Adjustments of pulmonary O₂ uptake and muscle deoxygenation during ramp incremental exercise and constant-load moderate-intensity exercise in young and older adults

Abstract

The matching of muscle O₂ delivery to O₂ utilization can be inferred from the adjustments in muscle deoxygenation (Δ[HHb]) and pulmonary O₂ uptake (V̇o_2p). This study examined the adjustments of V̇o_2p and Δ[HHb] during ramp incremental (RI) and constant-load (CL) exercise in adult males. Ten young adults (YA; age: 25 ± 5 yr) and nine older adults (OA; age: 70 ± 3 yr) completed two RI tests and six CL step transitions to a work rate (WR) corresponding to 1) 80% of the estimated lactate threshold (same relative WR) and 2) 50 W (same absolute WR). V̇o_2p was measured breath by breath, and Δ[HHb] of the vastus lateralis was measured using near-infrared spectroscopy. Δ[HHb]-WR profiles were normalized from baseline (0%) to peak Δ[HHb] (100%) and fit using a sigmoid function. The sigmoid slope (d) was greater (P < 0.05) in OA (0.027 ± 0.01%/W) compared with YA (0.017 ± 0.01%/W), and the c/d value (a value corresponding to 50% of the amplitude) was smaller (P < 0.05) for OA (133 ± 40 W) than for YA (195 ± 51 W). No age-related differences in the sigmoid parameters were reported when WR was expressed as a percentage of peak WR. V̇o_2p kinetics compared with Δ[HHb] kinetics for the 50-W transition were similar between YA and OA; however, Δ[HHb] kinetics during the transition to 80% of the lactate threshold were faster than V̇o_2p kinetics in both groups. The greater reliance on O₂ extraction displayed in OA during RI exercise suggests a lower O₂ delivery-to-O₂ utilization relationship at a given absolute WR compared with YA.

during transitions to constant-load (CL) moderate-intensity, or heavy-intensity exercise, pulmonary O₂ uptake (V̇o_2p) kinetics generally are slower in older adults (OA) compared with young adults (YA). The time constant (τ) for the fundamental component of the V̇o_2p response (τ_V̇o2p), which is considered to reflect the adjustment of muscle O₂ utilization (V̇o_2m) and oxidative energy production during the transition to exercise (3, 24, 33), is ∼40–50 s in OA and ∼20–30 s in YA (2, 14, 15, 18, 27, 40, 55). Studies (14–16, 18, 27) from our laboratory combining measures of V̇o_2p and near-infrared spectroscopy (NIRS)-derived muscle deoxygenation (Δ[HHb]) have shown that during transitions to CL moderate-intensity or heavy-intensity exercise, despite slowed V̇o_2p kinetics in OA compared with YA, overall muscle deoxygenation kinetics were generally similar to or faster than those for YA. In addition, the steady-state Δ[HHb]-to-ΔV̇o_2p relationship was greater in OA than in YA (15, 16, 18, 27, 40). As the NIRS-derived Δ[HHb] signal reflects local muscle O₂ extraction, and thus the ratio of local muscle O₂ utilization to microvascular blood flow, these observations suggest that the adjustment of muscle blood flow, specifically microvascular blood flow, during transitions to CL exercise is attenuated in OA compared with YA, thus requiring increased O₂ extraction to meet the muscle O₂ requirement for mitochondrial oxidative ATP production. The reported slower heart rate (HR; reflecting central O₂ delivery) (14–16, 18, 40) and leg (conduit artery) blood flow kinetics (18) as well as lower steady-state leg blood flow (36, 37, 46, 48, 50) and vascular conductance (36, 37, 46, 48, 50) in OA are consistent with an attenuated muscle blood flow response during transitions to and in the steady state of CL submaximal exercise. A consequence of slower adjustment in V̇o_2m (as reflected by V̇o_2p) in OA compared with YA is that, for a given ATP and O₂ requirement, a greater O₂ deficit and greater reliance on substrate-level phosphorylation are required, which would disrupt metabolic “stability” (25) and possibly compromise exercise tolerance in these individuals (25).

Unlike transitions to CL moderate-intensity and heavy-intensity exercise, where “steady-state” V̇o_2p, V̇o_2m, and muscle blood flow values eventually are achieved, this is not the case for ramp incremental (RI) exercise to the limit of tolerance. During RI exercise, V̇o_2p (1, 13) and muscle blood flow (1, 49, 53) increase in an apparently linear fashion as work rate (WR) increases across a wide range of exercise intensities and ATP requirements, but with V̇o_2m (and V̇o_2p) lagging behind the actual muscle O₂ requirement. For YA performing exercise transitions to CL exercise from low or raised baseline metabolic rates, the kinetics of V̇o_2p (and V̇o_2m) and limb (conduit artery) blood flow display dynamic nonlinearity across a range of increasing exercise intensities; the rate of adjustment of V̇o_2p (and V̇o_2m) (7, 31, 32, 38, 62, 63) and limb (conduit artery) blood flow become slower (38), the “steady-state” ΔV̇o_2p -ΔWR relationship (V̇o_2p gain) increases (7, 32, 38, 62, 63), and the “steady-state” Δmuscle blood flow-ΔV̇o_2p relationship becomes smaller (38) with exercise transitions initiated from elevated baseline intensities. Similar nonlinear responses in V̇o_2p kinetics and V̇o_2p gain were also observed in OA when transitions to CL exercise were initiated from baseline metabolic rates representing lower and upper regions of the moderate-intensity domain (57). In YA performing RI exercise to the limit of tolerance, NIRS-derived muscle deoxygenation displays a nonlinear, sigmoidal pattern of increase, reflecting a nonlinear increase in muscle O₂ extraction with increasing WR and V̇o_2p (6, 20, 47), which is consistent with a changing (nonlinear) relationship between microvascular blood flow and V̇o_2m throughout a range of exercise intensities. Given the slower V̇o_2p and similar Δ[HHb] kinetics observed in OA compared with YA during transitions to CL moderate-intensity or heavy-intensity exercise, with a greater reliance on O₂ extraction at any given V̇o_2m, a greater muscle deoxygenation response might be expected in OA compared with YA during RI exercise where steady-state conditions are not achieved. An apparently greater submaximal Δ[HHb]-versus-WR relationship in OA compared with YA was reported by Ferri et al. (22) during step-incremental knee extension (increases of 20%, 40%, and 60% of one repetition maximum every 3 min) and step-incremental cycling exercise (increments of 10 and 20 W every minute for OA and YA, respectively) to exhaustion, although in that study, V̇o_2p was not measured and detailed age- and WR-related comparisons of the Δ[HHb] profiles were not presented. Analysis of the adjustments of Δ[HHb] and V̇o_2p adjustments would provide inferences regarding the matching between muscle O₂ delivery and muscle O₂ utilization during nonsteady-state conditions spanning a wide range of exercise domains from light- to very heavy (or severe)-intensity exercise.

