Onset responses of ventilation and cerebral blood flow to hypercapnia in humans: rest and exercise

Abstract

The respiratory and cerebrovascular reactivity to changes in arterial Pco₂ () is an important mechanism that maintains CO₂ or pH homeostasis in the brain. It remains unclear, however, how cerebrovascular CO₂ reactivity might influence the respiratory chemoreflex. The purpose of the present study was therefore to examine the interaction between onset responses of the respiratory chemoreflex and middle cerebral artery (MCA) mean blood velocity (V_mean) to hypercapnia (5.0% CO₂-40% O₂-balance N₂) at rest and during dynamic exercise (∼1.0 l/min O₂ consumption). Each onset response was evaluated using a single-exponential regression model consisting of the response time latency [CO₂-response delay (t₀)] and time constant (τ). At rest, t₀ and τ data indicated that the MCA V_mean onset response was faster than the ventilatory (V̇e) response (P < 0.001). In contrast, during exercise, t₀ of V̇e and MCA V_mean onset responses were decreased. In addition, despite the enhanced response to CO₂ administration (P = 0.014), τ of MCA V_mean tended to increase during exercise (P = 0.054), whereas τ of V̇e decreased (P = 0.015). These findings indicate that 1) at rest, faster washout of CO₂ via cerebral vasodilation results in a reduced activation of the central chemoreflex and subsequent reduced V̇e onset response, and 2) during exercise, despite higher rates of increasing , the lack of change in the onset response of cerebral blood flow and reduced washout of CO₂ may act to augment the V̇e onset response.

the regulation of pH or CO₂ is a vital homeostatic function, because numerous enzyme and ion channels that influence neural activity are modulated by changes in pH (5). In the brain, there are two strong CO₂ regulatory mechanisms: the central respiratory chemoreflex and cerebrovascular CO₂ reactivity. Respiration is tightly controlled by changes in arterial Pco₂ () via the central chemoreflex (3) to maintain CO₂ homeostasis. In addition, cerebral blood flow (CBF) and its distribution are highly sensitive to changes in (16, 18, 26, 31). This CBF response, termed cerebrovascular CO₂ reactivity, is a vital homeostatic function that helps regulate and maintain constant pH (5). Previous studies (19, 25), however, have only considered the steady-state changes in cerebrovascular CO₂ reactivity and have not identified the potential dynamic nature of this response. Early reports (30) indicated that the CBF response started within 30 s of CO₂ inhalation and that ∼2 min were required to reach peak values (9). Recent studies (23, 24), incorporating transcranial Doppler (TCD) and more sophisticated methods to manipulate end-tidal gases, have demonstrated that the CBF response to step changes in CO₂ in humans was much faster (6-s delay). Because the response of immediately after CO₂ administration is rapidly increased (3), the onset (or dynamic) CBF response to changes in CO₂ is likely to be an important mechanism in the maintenance of CO₂ homeostasis in the brain.

Cerebrovascular CO₂ reactivity seems to influence the ventilatory response to CO₂ at rest (1, 4, 6, 21, 33, 34) and during exercise (19). However, these previous findings were based on the CBF (or CBF velocity) response to a steady-state change in . Therefore, it remains unclear how the onset (or dynamic) CBF response to changes in CO₂ might modulate dynamic ventilatory control via the respiratory chemoreflex, especially during exercise. The ventilatory- onset response is reported to be enhanced by exercise (17). In addition, exercise-induced hyperpnea may alter the ventilatory onset response and the subsequent interaction with CBF. Nevertheless, the dynamic response of CBF to changes in , especially during exercise, and the subsequent integration with the respiratory chemoreflex have not been characterized.

The purpose of this study was therefore to examine the onset responses of the respiratory chemoreflex and middle cerebral artery (MCA) mean blood velocity (V_mean) to hypercapnia at rest and during exercise. We examined the hypothesis that the onset response of CBF influences respiratory control via a central chemoreflex and exercise modifies the interaction between CBF and ventilatory onset responses.

