrance, 1996; Kumar & Bin-Jaliah, 2007). A lack of increase in ventilatory sensitivity to CO₂ with induction of hypoxia in highlanders suggests that their peripheral chemoreceptors are relatively insensitive to CO₂. However, the hypoxia-induced decrease in VRT in highlanders suggests that their ventilation is sensitive to hypoxia, albeit by a different mechanism than in lowlanders. One possible mechanism is a complete lack of peripheral chemoreceptor responsiveness in highlanders due to dysfunctional carotid bodies. In this case, hypoxia might induce a central lactic acidosis that would alter the relationship between [H⁺] and such that a given [H⁺] concentration would now be achieved at a lower , shifting the VRT to a lower value (Duffin, 2005). On the other hand, considering that the peripheral chemoreceptors and carotid bodies are intact in highlanders, the peripheral chemoreceptors may respond to hypoxia via CO₂-independent mechanisms by increasing their overall activity and shifting the VRT to lower values (Duffin, 2005). Other possible mechanisms of CO₂-independent peripheral responses to hypoxia may include a hypoxia-induced increase in the carotid body tonic drive to breathe, changes in systemic hormonal mediators, an altered cerebral spinal fluid-buffering capacity at the central chemoreceptor or an alteration in cerebral vascular reactivity leading to a higher [H⁺] at central chemoreceptor. What specific changes in carotid body function have occurred in highlanders to produce the observed differences in the chemoreflex response to hypoxia is beyond the scope of this investigation; however, other studies have shown that cellular and molecular mechanisms play a major role in adaptation to chronic hypoxia in many species (Powell, 2007), and that the carotid bodies of highlanders have altered gene expression and hypertrophy of carotid body glomus cells (Arias-Stella & Valcarcel, 1976; Kay & Laidler, 1977; Heath et al. 1985; Khan et al. 1988; Lahiri et al. 2000).

Decreased central and absent peripheral chemoreflex sensitivities to CO₂ in highlanders compared to lowlanders could account for the previously observed blunting of the HVR in highlanders (Moore, 2000; Leon-Velarde & Richalet, 2006). As seen from Fig. 5, the difference between hypoxic and hyperoxic rebreathing lines is less in highlanders than in lowlanders, which could account for a lower HVR in highlanders measured during previous studies. However, this is only an extrapolation of our results, as we did not measure HVR directly during the present study.

Genetic versus environmental considerations

An important question that arises from the current study is whether the observed differences in the control of breathing between Himalayan highlanders and Caucasian lowlanders residing at sea-level are due to genetic mechanisms or environmental factors. There is a mounting body of evidence suggesting that genetic mechanisms have an effect on the phenotype of ventilatory responses to hypoxia. For example, it has been shown that blunting of the HVR in Andean highlanders is linked to the degree of native Indian (Quechua) ancestry (Brutsaert et al. 2005), that certain HVR traits are heritable (Beall et al. 1997), and that the offspring of Tibetan and Chinese parents resemble the Tibetan population in terms of their resting ventilation and the Chinese population in terms of their HVR (Curran et al. 1997).

However, there is also evidence that environmental factors including altitude (Curran et al. 1995) and length of residence at high altitude (Lahiri, 1980) determine the magnitude of the HVR in Tibetan and Andean highlanders, although a more recent study (Beall et al. 1997) failed to identify an association between the length of life at high altitude and the HVR. Furthermore, although older studies reported sustained blunting of HVR in Andean highlanders when they moved to sea-level (Milledge & Lahiri, 1967; Sorensen & Severinghaus, 1968b; Lahiri et al. 1969), more recent studies suggest that blunting of the HVR in highlanders is at least partially reversed after at least 5 years residence at sea-level (Vargas et al. 1998), indicating plasticity of the mechanisms responsible for the HVR blunting.

There is therefore evidence for both genetic and environmental factors determining the control of breathing in highlanders. Although specific genes responsible for adaptation to high altitude have yet to be identified, there is evidence that the expression of hypoxic genes is probably related to the duration of high altitude residence in a given population, such that Tibetan highlanders, who have resided at high altitude for at least 25,000 years compared to 11,000 years for their Andean counterparts (Aldenderfer, 2003), had higher hypoxic gene expression than Andeans residing at similar altitude (Xing et al. 2008). Our study was not intended to discern the relative contribution of environmental and genetic factors on the control of breathing in highlanders, but rather to observe any phenotypic differences in the control of breathing between Himalayan highlanders and Caucasian lowlanders residing at sea-level. Whether the identified differences are due to environmental or genetic factors requires further evaluation with dedicated studies that examine both control of breathing parameters using Duffin's rebreathing method and genetic factors.

Methodological considerations

Previous control of breathing studies in highlanders used HVR and HCVR methods, and are briefly summarized below. More detailed discussion about these methods can be found elsewhere (Duffin, 2007).

HVR methods can be divided into isocapnic and non-isocapnic methods. Isocapnic methods clamp and allow the isolated effect of hypoxia on ventilation to be studied. However, these methods have several disadvantages. First, they take more time to complete, and since the ventilatory response to hypoxia varies with the length of hypoxic exposure (Powell et al. 1998), the measured ventilatory responses may be confounded by the ventilatory decline that occurs minutes after hypoxic exposure (Moore, 2000). However, recent studies suggest that the decline does not occur in the 5–7 min required to complete an isocapnic test (Huang et al. 1984). Second, the chosen level of isocapnia may affect the magnitude of the HVR (see Fig. 3) (Duffin, 2007). Third, the increase in CBF secondary to hypoxia or hypercapnia will washout the and lower [H⁺] at the central chemoreceptor, attenuating the measured ventilatory response (Berkenbosch et al. 1989; Mohan et al. 1999). Furthermore, the CBF reactivity affects the magnitude of this attenuation, and CBF reactivity to vasodilating stimuli has been shown to be different between different highlander populations (Norcliffe et al. 2005; Appenzeller et al. 2006; Claydon et al. 2008).