Therefore, the purpose of this study was to examine the adjustments of V̇o_2p and NIRS-derived leg muscle Δ[HHb] during nonsteady-state RI exercise to the limit of tolerance in OA and YA. Also, in the same subjects, responses were examined during transitions to steady-state CL moderate-intensity exercise of the same absolute intensity (i.e., 50 W) and the same relative intensity [i.e., 80% of the estimated lactate threshold (θ_L)]. We hypothesized that 1) the position of the sigmoid Δ[HHb]-WR response profile would be left shifted toward a lower WR in OA compared with YA and 2) the slope of the linear portion of the Δ[HHb]-WR relationship would be greater in OA compared with YA. Also, in agreement with previous findings at the same relative intensity, we hypothesized that 3) the rate of adjustment of V̇o_2p would be slowed in OA compared with YA, but that the adjustment of Δ[HHb] would be similar, when exercise transitions were performed at the same absolute intensity (same WR, 50 W) and the same relative intensity (80% θ_L) within the moderate-intensity domain. Together, the findings would support the suggestion that adjustments of local muscle microvascular blood flow are attenuated in OA compared with YA, thereby requiring a greater reliance on O₂ extraction to support oxidative ATP requirements.

METHODS

Subjects

Ten YA (age: 25 ± 5 yr, mean ± SD) and nine OA (age: 70 ± 3 yr, mean ± SD) men volunteered and gave written consent to participate in the study. All subjects were healthy, recreationally active, nonsmokers with no previously diagnosed respiratory, cardiovascular, metabolic, or musculoskeletal disease. All procedures were approved by the Ethics Committee for Research on Human Subjects of The University of Western Ontario. A detailed written explanation of the experimental protocol was given to all subjects before they were tested. Additionally, subjects were not taking any medications that could affect their cardiorespiratory or metabolic responses to exercise. OA subjects were cleared for participation in the study after first undergoing a physician-supervised medical screening and an exercise stress test to the limit of tolerance before the investigation.

Protocol

All subjects performed two RI exercise tests to the limit of tolerance on an electromagnetically braked cycle ergometer (H-3oc-R Lode; Lode, B.V., Groningen, The Netherlands); each test was separated by at least 48 h to allow recovery. The RI test began with 6 min of cycling at a light-intensity WR of 20-W baseline exercise, with a pedal cadence of 70–80 rpm, after which the WR increased linearly as a ramp function at a rate of 20 W/min until the subject could no longer maintain a cadence of 50 rpm. After each RI test, peak V̇o_2p and the V̇o_2p corresponding to θ_L were determined for each subject. Peak V̇o_2p was determined as the average V̇o_2p during the final 20 s of each RI test. θ_L was estimated using standard gas exchange and ventilatory indexes and was defined as the V̇o_2p at which pulmonary CO₂ output (V̇co_2p) and expired minute ventilation (V̇e) began to increase out of proportion to the rise in V̇o_2p. Additionally, there was a systematic rise in the ventilatory equivalent for V̇o_2p (V̇e/V̇o_2p) and end-tidal Po₂, whereas the ventilatory equivalent for V̇co_2p (V̇e/V̇co_2p) and end-tidal Pco₂ were stable. Based on the results from the RI tests, a WR was selected that would elicit a steady-state V̇o_2p corresponding to ∼80% θ_L.

Subjects returned to the laboratory on several occasions to complete repetitions of CL step transitions to WRs within the moderate-intensity exercise domain. The step transitions consisted of 6 min of cycling at a baseline of 20 W followed by an instantaneous increase in WR to 1) an absolute WR of 50 W or 2) a relative WR corresponding to 80% θ_L, with each step transition lasting 6 min. Subjects completed six step transitions to both absolute and relative WRs, with conditions being randomly assigned. Only a single step transition was completed by YA on each testing day, whereas OA performed two step transitions on the same testing day (to decrease the number of visits for OA), with each transition separated by a 30-min period of resting recovery (sitting on a chair) to allow cardiovascular and metabolic variables to return back toward preexercise baseline conditions.

Measurements

Gas exchange measurements were similar to those previously described by Babcock et al. (2). Briefly, inspired and expired airflow and volumes were measured throughout the exercise protocol using a low-dead space (90 ml) bidirectional turbine (3.0 liters, Hans Rudolph, Kansas City, MO), which was calibrated before each test using a syringe of known volume. Inspired and expired gases were sampled continuously (every 20 ms) at the mouth and analyzed for the fractional concentrations of O₂, CO₂, and N₂ by mass spectrometry (Innovision, AMIS 2000, Lindvedvej, Denmark) after calibration with precision-analyzed gas mixtures. Changes in gas concentrations were aligned with gas volumes by measuring the time delay (TD) for a square-wave bolus of gas passing the turbine to the resulting changes in fractional gas concentrations as measured by the mass spectrometer. The collected data were transferred to a computer, which aligned concentrations with volume information to build a profile of each breath. The algorithms of Beaver et al. (4) were used to calculate breath-by-breath alveolar gas exchange. Heart rate (HR) was continuously monitored by ECG using PowerLab (ML132/ML880, AD Instruments, Colorado Springs, CO) with a three-lead arrangement and recorded with LabChart (version 6.0, AD Instruments) on a separate computer.