METHODS

Seven healthy men were recruited to participate in the study as approved by the Human Subjects Committee of Morinomiya University of Medical Sciences (no. 001). The subjects were not performing endurance training on a regular basis. In addition, they were free of any known cardiovascular and pulmonary disorders and were not using any prescribed or over-the-counter medications. Before the experiment, each subject gave written informed consent and visited the laboratory for familiarization with the techniques and procedures. Subjects were requested to abstain from caffeinated beverages for 12 h and strenuous physical activity and alcohol for ≥24 h before the day of the experiment.

Measurements

All studies were performed at room temperature (23–24°C) with minimal external stimuli. A diagram of the experimental setup is presented in Fig. 1A. Heart rate (HR) was monitored using a lead II electrocardiogram. The MCA blood velocity was obtained by TCD ultrasonography (WAKI, Atys Medical, St. Genislaval, France). 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, Orange, NJ). To gain an optimum Doppler signal, the position and angle of the TCD probe were first adjusted at the same depth (5 cm from the skin surface of the temple window) for all subjects; optimization of the gain and power intensity of the signal was then modified accordingly for each subject. Ventilatory responses were measured using an open-circuit apparatus. The subjects breathed through a face mask attached to a low-resistance one-way valve with a flowmeter. The valve mechanism allowed subjects to inspire room air or a selected gas mixture from a 200-liter Douglas bag containing 0.0 or 5.0% CO₂-40% O₂-balance N₂. High O₂ concentration in the inspiration gas avoids peripheral chemoreflex influence (10). The total instrumental dead space was 200 ml. Respiratory and metabolic data during the experiments were recorded by an automatic breath-by-breath gas-analyzing system consisting of a differential pressure transducer, sampling tube, filter, suction pump, and mass spectrometer (model ARCO2000-MET, Arcosystem, Chiba, Japan). We digitized expired flow and CO₂ and O₂ concentrations and derived tidal volume (Vt), minute ventilation (V̇e), and end-tidal Po₂ and Pco₂ . We computed flow signals to single breath data and matched them to gas concentrations identified as single breaths using the peak , after accounting for the time delay in gas concentration measurements. The corresponding O₂ uptake (V̇o₂) and CO₂ output values for each breath were calculated from inspired-expired gas concentration differences and by expired ventilation (V̇e), with inspired ventilation being calculated by N₂ correction. During each protocol, HR, V̇e, , , and MCA blood velocity (V) were recorded continuously at 200 Hz.

Fig. 1.

A: experimental setup. Each subject breathed through a face mask attached to a low-resistance 1-way valve with a built-in hot-wire flowmeter. Valve mechanism allowed subjects to inspire room air or a selected gas mixture from a 200-liter Douglas bag. TCD, transcranial Doppler; , end-tidal Pco₂; , end-tidal Po₂; V̇e, minute ventilation; Vt, tidal volume. B: experimental protocol. CO₂ was administered to each subject at rest (i) and during exercise (ii). Order of the rest and exercise trials was randomized for each subject. V̇o₂, O₂ uptake.

Experimental Protocol

Exercise capacity.

On day 1, the subjects performed maximal cycling exercise for the measurement of maximal V̇o₂ (V̇o_2max). V̇o_2max was assessed with an incremental protocol on a cycle ergometer (model Corival1000SS, Lode, Groningen, Holland). The workload was set at 20 W and increased by 20 W every minute until the subjects could no longer maintain the pedaling frequency of ≥60 rpm, despite strong verbal encouragement. The subjects breathed through a face mask attached to a volume transducer, while gases were continuously sampled for analysis of fractional concentrations of O₂, CO₂, and N₂. The respiratory gas analysis system was calibrated before each test using known standard gases.

CO₂ administration protocol.

On the experimental day, the subjects arrived at the laboratory ≥2 h after a light meal. A diagram of the experimental protocol is presented in Fig. 1B. After instrumentation, the subjects sat on a comfortable chair. For characterization of dynamic CBF and respiratory responses to hypercapnia, the subjects breathed through a face mask and inspired a selected gas mixture from a 200-liter Douglas bag containing 5.0% CO₂-40% O₂-balance N₂ [fraction of inspired CO₂ = 0.05] at rest and during cycling exercise (∼1.0 l/min V̇o₂). Five minutes of baseline data were recorded during room air breathing from subjects wearing the face mask. After baseline recording, each hypercapnia trial was induced by a rapid change in the and lasted for 12 min, which is long enough to permit CO₂ to reach its new steady-state value at the central chemoreceptors (22). During the interval between rest and exercise protocols, the subjects inspired room air. The order of the rest and exercise trials was randomized for each subject.

