Experimental Physiology

Transcription

1 Exp Physiol (2013) pp Research Paper Slowed oxygen uptake kinetics in hypoxia correlate with the transient peak and reduced spatial distribution of absolute skeletal muscle deoxygenation T. Scott Bowen 1,2, Harry B. Rossiter 2,3,AlanP.Benson 2, Tatsuro Amano 4, Narihiko Kondo 4, John M. Kowalchuk 5,6 and Shunsaku Koga 7 1 Department of Internal Medicine & Cardiology, Heart Center, Leipzig University, Leipzig, Germany 2 School of Biomedical Sciences, University of Leeds, Leeds, UK 3 Rehabilitation Clinical Trials Center, Division of Respiratory & Critical Care Physiology & Medicine, Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center, Torrance, CA, USA 4 Laboratory for Applied Human Physiology, Graduate School of Human Development and Environment, Kobe University, Kobe, Japan 5 School of Kinesiology and 6 Department of Physiology and Pharmacology, The University of Western Ontario, London, Ontario, Canada 7 Applied Physiology Laboratory, Kobe Design University, Kobe, Japan Experimental Physiology New findings What is the central question of the study? Does a transient overshoot in skeletal muscle deoxygenation (reflecting a kinetic mismatch of microvascular O 2 delivery to consumption) and/or its spatial distribution slow the adjustment of oxidative energy provision at the onset of exercise? What is the main finding and its importance? Slowed oxidative energy provision at the onset of exercise was correlated with the transient skeletal muscle deoxygenation peak and the reduced spatial distribution, measured by quantitative near-infrared spectroscopy. It was not correlated with a microvascular O 2 delivery-to-consumption mismatch per se. This suggests that an absolute, rather than kinetic, mismatch of microvascular O 2 delivery and consumption limits the kinetics of muscular oxidative energy provision, but only when muscle deoxygenation reaches some critical level. It remains unclear whether an overshoot in skeletal muscle deoxygenation (HHb; reflecting a microvascular kinetic mismatch of O 2 delivery to consumption) contributes to the slowed adjustment of oxidative energy provision at the onset of exercise. We progressively reduced the fractional inspired O 2 concentration (F I,O2 ) to investigate the relationship between slowed pulmonary O 2 uptake ( V O2 ) kinetics and the dynamics and spatial distribution of absolute [HHb]. Seven healthy men performed 8 min cycling transitions during normoxia (F I,O2 = 0.21), moderate hypoxia (F I,O2 = 0.16) and severe hypoxia (F I,O2 = 0.12). V O2 uptake was measured using a flowmeter and gas analyser system. Absolute [HHb] was quantified by multichannel, time-resolved near-infrared spectroscopy from the rectus femoris and vastus lateralis (proximal and distal regions), and corrected for adipose tissue thickness. The phase II V O2 time constant was slowed (P < 0.05) as F I,O2 decreased (normoxia, 17 ± 3s;moderatehypoxia,22± 4s;and severe hypoxia, 29 ± 9 s). The [HHb] overshoot was unaffected by hypoxia, but the transient peak [HHb] increased with the reduction in F I,O2 (P < 0.05). Slowed V O2 kinetics in hypoxia were positively correlated with increased peak [HHb] in the transient (r 2 = 0.45; P < 0.05), but poorly related to the [HHb] overshoot. A reduction of spatial heterogeneity in peak [HHb] was inversely correlated with slowed V O2 kinetics (r 2 = 0.49; P < 0.05). These data suggest that aerobic energy provision at the onset of exercise may be limited by the following factors: DOI: /expphysiol

2 1586 T. S. Bowen and others Exp Physiol (2013) pp (i) the absolute ratio (i.e. peak [HHb]) rather than the kinetic ratio (i.e. [HHb] overshoot) of microvascular O 2 delivery to consumption; and (ii) a reduced spatial distribution in the ratio of microvascular O 2 delivery to consumption across the muscle. (Received 28 March 2013; accepted after revision 8 July 2013; first published online 12 July 2013) Corresponding author S. Koga: Applied Physiology Laboratory, Kobe Design University, Gakuennishi-machi, Nishi-ku, Kobe, , Japan. s-koga@kobe-du.ac.jp Introduction The dynamic profile of skeletal muscle microvascular deoxygenation, discerned from the kinetics of relative deoxyhaemoglobin (HHb) measured by near-infrared spectroscopy (NIRS), provides insight into the transient matching of microvascular O 2 delivery (the product of arterial O 2 concentration and blood flow; Q m )relative to muscle O 2 consumption ( V O2 m; Sperandio et al. 2009, 2012; Bowen et al. 2012). Individuals manifesting slowed phase II pulmonary O 2 uptake ( V O2 ) kinetics (τ) demonstrate a large, transient overshoot in the deoxygenation response at the onset of exercise (Bauer et al. 2007; Sperandio et al. 2009, 2012; Porcelliet al. 2010; Barbosa et al. 2011; Bowen et al. 2012); a feature that is attenuated or absent in individuals with more rapid V O2 kinetics (DeLorey et al. 2003; Grassi et al. 2003; Sperandio et al. 2009; Porcelli et al. 2010). Where Q m kinetics are experimentally slowed in the isolated canine hindlimb, the HHb overshoot is increased and V O2 m kinetics are slowed (Goodwin et al. 2011). In addition, the intervention of warm-up exercise (Bowen et al. 2012) and the administration of sildenafil (Sperandio et al. 2012) acutely speeded V O2 kinetics in chronic heart failure (CHF) patients, and this was associated with a reduction in the HHb overshoot, thus suggesting that a faster adjustment of microvascular O 2 delivery relative to demand may relieve an O 2 delivery limitation to V O2 m in these patients. While this HHb overshoot reflects a transient mismatch in the ratio of Q m -to- V O2 m (Diederich et al. 2002; Ferreira et al. 2005; Sperandio et al. 2009;Barbosaet al. 2010), whether it reflects a limitation to the rate of adjustment in oxidative energy provision is controversial. Previous studies typically assessed the HHb overshoot response at a single muscle site (Sperandio et al. 2009, 2012; Porcelli et al. 2010; Bowen et al. 2012). As the human quadriceps is anatomically and physiologically heterogeneous, and muscle activation during moderateintensity cycling is widely distributed, the degree to which the HHb overshoot is dispersed may provide important insight into the mechanisms of slowed V O2 kinetics, which itself reflects a mean response of the wide distribution in muscle metabolism. Homogeneity in the HHb response dynamics across the muscle could reflect microvascular conditions that are either conducive or opposing for fast V O2 kinetics; that is, a better maintenance of the Q m - to- V O2 m ratio throughout the muscle would be expected to be associated with a high capillary P O2 and present no limitation to the dynamic adaptation of intramuscular V O2 m in the transient. Alternatively, a uniform reduction of the Q m -to- V O2 m ratio and capillary P O2 across the muscle may contribute to slowing the V O2 m response. By the same logic, a more heterogeneous HHb response might also reflect conditions conducive (by maintaining the ratio of Q m -to- V O2 m in some muscle regions) or opposing to fast V O2 dynamics (by a large reduction of the Q m -to- V O2 m ratio in the active muscles). Which of these scenarios is associated with slowed V O2 kinetics is yet to be established. Importantly, all of the previous studies investigating these dynamic relationships (Sperandio et al. 2009, 2012; Goodwin et al. 2011; Bowen et al. 2012) were unable to quantify the absolute concentration of the deoxygenation ([HHb]) overshoot. Continuous-wave NIRS (CW-NIRS) makes only relative, normalized, measures of haem chromophore concentrations. This is likely to confound interpretation of the HHb kinetic response by concealing whether it is the Q m -to- V O2 m kinetic ratio (reflected in therelativehhbovershootfromcw-nirs)ortheq m -to- V O2 m absolute ratio (reflected in the absolute peak [HHb]) that may limit V O2 kinetics. This study, therefore, used graded hypoxia (by reducing the fractional inspired O 2 concentration; F I,O2 )toslow V O2 kinetics on transition to moderate exercise in health, in order to investigate its outcome on the dynamics and spatial heterogeneity of muscle [HHb] kinetics, as assessed by state-of-the art quantitative multichannel time-resolved (TRS) NIRS technology (Chin et al. 2011; Koga et al. 2011). This study design aimed to provide insight into whether the ratio of Q m -to- V O2 m acts as a peripheral mechanism limiting aerobic energy provision at the onset of exercise in humans. We hypothesized that hypoxia would slow V O2 kinetics, as previously shown (Linnarsson et al. 1974; Murphy et al. 1989; Springer et al. 1991; Hughson & Kowalchuk, 1995; Engelen et al. 1996; Spencer et al. 2012), and this slowing would be correlated with a greater overshoot magnitude and spatial distribution of [HHb]. Methods Participants, experimental design and protocols Seven young, healthy men (mean ± SD values: age, 22 ± 2 years; height, 172 ± 6cm; weight, 61± 6 kg; and V O2 peak,50± 8mlkg 1 min 1 ) provided written informed

3 Exp Physiol (2013) pp Effect of hypoxia on muscle deoxygenation and oxygen uptake kinetics 1587 consent to participate in the study approved by the Human Subjects Committee of Kobe Design University, in accordance with the Declaration of Helsinki. Exercise was performed in the following three F I,O2 conditions: 0.21 (normoxia; N); 0.16 (moderate hypoxia; MH); and 0.12 (severe hypoxia; SH). Tests were initiated with either a 2 min rest period when subjects breathed room air or a 15 min rest period during hypoxia for equilibrium of gases. The exercise protocols performed have been described in detail previously (Bowen et al. 2012). Briefly, a rampincremental exercise test (20 W min 1 ) to exhaustion was performed at 60 r.p.m. in each F I,O2 condition on different days for V O2 peak and lactate threshold (LT) estimation. During separate visits, participants performed four repeats of 8 min moderate-intensity, square-wave exercise transitions in each F I,O2 condition at 60 r.p.m. at an average power output of 81 ± 27 W ( 90% LT estimated fromsh).thesewerecompletedinarandomizedorder, separated by at least 60 min of rest, with no more than two repetitions performed in one day. Equipment and measurements A detailed description of the equipment and measurements has been published (Chin et al. 2011; Koga et al. 2011). Exercise was performed on an electromagnetically braked cycle ergometer ( C, Combi, Tokyo, Japan), heart rate (HR) was measured beat by beat from a three-lead electrocardiogram, and breath-by-breath alveolar gas exchange (Beaver et al. 1981) was measured using a flowmeter and gas analyser system (Aeromonitor AE-300S; Minato Medical Science, Osaka, Japan). Participants breathed through a mouthpiece connected to a low-resistance, two-way non-rebreathing valve (2700; Hans Rudolph, Shawnee, KS, USA), linked to rubber tubing that supplied humidified air from Douglas bags filled with room air (i.e. F I,O2 = 0.21) or room air diluted with N 2 to achieve the hypoxic gas mixture (i.e. F I,O2 = 0.16 or 0.12). Skeletal muscle deoxygenation of the right quadriceps was quantified by multichannel, TRS-NIRS (TRS-20; Hamamatsu Photonics KK, Hamamatsu, Japan), at three muscle sites, i.e. the distal (VL D ) and proximal (VL P ) regions of the vastus lateralis and the middle region of the rectus femoris (RF). Each TRS-20 probe provided picosecond light pulses at three different wavelengths (760, 795 and 830 nm) to measure absolute muscle deoxygenation ([deoxy(hb + Mb)]; [HHb]), oxygenation ([oxy(hb + Mb)]; [HbO 2 ]) and total haemoglobin concentration ([Hb + Mb]; [Hb tot ]). Temporal light intensityprofilesateachmeasurementpointwerefitted with a photon diffusion equation (Oda et al. 1999) for estimation of mean optical path length, scattering and absorption coefficients, which allowed quantification of NIRS chromophores in micromoles (Chin et al. 2011; Koga et al. 2011, 2012). To account for adipose tissue thickness (ATT) on the NIRS signal (Niwayama et al. 2000; Kogaet al. 2011), a novel correction factor based on the relationship between [Hb tot ]andattwasdetermined from participants in the rested, upright-seated position. The ATT (in millimetres) at each muscle site was determined using B-mode Doppler ultrasound (Logiq 400; GE-Yokogawa Medical Systems, Tokyo, Japan) and resting [Hb tot ] determined from a 2 min resting average. This normalization process is detailed in Fig. 1. Data analysis and kinetic modelling Breath-by-breath gas exchange was edited as previously described (Lamarra et al. 1987; Bowen et al. 2012). The LT was estimated by the V-slope method with corroborating ventilatory criteria (Whipp et al. 1986), and peak variables during ramp exercise were averaged over the final 20 s. Moderate-exercise repetitions in each F I,O2 condition for the V O2, [HHb] and HR responses were time aligned at exercise onset, interpolated and averaged into 1 s bins to improve the signal-to-noise ratio (Whipp & Rossiter, 2005). V O2 kinetics were modelled using Figure 1. The group relationship between subcutaneous adipose tissue thickness (ATT) and resting total Hb ([Hb tot ]) measured by time-resolved NIRS in the distal (VLL D )and proximal (VL P ) vastus lateralis regions, and rectus femoris (RF) To normalize the [HHb], we applied a linear regression to the group relationship between [Hb tot ] and ATT ([Hb tot ] = 21.4 (ATT) + 220; r 2 = 0.77; P < 0.001). Measured [HHb] values were then corrected to a common ATT of 0 mm. This correction was achieved by solving for [Hb tot ] (above) at the measured ATT at each individual muscle site, and expressing 220 μm (i.e. the y-intercept) as a proportion of the solved [Hb tot ]value. Corrected [HHb] was then calculated as the product of the correction factor and the measured [HHb] at each site (similar to Chin et al and Koga et al. 2011). This normalization process allowed absolute values of [HHb] to be compared between subjects and muscle sites differing in ATT.

