Original Paper

Cerebral hypoperfusion modifies the respiratory chemoreflex during orthostatic stress

Shigehiko Ogoh, Hidehiro Nakahara, Kazunobu Okazaki, Damian M. Bailey, Tadayoshi Miyamoto

Abstract

The respiratory chemoreflex is known to be modified during orthostatic stress although the underlying mechanisms remain to be established. To determine the potential role of cerebral hypoperfusion, we examined the relationship between changes in MCA Vmean (middle cerebral artery mean blood velocity) and V̇E (pulmonary minute ventilation) from supine control to LBNP (lower body negative pressure; −45mmHg) at different CO2 levels (0, 3.5 and 5% CO2). The regression line of the linear relationship between V̇E and PETCO2 (end-tidal CO2) shifted leftwards during orthostatic stress without any change in sensitivity (1.36±0.27 l/min per mmHg at supine to 1.06±0.21 l/min per mmHg during LBNP; P=0.087). In contrast, the relationship between MCA Vmean and PETCO2 was not shifted by LBNP-induced changes in PETCO2. However, changes in V̇E from rest to LBNP were more related to changes in MCA Vmean than changes in PETCO2. These findings demonstrate for the first time that postural reductions in CBF (cerebral blood flow) modified the central respiratory chemoreflex by moving its operating point. An orthostatically induced decrease in CBF probably attenuated the ‘washout’ of CO2 from the brain causing hyperpnoea following activation of the central chemoreflex.

  • arterial blood pressure
  • end-tidal CO2
  • middle cerebral artery blood velocity
  • lower body negative pressure (LBNP)

CLINICAL PERSPECTIVES

  • Autonomic dysfunction and impaired cerebral autoregulation are characteristic features of patients suffering from the combined effects of hypercapnia and hypoxaemia, including those with COPD and OSA. Furthermore, cerebral vasodilator responses to hypoxia and hypercapnia have been shown to be attenuated in these patients, who are at an increased risk of developing stroke.

  • The present findings have identified that orthostatic stress modified respiratory function due primarily to changes in CBF. During orthostatic stress, the reduction in CBF has the potential to stimulate V̇E through activation of the central chemoreflex, which we suggest may be due to attenuation of CO2 ‘washout’.

  • Although our findings are confined to healthy subjects free of any underlying pathology, they nonetheless provide important mechanistic insight that can be translated to the ‘at-risk’ patient, in particular those suffering with COPD and OSA.

INTRODUCTION

Orthostatic-stress-induced hypocapnia [14] is expected to decrease V̇E (pulmonary minute ventilation) via the central controller [controlling element; the response of ventilation to PaCO2 (arterial partial pressure of CO2)] of the respiratory chemoreflex; however, V̇E is slightly increased rather than decreased [14], suggesting that orthostatic stress re-sets the central controller to respond to lower PaCO2 or PETCO2 (end-tidal CO2). On the other hand, orthostatic stress decreases CBF (cerebral blood flow) [47]. The cerebrovasculature is exquisitely sensitive to PaCO2 [8,9], thus the decrease in CBF from supine to upright positions seems to be explained partially by hypocapnia [10,11]. However, the reduction in PaCO2 cannot fully explain the decrease in CBF [2,12,13]. Thus the precise mechanisms underlying postural cerebral hypoperfusion and its potential modification by the chemoreflex continue to remain unclear.

From a clinical perspective, autonomic dysfunction and impaired cerebral autoregulation are characteristic features of patients suffering from COPD (chronic obstructive pulmonary disease), who are typically hypercapnic and hypoxaemic [14]. Furthermore, cerebral vasodilatory responses to hypoxia and hypercapnia have been shown to be attenuated [15] and these patients have an increased risk of stroke [16]. Collectively, these findings highlight an intimate relationship between the respiratory system and the cerebral circulation. For example, a decrease in CBF reduces the diffusion of CO2 from the cerebrospinal fluid and brain extracellular fluid to the cerebral vessels causing H+ ions to increase at the level of the central chemoreceptors [1719]. Thus unexpected hyperventilation during orthostatic stress may be related to cerebral hypoperfusion, thereby enhancing central respiratory drive.