Data Analysis

From the 200-Hz sampling data, , Vt, V̇e, , and MCA V_mean were averaged at 0.2 Hz. Because is known to not accurately reflect during conditions of exercise or hypercapnia (13), was calculated from and Vt using the following equation

To address our hypothesis, onset responses of V̇e, predicted , and MCA V_mean were evaluated using a one-compartment nonlinear least-squares optimization method. The remaining data of onset responses of , V̇e, predicted , and MCA V_mean were fitted to the following single-exponential regression equation consisting of the response time latency [CO₂-response delay (t₀)], baseline value, gain term (G), and time constant (τ) fitted to the CO₂ administration protocol

where y is response, t is time, and y₀ is a baseline value. Time 0 was the beginning of CO₂ administration.

Statistical Analysis

A paired t-test was used to assess the differences in G of MCA V_mean, V̇e, and predicted between rest and exercise. Hemodynamic values and t₀, τ, and t₀ + τ were compared using a two-way repeated-measures ANOVA, and a Student-Newman-Keuls test was employed post hoc to investigate main effects and interactions. Statistical significance was set at P < 0.05. Values are means ± SE, and analyses were conducted using SigmaStat (Jandel Scientific Software, SPSS, Chicago, IL).

RESULTS

Steady-State Responses to CO₂ Administration: Rest and Exercise

Table 1 outlines the steady-state hemodynamic values at rest and during exercise before and after 5% CO₂ administration. Averaged V̇o₂ during the cycling exercise was 33 ± 6% of V̇o_2max. This mild exercise increased HR, MCA V_mean, V̇e, Vt, , and predicted (Table 1). During 5% CO₂ administration, as expected, , , and predicted were elevated at rest and during exercise. Although hypercapnia elevated V̇e and MCA V_mean at rest and during exercise, these responses were much higher during exercise than at rest (P = 0.009 for V̇e, P = 0.010 for MCA V_mean).

Table 1.

Ventilatory and hemodynamic variables before hypercapnia and after attainment of steady state at rest and during exercise

	Rest		Exercise		P
	Pre	Steady state	Pre	Steady state	Pre vs. steady state	Condition	Interaction
Fraction of inspired CO2, %	0.07±0.01	5.32±0.01	0.05±0.01	5.37±0.08	<0.001	0.604	0.356
HR, beats/min	70±6	73±6	97±8*	112±9†‡	0.002	<0.001	0.002
MCA Vmean, cm/s	51±4	64±4*	60±4*	83±4†‡	<0.001	<0.001	0.01
Minute ventilation, l/min	10.8±0.7	24.9±2.9*	23.6±1.7*	47.0±4.6†‡	<0.001	<0.001	0.009
VT, ml	910±118	1,512±196	1,357±103	2,100±113	<0.001	0.006	0.445
End-tidal PCO2, Torr	37.7±1.8	50.2±1.0	44.5±2.2	59.7±3.2	<0.001	0.003	0.183
Predicted arterial PCO2, Torr	37.4±1.4	47.5±0.8*	42.9±2.0*	54.7±2.7†‡	<0.001	0.013	0.271

Values are means ± SE. HR, heart rate; MCA V_mean, middle cerebral artery mean blood velocity; V_T, tidal volume.

^*P < 0.05 vs. Pre at rest.

^†P < 0.05 vs. steady state at rest.

^‡P < 0.05 vs. Pre during exercise.