4 1588 T. S. Bowen and others Exp Physiol (2013) pp non-linear least squares regression (OriginPro 7.5; OriginLab Corp., Northampton, MA, USA). The fundamental phase (phase II) was isolated from the first breath following the initial cardiodynamic response (phase I) through to the end of exercise; the goodness of fit was determined by visual inspection of the flatness of the residuals and primarily by optimization of the 95% confidence interval (C 95 ) for τ V O2 (Rossiter et al. 2001). The response was fitted to an exponential equation of the following form: Y (t) = Y bl + Y ss ( 1 e (t TD)/τ ) (1) where Y bl indicates the baseline value (2 min average) before exercise onset, Y ss indicates the amplitude between Y bl and the steady state, TD is the time delay and τ the time constant of the exponential function. The C 95 for τ was also calculated (Whipp & Rossiter, 2005). Heart rate responses were fitted to eqn (1) using the entire exercise duration, with TD fixed at 0 s. The mean response time (MRT) of V O2 was fitted in a similar fashion and used to calculate the O 2 deficit (MRT V O2 ss; Rossiter et al. 1999). The [HHb] response was modelled by the following two-component exponential equation (Ferreira et al. 2005; Barbosa et al. 2010; Porcelli et al. 2010; Barbosa et al. 2011): Y (t) = Y bl + Y ss u ( 1 e (t TDu)/τu) Y ss d ( 1 e (t TDd)/τd) (2) where u indicates the upward exponential component and d the downward exponential component [see eqn (1) for symbol definitions]. When the fitting algorithm could not resolve the downward parameters (i.e. where the measured response increased continually throughout exercise to steady state), a single exponential was fitted by fixing Y ss d to zero. The start of the [HHb] response was fitted as previously described (DeLorey et al. 2003) and stopped at 180 s (i.e. the time at which the participant with the slowest V O2 kinetics attained a steady state). As well as the fitted parameters, the following variables were measured: the peak [HHb] value; the time at which the peak [HHb] occurred; the [HHb] overshoot height, i.e. the difference between the peak [HHb] and [HHb] at 180 s; and the [HHb] overshoot area, i.e. the area bounded by the overshoot in [HHb], where present (Barbosa et al. 2010). The immediate decrease in [HHb] following exercise onset was assessed for both amplitude and duration (DeLorey et al. 2003). End-exercise values for cardiopulmonary and muscle deoxygenation variables (Y ee ) were determined from the average of the final 60 s of exercise. The present study made no comparative analysis of [HHb] and V O2 kinetics because the τ parameters in eqns (1) and (3) are not similar and therefore comparison is contraindicated (Barbosa et al. 2010, 2011). The effects of F I,O2 on spatial heterogeneity of [HHb] were determined by the coefficient of variation [CV (%): 100 SD/mean of three muscle regions] for each subject in each condition. Statistical analyses Data are presented as means ± SD unless stated. The [HHb] variables and parameters were analysed by twoway repeated measures ANOVA (F I,O2 muscle region). Other data were analysed by one-way repeated measures ANOVA. Post hoc Bonferroni corrected t tests determined the location of the differences were appropriate. Significance was accepted at P < Analyses were completed using SPSS v.16.0 (SPSS Inc., Chicago, IL, USA). Results Effect of hypoxia on V O2 kinetics and spatial muscle deoxygenation Group mean pulmonary gas exchange, ventilation and heart rate responses during exercise in the different F I,O2 conditions are presented in Table 1. As hypothesized, phase II V O2 kinetics slowed (P < 0.05) in direct proportion (r 2 = 0.97) to the decrease in F I,O2 (N, MH and SH: 17 ± 3, 22 ± 4and29± 9 s, respectively; Table 1 and Fig. 2). Absolute quadriceps muscle deoxygenation variables, parameters and spatial heterogeneity (CV) in each F I,O2 condition are presented in Table 2. Group mean muscle deoxygenation profiles (both muscle region specific and within-subject quadriceps mean) between F I,O2 conditions are presented in Fig. 3, and the results from kinetic modelling are shown in Fig. 4. Group mean [Hb tot ] was not different amongst F I,O2 conditions at baseline (218 ± 19 versus 218 ± 15 versus 218 ± 14 μm for N, MH and SH, respectively; P > 0.05) or at end exercise (236 ± 24 versus 234 ± 18 versus 232 ± 17 μm for N, MH and SH, respectively; P > 0.05), but there was a significant increase in [Hb tot ] between baseline and end exercise (16 ± 6 μm; P < 0.05). There was a main effect for both F I,O2 (P < 0.05) and muscle region (P < 0.05) on baseline, amplitude and end-exercise [HHb] (Table 2 and Figs 3 and 4). The amplitude of the initial, immediate reduction in [HHb] at the onset of exercise was similar (P > 0.05) amongst F I,O2 conditions and muscle regions ( 4 ± 2 μm). The duration of the initial [HHb] reduction was also not different (P > 0.05) between F I,O2 conditions or regions ( 7 ± 3 s). While an overshoot in [HHb] (Fig. 5A) was seen in most F I,O2 conditions and muscle regions (with the exception of RF muscle during N; Table 2 and Figs 3 and 4), the incidence of an overshoot was variable within a given subject and inconsistent amongst subjects (Fig. 5A). The incidence of an overshoot was, therefore,