Therefore, in light of the existing controversy, we hypothesized that postural reductions in CBF would modify the central ventilatory chemoreflex (i.e. alter the operating point of the V̇EPETCO2 relationship). To test this hypothesis, we investigated the effect of a decrease in MCA Vmean (middle cerebral artery mean blood velocity) on the central controller of ventilatory chemoreflex at different CO2 levels from supine to orthostatic stress [LBNP (lower body negative pressure); −45 mmHg].

MATERIALS AND METHODS

Subjects

A total of nine healthy men, aged 24±1 years old and weighing 67±2 kg (values are means±S.E.M.), volunteered for the present study. Each subject provided written informed consent after all of the potential risks and procedures were explained. All experimental procedures and protocols conformed to the Declaration of Helsinki and were approved by the Human Subjects Committee of Morinomiya University of Medical Sciences (No. 016). The subjects were considered physically active but not trained since they did not perform endurance training on a regular basis (<5 h/week) [20]. In addition, they were free of any known cardiovascular and pulmonary disorders, and were not using any prescribed medication. Before the formal experimentation, each subject was familiarized with the techniques and procedures. They were requested to abstain from caffeinated beverages for 12 h and from strenuous physical activity and alcohol for at least 24 h before the day of the experiment.

Measurements

All studies were performed at a room temperature between 23–24°C with minimal external stimuli. HR (heart rate) was monitored using a lead II ECG. Beat-to-beat ABP (arterial blood pressure) was measured via a tonometer placed over the left radial artery (BP-608; Omron). The MCA blood velocity was measured using TCD (transcranial Doppler) ultrasonography (WAKI; Atys Medical). A 2-MHz Doppler probe was placed over the temporal ultrasound window and fixed with an adjustable headband and adhesive ultrasonic gel (Tensive; Parker Laboratories). To gain an optimum Doppler signal, the position and angle of the TCD probe was first adjusted at the same depth for all subjects; optimization of the gain and power intensity of the signal was then modified accordingly for each subject. Ventilatory responses were measured breath-by-breath using an open-circuit apparatus (ARCO2000-MET; Arcosystem). The subjects breathed through a face mask attached to a low-resistance one-way valve with a flow meter. The valve mechanism allowed subjects to inspire room air or a gas mixture from a 200 litre Douglas bag containing 0.0, 3.5 or 5.0% CO2 in 40% O2 with N2 balance. High O2 concentrations in the inspiration gas avoids potentially confounding (respiratory) influences of the peripheral chemoreflex [21]. The total instrumental dead space was 200 ml. Respiratory and metabolic data were recorded using an automatic breath-by-breath gas analyser that housed a differential pressure transducer, sampling tube, filter, suction pump and mass spectrometer (ARCO2000-MET; Arcosystem). We digitized expired flow, and CO2 and O2 concentrations, and derived VT (tidal volume), V̇E and PETCO2. Flow signals were matched to gas concentrations identified as single breaths using the peak PETCO2 after accounting for the time delay (350 ms) in gas concentration measurements. The corresponding O2 uptake and CO2 output values for each breath were calculated from inspired–expired gas concentration differences and by expired ventilation, with inspired ventilation calculated by N2 correction. During each protocol, HR, V̇E, PETCO2 and MCA V were recorded continuously at 200 Hz.

Experimental Protocol

On the experimental day the subjects arrived at the laboratory at least 2 h after a light meal. After instrumentation, the subjects were placed in an LBNP box. To characterize MCA V and respiratory responses to hypercapnia, the subjects breathed through a face mask and inspired a selected gas mixture from a 200 litre Douglas bag containing 0.0, 3.5 or 5.0% CO2, 40% O2 balanced with N2 [FiCO2 (fraction of inspired CO2)=0.000, 0.035 and 0.050] at rest and during LBNP (−45 mmHg). A total of 5 min of baseline data were recorded at supine rest or during LBNP while the subjects breathed from the Douglas bag containing 0% CO2, 40% O2 balanced with N2. After baseline recordings, each hypercapnia trial was induced by rapidly changing the FiCO2 and lasted for 8 min. This duration is sufficiently long to permit CO2 to reach its new steady state at the level of the central chemoreceptors [22]. During the interval period between experimental trials (>20 min), the subjects inspired room air. The order of the rest and LBNP trials was randomized for each subject (Figure 1).