Changes in the concentration of local Δ[HHb], oxy-, and total hemoglobin + myoglobin (with the absorbance spectra of hemoglobin and myoglobin being virtually indistinguishable within the NIR spectral range) of the vastus lateralis quadriceps muscle were measured continuously (sampling rate: 2 Hz) using NIRS (NIRO 300, Hamamatsu Photonics). Optodes were placed on the belly of the muscle midway between the lateral epicondyle and greater trochanter of the femur. Optodes were separated by 5 cm and housed in an optically dense rubber holder that was secured to the skin surface with tape and covered with an optically dense, black vinyl sheet, thus minimizing the intrusion of extraneous light. To further secure the position of the optodes during the exercise protocol, an elastic bandage was wrapped around the leg to prevent any movement of the optode assembly without restricting the range of motion of the leg during the cycling exercise. While subjects were at rest, before beginning any exercise, NIRS signals were monitored until steady-state baseline levels were established, at which time the signals were “zero set” such that further changes in the signal were reported as a change (Δ) from this relative “zero” baseline.

The physical principles of tissue spectroscopy and the manner in which these are applied have been previously explained by DeLorey et al. (17). Briefly, four laser diodes that produce different wavelengths (775, 810, 850, and 910 nm) are pulsed in rapid succession, and the NIR light produced by the diodes is transmitted through a fibre optic bundle to the tissue of interest. The transmitted light is returned through a separate fibre optic bundle and detected by a photomultiplier tube for online estimation and display of the concentration changes from the resting baseline values. As the differential path length in the quadriceps muscle at rest and during exercise is presently unknown, the NIRS-derived measures are reported in arbitrary units (AU). The raw attenuation signals were transferred and stored on a separate computer for later analysis.

Data Analysis

V̇o_2p and HR data sets were edited for each individual trial by removing aberrant data that lay outside 4 SD of the local mean as they did not conform to a Gaussian distribution as previously described by Lamarra et al. (35). The data for each RI or CL transition were linearly interpolated on a second-by-second basis and time aligned to the onset of the RI or CL transition, labeled as time 0. The individual repetitions of RI and CL transitions were ensemble averaged to yield a single “average” profile for each subject and time averaged into 5-s bins.

The slope of the time-averaged V̇o_2p-WR profiles for RI exercise [reflecting the functional V̇o_2p gain (ΔV̇o_2p/ΔWR)] were determined in YA and OA using linear regression analysis. The V̇o_2p response was time aligned with the onset of RI exercise by shifting the V̇o_2p profile data back by a calculated TD for each individual, as previously described by Davis et al. (12), to account for the delay in the rise of V̇o_2p (a consequence of V̇o_2p kinetics) relative to the increase in WR. Therefore, the left-shifted V̇o_2p data were fit from the onset of RI exercise to a WR corresponding to ∼80% θ_L, thereby reflecting the linear V̇o_2p-WR relationship within the moderate-intensity domain of the RI protocol.

CL on-transient responses for the 5-s averaged V̇o_2p, Δ[HHb], and HR profiles were modeled using nonlinear regression analysis and a monoexponential model of the following form:

$$Y_{\left(t\right)}=Y_{\text{Bsln}}+\text{Amp}\left[1-e^{-\left(t-\text{TD}\right)/\tau }\right]$$ (1)

where Y_(t) is V̇o_2p at any time (t), Y_Bsln is baseline V̇o_2p during 20-W cycling before the step increase in WR, Amp is the steady-state increase in V̇o_2p above the baseline value, and τ is the duration of time for V̇o_2p to increase to 63% of the steady-state Amp. V̇o_2p data were modeled from the phase 1 to phase 2 transition, determined as previously described by Rossiter et al. (54) and Gurd et al. (29), to the end of the exercise transition.

Beat-by-beat HR data were edited and averaged as described above for V̇o_2p. The monoexponential model described in Eq. 1 was used to fit the on-transient HR response, with the TD constrained to time 0, such that the entire HR response from the onset to the end of the exercise was modeled.

Second-by-second NIRS-derived Δ[HHb] data were time aligned, ensemble averaged, and time averaged into 5-s bins to yield a single, averaged response for each subject in the RI and CL protocols. The Δ[HHb] profile for CL has been previously described to consist of a TD (TD-Δ[HHb]) at the onset of exercise followed by an increase in the Δ[HHb] signal that followed an “exponential-like” time course (17, 23). TD-Δ[HHb] was determined using second-by-second data and corresponded to the time, after the onset of exercise, at which the Δ[HHb] signal began to increase systematically above an early nadir value in the signal. TD-Δ[HHb] was determined for every CL trial and averaged to yield an average TD-Δ[HHb] for each subject. The time course for Δ[HHb] was modeled using an exponential function of the form described in Eq. 1 starting from the time corresponding to TD-Δ[HHb] and ending at a time corresponding to the beginning of the V̇o_2p steady state (i.e., 5 × τ_V̇o2p). The time course for the increase in Δ[HHb] was described using the Δ[HHb] τ (τ_Δ[HHb]), whereas the overall time course of Δ[HHb] from the onset of moderate-intensity exercise was described using the “effective” τ (τ_Δ[HHb] = TD-Δ[HHb] +τ_Δ[HHb]).

During the RI exercise tests, Δ[HHb] data were normalized to the peak (calculated as a 5-s average) amplitude of the response. Baseline Δ[HHb] (0%) was established as the average steady state for the 20-W baseline cycling. The normalized Δ[HHb] response for both age groups during RI exercise was described as a function of 1) absolute WR (in W) and 2) relative WR (as a percentage of peak WR) and fit using a sigmoid model of the following form:

$$F\left(x\right)=f_0+A/\left[1+e^{-\left(-c+dx\right)}\right]$$ (2)

where F(x) is the normalized Δ[HHb] value at a given x value (e.g., WR), ƒ₀ is the baseline, A is the total amplitude, c is a constant dependent on d, d is the slope of the sigmoid, and the c/d value is the x value corresponding to 50% of A. The sigmoid model has previously been shown, in young healthy adults, to provide a significantly better fit of the Δ[HHb]-WR response compared with a hyperbola model (6, 20).