5 Exp Physiol (2013) pp Effect of hypoxia on muscle deoxygenation and oxygen uptake kinetics 1589 Table 1. Group mean (±SD) pulmonary and heart rate responses to moderate-intensity exercise in the different F I,O2 conditions F I,O Pulmonary measurements Resting V O2 (l min 1 ) 0.27 ± ± ± 0.04 V O2 bl (l min 1 ) 0.42 ± ± ± 0.05 V O2 ss (l min 1 ) 0.72 ± ± ± 0.26 V O2 ee (l min 1 ) 1.13 ± ± ± 0.28 τ V O2 (s) 17 ± 3 22 ± 4 29 ± 9 C 95 (s) 2 ± 1 2 ± 1 2 ± 1 TD (s) 22 ± 7 18 ± 4 15 ± 3 V O2 ss/ WR 9.42 ± ± ± 0.73 V O2 MRT (s) 35 ± 2 38 ± 5 45 ± 8 O 2 deficit (ml) 409 ± ± ± 217 RER bl 0.86 ± ± ± 0.06 RER ee 0.93 ± ± ± 0.05 V CO2 bl (l min 1 ) 0.36 ± ± ± 0.05 V CO2 ss (l min 1 ) 0.70 ± ± ± 0.30 V CO2 ee (l min 1 ) 1.06 ± ± ± 0.03 V Ebl (l min 1 ) 15 ± 2 15 ± 1 17 ± 1 V E ee (l min 1 ) 33 ± 6 35 ± 6 44 ± 9 Heart rate Resting HR (beats min 1 ) 68 ± ± 9 83 ± 11 HR bl (beats min 1 ) 73 ± ± ± 13 HR ee (beats min 1 ) 105 ± ± ± 17 HR MRT (s) 31 ± ± ± 19 O 2 pulse (ml per beat) 10.9 ± ± ± 2.1 Abbreviations and symbols are as follows: C 95, 95% confidence interval; F I,O2, fractional inspired O 2 concentration; HR, heart rate; MRT, mean response time; RER, respiratory exchange ratio; τ, time constant; TD, time delay; V CO2, carbon dioxide output; V E, minute ventilation; V O2, pulmonary oxygen uptake; V O2 ss, increment in V O2 above baseline; WR, increment in work rate; and subscripts bl, ss and ee indicate baseline, steady state and end exercise, respectively. P < 0.05 between conditions; P < 0.05 versus F I,O2 = 0.21; and P < 0.05 versus F I,O2 = 0.21 and not different (P >0.05) between N (24%), MH (48%) andsh(43%).assuch,therewasnomaineffectof F I,O2 condition (P > 0.05) or muscle region (P > 0.05) on the [HHb] overshoot area. In contrast, there was a significant main effect of F I,O2 (P < 0.05) and muscle region (P < 0.05) on the peak [HHb] achieved during the exercise transient (Table 2 and Figs 3 and 4), such that peak deoxygenation progressively increased with the severity of hypoxia. The absolute peak [HHb] was independent of the [HHb] overshoot area (r 2 = 0.01; P > 0.05) and overshoot height (r 2 = 0.01; P > 0.05), and was less (80 ± 10 μm) than the maximal [HHb] achieved during cuff occlusion (149 ± 42 μm; P < 0.05). In addition, the time at which the peak [HHb] occurred was not different (P >0.05) amongst F I,O2 conditions (N, 75 ± 11 s; MH, 80 ± 10 s; and SH, 97 ± 10 s), but there was a main effect (P < 0.05) Table 2. Group mean (±SD) quadriceps muscle deoxygenation (HHb; micromolar) parameters and variables across muscle regions during moderate-intensity exercise during different F I,O2 conditions Muscle deoxygenation ([HHb]) Coefficient of variation (%) F I,O2 F I,O Baseline (μm) 56 ± 8 61 ± 7 70 ± 7 10 ± 6 9 ± 5 9 ± 4 Amplitude (μm) 10 ± 5 11 ± 5 14 ± 7 47 ± ± ± 21 End exercise (μm) 66 ± 8 72 ± 8 84 ± 9 7 ± 4 9 ± 4 6 ± 5 Overshoot incidence (%) Overshoot area (μm s 1 ) 66 ± ± ± ± ± ± 64 Overshoot height (μm) 1 ± 2 2 ± 2 2 ± 3 52 ± ± ± 64 Peak (μm) 72 ± 9 77 ± 9 90 ± 10 8 ± 3 10 ± 4 8 ± 5 P < 0.05 between all conditions; P < 0.05 versus F I,O2 = 0.16; and P = versus F I,O2 = 0.16.

6 1590 T. S. Bowen and others Exp Physiol (2013) pp of muscle region on this variable (VL D,80± 12 s; VL P, 74 ± 6s;andRF,97± 10 s). There was substantial spatial heterogeneity (measured as CV) of muscle deoxygenation amongst quadriceps muscle regions in each F I,O2 condition (Table 2 and Figs 3 and 4). In general, however, F I,O2 did not alter (P > 0.05) the spatial heterogeneity for most [HHb] variables and parameters, except for the spatial CV of peak [HHb] (P = 0.056) and end-exercise [HHb] (P < 0.05), which were reduced in SH compared with MH (Table 2). Relationship between V O2 kinetics and muscle deoxygenation heterogeneity Relative to N, there was an inverse correlation between the increase in τ V O2 and the CV of peak [HHb] (r 2 = 0.49; P < 0.05; Fig. 7) and end-exercise [HHb] (r 2 = 0.32; Relationship between V O2 kinetics and transient muscle deoxygenation There was no significant relationship between τ V O2 and either the [HHb] overshoot area (r 2 = 0.22; P > 0.05) or the [HHb] overshoot height (r 2 = 0.15; P > 0.05) across the F I,O2 conditions. Interestingly, however, the increase in absolute peak [HHb] during the transient was positively correlated with the increase in τ V O2 relative to N (r 2 = 0.45; P < 0.05; Fig. 6), regardless of whether an overshoot was present or not. This revealed that, amongst subjects, peak muscle [HHb] increased as τ V O2 was progressively slowed by hypoxia. This increase in τ V O2 was also related positively to the increase in [HHb] baseline (r 2 = 0.39; P < 0.05), [HHb] amplitude (r 2 = 0.25; P < 0.05) and end-exercise [HHb] (r 2 = 0.53; P < 0.05). Figure 2. Pulmonary oxygen uptake ( V O2 ) response to the different fractional inspired oxygen (F I,O2 ) conditions (normoxia, N = 0.21; moderate hypoxia, MH = 0.16; and severe hypoxia, SH = 0.12) on transition to moderate-intensity cycling exercise in a representative subject Note that phase II V O2 kinetics were slowed with hypoxia (see inset of first 180 s of exercise), yet the end-exercise V O2 attained between conditions was similar. Continuous lines represent fitted responses (baseline and phase II kinetics), and dashed lines represent these fits extrapolated during the phase I region. The vertical dashed line indicates the transition from 5 W to moderate-intensity exercise. Figure 3. Group mean (±SEM, every 60 s) absolute muscle deoxygenation ([HHb]) responses to the different F I,O2 conditions (N = 0.21, MH = 0.16 and SH = 0.12) during moderate-intensity cycling exercise across the three quadriceps sites measured The panels show the mean of the three measurement sites, as well as individual measurements in distal (VL D ) and proximal vastus lateralis (VL P ) and the rectus femoris (RF). Note that reductions in F I,O2 result in increased muscle deoxygenation in all muscle regions, and there remains a considerable degree of spatial heterogeneity for deoxygenation between regions.

