Determinants of skeletal muscle oxygen consumption assessed by near-infrared diffuse correlation spectroscopy during incremental handgrip exercise

Abstract

Near-infrared diffuse correlation spectroscopy (DCS) is a rapidly evolving optical imaging technique for the assessment of skeletal muscle O₂ utilization (mVO₂). We compared DCS-derived determinants of mVO₂ with conventional measures [blood flow by brachial artery Doppler ultrasound and venous O₂ saturation ($\text{S}{\text{V}}_{{\text{O}}_2}$)] in eight volunteers at rest and during incremental handgrip exercise. Brachial artery blood flow and DCS-derived blood flow index (BFI) were linearly related (R² = 0.57) and increased with each workload, whereas $\text{S}{\text{V}}_{{\text{O}}_2}$ decreased from 65.3 ± 2.5% (rest) to 39.9 ± 3.0% (light exercise; P < 0.01) with no change thereafter. In contrast, DCS-derived tissue O₂ saturation decreased progressively with each incremental stage (P < 0.01), driven almost entirely by an initial steep rise in deoxyhemoglobin/myoglobin, followed by a linear increase thereafter. Whereas seemingly disparate at first glance, we believe these two approaches provide similar information. Indeed, by plotting the mean convective O₂ delivery and diffusive O₂ conductance, we show that the initial increase in mVO₂ during the transition from rest to exercise was achieved by a greater increase in diffusive O₂ conductance versus convective O₂ delivery (10-fold vs. 4-fold increase, respectively), explaining the initial decline in $\text{S}{\text{V}}_{{\text{O}}_2}$. In contrast, the increase in mVO₂ from light to heavy exercise was achieved by equal increases (1.8-fold) in convective O₂ delivery and diffusive O₂ conductance, explaining the plateau in $\text{S}{\text{V}}_{{\text{O}}_2}$. That DCS-derived BFI and deoxyhemoglobin/myoglobin (surrogate measure of O₂ extraction) share the same general biphasic pattern suggests that both DCS and conventional approaches provide complementary information regarding the determinants of mVO₂.

NEW & NOTEWORTHY Near-infrared diffuse correlation spectroscopy (DCS) is an emerging optical imaging technique for quantifying skeletal muscle O₂ delivery and utilization at the microvascular level. Here, we show that DCS provides complementary insight into the determinants of muscle O₂ consumption across a wide range of exercise intensities, further establishing the utility of DCS.

INTRODUCTION

Prior research, using invasive hemodynamic measures, has assessed the determinants of skeletal muscle O₂ consumption (mVO₂) during large (cycling) and small (single leg-knee extension) muscle-mass exercise in healthy individuals and clinical populations (12, 19, 31, 35). Near-infrared diffuse correlation spectroscopy (DCS) is an emerging, noninvasive technique for the simultaneous assessment of O₂ delivery and use at the microvascular level (1, 7, 15, 16), with recent application in exercising skeletal muscle (2, 15, 36). Our group has reported that the DCS-derived skeletal muscle blood flow index (BFI; i.e., muscle perfusion) and conventional measures of convective O₂ delivery (measured by Doppler ultrasound) are highly related in terms of the magnitude and rate of change during exercise (2, 36). Moreover, we have shown good agreement between DCS-derived measures of mVO₂ compared with conventional measurements made at the macrovascular level (36), at least during a single-workload exercise challenge.

Despite the above-mentioned similarities between DCS and conventional approaches, however, we and others have also observed several key differences that highlight the added utility of this technology. For example, when perfusion pressure was experimentally reduced (by raising the exercising limb above the heart), steady-state, Doppler-derived bulk conduit blood flow decreased significantly, whereas DCS-derived, steady-state microvascular perfusion was maintained across conditions, which we interpreted as evidence of microvascular autoregulation (36). A similar observation was also made by Hammer et al. (15), who found DCS-derived microvascular perfusion to be uncoupled from convective O₂ delivery during heavy incremental exercise.