Figure 1 Experimental protocol under control and LBNP conditions

The order of trials was randomized and the interval period between trials was more than 20 min.

Data analysis

Respiratory controller

The central chemoreflex was identified by linear regression analysis between PETCO2 and V̇E. To characterize the controller at rest and during LBNP, we used a protocol of CO2 administration (three levels; FiCO2, 0, 3.5, 5.0%), a conventional linear equation, V̇E=S·(PETCO2B), and determined the slope S and intercept B using least-squares regression. The slope S reflects the sensitivity of the controller [23].

Effect of changes in CO2 on CBF

Cerebrovascular reactivity to changes in CO2 was identified by exponential and linear regression analysis between changes in PETCO2 and MCA Vmean. To characterize cerebrovascular reactivity to CO2 at rest and during LBNP, we used a protocol of CO2 administration (three levels; FiCO2, 0, 3.5, 5.0%), an exponential function, MCA Vmean=K·exp(R·PETCO2), and determined the values K and R [9,25]. In addition, the cerebral CO2 reactivity was calculated by linear model (%MCA Vmean=k+α·PETCO2) [25]. The values R and α also reflect cerebral CO2 reactivity.

Effect of LBNP-induced changes in PETCO2 or MCA Vmean on V̇E

The relationship between changes in V̇E, PETCO2 and MCA Vmean from supine rest to LBNP was described using simple and multiple linear regression analysis for each subject. In these analyses, we used the recorded data (1 min average from 3–8 min steady-state period at each condition; 15 data points for one subject). The MCA Vmean response to changes in CO2 attained a steady state within 10 s. In contrast, PETCO2 did not reach steady state within this time frame (see Figure 2), which may have influenced V̇E. Thus, given the inevitable differences in on-kinetics, we chose to exclude the 0–3 min period (to ensure that PETCO2 had indeed attained a steady state) to identify the inter-relationships between V̇E and CO2 or MCA Vmean. The data from these regression analyses show the time-dependent variability of changes in MCA Vmean from supine to LBNP on changes in V̇E at steady state. In multiple linear regression analysis, we only used MCA Vmean and PETCO2 as dependent variables, since MAP (mean arterial pressure) was unchanged by LBNP (P=0.584) or HR and thus they were unlikely to have had an effect on the central respiratory chemoreflex.

Figure 2 Continuous recordings of the changes in PETCO2, V̇E, MCA Vmean during supine (solid line) and LBNP (broken line) at 0% (control) (left-hand panel) and 5% (right-hand panel) CO2 administration in a representative subject

Statistical analysis

All analyses were conducted using SigmaStat (Jandel Scientific Software; SPSS). A normal distribution was confirmed via repeated Shapiro–Wilk W tests. Data were subsequently analysed using a repeated-measures two-way ANOVA [condition: control compared with LBNP×FiCO2 (0% compared with 3.5% compared with 5% CO2)] and post-hoc Student–Newman–Keul tests. The relationship between changes in V̇E, PETCO2 and MCA Vmean from supine rest to LBNP was described using simple and multiple linear regression analysis. Results are expressed as means±S.E.M., and significance for all two-tailed tests was set at P<0.05.

RESULTS

PETCO2 was manipulated well by different respiratory CO2 levels, whereas LBNP decreased PETCO2 with the exception of the 5% trial (P<0.05; Table 1 and Figure 2). In both conditions, HR was unaffected by FiCO2, whereas LBNP caused tachycardia. MAP was unchanged by both LBNP and different respiratory CO2 levels. MCA Vmean was decreased during LBNP at each CO2 level. In contrast, V̇E was increased by increases in FiCO2, whereas V̇E tended to increase during LBNP at each CO2 level compared with the supine control condition. VT increased in response to an increase in FiCO2, whereas this was blunted during LBNP.

