Cerebral Blood Flow and Oxygenation at Maximal Exercise: The Effect of Clamping Carbon Dioxide

Abstract

During exercise, as end-tidal carbon dioxide (PET_CO2) drops after the respiratory compensation point (RCP), so does cerebral blood flow velocity (CBFv) and cerebral oxygenation. This low-flow, low-oxygenation state may limit work capacity. We hypothesized that by preventing the fall in PET_CO2 at peak work capacity (W_max) with a newly-designed high-flow, low-resistance rebreathing circuit, we would improve CBFv, cerebral oxygenation, and W_max. Ten cyclists performed two incremental exercise tests, one as control and one with PET_CO2 constant (clamped) after the RCP. We analyzed PET_CO2, middle cerebral artery CBFv, cerebral oxygenation, and cardiopulmonary measures. At W_max, when we clamped PET_CO2 (39.7 ± 5.2 vs. 29.6 ± 4.7 mmHg, P<0.001), CBFv increased (92.6 ± 15.9 vs. 73.6 ± 12.5 cm/s, P<0.001). However, cerebral oxygenation was unchanged (ΔTSI -21.3 ± 13.1% vs. -24.3 ± 8.1%, P=0.33), and W_max decreased (380.9 ± 20.4 vs. 405.7 ± 26.8 Watts, P<0.001). At W_max, clamping PET_CO2 increases CBFv, but this does not appear to improve W_max.

1. Introduction

During intense aerobic exercise, hyperventilation reduces the arterial partial pressure of carbon dioxide (Pa_CO2). As a result, cerebral blood flow (CBF) and oxygenation may fall below a critical level needed to maintain motor output (Nybo and Rasmussen, 2007). Several studies have demonstrated that increasing inspired oxygen (PI_O2) can improve cerebral oxygenation and prolong intense aerobic exercise (Amann et al., 2007; Nielsen et al., 1999; Subudhi et al., 2008), but no studies have been conducted to maintain CBF as a method of improving cerebral oxygenation and exercise performance.

Since CBF is highly dependent on Pa_CO2 during exercise (Ogoh and Ainslie, 2009), we sought to maintain CBF by precisely controlling end-tidal carbon dioxide levels (PET_CO2). While rebreathing devices (Banzett et al., 2000; Slessarev et al., 2007) can be used to hold PET_CO2 constant (clamp) and thus control CBF at rest, they are not engineered for high-intensity exercise. We felt that by modifying the circuit designed by Banzett et al.(Banzett et al., 2000) to accommodate elevated minute ventilation and CO₂ production, we could clamp PET_CO2 and study the effect of CBF on cerebral oxygenation and peak work capacity (W_max).

The primary intent of this study was to determine whether the new rebreathing circuit could be used to clamp PET_CO2 at exercise intensities above the respiratory compensation point (RCP). To evaluate the circuit's performance, we conducted a pilot study to test the hypothesis that clamping PET_CO2 after the RCP would increase CBF, cerebral oxygenation, and W_max.

2. Methods

2.1 Sequential gas delivery circuit

To clamp PET_CO2 we designed a rebreathing circuit conceptually similar to the one designed by Banzett et al. (Banzett et al., 2000), with modifications to accommodate variable CO₂ production and high ventilatory volumes at maximal exercise. On the inspiratory limb, there is an investigator-controlled supply of compressed air with adjustable flow, a 14 liter reservoir, and a one-way, low-resistance valve connected to a T connector (Hans Rudolf, model 2700; Figure 1). We did not include a heated humidifier as used in the Banzett circuit. On the expiratory limb (Figure 1), there is a 3 meter section of 3.3 cm internal diameter tubing, acting as an inspiratory reservoir of exhaled gas, which then divides into five limbs capped with positive end expiratory pressure (PEEP) valves (Instrumentation Industries, Inc, Bethel Park, PA). Two low-resistance expiratory valves were placed in close proximity to the PEEP valves. We calibrated the PEEP valves to approximately 2-3 cmH₂0 and aligned them so that air would be drawn into the circuit from the room if there was negative pressure in the tubing, which occurs only when rebreathing exhaled gas. We used the multiple limb design to provide high peak inspiratory flows and low resistance needed during maximal exercise.