Statistics

The dependent variables (V̇o_2p, HR, and Δ[HHb]) were analyzed using unpaired t-tests and two-way repeated-measures ANOVA. Each analysis was performed using SPSS (version 17.0, SPSS, Chicago, IL). A significant F-ratio was identified using Fisher's least significant difference post hoc analysis, with statistical significance accepted at P < 0.05. All values are expressed as means ± SD.

RESULTS

Nonsteady-State RI Exercise

V̇o_2p response.

Subject characteristics and peak responses during RI exercise in YA and OA are shown in Table 1, and group mean V̇o_2p response profiles during RI exercise are shown in Fig. 1. After an initial transient delay, V̇o_2p increased linearly relative to time (and thus WR) (Fig. 1) in both YA (r = 0.89 ± 0.07) and OA (r = 0.76 ± 0.11). Absolute and relative peak V̇o_2p were greater (P < 0.05) in YA (4.06 ± 0.43 l/min and 49 ± 5 ml·kg⁻¹·min⁻¹) compared with OA (2.63 ± 0.35 l/min and 30 ± 6 ml·kg⁻¹·min⁻¹; Table 1), a consequence of a greater (P < 0.05) peak WR in YA (338 ± 30 W) than in OA (215 ± 31 W; Table 1). Also, the V̇o_2p corresponding to θ_L was greater (P < 0.05) in YA (2.3 ± 0.2 l/min) than in OA (1.6 ± 0.1 l/min; Table 1). During RI exercise, the functional gain (ΔV̇o_2p/ΔWR) calculated for the V̇o_2p-WR relationship (with WRs restricted to the moderate-intensity region of RI exercise) was greater (P < 0.05) in YA (10.0 ± 0.4 ml·min⁻¹·W⁻¹) than in OA (8.7 ± 1.2 ml·min⁻¹·W⁻¹); however, during CL exercise, the steady-state functional gain was not different between YA (9.5 ± 0.7 ml·min⁻¹·W⁻¹) and OA (9.9 ± 1.2 ml·min⁻¹·W⁻¹).

Table 1.

Subject characteristics and peak exercise responses

					V̇o2peak
	n	Age, yr	Body mass, kg	Body mass index, kg/m	In l/min	In ml·kg−1·min−1)	Peak WR, W	V̇o2p at θL, l/min	WR at 80% θL, W
YA	10	25 ± 5	84 ± 11	25 ± 3	4.06 ± 0.43	49 ± 5	338 ± 30	2.3 ± 0.2	118 ± 25
OA	9	70 ± 3*	88 ± 12	28 ± 3*	2.63 ± 0.35*	30 ± 6*	215 ± 31*	1.6 ± 0.1*	64 ± 10*

Values are means ± SD; n, subjects/group. YA, young adults; OA, older adults; V̇o_2peak, peak O₂ consumption; WR, work rate; V̇o_2p, pulmonary O₂ consumption; θ_L, estimated lactate threshold.

^*P < 0.05 compared with YA.

Fig. 1.

Group mean (±SD) pulmonary O₂ consumption (V̇o_2p) response profiles during ramp incremental (RI) exercise as a function of time (in s) for young adults (YA; A, ●) and older adults (OA; B, ○). Continuous curves represent mean values for all subjects; single points represent group mean (±SD) values at the point of fatigue during RI exercise. The dashed line indicates the onset of RI exercise.

Δ[HHb] response.

The group mean adjustment of Δ[HHb] during RI exercise in YA and OA groups is shown in Fig. 2A (absolute values) and Fig. 2B (normalized to the amplitude measured during RI exercise); Δ[HHb] data for one YA and three OA were removed due to poor quality during RI tests. The modeled sigmoid regression profile of the Δ[HHb]-WR response during RI exercise is shown for two age groups (Fig. 3A) and for a representative YA and OA (Fig. 3B), and the parameter estimates for Δ[HHb]-WR profiles during RI exercise for YA and OA are shown in Table 2. The slope of the Δ[HHb]-WR response was greater (P < 0.05) in OA (0.027 ± 0.01%/W) compared with YA (0.017 ± 0.01%/W), and the WR corresponding to 50% amplitude was lower (P < 0.05) in OA (133 ± 40 W) than in YA (195 ± 51 W; Fig. 3A and Table 2). No age-related differences were found for any of the sigmoid parameters when the relationship was expressed as a percentage of peak WR (Fig. 4 and Table 2).

Fig. 2.

Group mean (±SD) absolute (A) and normalized (B; percentage of the peak amplitude during RI exercise) muscle deoxygenation (Δ[HHb]) response profiles to RI exercise in YA (●; n = 9) and OA (○; n = 6). Continuous curves represent mean values for all subjects; single points represent group mean (±SD) values at the point of fatigue during RI exercise. The dashed line indicates the onset of RI exercise.

Fig. 3.

Sigmoid models of normalized Δ[HHb] (percentage of the peak amplitude during RI exercise) as a function of absolute work rate (WR) for the group mean (A) and a representative subject (B) for YA (solid line, ●) and OA (dashed line, ○). The sigmoid models in A were recreated from the parameters shown in Table 2.

Table 2.

Parameter estimates for normalized Δ[HHb] as a function of absolute and relative WRs during ramp incremental exercise in YA and OA

	YA	OA
Absolute WR
ƒ0, %	−2.6 ± 7.1	−1.8 ± 18.4
A, %	118 ± 20	112 ± 32
d, %/W	0.017 ± 0.01	0.027 ± 0.01*
c/d, W	195 ± 51	133 ± 40*
Relative WR
ƒ0, %	−2.2 ± 7.3	−1.7 ± 18.3
A, %	116 ± 19	112 ± 32
d, %/%peak	0.056 ± 0.02	0.056 ± 0.02
c/d, %peak	57.1 ± 9.5	64.1 ± 19.8

Values are means ± SD of the percentage of peak muscle deoxygenation (Δ[HHb]) during ramp incremental exercise as a function of absolute (in W) and relative (in %) WRs; n = 10 YA and 9 OA. ƒ₀, baseline; A, amplitude; d, slope; c/d, WR corresponding to 50% of the total amplitude.