7 Exp Physiol (2013) pp Effect of hypoxia on muscle deoxygenation and oxygen uptake kinetics 1591 P < 0.05). Therefore, as τ V O2 was progressively slowed with hypoxia, the spatial variance in peak and endexercise [HHb] between muscle regions was reduced. No other relationships between changes in τ V O2 and spatial heterogeneity of [HHb] were detected. Discussion This study progressively slowed V O2 kinetics by hypoxia to determine the effect on skeletal muscle [HHb] dynamics and its spatial heterogeneity as assessed by quantitative multichannel time-resolved NIRS. We hypothesized that slowed V O2 kinetics in hypoxia would be associated with a greater kinetic mismatch in the ratio of Q m -to- V O2 m at the onset of exercise. However, slowed V O2 kinetics were not related to the magnitude of the [HHb] overshoot, which was inconsistent within and between individuals across both F I,O2 conditions and muscle regions. In contrast to our hypothesis, therefore, the overshoot in [HHb] does not appear to be the sole factor relating muscle deoxygenation to slowed V O2 kinetics in hypoxia. Instead, a progressive slowing of V O2 kinetics by hypoxia had the following characteristics: (i) it was associated with an overall increase in muscle [HHb] at rest and during exercise (Fig. 3); (ii) it was linearly correlated with the highest absolute [HHb] over the exercise transient (Fig. 6); and (iii) it was inversely correlated with the reduction in the spatial distribution of the transient peak [HHb] (Fig 7). These data support the notion that the absolute ratio of Q m -to- V O2 m (i.e. the peak [HHb] attained during the work-rate transition) may limit aerobic energy provision at exercise onset, and not the kinetic ratio of the two variables as previously suggested (i.e. reflected by the height or area of the relative HHb overshoot). These findings also suggest that a more uniform reduction in the Q m -to- V O2 m ratio across the muscle (i.e. a reduced CV of the peak [HHb]) is associated with a slowing of the V O2 m response, perhaps via a uniform reduction in capillary Figure 4. Group mean (±SD) muscle deoxygenation variables in response to the different F I,O2 conditions (N = 0.21, MH = 0.16 and SH = 0.12) during moderate-intensity cycling exercise for the distal (VL D ) and proximal regions of the vastus lateralis (VL P ) and the rectus femoris (RF) muscles There was wide heterogeneity in muscle deoxygenation responses between muscle regions, but this remained similar between F I,O2 conditions. Main effect (P < 0.05) of F I,O2 ; main effect (P < 0.05) of muscle region; and interaction (P < 0.05; F I,O2 muscle region).

8 1592 T. S. Bowen and others Exp Physiol (2013) pp P O2 throughout the active tissues. Collectively, this study provides novel insight into the peripheral mechanisms that are associated with slowed V O2 kinetics. Effects of hypoxia on V O2 kinetics and muscle deoxygenation Consistent with previous studies (Linnarsson et al. 1974; Murphy et al. 1989; Springer et al. 1991; Hughson & Kowalchuk, 1995; Engelen et al. 1996; Spencer et al. 2012), our findings demonstrate that V O2 kinetics are slowed with reductions in F I,O2 (Table 1 and Fig. 2). The mechanism of this delayed response remains unclear, but interventions designed to limit the transport of O 2 generally slow V O2 kinetics [e.g. β blockade (Hughson & Smyth, 1983), supine exercise (Hughson et al. 1991) or slowed pump perfusion (Goodwin et al. 2011)]. This expected response formed an integral part of our experimental design, because it allowed us to assess the subsequent relationship amongst slowed V O2 kinetics and absolute [HHb] and its spatial heterogeneity. We found that the [HHb] before exercise Figure 6. The change ( ) in peak muscle deoxygenation ([HHb]) as a function of pulmonary oxygen uptake kinetics (τ) on transition to moderate-intensity exercise Values are the difference between those in normoxia and either moderate (F I,O2 = 0.16) or severe hypoxia (F I,O2 = 0.12). Figure 5. An example of the absolute (A) or normalized muscle deoxygenation ([HHb]) profile (B), where an overshoot (open circles) or no overshoot (filled circles) is demonstrated at exercise onset in the distal region of the VL muscle (VL D )in representative subjects Data were modelled over the first 180 s of exercise, and where an overshoot was present a two-component (up and down) exponential was used to characterize the response. Where no overshoot occurred, a single exponential was used to fit the data adequately. Note that normalizing the [HHb] response (as might be seen with continuous-wave or frequency-domain near-infrared spectroscopy) overemphasizes the role of the overshoot in absolute muscle deoxygenation. Figure 7. The change ( ) in the coefficient of variation (CV) in peak muscle deoxygenation ([HHb]) among muscle regions as a function of pulmonary oxygen uptake kinetics (τ) on transition to moderate-intensity exercise Values are the difference between those in normoxia and either moderate (F I,O2 = 0.16) or severe hypoxia (F I,O2 = 0.12). Note that with increasing hypoxia the spatial variance in peak [HHb] is reduced.