Onset Responses to CO₂ Administration: Rest and Exercise

The onset response of to CO₂ administration was faster than that of other variables (i.e., MCA V_mean and V̇e), with no differences (P = 0.147) between rest and exercise (Fig. 2). Similarly, the response of predicted to these changes in was faster than the responses of MCA V_mean and V̇e. At rest, the average of t₀ and τ of from the exponential fitting curve was 1.1 ± 0.83 and 22.2 ± 5.1 s, respectively. Exercise did not change t₀ (1.4 ± 0.7 s, P = 0.703). In contrast, averaged τ was faster during exercise (11.9 ± 4.4 s vs. rest, P = 0.014).

Fig. 2.

A and B: continuous recordings of fraction of inspired CO₂ () and predicted arterial Pco₂ () responses to hypercapnia (5% CO₂) at rest and during exercise (Ex) in 1 representative subject. Hypercapnia was started at time 0. C: group-averaged sum of CO₂-response delay (t₀) and time constant (τ) of and predicted exponential fitting curves at rest and during exercise. Values are means ± SE. *P < 0.05 vs. . †P < 0.05 vs. rest.

At rest, the average of t₀ and τ of the MCA V_mean exponential fitting curve was 5.3 ± 1.0 and 34.3 ± 9.6 s, respectively (Fig. 3). These values were much faster than t₀ (13.9 ± 2.0 s, P < 0.001) and τ (147.7 ± 18.0 s, P < 0.001) of V̇e. In addition, averaged τ of MCA V_mean was not different from that of (P = 0.49). During exercise, t₀ of MCA V_mean was unchanged from rest (−1.7 ± 1.0 s, P = 0.235), whereas t₀ of V̇e was markedly reduced (−8.1 ± 1.6 s, P < 0.001). In addition, exercise increased τ of MCA V_mean from rest, although this did not quite reach statistical significance (+51 ± 18 s, P = 0.054). In contrast, exercise decreased τ of V̇e (−68 ± 28 s, P = 0.015). G of the exponential curves of the predicted (P = 0.147) and MCA V_mean (P = 0.104) was unchanged during exercise, whereas G of V̇e was increased from 14.6 ± 2.2 to 25.8 ± 3.3 l/min (P = 0.008; Table 2).

Fig. 3.

A and B: continuous recordings of middle cerebral artery (MCA) mean blood velocity (V_mean) and V̇e during hypercapnia at rest and during exercise in 1 representative subject. Hypercapnia was started at time 0. C and D: group-averaged t₀ and τ of MCA V_mean and V̇e exponential fitting curves at rest and during exercise. Values are means ± SE. *P < 0.05 vs. MCA V_mean. †P < 0.05 vs. rest.

Table 2.

Gain of MCA V_mean, minute ventilation, and predicted arterial PCO₂ exponential fitting curves at rest and during exercise

	Rest	Exercise	P
MCA Vmean, cm/s	16.3±2.0	19.0±2.0	0.104
Minute ventilation, l/min	14.6±2.2	25.8±3.3*	0.008
Predicted arterial PCO2, Torr	10.2±1.0	11.5±1.3	0.147

Values are means ± SE.

^*P < 0.05 vs. rest.

DISCUSSION

At rest, during hypercapnia, the MCA V_mean onset response was much faster than the V̇e onset response. However, although the V̇e onset response was augmented, exercise did not enhance the MCA V_mean onset response, despite the high rate of increasing . These findings suggest that dynamic CBF regulation is tightly linked with respiratory control via a central chemoreflex.

At rest, t₀ and τ of the MCA V_mean exponential fitting curve were much smaller than those of V̇e (P < 0.001; Fig. 3). In addition, τ did not differ between and MCA V_mean (22 ± 5 and 34 ± 10 s, respectively, P = 0.49). These findings indicate that the MCA V_mean onset response was much faster and more directly reflected changes in than the V̇e onset response via the central chemoreflex. had increased to 51 ± 8% of the steady-state value after CO₂ administration at t₀ of V̇e (14 s) at rest. Thus the fast response of CBF is potentially an important mechanism that acts to cause the time delay in the V̇e response at the onset of hypercapnic stimulation.