^*Significantly different from YA (P < 0.05).

Fig. 4.

Sigmoid models of normalized Δ[HHb] (percentage of the peak amplitude during RI exercise) as a function of relative (percent peak) WR for the group mean for YA (solid line, ●) and OA (dashed line, ○). Sigmoid models were recreated from the parameters shown in Table 2.

Steady-State CL Exercise

V̇o_2p kinetics.

The steady-state increase in V̇o_2p (V̇o_2p amplitude) for CL transitions to 50 W was similar in YA (0.26 ± 0.04 l/min) and OA (0.29 ± 0.03 l/min; Fig. 5A). During transitions to 80% θ_L, the V̇o_2p amplitude was greater (P < 0.05) in YA (0.97 ± 0.27 l/min) than in OA (0.44 ± 0.12 l/min; Fig. 5B), reflecting the greater WR in YA (118 ± 25 W) than in OA (64 ± 10 W; Table 1).

Fig. 5.

Group mean absolute V̇o_2p response profiles to 50 W (A) and 80% of the lactate threshold (θ_L; B) constant-load (CL) exercise in YA (●; n = 10) and OA (○; n = 9). The dashed line indicates the onset of CL exercise.

There were no age-related differences (age main effect, P = 0.093) in phase II τ_V̇o2p between YA and OA for transitions to an absolute WR of 50 W (Fig. 5A) and a relative WR of 80% θ_L (Fig. 5B); however, τ_V̇o2p was greater (intensity main effect, P < 0.05) for transitions to 80% θ_L than to 50 W (Table 3). There was a trend for an age × intensity interaction (P = 0.053), suggesting that in YA, τ_V̇o2p during the 50-W transition was less than τ_V̇o2p for the 80% θ_L transition and less than τ_V̇o2p for exercise transitions in OA (Fig. 5).

Table 3.

Parameter estimates for V̇o_2p, HR, and Δ[HHb] kinetics during the “on” transition to constant-load, moderate-intensity exercise at 50 W and 80 % θ_L in YA and OA

	Phase II τV̇o2p, s	τHR, s	τΔ[HHb], s	TD-Δ[HHb], s	τ′Δ[HHb], s
50 W
YA	22 ± 11	15 ± 8	12 ± 9	15 ± 3	27 ± 7
OA	38 ± 17	46 ± 32*	15 ± 12	15 ± 2	30 ± 13
80% θL
YA	33 ± 11	25 ± 9†	11 ± 4	10 ± 3†	21 ± 6‡
OA	39 ± 17	41 ± 21*	16 ± 10	13 ± 2*	30 ± 10‡

Values are means ± SD. τ, Time constant of the response; HR, heart rate; TD, time delay; τ′_Δ[HHb], sum of effective τ′_Δ[HHb], effective time constant (τ_Δ[HHb] + TD-Δ[HHb]). Parameter estimates for V̇o_2p and HR were based on fitting to the end of exercise, whereas estimates for Δ[HHb] were based on fitting to a time corresponding to 5 × t_V̇o2p.

^*Significantly different from YA (P < 0.05);

^†significantly different from 50 W (P < 0.05);

^‡significantly different from phase II t_V̇o2p (P < 0.05).

HR kinetics.

The rate of adjustment of HR (τ_HR) was greater (P < 0.05) in OA compared with YA during transitions to both 50 W and 80% θ_L. There were no differences in τ_HR between WR conditions in OA; however, YA displayed a greater τ_HR (P < 0.05) during the transition to 80% θ_L compared with 50 W (Table 3).

Δ[HHb] kinetics.

Baseline Δ[HHb] was similar between age groups for CL transitions to 50 W (YA: −5.11 ± 5.29 AU and OA: −6.46 ± 3.99 AU) and 80% θ_L (YA: −4.83 ± 3.77 AU and OA: −6.75 ± 3.63 AU). The Δ[HHb] amplitude was greater (P < 0.05) in YA than in OA; the Δ[HHb] amplitude for YA at 80% θ_L (9.13 ± 5.41 AU) was greater than for YA (P < 0.05) at 50 W (3.23 ± 1.79 AU), and both values were greater than for OA (P < 0.05) at 50 W (1.26 ± 0.66 AU) and 80% θ_L (1.99 ± 0.84 AU), which were not different between exercise conditions (Fig. 6). When expressed relative to the Δ[HHb] amplitude for RI exercise, the normalized Δ[HHb] amplitude for CL corresponded to 12% (50 W) and 35% (80% θ_L) for YA and to 7% (50 W) and 11% (80% θ_L) for OA.

Fig. 6.

Group mean absolute Δ[HHb] response profiles to 50 W (A) and 80% θ_L (B) CL exercise in YA (●; n = 10) and OA (○; n = 9). The dashed line indicates the onset of CL exercise.

The rates of adjustment of Δ[HHb] (τ_Δ[HHb] and τ′_Δ[HHb]) were not different with regard to age and exercise intensity (Fig. 6 and Table 3); in YA, TD-Δ[HHb] was shorter (P < 0.05) at 80% θ_L (10 ± 3 s) than at 50 W (15 ± 3 s), but in OA, there was no difference in TD-Δ[HHb] between exercise intensities (Table 3).

DISCUSSION

The present study is the first to examine the adjustments in Δ[HHb] and V̇o_2p during nonsteady-state RI and steady-state CL exercise in YA and OA. The main findings of the present study were that 1) during RI exercise, after the initial kinetic phase, V̇o_2p (reflecting muscle O₂ utilization) increased linearly with increasing WR, whereas Δ[HHb] increased with a sigmoidal profile in both YA and OA; 2) during RI exercise, the normalized (relative to peak Δ[HHb]) Δ[HHb]-WR profile increased with a greater slope and was shifted leftward (toward lower WRs) in OA compared with YA; 3) the overall rate of adjustment for Δ[HHb] (measured as τ′_Δ[HHb]) was not different between age groups for CL exercise intensities and was faster than the rate of adjustment of V̇o_2p for YA and OA at 80% θ_L, whereas Δ[HHb] and V̇o_2p kinetics were similar in each group during transitions to 50 W; and 4) the fundamental gain (ΔV̇o_2p/ΔWR, the slope of the V̇o_2p-WR relationship within the moderate-intensity domain), a reflection of the O₂ cost of exercise, was lower in OA than in YA during nonsteady-state RI exercise but was not different between OA and YA for both intensities of steady-state CL exercise.