Figure 1.

Rebreathing circuit used to clamp end-tidal carbon dioxide. During the clamped trial, we forced rebreathing from the expiratory limb by limiting the flow of source gas, 21% O₂, 0.03% CO₂. Airflow, denoted by broad arrows, is differentiated as follows: The unbroken hollow arrow indicates continuous inspiratory flow of source gas. The broken hollow arrow indicates cyclic inspiratory flow of source gas. The broken black arrows indicate cyclic inspiratory and expiratory flow of exhaled gas. The broken gray arrow indicates cyclic inspiratory flow of mixed gas.

Valves are denoted by thin arrows with perpendicular bases. Those with a single perpendicular base are near-zero resistance one-way valves. Those with a double perpendicular base are low resistance one-way PEEP valves.

We controlled the relative amounts of inspired gas from the inspiratory and expiratory limbs, in response to real-time PET_CO2 measurements, by manually adjusting the flow rate of compressed room air into the inspiratory limb. Rebreathing occurred when the compressed air flow rate was less than alveolar ventilation (Banzett et al., 2000).

2.2 Experimental testing of rebreathing circuit

In this single-blind, crossover experiment approved by the Colorado Multiple Institutional Review Board, 10 male competitive cyclists (professional or category 1), who provided written informed consent, performed two sequential maximal exercise tests, with and without externally-controlled PET_CO2 after the RCP, in a randomly assigned and counterbalanced order. Each participant, instrumented with a custom fit headset with a transcranial Doppler (TCD, Spencer ST3) probe and multichannel NIRS (Oxymon MkIII, Artinis) sensors, completed a 25-30 min self-paced warm up on a Velotron Elite cycle ergometer (Racermate, Seattle, WA). After a brief rest to optimize sensor placement, we increased the workrate from 25 W at 25 W/min until volitional exhaustion or pedal cadence dropped below 50 rpm. Subjects repeated the protocol after 35 min of self-paced recovery. We minimized sensor placement and replacement errors by leaving all instrumentation in place, making fine adjustments between trials.

2.3 RCP selection and PET_CO2 clamping

We identified the RCP in the control trials post hoc by averaging three trained investigators’ analyses of the PET_CO2 and ventilatory equivalent trends. Due to the random assignment of trial order, we were unable to prospectively identify the RCP during the clamped trials. Instead, we clamped PET_CO2 (defined as keeping PET_CO2 levels within ± 2 mmHg) during moderate exercise at the point we observed the first decrease in PET_CO2 levels of 2mmHg lasting greater than 10 s.

2.4 Measurements of cerebral oxygenation, cerebral blood flow, and metabolism

A multichannel, continuous wave NIRS instrument calculated a tissue saturation index (TSI) from measured oxygenated (O₂Hb), de-oxygenated (HHb), and total hemoglobin (THb, O2Hb+HHb). We expressed changes (Δ) in TSI, O₂Hb, HHb, and THb relative to the start of each test. Transcranial Doppler, ipsilateral to the NIRS probes, monitored middle cerebral artery blood flow velocity (CBFv) at depths between 40 to 54 mm. We processed NIRS, transcranial Doppler, and cardiopulmonary variables as previously reported.(Heine et al., 2009; Subudhi et al., 2009)

2.5.1 Calculations and analyses

All measurements were averaged over 15-s intervals. Oxygen consumption (VO₂) could only be determined during the control trial when a mixing chamber was utilized instead of the rebreathing circuit. We used the ratio of CBFv to PET_CO2 as a cerebral vasomotor reactivity index and the ratio of CBFv/ABP to PET_CO2 as a cerebrovascular conductivity index across workrates to further investigate the effect of the rebreathing circuit on cerebral hemodynamics during exercise.

2.5.3 Statistical analyses

Data from specific 15-s intervals of interest corresponding to 0, 25, 50, 75, and 100 percent of maximal power were analyzed with 2 × 5 repeated-measures ANOVA to evaluate effects of trial across workrate. Criterion for significance was set at P < 0.05. Post hoc, pairwise comparisons were made using paired t-tests with Holm's sequential method to control for type I error. Values are means ± standard deviation.