9 Exp Physiol (2013) pp Effect of hypoxia on muscle deoxygenation and oxygen uptake kinetics 1593 and the profile of dynamic changes during exercise were increased with the reduction in F I,O2 (Figs 3 and 4). These findings are in agreement with some (Costes et al. 1996; Maehara et al. 1997; Wang et al. 2010) but not all studies that have assessed relative HHb dynamics in hypoxia during square-wave exercise (DeLorey et al. 2004); discrepancies that are likely to be explained by a number of factors related to the NIRS device, adipose tissue thickness and the exercise mode and protocol. The present study used TRS-NIRS to quantify absolute [HHb] (for details, see Chin et al. 2011; Koga et al. 2011). Earlier studies investigating the effects of hypoxia in exercise reported values of NIRS-derived HHb that are normalized within each individual during the preexercise phase of the experiment by the workings of the instrument (Costes et al. 1996; Maehara et al. 1997; DeLorey et al. 2004; Wang et al. 2010; Spencer et al. 2012); therefore, most were limited to estimating HHb change in arbitrary units that do not allow simple comparison between individuals (Costes et al. 1996; Maehara et al. 1997; DeLorey et al. 2004; Wang et al. 2010). Thus, the absolute quantification of [HHb] in the present study is likely to have important implications for the interpretation of the [HHb] response, particularly with regard to its amplitude-related characteristics, such as the overshoot height and area (for example, see Fig. 5A and B). In addition, both the relative HHb and the absolute [HHb] signal are influenced by subcutaneous adipose tissue thickness (Niwayama et al. 2000; Koga et al. 2011). The present study used a novel ATT correction factor to account for the influence of ATT (Fig. 1). The [Hb tot ] ATT correction method was based on the NIRS signal being derived predominantly from the muscle and skin (assuming minimal signal from Hb in adipose tissue), such that attenuation of the [Hb tot ] signal due to light scattering would be proportional to the volume of adipose tissue within the sample (represented by ATT). This assumption was supported by the strong and inverse correlation between resting [Hb tot ]andatt(r 2 = 0.77; P < 0.001; Fig 1.). The [Hb tot ] ATT relationship provided a stronger correlation than a previously reported ATT correction method (r 2 = 0.37 for our subject group; Chin et al. 2011; Koga et al. 2011), supporting its future application. Relationship between slowed V O2 kinetics and the [HHb] overshoot A dynamic overshoot in skeletal muscle microvascular deoxygenation during the exercise transient has been observed in conditions where V O2 kinetics are slowed (Bauer et al. 2007; Sperandio et al. 2009, 2012; Porcelli et al. 2010; Barbosa et al. 2011; Bowen et al. 2012). This overshoot area is suggested to provide a close correlate of Q m delivery dynamics; the greater the relative HHb overshoot area, the slower the dynamics of Q m relative to V O2 m (Barbosa et al. 2010). Although variable, the absolute occurrenceofan[hhb]overshootwaslessinthepresent study when subjects breathed room air compared with hypoxic conditions, which suggests that it may be sensitive to O 2 delivery. In support of this, interventions speeding V O2 kinetics in CHF patients have reported a concomitant reduction in the relative HHb overshoot magnitude (Bowen et al. 2012; Sperandio et al. 2012), suggesting the attenuation of a transient O 2 delivery limitation to muscular oxidative energy production in these patients. Accordingly, we hypothesized that slowed V O2 kinetics would correlate with an increase in the absolute [HHb] overshoot. However, we found no association between the presence, absence, or magnitude of the [HHb] overshoot and the slowing of V O2 kinetics. Rather, this study found a direct relationship between slowed V O2 kinetics and the highest absolute (peak) [HHb] value achieved during the transient (Fig. 6). Naturally, the [HHb] peak and overshoot are intricately linked during the exercise transient, such that a greater overshoot will necessarily lead to a higher peak [HHb]. However, only through the use of novel quantitative TRS-NIRS were we able to reveal that the absolute ratio of Q m - to- V O2 m (reflected by the peak [HHb] attained during the transition to exercise) was associated with slowed aerobic energy provision at work-rate onset, and that the kinetic ratio (reflected by the [HHb] overshoot) was not. Therefore, whilst the magnitude of the [HHb] overshoot reflects the kinetic mismatch of Q m -to- V O2 m, it is the absolute peak [HHb] that closely correlates with slowed oxidative energy provision. This presumably reflects the encroachment of some critical limiting value for microvascular oxygenation, beyond which V O2 kinetics become slowed. This suggestion is based on the dynamic relationship between Q m and V O2 m, which is a key determinant of the O 2 driving pressure (i.e. P O2 )required for diffusion between the capillary and myocyte (Behnke et al. 2001; Diederich et al. 2002). The [HHb] kinetics measured by TRS-NIRS reflect the dynamic profile of microvascular P O2 in the rat, which have been assessed directly by phosphorescence quenching (Koga et al. 2012). Thus, the quantitative TRS-NIRS technology used in this study provides initial non-invasive support for the notion that a reduced capillary-to-myocyte O 2 flux in humans during the transient of moderate-intensity exercise in hypoxia may limit V O2 kinetics, as indicated by the transient attainment of a high muscle HHb concentration. The absolute Q m -to- V O2 m ratio includes not only the τ but also the baseline, steady-state, time-delay and amplitude responses that in combination are associated with conditions that oppose fast V O2 response kinetics. As such, both baseline and end-exercise [HHb] were also

10 1594 T. S. Bowen and others Exp Physiol (2013) pp associated with slowed V O2 kinetics in the present study. We have no data to suggest that V O2 m was limited in the steady state prior to or during exercise (e.g. pulmonary V O2 was similar at baseline and end exercise across F I,O2 conditions), and the peak [HHb] was achieved during the V O2 kineticphase.clearly,however,restingoxygenation wouldbeakeyfactorinthisrelationship.exerciseinitiated from a higher baseline [HHb] may cause critical levels of P O2 to be reached more rapidly, thereby limiting muscle V O2. But our data also show that the time to achieve peak [HHb] was not different among F I,O2 conditions, despite V O2 kinetics becoming slower in each hypoxic increment. The combination of low baseline oxygenation, similar time-to-peak deoxygenation and slower muscle V O2 kinetics therefore suggests that a critical limiting value for microvascular oxygenation would be achieved more rapidly in the two hypoxic conditions. In accordance with this idea, we have previously shown using NIRS that a low pre-exercise muscle oxygenation in CHF patients coincides with slowed V O2 kinetics (Bowen et al. 2012). These findings suggest that the pretransition Q m -to- V O2 m ratio, as well as the