Nonsteady-State RI Exercise

In the present study, whereas V̇o_2p increased linearly relative to WR, Δ[HHb] increased in a nonlinear, approximately sigmoidal manner with WR in both YA and OA. A sigmoidal increase in Δ[HHb] during RI exercise has previously been reported in YA (6, 20), but, to our knowledge, there have been no previous studies describing this pattern of response in OA; although an apparently higher deoxygenation was reported in OA compared with YA during step-incremental exercise by Ferri et al. (22), a detailed analysis of the response pattern was not reported. Assuming that V̇o_2m increases linearly with increasing WR [as shown by the linear increase in V̇o_2p, albeit with a lower slope in OA (see below)], the sigmoidal increase in muscle Δ[HHb] [reflecting the ratio of muscle blood flow (Q_cap) to V̇o_2m] implies nonlinear dynamics between microvascular O₂ delivery (i.e., Q_cap) and V̇o_2m during nonsteady-state RI exercise in both YA and OA, as suggested by Ferreira et al. (20) for YA. While we acknowledge that this pattern of muscle microvascular blood flow changes during nonsteady-state RI exercise is speculative based, indirectly, on the relationship between Δ[HHb] and V̇o_2p data, we are not aware of any current technology that allows continuous, nonsteady-state measures of muscle microvascular blood flow dynamics in exercising humans.

When the Δ[HHb] signal was normalized to the peak Δ[HHb] amplitude during RI exercise, the Δ[HHb] response increased at a faster rate in OA than in YA when expressed relative to absolute WR and was shifted leftward toward a lower absolute WR (Table 2 and Fig. 3A). Similar data have previously been reported by Ferri et al. (22), where a greater Δ[HHb] (vs. WR) was shown in OA compared with YA during step-incremental leg cycling and knee extension exercise to exhaustion. In the present study, at low WRs at the start of RI exercise, when the absolute demand for ATP and O₂ were low, the increase in Δ[HHb] was not different in OA and YA. At higher WRs and ATP demands, the rate of increase in Δ[HHb] was greater in OA than in YA, reflecting greater fractional O₂ extraction (Fig. 2B) and thus a lower ratio of Q_cap to V̇o_2m in OA as V̇o_2m increased.

Therefore, as demonstrated by Ferreira et al. (20), the sigmoidal increase in the Δ[HHb]-WR relationship implies that the adjustment of Q_cap was faster than that of V̇o_2m (which is assumed to increase linearly with increasing WR) at the lower (light-intensity) WRs during RI exercise. The adjustment of Q_cap slowed progressively during heavier intensity WRs until a plateau in the Δ[HHb]-WR relationship occurred, reflecting a relative “matching” in the rise of both Q_cap and V̇o_2m. A greater Δ[HHb] in OA compared with YA at any given absolute WR would be associated with a lower microvascular Po₂ and, according to Fick's law of diffusion, would lower O₂ diffusion into the muscle cell, as suggested by previous findings of Behnke et al. (5) in young and old rats. A slower rate of O₂ diffusion to the muscle mitochondria in OA, presumably, might require a greater delivery of other oxidative substrates (including reducing equivalents, H⁺, ADP, and P_i) and lead to a greater disruption of metabolic “stability” (64), possibly contributing to premature fatigue.

The sigmoid-shaped increase in Δ[HHb] in both OA and YA (Figs. 3 and 4) is dependent on factors influencing the Q_cap-to-V̇o_2m relationship. It has been suggested that at the onset of exercise, an immediate increases in Q_cap (relative to VO_2m) occurs because of effects related to the muscle pump and rapid vasodilatation (8, 42, 56, 60). While these mechanisms may play a role at the onset of a step transition to CL exercise, their effects may be attenuated during RI exercise because the increase in WR above a baseline of 20-W cycling is not “instantaneous” but increases gradually in a “ramp-like” manner (20 W/min in the present study).

With increasing WR during RI exercise, there may be a progressive recruitment of muscle fibers having lower efficiency, as inferred from a greater steady-state O₂ and phosphocreatine (PCr) cost per given change in WR (i.e., V̇o_2p and PCr functional gain, respectively) and slower O₂ utilization kinetics (7, 31, 32, 38, 62), even within a given muscle fiber population. In addition, the steady-state increase in the conduit artery blood flow-to-V̇o_2p ratio is lower and the steady-state Δ[HHb]-to-V̇o_2p ratio is higher during transitions to higher WRs, reflecting increased reliance on O₂ extraction as WR increases, at least within the moderate-intensity domain (38). However, because the O₂ extraction-O₂ utilization relationship increases hyperbolically with increasing WR (21), a gradual “plateauing” in O₂ extraction is expected at higher WRs (and V̇o_2p), which is supported by the flattening of the Δ[HHb]-WR relationship in the present study (Fig. 3, A and B).

Additionally, with increasing WR, there is a progressive recruitment of type II muscle fibers according to the Henneman size principle. Ferreira et al. (21) examined the steady-state Q_cap-to-V̇o_2m relationship in rat muscles having a range of muscle fiber types and oxidative capacities and reported that, although the slope of the relationship was similar among different muscles, the intercept on the Q_cap axis was lower in muscles having a predominantly fast-twitch fiber composition. As a consequence, the hyperbolic O₂ extraction-to-V̇o_2m relationship was shifted upward to a higher O₂ extraction at lower levels of V̇o_2m in these “fast-twitch” muscles (21). Therefore, as muscle type II fiber recruitment increases with increasing WR and V̇o_2m during RI exercise, muscle O₂ extraction might be expected to increase (for a given rate of V̇o_2m) and then plateau, as would be expected given a hyperbolic O₂ extraction-V̇o_2m relationship.