View this table:
Table 1 Haemodynamic and respiratory responses to changes in CO2 and orthostatic stress (LBNP)

Values are means±S.E.M. *P<0.05 compared with control; †P<0.05 compared with 0% CO2; ‡P<0.05 compared with 3.5% CO2.

The group-averaged data of the relationship between PETCO2 and V̇E are shown in Figure 3. V̇E was linearly related to PETCO2 during the supine control (r2=0.976) and LBNP (r2=0.953) conditions, whereas the LBNP condition decreased the PETCO2 intercept (B) (Table 2; P=0.012). LBNP induced a leftward shift in the V̇EPETCO2 relationship without changes in controller sensitivity (S) (1.36±0.27 l/min per mmHg for control compared with 1.06±0.21 l/min per mmHg for LBNP; P=0.087).

Figure 3 Characteristics of the central controller of respiratory chemoreflex (V̇EPETCO2 relationship) derived from the averaged data at control supine (○) and during LBNP (●)
View this table:
Table 2 Characteristics of respiratory chemoreflex controller and cerebrovascular reactivity at rest and during LBNP

Values are means±S.E.M. Central controller: V̇E=S·(PETCO2B); cerebrovascular reactivity to CO2: exponential model, MCA Vmean=exp(R·PETCO2); linear model, %MCA Vmean=k+α·PETCO2. *P<0.05 compared with cotnrol.

Hypercapnia resulted in an exponential increase in MCA Vmean during LBNP (r2=0.99), as well as at control supine (r2=0.93; Figure 4). However, both R (P=0.279) and K (P=0.232) of the cerebrovascular CO2 reactivity exponential curves (PETCO2–MCA Vmean relationship) were unchanged during LBNP (Table 2 and Figure 4). In addition, cerebrovascular CO2 reactivity calculated from the linear model during LBNP was similar to control (2.5±0.3%/mmHg in control compared with 2.4±0.4%/mmHg in LBNP; P=301). These findings indicate that the functional curve of cerebral CO2 reactivity was not shifted during LBNP.

Figure 4 Characteristics of cerebrovascular CO2 reactivity (PETCO2–MCA Vmean relationship) derived from the averaged data at control supine (○) and during LBNP (●)

Figure 5 shows the typical (individual) relationship between changes in V̇E and PETCO2 or MCA Vmean from supine to LBNP. Changes in V̇E were not associated with changes in PETCO2 (P=0.511); in contrast, changes in V̇E were inversely associated with changes in MCA Vmean from control to LBNP (P=0.008). Similarly, in simple linear regression analysis, correlation between changes in MCA Vmean and V̇E from supine to LBNP was observed in all subjects (P=0.007–0.044), whereas only three subjects had a correlation between PETCO2 and V̇E (Table 3). This was confirmed further by multiple linear regression analysis. In six subjects, the changes in V̇E were related to the changes in MCA Vmean, but not to changes in PETCO2, whereas in one subject the changes in V̇E were related to the changes in PETCO2

View this table:
Table 3 Simple or multiple linear regression analysis between LBNP-induced changes in V̇E and PETCO2 or MCA Vmean

P values are shown, with asterisks representing significant relationships (P<0.05).

DISCUSSION

The present study is the first of its kind to address how changes in CBF influence the respiratory controller during LBNP-induced orthostatic stress in humans. Two major findings were apparent. First, orthostatic stress was shown to cause a leftward shift in the V̇EPETCO2 relationship, thereby suggesting that the operating point of the central controller of the respiratory chemoreflex had been modified. Secondly, this shift appeared to be more related to changes in MCA Vmean as opposed to changes in PETCO2. In combination, these findings suggest that, during orthostatic stress, the reduction in CBF has the potential to stimulate V̇E through activation of the central chemoreflex, which we speculate may be due to attenuation of CO2 ‘washout’.