Given that our prior observation was limited to a single, moderate-intensity exercise workload (36) and because Hammer et al. (15) did not directly measure skeletal mVO₂, we hypothesized that changes in skeletal muscle O₂ delivery (related to macro- vs. microvascular uncoupling or otherwise) would be compensated for by greater O₂ extraction and that DCS would closely track changes measured directly from the venous effluent. To test this hypothesis, we compared direct macrovascular (Doppler-derived flow and O₂ extraction) and indirect microvascular (DCS-derived BFI and tissue saturation) measures across a wide range of exercise intensities using an incremental handgrip exercise test. The data highlight key limitations associated with reliance solely on the arterial-venous O₂ difference to reflect changes in O₂ extraction and why DCS-derived tissue measures provide important, complementary insight into the determinants of mVO₂.

METHODS

Ethical Approval and Subjects

The study was approved by the Institutional Review Board for research involving human subjects at the University of Texas at Arlington. All subjects provided signed consent after receiving a verbal and written description of the experimental protocol and potential risks.

Ten healthy, recreationally active, young (18- to 35-yr-old) men, all of whom volunteered for our previous study (36), participated in the present investigation. Aside from the participant characteristics, none of the data reported previously are herein included. Moreover, unlike the single-workload data presented previously, the data herein evaluate the determinants of mVO₂ across a range of exercise intensities (i.e., incremental exercise test). Exclusion criteria included the following: body mass index < 18.5 or > 35 kg/m²; handgrip maximal voluntary contraction (MVC) < 40 kg or > 65 kg (to limit subject heterogeneity); female sex (to avoid large interindividual differences in the relative exercise intensities across the incremental exercise test); history of cardiovascular, pulmonary, or metabolic disease; anemia; current medication or tobacco use; orthopedic limitations; poor venous access; and poor acoustic window for brachial artery imaging by ultrasound.

Experimental Design

Screening and familiarization.

All subjects participated in a separate familiarization visit and underwent dual-energy X-ray absorptiometry (DXA) to quantify fat/lean mass.

Experimental testing visit.

Experimental testing visits were performed on a separate day, after the initial screening/familiarization visit. All studies were performed in a quiet, temperature (~22°C)- and ambient light-controlled room. Subjects were studied in a fasted state, having abstained from alcohol and vigorous exercise for >24 h and caffeine for >12 h. Upon arrival at the laboratory, subjects were positioned supine on a bed and asked to rest quietly while a retrograde intravenous catheter was inserted into a deep vein in the forearm of the exercising (nondominant) arm. Subjects were then instrumented for measurement of continuous blood pressure, heart rate, arterial O₂ saturation ($\text{S}{\text{a}}_{{\text{O}}_2}$), and brachial blood flow (described in detail below). A Smedley handgrip dynamometer was positioned next to the subject’s exercising arm so that the arm could comfortably be extended and supported. Once fully instrumented, subjects performed an incremental rhythmic handgrip exercise test, using a 50% contraction duty cycle (2 s contraction, 2 s relaxation) at a rate of 15 contractions per min. The first workload started at 10 kg and progressed 3 kg every 3 min thereafter. Simultaneous measurements of O₂ delivery and use by both approaches (conventional and DCS derived) were performed throughout. Exercise was guided by a monitor that displayed the handgrip force output to give the subject visual feedback on force production and a recorded voice prompt to guide contraction/relaxation. The exercise test was terminated volitionally by the subject or by the investigator when brachial artery image quality was compromised, or the subject was unable to achieve the prescribed force target despite verbal encouragement to do so.

Instrumentation and Measurements

Anthropometrics.

DXA was used to determine whole-body body-fat percent, fat mass, and fat-free mass (DXA Lunar Prodigy; GE Healthcare, Little Chalfont, UK). A DXA-certified radiology technician also constructed a region of interest from the left arm antecubital fossa through the fingertips on the left hand to quantify forearm lean mass. Height and weight were measured with a dual-function stadiometer and weighing scale (Professional 500KL; Health o meter, McCook, IL). Forearm adipose tissue thickness was measured with skinfold calipers (Slim Guide; Creative Health Systems, Plymouth, MI) on the left arm at the location of the flexor digitorum profundus.