During exercise, the onset response of after hypercapnia was not different from that at rest (Fig. 2). Thus the differences in the , MCA V_mean, and V̇e onset responses between rest and exercise were independent of changes in . During exercise, τ of the exponential fitting curve (12 s) was significantly decreased from rest (22 s, P = 0.014), indicating that the onset response was enhanced by exercise. This enhancement was attributed not only to the increased hemodynamics of exercise, but also to the possibility that, during exercise, the improved rapid changes in respiratory gas composition were transmitted to the arterial blood with greater fidelity than at rest (32). Cerebrovascular CO₂ reactivity, especially to hypercapnia, is enhanced during steady-state exercise (19, 25). Although the onset responses of and cerebrovascular CO₂ reactivity were augmented, the onset response of MCA V_mean was not enhanced during exercise. The τ of the MCA V_mean exponential fitting curve tended to increase (P = 0.054), rather than decrease, during exercise. However, τ includes information of not only response speed, but also G, which is the difference in steady-state value between pre- and post-CO₂ administration. For example, a rise in G increases τ, despite no change in the response speed. Thus, increased τ of the MCA V_mean exponential fitting curve could be partly due to an increased G induced by an enhanced cerebral CO₂ reactivity, rather than a reduction in the response speed. Interestingly, exercise-induced enhancement of cerebral CO₂ reactivity (steady-state) (19, 25) did not correlate with the onset response of MCA V_mean (dynamic) to hypercapnia.

In contrast to rest, exercise markedly decreased the response time delay of V̇e (t₀; −8.1 s, P < 0.001). In addition, exercise decreased τ of the V̇e exponential fitting curve (−68 s, P = 0.015). These time delays are comparable to those reported in the previous studies (3, 17). Bellville et al. (3) assumed that the transport delay time to the central chemoreflex was estimated as a model parameter by curve-fitted analysis and reported that this value at rest was 13.7 ± 3.4 s. In addition, MacFarlane and Cunningham (17) reported that central chemoreflex delay during exercise was 9.9 ± 1.7 s. The shortened central chemoreflex time delay during exercise seems to be associated with the faster onset response of (Fig. 2), potentially caused by the exercise-induced fast transmission of CO₂ from the lungs to the arterial blood (32). Moreover, G of the exponential curves of V̇e was increased from 14.6 ± 2.2 to 25.8 ± 3.3 l/min (P = 0.008; Table 2). Together, these findings highlight the marked enhancement of the V̇e onset response to hypercapnia during exercise.

There have been previous conflicting reports of an augmentation of the central chemoreflex during exercise. Poon and Greene (22) showed that the slope of the V̇e-Pco₂ relationship was increased during exercise. In contrast, Asmussen and Nielsen (2) demonstrated that the V̇e-Pco₂ relationship line shifted to the left without a change in its sensitivity (or gain). The findings of the present study indicate that the V̇e- relationship was unchanged from rest to exercise (1.5 ± 0.3 vs. 2.3 ± 0.6 l·min⁻¹·Torr⁻¹). Thus, because of the lack of augmentation of the central chemoreflex during exercise, it seems unlikely that this mechanism might act to enhance the V̇e onset response.

From the results of the present study, enhancement of the V̇e onset response during exercise seems to be related to unchanged CBF onset at this time point, despite high rates of increasing . For example, the faster elevations in will activate the central chemoreflex to a greater extent during exercise than at rest because of the lack of “buffering” of brain tissue Pco₂ via cerebral vasodilation and subsequent elevations in CBF. This notion is broadly consistent with previous investigations (1, 4, 6, 21, 33, 34), which indicate that cerebral CO₂ reactivity is linked to the ventilatory response to CO₂. Peebles et al. (21) reported that hypercapnic cerebral CO₂ reactivity was inversely related to the increase in V̇e change. In other words, a reduced cerebral CO₂ reactivity results in less central CO₂ washout and a greater V̇e stimulus. This finding is consistent with the onset data of the present study.

Because the starting time point of the attenuation of the elevation in is around t₀ of V̇e (Figs. 2 and 3), it seems possible that the increase in V̇e attenuated the elevations in . The onset responses of MCA V_mean followed ; therefore, the increase in V̇e also might cause a marked attenuation of increasing MCA V_mean at rest as well as during exercise. Since hyperventilation constrains subsequent elevations in , the ventilatory chemoreflex seems to indirectly regulate MCA V_mean at rest and during exercise. Collectively, the onset responses of , MCA V_mean, and V̇e clearly indicate that CBF regulation is linked to respiratory control via the central chemoreflex.