Cerebral haemodynamic function

The findings of our present study have generally confirmed previous findings [2,26,27] in that, as anticipated, MCA Vmean was decreased during orthostatic stress, i.e. LBNP, head-up tilt and standing-up during normocapnia. It has been suggested that hyperventilation-induced hypocapnia and associated cerebral vasoconstriction may be responsible, at least in part, for the cerebral hypoperfusion observed during the transition from supine to the free-standing upright position [1,11]. However, this remains equivocal given that other investigators have observed an apparent dissociation between the decrease in PaCO2 and CBF during orthostatic stress, suggesting that the postural decrease in CBF is unlikely to be accounted for by cerebral CO2 reactivity alone [2,13]. For example, Serrador et al. [13] demonstrated that the decline in MCA Vmean was not related to hypocapnia during 10 min of head-up tilt. In the present study, LBNP shifted cerebrovascular CO2 reactivity leftwards without affecting both K and R, indicating that the decrease in CBF is independent of changes in CO2. These findings strongly suggest that hyperventilation-induced hypocapnia is unlikely to be the primary mechanism underlying cerebral hypoperfusion.

One possible mechanism that may contribute to altered cerebral perfusion relates to the hydrostatic effects imposed by gravity. We [28] have previously examined the relationship between cardiac output and MCA Vmean by differentially manipulating cardiac output by LBNP (reduction) and plasma volume expansion (increase). We identified that the cardiac output associated with the changes in central blood volume influence MCA Vmean at rest and during exercise, and its regulation is independent of cerebral autoregulation. Therefore orthostatic-stress (head-up tilt)-induced central hypovolaemia [29] may be related to the decline in CBF during LBNP. These findings clearly emphasize that the fundamental mechanisms underlying orthostatically induced cerebral hypoperfusion remain to be established.

Figure 5 Representative correlational analyses between changes in V̇E and PETCO2 (left-hand panel), and changes in V̇E and MCA Vmean (right-hand panel) from control supine to LBNP

Results are taken from subject 1.

Respiratory function: ventilation and PETCO2

During the transition from supine to the upright position, PETCO2 decreased subsequent to an increase in V̇E [10]. Similarly, in the present study, V̇E increased during LBNP with a reduction in PETCO2 (Figure 2). More importantly, changes in V̇E were shown to be unrelated to changes in PETCO2 during LBNP across all CO2 conditions, indicating that orthostatically induced hyperventilation could not be explained by the central chemoreflex (Figure 3). These findings raise the possibility that orthostatic stress modifies the respiratory chemoreflex, although the precise mechanisms remain to be established.

One such mechanism that may have modified the controller property relates to central nervous drive including afferent neural inputs from respiratory muscle mechanoreceptors [30], given the gravitational pull on the diaphragm [1,31]. The present study provides another possible mechanism regarding V̇E increases during orthostatic stress. Interestingly, changes in V̇E were associated with changes in MCA Vmean from control to LBNP rather than changes in PETCO2 (Table 3 and Figure 5).

An increase in CBF increases the diffusion of CO2 from cerebrospinal fluid and brain extracellular fluid to the cerebral vessels, causing a decrease in H+ ions at the level of the central chemoreceptors [32]. As both the cerebrovasculature and the central chemoreceptor are sensitive to CO2, CBF and the ventilatory response to CO2 seem to be tightly linked [9,1719,3335]. For example, changes in CBF might have an important role in stabilizing the breathing pattern [19]. Chapman et al. [33] reported that severe brain ischaemia blunted the ventilatory response to CO2. Clinically, respiratory dysfunction attenuates CBF regulation [14,16,36]. Thus the corollary (i.e. orthostatic reduction in CBF), as observed in the present study, may have reduced the elimination rate of CO2 from the brain thereby causing hyperpnoea subsequent to activation of the central chemoreflex. Orthostatic-induced decreases in CBF may therefore modify the central controller of respiratory chemoreflex.