The steady-state conduit artery blood flow-to-V̇o_2p (or WR) relationship, generally, is attenuated in OA compared with YA (especially as intensity increases) (37, 45, 46, 48, 50, 61), although an unchanged relationship has been previously reported (49), especially in older men (45). Also, during the transition to a higher metabolic demand, the adjustment of conduit artery blood flow, HR, and vascular conductance are slower in OA compared with YA (18). Taken together, the lower blood flow and slower conduit artery blood flow kinetics in OA for a given rate of V̇o_2m (or WR) suggests that a higher rate of O₂ extraction (with greater Δ[HHb] and widening of the arterial-venous O₂ content difference) would be required to support the O₂ requirement of the muscle, as previously reported by others during steady-state exercise (15, 18, 27, 37, 46, 50, 61) and during transitions to WRs above moderate intensity (16, 27).

The differences in muscle O₂ extraction (Δ[HHb]) between YA and OA could be attributed to the disparity in relative WR intensities (i.e., a given absolute WR will represent a higher relative intensity and greater percentage of maximal exercise for OA compared with YA). In the present study, the peak WR during RI exercise for OA (215 ± 31 W; Table 2) represents only 64% of the peak WR attained by YA during RI exercise (338 ± 30 W; Table 2). When normalized Δ[HHb] was expressed as a function of relative WR during RI exercise (Table 2 and Fig. 4), there were no differences in the parameter estimates for Δ[HHb] between YA and OA (Table 2), which could be related to the loss of lean muscle tissue that can occur with ageing. For example, it has been reported that for a similar thigh cross-sectional area, total muscle area and quadriceps muscle area are reduced, whereas the area of fat (and nonmuscle) tissue is increased, in OA, reflective of a reduced muscle mass in OA. Also, there is an increased infiltration of fat tissue within the muscle, reflective of reduced lean muscle in OA (43, 44, 51, 52). However, it was shown in single muscle fibers that when peak force and peak power were normalized for fiber size, there were no differences between YA and OA (59). Therefore, for any given absolute WR, in OA, where total lean muscle mass may be reduced compared with YA, this may require a greater percentage of the total lean muscle mass, but, presumably, in the submaximal range of absolute WRs studied here, adequate muscle recruitment is possible. Also, based on the results of the present study, during the RI protocol, the V̇o_2p at any given WR was similar for YA and OA. During the CL protocol, the steady-state V̇o_2p for exercise at the absolute WRs of 20 and 50 W were not different between YA and OA (0.92 and 1.19 l/min, respectively, for YA and 0.89 and 1.18 l/min, respectively, for OA). Thus, insofar as V̇o_2p reflects the rate of O₂ utilization in the exercising muscle, the total muscle mass likely is not an issue, as both OA and YA would recruit a similar muscle mass to generate the power required to complete a given absolute WR. Rather, at the same “relative” exercise intensity during nonsteady-state RI exercise, the relationship between sympathetic activation and parasympathetic withdrawal, or the production and release of vasodilatory metabolites from the exercising muscle, results in a similar reliance on muscle O₂ extraction in YA and OA.

Steady-State CL Exercise

In the present study, although there was a tendency for τ_V̇o2p to be greater in OA (age main effect, P = 0.093), differences in τ_V̇o2p were not significant (50 W: 38 s in and 22 s in YA and 80% θ_L: 39 s in OA and 33 s and YA; Fig. 5 and Table 3). Slowed V̇o_2p kinetics in OA are typically reported for step transitions in WR within the moderate-intensity domain (2, 15, 18, 27, 40, 55, 57). However, in the present study, estimated τ_V̇o2p in 5 of 10 YA was >30 s, and although τ_V̇o2p for healthy YA is generally reported to be ∼20–30 s (15, 55), recent studies (28, 29, 40) have reported τ_V̇o2p estimates of greater than ∼30 s for some YA.

In the present study, we also determined a substantial reserve for muscle O₂ extraction, that is, during moderate-intensity CL exercise for both OA and YA, the steady-state Δ[HHb] amplitude represented only 10–35% of the Δ[HHb] amplitude measured during RI exercise. During the transition to 50-W exercise, Δ[HHb] kinetics (Fig. 6A and Table 3) were not different between YA and OA despite an apparently slower adjustment of V̇o_2m in OA (Fig. 5A and Table 3), reflecting a poorer matching in the adjustment of Q_cap (relative to V̇o_2m) in OA, at least for “instantaneous” step transitions to a higher WR compared with more “gradual” increases in WR associated with RI exercise. However, for both age groups, step transitions to 80% θ_L were associated with a greater τ_V̇o2p (Fig. 5B and Table 3) relative to the overall adjustment of Δ[HHb] (τ′_Δ[HHb]; Fig. 6B and Table 3), suggesting that at higher WRs, adjustments in Q_cap may impose a limitation to the adjustment of V̇o_2m. The mechanisms responsible for this limitation cannot be determined from the measurements of the present study. However, in YA, faster HR kinetics compared with V̇o_2p kinetics were observed for the two CL WRs, whereas HR kinetics were similar or slower than V̇o_2p kinetics during both CL exercise protocols for OA. These data suggest that in OA but not in YA, adjustments of cardiac output (and conduit artery blood flow) may limit V̇o_2m kinetics (and V̇o_2m), whereas the peripheral distribution of blood flow and O₂ delivery within the active muscle microvasculature along with the delivery of oxidative substrate may constrain the adjustment of V̇o_2p in both OA and YA (see Refs. 26–29, 40, and 41).