HHb kinetics, could be a key determinant of transient muscle deoxygenation and the V O2 m kinetic response. Collectively, the importance of the absolute Q m -to- V O2 m ratio may help to explain, for example, why not all CHF patients demonstrate an overshoot in HHb despite slowed V O2 kinetics (Sperandio et al. 2009, 2012; Bowen et al. 2012) or, in contrast, why some healthy trained subjects demonstrate an HHb overshoot despite fast V O2 kinetics (as in the present study). While it is thought that endurance-trained individuals in whom an overshoot in [HHb] is seen are unlikely to reach an absolute ratio of Q m -to- V O2 m low enough to impact aerobic energy provision, it has yet to be demonstrated directly whether reducing this peak [HHb] would further speed their V O2 kinetics. In contrast, CHF patients are more likely to reach a low absolute ratio of Q m -to- V O2 m across the exercise transient that limits aerobic energy provision, despite the lack of appearance of an [HHb] overshoot. This low Q m -to- V O2 m ratio may be due to low resting perfusion, low exercise perfusion, slow perfusion kinetics or a combination, which could be revealed by application of TRS-NIRS. It is also unknown whether the magnitude of the [HHb] overshoot may be exacerbated during exercise above the LT (Sperandio et al. 2009; Sperandio et al. 2012), where the potential for an O 2 delivery limitation is expected to be greater. Importantly, as shown in the comparison in Fig. 5A and B, the use of CW-NIRS to determine relative changes in HHb during the transition to exercise (Sperandio et al. 2009; Porcelliet al. 2010) may obscure these differences between the magnitude of the overshoot in HHb and the absolute peak [HHb] value achieved. Despite these limitations, however, reductions in the magnitude of the HHb overshoot, as measured during interventional studies and within subjects by CW- NIRS (Sperandioet al. 2009, 2012), would still presumably reflect a reduction in the absolute peak [HHb], but this would have to be assumed. Does the spatial heterogeneity of muscle deoxygenation limit V O2 kinetics? Similar to previous studies (Koga et al. 2007, 2011; Saitoh et al. 2009; Prieur et al. 2010; Chin et al. 2011), we found that [HHb] across the quadriceps was spatially heterogeneous at rest and during exercise (Table 2 and Figs 3 and 4). Considering that prior exercise speeds V O2 kinetics and reduces the spatial heterogeneity of muscle deoxygenation (Saitoh et al. 2009; Prieur et al. 2010), we hypothesized that slowed V O2 kinetics would be associated with greater spatial heterogeneity in the [HHb] response consequent to a large reduction of the Q m -to- V O2 m ratio in active muscles. In contrast, we found that slowed V O2 kinetics were associated with reduced overall spatial heterogeneity of absolute [HHb] (Table 2 and Figs 3 and 4), but specifically reduced spatial heterogeneity of peak [HHb] (Fig. 7). This finding suggests that a more homogeneous peak [HHb] response (by a uniform reduction of Q m -to- V O2 mratio across the muscle) may oppose fast V O2 kinetics, whilst a more heterogeneous [HHb] response seems conducive to fast V O2 kinetics (presumably by maintaining a higher Q m -to- V O2 mratio in some muscle regions). Alternatively, hypoxia may have reduced the O 2 driving pressure between capillary and myocyte and possibly modified the local appearance of vasodilatory metabolites (e.g. adenosine), which caused a more homogeneous spatial distribution of quadriceps Q m (Heinonen et al. 2007). Clearly, these findings warrant further investigation. Overall, however, these data highlight that the regional ratio of Q m -to- V O2 m is not well reflected in whole-muscle measurements (Whipp et al. 1995; Kalliokoski et al. 2005) and that single-site measurements (e.g. VL D )poorlyreflectoverallquadriceps Q m -to- V O2 m ratio (Figs 3 and 4). Such disparities are probably the result of regional variations in muscle recruitment (Chin et al. 2011), fibre type and/or blood flow distribution (Koga et al. 2011), and remind us that techniques with improved spatial and temporal resolution are needed to assess both macroscopic variability (among muscles or muscle regions) and microscopic variability (among capillaries and fibres) in muscle metabolic and vascular responses during exercise in humans. Conclusions This study provides the first evidence in humans that the magnitude of the transient peak in absolute [HHb] (reflecting a low absolute ratio of Q m -to- V O2 m) may

11 Exp Physiol (2013) pp Effect of hypoxia on muscle deoxygenation and oxygen uptake kinetics 1595 limit oxidative energy provision during the transition to moderate exercise in hypoxia. In contrast, the relative kinetic ratio of Q m -to- V O2 m was not well related to the slowing of V O2 kinetics in hypoxia, suggesting that the [HHb] overshoot per se does not reflect a limitation to V O2 kinetics. The [HHb] response across the quadriceps showed wide spatial distribution within and between muscle regions, but this was reduced as V O2 kinetics were slowed. This suggests that a more uniform reduction of the Q m -to- V O2 m ratio across the muscle creates conditions that oppose fast V O2 kinetics. Importantly, these conclusions could only be drawn based on the novel quantitative multichannel measurements of absolute [HHb]. References Barbosa PB, Bravo DM, Neder JA & Ferreira LF (2010). Kinetics analysis of muscle arterial venous O 2 difference profile during exercise. Respir Physiol Neurobiol 173, BarbosaPB,FerreiraEM,ArakakiJS,TakaraLS,MouraJ, Nascimento RB, Nery LE & Neder JA (2011). Kinetics of skeletal muscle O 2 delivery and utilization at the onset of heavy-intensity exercise in pulmonary arterial hypertension. Eur J Appl Physiol 111, Bauer TA, Reusch JE, Levi M & Regensteiner JG (2007). Skeletal muscle deoxygenation after the onset of moderate exercise suggests slowed microvascular blood flow kinetics in type 2 diabetes. Diabetes Care 30, Beaver WL, Lamarra N & Wasserman K (1981). Breath-by-breath measurement of true alveolar gas exchange. JApplPhysiol51, Behnke BJ, Kindig CA, Musch TI, Koga S & Poole DC (2001). Dynamics of microvascular oxygen pressure across the rest-exercise transition in rat skeletal muscle. Respir Physiol 126, Bowen TS, Cannon DT, Murgatroyd SR, Birch KM, Witte KK & Rossiter HB (2012). The intramuscular contribution to the slow oxygen uptake kinetics during exercise in chronic heart failure is related to the severity of the condition. JAppl Physiol 112, Chin LM, Kowalchuk JM, Barstow TJ, Kondo N, Amano T, Shiojiri T & Koga S (2011). The relationship between muscle deoxygenation and activation in different muscles of the quadriceps during cycle ramp exercise. JApplPhysiol1111, Costes F, Barthelemy JC, Feasson L, Busso T, Geyssant A & Denis C (1996). Comparison of muscle near-infrared spectroscopy and femoral blood gases during steady-state exercise in humans. JApplPhysiol80, DeLorey DS, Kowalchuk JM & Paterson DH (2003). Relationship between pulmonary O 2 uptake kinetics and muscle deoxygenation during moderate-intensity exercise. J Appl Physiol 