Technical Considerations

The changes in after CO₂ administration are related to the dead space of respiratory equipment, the sampling and response time of gas analyzers, and the time in which the change in is reflected at the sampling site (i.e., the mouth), which is, in part, determined by pulmonary ventilation. Thus exercise-induced elevation of V̇e would be expected to increase the response time. However, because of the specially designed fast gas delivery system used in the present study, there was no difference in the response between rest and exercise (Fig. 2). In contrast to the response time of , the response time of differed between rest and exercise. overestimates when metabolic CO₂ production and V̇e are increased during exercise (13). In addition, cerebrovascular CO₂ reactivity is underestimated by compared with (21). Thus, to better reflect (i.e., the stimulus for ventilatory and CBF alterations), we corrected our data to on the basis of established equations (13). CO₂ is an important controlled variable or mediator, but it does not directly influence the central chemoreflex. For example, CO₂ diffuses freely to the cerebrospinal fluid and influences H⁺ concentration, which drives ventilation via central chemoreceptors (19, 28, 29). Indexes of pH regulation were not measured in the present study; however, it is unfortunately impossible in humans to assess the dynamic regulation of cerebrospinal fluid H⁺ concentration. Another potential limitation of MCA V estimation using TCD ultrasonography is that changes in the diameter of the isonated vessels could modulate MCA V independently of flow. However, the MCA diameter appears to remain relatively constant in humans during moderate variations in blood pressure and CO₂ (11), the physiological stimuli of moderate lower body negative pressure, or changes in (27). Poulin et al. (23) reported that there is little change in total power of the Doppler signal with hypercapnic stimuli, indicating that MCA diameter changes little. However, this index has never been validated against a direct method, and the use of reflected power has been dismissed as a reliable indicator of vascular size changes (7). Nevertheless, the changes in MCA V_mean during submaximal dynamic exercise appear to be similar to the changes in CBF determined by other techniques, i.e., internal carotid artery blood flow (12) and ¹³³Xe clearance (14, 15). Importantly, since determinations of cerebrovascular CO₂ reactivity are based on stimulus-response principles, absolute CBF values are not as important as reliable and repeatable recordings with short (beat-to-beat) time resolution. For this purpose, TCD is a well-suited (and, indeed, the only) technique for this type of experimental research. Finally, when CBF increases during exercise or with hypercapnia (or both), the changes in cerebral vessel diameter occur downstream of the MCA. For example, the variable resistance to flow is encountered mostly in the cerebral arteriolar and capillary bed (8). In contrast, the large cerebral arteries (e.g., MCA) and veins are dedicated to blood distribution and the storage of blood volume and are, in principle, noncompliant and act merely as a conduit for the pulsatile arterial flow from the aorta to the brain (20). Taken together, cerebral vasoconstriction and vasodilatation occur via changes in the caliber of the smaller cerebral vessels, whereas MCA diameter likely remains unchanged.

The onset response of MCA V_mean to hypercapnia was faster than that of respiratory chemoreflex. Dynamic exercise did not enhance the MCA V_mean onset response to hypercapnia, despite the greater elevation of and subsequent elevation of the ventilatory onset response. These findings indicate that the CBF onset response influences respiratory control via a central chemoreflex and exercise modifies the interaction between CBF and ventilatory onset responses. Clinically, a range of pathological conditions alter the normal chemoreflex control of breathing (e.g., chronic lung disease, heart failure, and sleep apnea) and, therefore, may alter dynamic CBF regulation. In other words, in these patient groups, any alterations in the interaction between respiratory control and CBF regulation may be associated with adverse cerebral events and/or may increase the likelihood of abnormal breathing. Further studies, however, are needed to identify the extent to which dynamic CBF regulation may be associated with respiratory dysfunction in these pathological states. Application of our described model to integration of the onset and steady-state responses of ventilation and CBF provides a new tool for such investigations.

GRANTS

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