3. Results

3.1 Subject characteristics

Ten subjects (age: 25.9 ± 7.7 years, weight: 71.2 ± 7.7 kg, height: 181.1 ± 7.4 cm) completed the control and intervention trials.

3.2 Control trial

During the control trial, we observed the expected decrease in PET_CO2 after the RCP (29.6 ± 4.7 mmHg at W_max vs. 36.8 ± 6.6mmHg at RCP, P<0.001) without a significant decrease in CBFv (73.6 ± 12.5 cm/s at W_max vs. 76.7 ± 15.4 cm/s at RCP, P=0.86). ΔTSI decreased after the RCP as expected (-21.3 ± 13.1% at W_max vs. -9.7 ± 6.7% at RCP, P=0.008). The RCP occurred at 79.5 ± 4.7 %W_max. W_max (405 ± 26.8 W), ventilation (200.8 ± 24.6 L/min), and cardiopulmonary responses (VO_2max 4.8 ± 0.6 L/min) were representative of altitude specific values for well-trained cyclists (Subudhi et al., 2007).

3.3 Clamp trial

In the clamp trial, we successfully maintained PET_CO2 after the RCP (39.7 ± 5.2 mmHg at W_max vs. 40.1 ± 4.9 mmHg at RCP, P=0.63). PET_CO2 at W_max was higher than that in the control trial (P<0.001; Figure 2a). In contrast to the control trial, CBFv continued to increase after the RCP, ending 26% higher (92.6 ± 15.9 cm/s at W_max vs. 85.0 ± 17.9 cm/s at RCP, P=0.001,Figure 2b), despite a lower W_max (380.9 ± 20.4W, P<0.001). ΔTSI decreased after the RCP (-24.3 ± 8.1 at W_max vs. -12.8 ± 10.7 at RCP, P=0.005), but values at W_max were similar between trials (P= 0.33; Figure 2c). At W_max, there were no differences in heart rate, blood pressure, ventilation, tidal volume, respiratory rate, or oxygen saturation (Table 1). There were no differences at any point across trials in NIRS measurements (ΔO2Hb, ΔHHb, ΔTHb) on either the cerebral cortex or vastus lateralis (muscle data not shown).

Figure 2.

The effect of clamping carbon dioxide on cerebral blood flow and cerebral oxygenation. A. PET_CO2 was effectively clamped at maximal exercise. B. Middle cerebral artery blood flow velocity continually increased to maximal exertion with PET_CO2 clamped. C. Frontal cortex tissue saturation index was unchanged with PET_CO2 clamped. D. Cerebral vasomotor reactivity index was not affected by PET_CO2 clamping. E. Cerebral conductivity showed minimal change across workrates in both trials. (control, ○; clamped, •) Values are means ± SD.

Table 1.

Comparison of metabolic, respiratory, cerebral blood flow, and cerebral oxygenation responses to incremental exercise with unclamped (control) and clamped end-tidal carbon dioxide.