95, DeLorey DS, Shaw CN, Shoemaker JK, Kowalchuk JM & Paterson DH (2004). The effect of hypoxia on pulmonary O 2 uptake, leg blood flow and muscle deoxygenation during single-leg knee-extension exercise. Exp Physiol 89, Diederich ER, Behnke BJ, McDonough P, Kindig CA, Barstow TJ, Poole DC & Musch TI (2002). Dynamics of microvascular oxygen partial pressure in contracting skeletal muscle of rats with chronic heart failure. Cardiovasc Res 56, Engelen M, Porszasz J, Riley M, Wasserman K, Maehara K & Barstow TJ (1996). Effects of hypoxic hypoxia on O 2 uptake and heart rate kinetics during heavy exercise. JApplPhysiol 81, Ferreira LF, Poole DC & Barstow TJ (2005). Muscle blood flow O 2 uptake interaction and their relation to on-exercise dynamics of O 2 exchange. Respir Physiol Neurobiol 147, Goodwin ML, Hernández A, Lai N, Cabrera ME & Gladden LB (2011). V O2 on-kinetics in isolated canine muscle in situ during slowed convective O 2 delivery. JApplPhysiol112, Grassi B, Pogliaghi S, Rampichini S, Quaresima V, Ferrari M, Marconi C & Cerretelli P (2003). Muscle oxygenation and pulmonary gas exchange kinetics during cycling exercise on-transitions in humans. JApplPhysiol95, Heinonen I, Nesterov SV, Kemppainen J, Nuutila P, Knuuti J, Laitio R, Kjaer M, Boushel R & Kalliokoski KK (2007). Role of adenosine in regulating the heterogeneity of skeletal muscle blood flow during exercise in humans. JApplPhysiol 103, Hughson RL & Kowalchuk JM (1995). Kinetics of oxygen uptake for submaximal exercise in hyperoxia, normoxia, and hypoxia. Can J Appl Physiol 20, Hughson RL & Smyth GA (1983). Slower adaptation of V O2 to steady state of submaximal exercise with beta-blockade. Eur J Appl Physiol Occup Physiol 52, Hughson RL, Xing HC, Borkhoff C & Butler GC (1991). Kinetics of ventilation and gas exchange during supine and upright cycle exercise. Eur J Appl Physiol Occup Physiol 63, Kalliokoski KK, Knuuti J & Nuutila P (2005). Relationship between muscle blood flow and oxygen uptake during exercise in endurance-trained and untrained men. JAppl Physiol 98, Koga S, Kano Y, Barstow T, Ferreira L, Ohmae E, Sudo M & Poole D (2012). Kinetics of muscle deoxygenation and microvascular PO 2 during contractions in rat: comparison of optical spectroscopy and phosphorescence-quenching techniques. JApplPhysiol112, Koga S, Poole DC, Ferreira LF, Whipp BJ, Kondo N, Saitoh T, Ohmae E & Barstow TJ (2007). Spatial heterogeneity of quadriceps muscle deoxygenation kinetics during cycle exercise. JApplPhysiol103, Koga S, Poole DC, Fukuoka Y, Ferreira LF, Kondo N, Ohmae E & Barstow TJ (2011). Methodological validation of the dynamic heterogeneity of muscle deoxygenation within the quadriceps during cycle exercise. AmJPhysiolRegulIntegr Comp Physiol 301, R534 R541. Lamarra N, Whipp BJ, Ward SA & Wasserman K (1987). Effect of interbreath fluctuations on characterizing exercise gas exchange kinetics. JApplPhysiol62, Linnarsson D, Karlsson J, Fagraeus L & Saltin B (1974). Muscle metabolites and oxygen deficit with exercise in hypoxia and hyperoxia. JApplPhysiol36,

12 1596 T. S. Bowen and others Exp Physiol (2013) pp Maehara K, Riley M, Galassetti P, Barstow TJ & Wasserman K (1997). Effect of hypoxia and carbon monoxide on muscle oxygenation during exercise. Am J Respir Crit Care Med 155, Murphy PC, Cuervo LA & Hughson RL (1989). A study of cardiorespiratory dynamics with step and ramp exercise tests in normoxia and hypoxia. Cardiovasc Res 23, Niwayama M, Jun Shao LL, Kudo N & Yamamoto K (2000). Quantitative measurement of muscle hemoglobin oxygenation using near-infrared spectroscopy with correction for the influence of a subcutaneous fat layer. Rev Sci Instrum 71, Oda M, Yamashita Y, Nakano T, Suzuki A, Shimizu K, Hirano I, Shimomura F, Ohmae E, Suzuki T & Tsuchiya Y (1999). Nearinfrared time-resolved spectroscopy system for tissue oxygenation monitor. Proc SPIE 3597, Porcelli S, Marzorati M, Lanfranconi F, Vago P, Pisot R & Grassi B (2010). Role of skeletal muscles impairment and brain oxygenation in limiting oxidative metabolism during exercise after bed rest. JApplPhysiol109, Prieur F, Berthoin S, Marles A, Blondel N & Mucci P (2010). Heterogeneity of muscle deoxygenation kinetics during two boutsofrepeatedheavyexercises.eur J Appl Physiol 109, Rossiter HB, Ward SA, Doyle VL, Howe FA, Griffiths JR & Whipp BJ (1999). Inferences from pulmonary O 2 uptake with respect to intramuscular [phosphocreatine] kinetics during moderate exercise in humans. JPhysiol518, Rossiter HB, Ward SA, Kowalchuk JM, Howe FA, Griffiths JR & Whipp BJ (2001). Effects of prior exercise on oxygen uptake and phosphocreatine kinetics during high-intensity knee-extension exercise in humans. JPhysiol537, Saitoh T, Ferreira LF, Barstow TJ, Poole DC, Ooue A, Kondo N & Koga S (2009). Effects of prior heavy exercise on heterogeneity of muscle deoxygenation kinetics during subsequent heavy exercise. Am J Physiol Regul Integr Comp Physiol 297, R615 R621. Spencer MD, Murias JM, Grey TM & Paterson DH (2012). Regulation of V O2 kinetics by O 2 delivery: insights from acute hypoxia and heavy-intensity priming exercise in young men. JApplPhysiol112, Sperandio PA, Borghi-Silva A, Barroco A, Nery LE, Almeida DR & Neder JA (2009). Microvascular oxygen delivery-to-utilization mismatch at the onset of heavy-intensity exercise in optimally treated patients with CHF. Am J Physiol Heart Circ Physiol 297, H1720 H1728. Sperandio PA, Oliveira MF, Rodrigues MK, Berton DC, Treptow E, Nery LE, Almeida DR & Neder JA (2012). Sildenafil improves microvascular O 2 delivery-to-utilization matching and accelerates exercise O 2 uptake kinetics in chronic heart failure. Am J Physiol Heart Circ Physiol 303, H1474 H1480. Springer C, Barstow TJ, Wasserman K & Cooper DM (1991). Oxygen uptake and heart rate responses during hypoxic exercise in children and adults. Med Sci Sports Exerc 23, Wang JS, Wu MH, Mao TY, Fu TC & Hsu CC (2010). Effects of normoxic and hypoxic exercise regimens on cardiac, muscular, and cerebral hemodynamics suppressed by severe hypoxia in humans. JApplPhysiol109, Whipp BJ, Lamarra N & Ward SA (1995). Obligatory anaerobiosis resulting from oxygen uptake-to-blood flow ratio dispersion in skeletal muscle: a model. Eur J Appl Physiol Occup Physiol 71, Whipp BJ & Rossiter HB (2005). The kinetics of oxygen uptake: physiological inferences from the parameters. In Oxygen Uptake Kinetics in Sport, Exercise and Medicine, ed. Jones AM & Poole DC, pp Routledge, London. Whipp BJ, Ward SA & Wasserman K (1986). Respiratory markers of the anaerobic threshold. Adv Cardiol 35, Additional information Competing interests None declared. Funding Support for this study was provided by The Medical Research Council UK (studentship to T.S.B.), The BBSRC UK (BB/ /1; BB/I00162X/1) and The Japan Society for the Promotion of Science, the Ministry of Education, Science and Culture of Japan (Grant-in-Aid to S.K.). Acknowledgements The authors would like to thank Dr Rob Wüst for his constructive suggestions during the preparation of this manuscript, as well as the participants who volunteered to take part in the study.