Clinical implications

Patients suffering from COPD exhibit functional impairments in cardiovascular and respiratory control [37] and are characterized by neurocognitive dysfunction [38], tentatively suggesting a functional link between respiratory control abnormalities and cerebral blood flow regulation. In support of this, Bernardi et al. [14] have demonstrated that autonomic dysfunction in COPD patients was associated with impaired cerebrovascular regulation that could be restored by improving cardio- and cerebro-vascular modulation. In addition, cerebral vasodilatory responses to hypoxia and hypercapnia were attenuated in patients with OSA (obstructive sleep apnoea) [15], and these patients are at a significantly greater risk of suffering a stroke [16]. Although orthostatic-stress-induced central hypovolaemia was shown to modify respiratory function, albeit in healthy subjects free of any underlying pathology, our findings provide important insight into the fundamental mechanisms that predispose to the systemic complications observed in COPD or OSA and the potential therapeutic interventions available.

Experimental limitations

The postural reductions in PETCO2 probably overestimate the decrease in PaCO2 [2] and thus the sensitivity of the central chemoreflex or cerebral CO2 reactivity may have been underestimated. However, this is unlikely, given that our conclusions were based on the assessment of V̇E or MCA Vmean independently of PETCO2. TCD ultrasonography may be considered as an additional limitation, since it measures cerebral flow velocity and not blood flow. However, in humans, the MCA diameter appears to remain relatively constant under a variety of conditions [39,40]. Thus we would contend that beat-to-beat changes in MCA Vmean reflect changes in flow. In addition, vertebral blood flow is a more accurate reflection of the central cardiovascular and respiratory systems compared with MCA blood flow [41,42]. Finally, under closed-loop conditions, MCA Vmean did not stabilize until V̇E had achieved a steady state by a hyperventilation-induced reduction in PETCO2. Thus, given the inevitable differences in on-kinetics, we chose to exclude the 0–3 min period (to ensure that PETCO2 had indeed attained a steady state) to identify the inter-relationships (we duly acknowledge that this does not disassociate cause from effect) between V̇E, CO2 and MCA Vmean. Although the effect of acute changes in MCA Vmean during hypercapnia on V̇E is unclear from our results, an LBNP-induced reduction in MCA Vmean (steady state) was clearly associated with an increase in V̇E, which we certainly feel is worthwhile reporting.

Conclusions

In combination, these findings tentatively suggest that, during orthostatic stress, the reduction in CBF has the potential to stimulate V̇E through activation of the central chemoreflex, which may be due to attenuation of CO2 ‘washout’.

FUNDING

This study was supported, in part, by a Grant-in-Aid for Scientific Research [grant number 19500574] from the Japanese Ministry of Education, Culture, Sports, Science and Technology, the Descente and Ishimoto Memorial Foundation for the Promotion of Sports Science and the Kouzuki Foundation for Sports and Education.

AUTHOR CONTRIBUTION

Shigehiko Ogoh designed the study, performed the experiments, interpreted the data, performed the statistical analysis and wrote/revised the paper. Hidehiro Nakahara performed the experiments. Kazunobu Okazaki performed the experiments, interpreted the data and revised the paper. Tadayoshi Miyamoto developed the methods, interpreted the data and wrote/revised the paper. Damian Bailey interpreted the data and wrote/revised the paper.

Acknowledgments

We appreciate the time and effort invested by the volunteers.

Abbreviations: CBF, cerebral blood flow; COPD, chronic obstructive pulmonary disease. FiCO2, inspired fraction of CO2; HR, heart rate; LBNP, lower body negative pressure; MAP, mean arterial pressure; MCA, Vmean, middle cerebral artery mean blood velocity; OSA, obstructive sleep apnoea; PaCO2, arterial partial pressure of CO2; PETCO2, end-tidal CO2; TCD, transcranial Doppler; V̇E, pulmonary minute ventilation; VT, tidal volume

References

View Abstract