	control			clamp
variable	rest	RCP	max	rest	% control	RCP	% control	max	% control
Power, W	0.0 ± 0.0	324.9 ± 33.2	405.7 ± 26.8	0.0 ± 0.0		324.9 ± 33.2	100%	380.9 ± 20.4	94%
HR, beats/min	93.0 ± 15.4	172.9 ± 12.3	191.9 ± 14.7	88.9 ± 14.0	96%	173.5 ± 12.5	100%	184.1 ± 9.5	96%
BP, mmHg	102.1 ± 11.2	132.8 ± 10.4	135.9 ± 17.8	104.6 ± 5.5	102%	132.4 ± 14.2	100%	139.9 ± 11.4	103%
MV, L/minBTPS	21.3 ± 4.0	129.7 ± 17.6	200.8 ± 24.6	26.2 ± 6.5	123%	157.3 ± 27.3*	121%	192.2 ± 37.2	96%
f, breaths/minute	15.7 ± 5.8	40.9 ± 8.9	62.0 ± 10.9	18.3 ± 7.5	117%	48.3 ± 7.4*	118%	61.3 ± 9.3	99%
Tidal volume, L	1.4 ± 0.5	3.3 ± 0.7	3.3 ± 0.4	1.5 ± 0.3	107%	3.2 ± 0.6	97%	3.1 ± 0.5	94%
PETCO2, mmHg	34.2 ± 3.2	36.8 ± 6.6	29.6 ± 4.7	36.7 ± 3.1	107%	40.1 ± 4.9	109%	39.7 ± 5.2†	134%
VE/VCO2	29.2 ± 5.5	20.0 ± 2.6	23.3 ± 2.1
O2 uptake, L/min	0.5 ± 0.1	4.0 ± 0.4	4.8 ± 0.6
MCA Vmean, cm/s	58.0 ± 6.4	76.7 ± 15.4	73.6 ± 12.5	61.5 ± 6.4	106%	85.0 ± 17.9	111%	92.6 ± 15.9†	126%
O2 saturation, %	97.0 ± 1.0	90.6 ± 3.8	87.7 ± 5.5	95.8 ± 2.4	99%	89.0 ± 3.4	98%	85.1 ± 4.7	97%
cerebral oxygenation
ΔTSI, %	0.3 ± 1.6	-9.7 ± 6.7	-21.3 ± 13.1	0.3 ± 1.9		-12.8 ± 10.7		-24.3 ± 8.1
ΔO2Hb, μM	0.7 ± 1.2	0.6 ± 6.9	-3.4 ± 10.0	1.4 ± 3.1		1.6 ± 8.9		-1.8 ± 8.7
ΔHHb, μM	0.2 ± 0.7	7.0 ± 3.6	14.1 ± 6.2	0.6 ± 0.9		8.6 ± 2.5		15.9 ± 4.9
ΔTHb, μM	1 ± 1.1	7.6 ± 7.8	10.7 ± 12.7	2.0 ± 3.7		10.2 ± 8.7		14.1 ± 11.3

Data are means ± SD

All NIRS values expressed relative to the start of each test

^*P=0.002 vs. control by ANOVA

^†P<0.001 vs. control by ANOVA

3.5 Vasomotor reactivity and conductivity

The reactivity index (CBFv·PET_CO2^-1) gradually increased across workrates to exhaustion and was not affected by clamping PET_CO2 (Figure 2d). When we adjusted the reactivity index for changes in ABP, the resulting conductivity index (CBFv·ABP^-1·PET_CO2^-1) was unchanged across workrates (Figure 2e).

4. Discussion

In this study, we designed a novel rebreathing circuit and successfully clamped PET_CO2 during incremental exercise to maximal exertion. Using this circuit, we showed that a relative increase in PET_CO2 at W_max causes an increase in CBFv. However, despite the increased CBFv, neither cerebral oxygenation nor W_max were elevated.

4.1 Success and limitations of rebreathing circuit

With our circuit we achieved our primary goal of clamping PET_CO2, holding levels within ± 2 mmHg, despite rapidly changing minute ventilation and CO₂ production. We accommodated tidal volumes ranging from 0.7 to 4L and minute ventilations in excess of 200 L/min. Minute ventilation was unchanged versus control values and the circuit did not affect heart rate or blood pressure. Because cardiac parameters were unchanged across trials, we believe that the circuit provides a novel approach to isolating and controlling PET_CO2, and thus to manipulate CBF.

We noticed a few technical limitations with the circuit, however. We used several branches on the expiratory limb to minimize inspiratory resistance, but all subjects noted a degree of inspiratory resistance at maximal exercise. This did not appear to cause a distinct ventilatory ceiling in tidal volume or minute ventilation statistically, but the effort required to ventilate may have been higher in the clamped trial. Additionally, rebreathing exhaled gas limits the amount of oxygen available for diffusion. As detailed by Slessarev et al., (Slessarev et al., 2007) the fraction of gas inhaled from the expiratory limb is physiologically neutral and does not participate in diffusion of oxygen or carbon dioxide. Despite this, we did not see differences in oxygen saturations or NIRS measurements across trials, and values were similar to those seen in maximal exercise tests under similar ambient conditions. Calculations of oxygen uptake and carbon dioxide production with this rebreathing system were not possible as the content of inspired gas varied both between and within breaths. This limitation does not affect our conclusions, and technical improvements will make these calculations possible in future studies.