HCVR methods can be separated into steady-state and rebreathing methods, with the former measuring the ventilatory response to progressively increasing isocapnic CO₂ levels, and the latter measuring the ventilatory response to continuously rising . Rebreathing tests have an advantage over steady-state methods, because the difference between arterial and tissue is probably reduced during rebreathing (Read, 1967; Read & Leigh, 1967), so that the effect of changes in CBF on [H⁺] at the central chemoreceptor are reduced (Berkenbosch et al. 1989; Mohan et al. 1999).

In the present study, we used the Duffin's rebreathing technique to compare individual components of control of the breathing model (Fig. 1) between Himalayan highlanders and lowlanders at sea-level. Unlike previous HVR and HCVR methods, Duffin's technique allowed for independent measurement of the central and peripheral chemoreflex drives, and the non-chemoreflex drives to breathe and comparison of these between Himalayan highlanders and sea-level lowlanders – two populations living in their native environments. Measurement of the change in control of breathing with sustained (hours to days) hypoxic exposure in these populations was beyond the scope of this study.

As mentioned earlier in the Methods section, in Duffin's technique, rebreathing is carried out against a background of both hyperoxia ( of 150 mmHg) and hypoxia ( of 50 mmHg). Hyperoxia blunts peripheral chemoreflexes and therefore emphasises measurement of the central chemoreflex response, while hypoxic rebreathing measures the combined central and peripheral chemoreflex responses. In sea-level lowlanders, hypoxic rebreathing both increases the slope of the ventilation–CO₂ relationship and shifts the VRT to the left compared to hyperoxic rebreathing (Mohan & Duffin, 1997). The change in VRT with hypoxia is thought to be brought about by the increase in slope rather than an increased tonic drive from the peripheral chemoreceptors according to the current model of the chemoreflexes in lowlanders (Duffin, 2005; Duffin, 2007). As the threshold of the peripheral chemoreceptors is below the chemoreflex drive threshold in lowlanders the increased slope of the peripheral response to CO₂ in hypoxia results in a decreased VRT. Further detailed discussion of this phenomenon can be found elsewhere (Duffin et al. 2000).

Future directions

In future studies, it will be important to compare the present results to those from other high altitude populations, including Andean and Ethiopian highlanders, as well as acclimatizing lowlanders at different stages of acclimatization (Moore, 2000). Control of breathing parameters should also be evaluated in highlanders with a varying degree of ancestry to look for correlations (Brutsaert et al. 2005), as well as compared between generations to determine heritability (Curran et al. 1997). Lastly, the interaction between ventilation and other physiological parameters, such as cerebral blood flow, should also be considered (Ainslie & Duffin, 2009).

Relevance of our findings

This is the first study to examine the individual components of the Oxford control of breathing model (Fig. 1) in highlanders. We showed that the chemoreflex drives to breathe of Himalayan highlanders differ from those of lowlanders, while their non-chemoreflex drives to breathe are similar. Specifically, we showed that Himalayan highlanders have decreased central and absent peripheral chemoreceptor sensitivity to CO₂, and that they are sensitive to hypoxia, albeit via a different mechanism than that observed in lowlanders at sea-level. A blunted central and an absent peripheral ventilatory sensitivity to CO₂ in Himalayan highlanders may stabilize their ventilatory controller by reducing the overall gain in the feedback part of the controller circuit, thereby reducing altitude-related breathing instability (Ainslie & Duffin, 2009) Furthermore, reduced CO₂ sensitivity may reduce their work of breathing and sensation of dyspnoea, providing an energy-conserving adaptation in highlanders (Moore, 2000). However, this adaptation should be compensated for by other adaptations in the O₂ delivery chain to ensure that O₂ delivery to the tissues is maintained, such as changes in O₂ carrying capacity of the blood, blood flow to the organs, oxygen extraction, and energy production and utilization at the tissues (Moore, 2000; Beall, 2007).

Conclusion

Himalayan highlanders have an altered control of breathing relative to lowlanders residing at sea-level. They have lower ventilatory recruitment thresholds, decreased central chemoreflex sensitivity to CO₂, and little or no peripheral chemoreflex sensitivity to CO₂. The latter finding may account for the previous findings of decreased HVR among highlanders. Hypoxia causes a decrease in the ventilatory recruitment threshold in Himalayan highlanders without a concomitant increase in the sensitivity to CO₂ that is seen in lowlanders, suggesting a CO₂-independent O₂ sensing mechanism by the peripheral chemoreceptor, or an increased central chemoreceptor drive secondary to hypoxia-induced central lactic acidosis. The non-chemoreflex drives to breathe were similar between Himalayan highlanders and sea-level lowlanders.

Glossary

Abbreviations

CBF

cerebral blood flow

HCVR

hypercapnic ventilatory response

HVR

hypoxic ventilatory response

arterial

end-tidal

end-tidal

sub-ventilatory recruitment threshold basal ventilation

ventilatory sensitivity to CO₂

VRT

ventilatory recruitment threshold

Author contribution

All the authors were involved in all aspects of this work and all authors approved the final version of the manuscript.