Age, Functional Gain, and Exercise Efficiency

A novel observation in this study was that the relationship between functional gain (ΔV̇o_2p/ΔWR) and age was different for nonsteady-state RI and steady-state CL exercise. The functional gain reflects the O₂ cost of exercise and is the inverse of exercise efficiency. In the present study, the functional gain calculated during CL exercise was not different between OA (9.9 ml·min⁻¹·W⁻¹) and YA (9.5 ml·min⁻¹·W⁻¹), in agreement with our previously published findings (14–16, 18, 27, 40), and suggests that exercise efficiency is not different between OA and YA. However, for the same group of adults performing RI exercise, the functional gain (calculated as the slope of the V̇o_2p-WR relationship restricted to the moderate-intensity region of RI exercise) was lower in OA (8.7 ml·min⁻¹·W⁻¹) than in YA (10.0 ml·min⁻¹·W⁻¹), reflecting an apparently greater exercise efficiency in OA. We believe this to be the first study to report functional gain in OA during RI exercise. The lower gain in OA than in YA is surprising considering that we consistently found no differences between age groups (14–16, 18, 27, 40), at least when calculated during the steady state of CL exercise. There does not appear to be a consistent trend regarding the relationship between age and exercise efficiency. For example, Tevald and colleagues (58) reported a lower energy cost for twitch and tetanic muscle contractions in OA. Hepple and coworkers (30) found that compared with young rats (8–9 mo), O₂ and ATP costs of contractions were greater in old rats (28–29 mo) but became lower in senescent rats (36 mo), suggesting that changes in efficiency with age might be variable. Conley and colleagues reported lower muscle oxidative capacity, lower mitochondrial volume density, and lower oxidative capacity per mitochondria (11) and reduced mitochondrial coupling efficiency (ATP/O) (9, 39) in older compared with younger animals and humans and that these changes may not be consistent across all muscles studied (9). In a recent review, Conley and coworkers (10) suggested that because of a lower mitochondrial coupling efficiency, the slope of the V̇o_2p-WR relationship would be greater in OA compared with YA (see Fig. 3 in Ref. 10), not lower as in the present study. The lower slope for the V̇o_2p-WR relationship observed in OA in the present study may be related to slower V̇o_2p kinetics in OA such that the rise in V̇o_2p was not able keep pace to the WR increment [and progressively rising O₂ cost per WR increment; see above and Rossiter (53a) for discussion] used during RI exercise in both OA and YA in the present study (i.e., 20 W/min). Therefore, based on the similar functional gain in OA and YA reported in this and our other studies during steady-state CL exercise, we suggest that exercise efficiency is not adversely affected in the apparently healthy OA observed in the present study (age: ∼70 yr). It may be that the impairments in mitochondrial coupling efficiency reported by Conley and coworkers are not manifest at the whole body level until a much older age.

Limitations

A limitation with the NIRS technology used in the present study is that direct comparisons of “absolute” concentration and changes in Δ[HHb] between YA and OA groups and among adults within each of the age groups is not possible because of uncertainties regarding 1) initial values for path length, absorption, and scattering coefficients (i.e., as required when applying the Beer-Lambert law) and 2) whether these values change with exercise and intensity (19). Attenuation of NIR light as it passes through tissue is dependent not only on absorption by the chromophores of interest but also by attenuation due to photon scattering. Because there is a prevalence of scattering within biological tissues, the exact path length traveled by the photons through the tissue (i.e., between the NIRS emitter and detector optodes) is unknown. Whether these coefficients and responses are affected by aging has not been established.

Also, light must first pass through a relatively NIRS-inert adipose tissue layer before penetrating the active muscle layer. The depth of penetration is determined by the separation of NIRS emitter and detector optodes on the skin surface, and thus the actual volume of muscle being interrogated by NIRS is dependent inversely on the adipose tissue (and nonmuscle tissue) thickness. In general, it has been reported that aging is associated with an increase in adiposity (34), a redistribution in the pattern of adiposity (34), an increase in inter- and intramuscular fat (and nonmuscle tissue) deposits (34, 43, 44, 51, 52), and an increase in subcutaneous fat deposits (43, 44, 51, 52), although this may be influenced by the measurement site (34, 52). In the present study, although adipose tissue thickness was not measured, the attenuated Δ[HHb] response in OA may be related, in part, to a greater underlying subcutaneous, inter-, and intramuscular adipose tissue thickness, resulting in a smaller active muscle volume being interrogated.

Therefore, given these uncertainties, in the present study, the NIRS data were normalized for each subject in an effort to account for these effects. As described in methods, data were measured relative to a predetermined steady-state resting baseline and then normalized within the region bounded by values measured for baseline (20 W) cycling and the peak value measured during the RI protocol, thus providing a “functional” physiological normalization. The peak deoxygenation at the limit of tolerance will reflect conditions within a muscle contracting dynamically and having repeated periods (as dictated by the cadence associated with the cycling movement) of very high muscle force production near “maximal/peak” power outputs associated with very heavy-intensity exercise.

Conclusions

During RI exercise, Δ[HHb] increased as a near-sigmoid function relative to WR in both OA and YA. In OA, the slope of the Δ[HHb]-WR relationship was greater and the response was shifted to the left, reflecting greater deoxygenation for a given absolute WR and V̇o_2p. Also, during CL exercise, V̇o_2p kinetics tended to be slower in OA than in YA, whereas Δ[HHb] kinetics were not different between OA and YA. Therefore, the greater Δ[HHb] (reflecting muscle fractional O₂ extraction) for a given change in V̇o_2p (and V̇o_2m) seen in OA than in YA during both RI and CL exercise supports the suggestion that microvascular blood adjustments are attenuated in OA relative to YA during transitions associated with RI and CL exercise. The slowed V̇o_2p kinetics in OA likely also contribute to the lower functional gain calculated during nonsteady-state RI exercise as the functional gain calculated during steady-state CL exercise is not different between OA and YA, suggesting that there is no measureable reduction in exercise efficiency in healthy OA.

GRANTS

This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada, the University of Western Ontario Academic Development Fund, the Canadian Foundation for Innovation, and the Ontario Innovation Trust. Additional support was provided by Standard Life Assurance Company of Canada. J. M. Murias was supported by a doctoral research scholarship from the Canadian Institutes of Health Research.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: B.M.G., J.M.M., M.D.S., D.H.P., and J.M.K. conception and design of research; B.M.G., J.M.M., and M.D.S. performed experiments; B.M.G., J.M.M., and M.D.S. analyzed data; B.M.G., J.M.M., M.D.S., D.H.P., and J.M.K. interpreted results of experiments; B.M.G. prepared figures; B.M.G. drafted manuscript; B.M.G., J.M.M., M.D.S., D.H.P., and J.M.K. edited and revised manuscript; B.M.G., J.M.M., M.D.S., D.H.P., and J.M.K. approved final version of manuscript.