Additionally, because we randomly assigned the order of the trials, we were unable to prospectively identify the RCP for use in the clamp trial. We instead clamped PET_CO2 levels when we observed the first sustained decrease, generally in the 250-300 Watt range. For this reason, the RCP identified post hoc in the control trial may not precisely correlate with the work rate at which clamping began in the intervention trial. However, the conclusions of this trial are based on responses at maximal exertion, and are unlikely to be affected by the precise start time of clamping. Despite the above concerns, subjects reached near-maximal values for all cardiopulmonary measures in the clamped trial (Table 1), which demonstrates that our rebreathing circuit can be an effective tool in the study of CBF during high intensity exercise.

4.2 Cerebrovascular implications of clamped PET_CO2

Utilization of the rebreathing circuit to clamp PET_CO2 revealed some interesting cerebrovascular physiology which has not been previously described. As expected, during the control condition, PET_CO2 fell at very high workrates. However, we did not observe a parallel decrease in CBFv, which would be expected if PET_CO2 was the sole regulator of CBFv. Previously, we have questioned the absolute strength of relationship between PET_CO2 and CBFv during high intensity exercise since not all subjects show a decrease in CBFv with PET_CO2 (Heine et al., 2009). The circuit gives us new insight into the regulation of CBFv after the RCP, and our findings validate recent studies on the determinants of CBF in exercise at submaximal and maximal exertion (Nybo and Rasmussen, 2007). When we clamped PET_CO2, CBFv continued to increase to values 26% higher than control values, demonstrating that while PET_CO2 is a key regulator, it is not the only variable exerting an influence on CBFv.

We used a reactivity index to evaluate the relationship further. Results demonstrate that the inherent relationship between PET_CO2 and CBFv was similar in the control and rebreathing conditions (Figure 2d). However, since the reactivity index (CBFv·PET_CO2^-1) increases with exercise intensity, other factors must exert an effect. Evaluation of the conductivity index (CBFv·ABP^-1·PET_CO2^-1) suggests that the increase in ABP may be the key response driving CBFv after the RCP because correcting the reactivity for ABP resulted in a conductivity graph that did not change across workrates (Figure 2e).

We hypothesized that an increase in CBFv would increase cerebral oxygenation, but found evidence to the contrary. The increase in CBFv after the RCP did not affect frontal cortex oxygenation. We have previously noted an imperfect correlation between CBFv and oxygenation (Heine et al., 2009), but were surprised that there were no measurable effects of elevated CBFv on frontal cortex oxygenation in the present study. Since the manipulation of did not affect oxygen saturation, it is possible that our analogous measurements of TSI in the brain were also unaffected. Additionally, because increased middle cerebral artery flow is not all directed to the frontal cortex, expected increases in THb may have been blunted by regional autoregulatory mechanisms.

Finally, our data refute the hypothesis that increased CBFv would improve W_max. In fact, W_max was significantly impaired (< 20 watts) in the clamped trial. We suspect that the subjective sensations of dyspnea and increased inspiratory resistance led to early termination of exercise with clamped PET_CO2. However, we cannot exclude the possibility that increased PET_CO2 influenced the decision to stop exercising via other physiologic mechanisms which were not measured in this pilot study, such as changes in sympathetic activity, acid/base balance, and distribution of cardiac output.

5. Conclusions

In conclusion, we developed a novel rebreathing circuit that allows us to clamp PET_CO2 at exercise intensities above the RCP in well-trained athletes. We showed that the circuit provides a means of manipulating CBFv and that CBFv increases relative to PET_CO2 at maximal exertion. Despite increases in CBFv, we did not observe similar increases in cerebral oxygenation or W_max. Based on these preliminary findings, we believe that the circuit may be used to control PET_CO2 in future studies of cerebral blood flow during exercise and offer new insights into factors limiting maximal exercise performance.