Central hemodynamic responses.

Beat-by-beat arterial blood pressure was measured from a small finger cuff placed around the middle finger of the subject’s nonexercising hand using photoplethysmography (Finometer PRO; Finapres Medical Systems, Arnhem, The Netherlands) that was calibrated to an automated brachial artery blood pressure cuff (71WX-B Connex Spot Monitor; Welch Allyn Connex, Skaneateles Falls, NY). Mean arterial pressure (MAP) was calculated as the mean pressure across a continuous average of arterial waveforms. Heart rate was measured via three-lead electrocardiography using standard CM₅ placement of electrocardiography electrodes (MLA 0313; ADInstruments, Colorado Springs, CO). $\text{S}{\text{a}}_{{\text{O}}_2}$ was noninvasively obtained by placement of a pulse oximeter (ML 320; ADInstruments) over the index finger of the subject’s nonexercising hand. The analog outputs for arterial blood pressure, heart rate, and $\text{S}{\text{a}}_{{\text{O}}_2}$ modules were connected to a high-performance physiological data acquisition system (PowerLab 16/35; ADInstruments) for simultaneous data recording.

Forearm blood flow.

Brachial artery blood flow velocity and diameter were measured with a duplex ultrasound system (Vivid i; GE Healthcare) on the upper-left portion of the exercising arm (proximal to the antecubital fossa). This ultrasound system had a 12-MHz linear array probe with 60° of insonation. The ultrasound gate was optimized to ensure complete insonation of the entire vessel cross-section with constant intensity. The continuous Doppler audio signal was converted to real-time blood flow velocity waveforms using a validated Doppler audio converter (17) and recorded using a PowerLab data acquisition system (ADInstruments). Brachial artery diameter was measured with B-mode ultrasound imaging, with measurements made during the resting baseline and during the final 30 s of each 3-min exercise stage. The analog outputs from the Doppler ultrasound module, along with the amplified signal from the handgrip dynamometer, were connected to the data acquisition system, detailed above, for simultaneous data recording.

Forearm deep venous blood sampling.

An 18-gauge intravenous catheter was inserted retrograde to venous blood flow into a deep forearm vein. Confirmation that the selected vein drained from the active muscle group of interest (flexor digitorum profundus) was obtained using ultrasound and/or a portable vein illuminator imaging device (VeinViewer Flex; Christie Medical Holdings Inc., Memphis, TN) before insertion of the catheter. The catheter was placed in this manner to improve the likelihood of sampling venous effluent draining from the exercising muscle, whereas also decreasing the potential contribution from the nonexercising muscle and/or the skin. Blood samples were taken during the resting baseline and during the final 30 s of each 3-min exercise stage of the incremental handgrip exercise test to measure venous O₂ content ($\text{C}{\text{v}}_{\text{O}}{}_{{}_2}$). A 3-mL discard was drawn before the 2-mL blood sample. A 2-mL saline flush followed each sample to prevent the catheter from clotting.

Near-Infrared DCS

Detailed methodology of our in-house DCS system has previously been published (2, 36). Briefly, the system consists of two continuous-wave, long coherence-length laser diodes (785 and 852 nm; CrystaLaser Inc., Reno, NV) that are alternatively switched (Micro-Electro-Mechanical System 2 × 2 Blocking Switch; DiCon Fiberoptics Inc., Richmond, CA) and a single photon-counting avalanche photodiode (APD) as the photon detector (SPCM-AQRH-14-FC; Pacer USA LLC, Palm Beach Gardens, FL). The output of the APD is connected to a computer with a 32-bit, eight-channel data acquisition card (PCI-6602; National Instruments Corp., Austin, TX). A LabVIEW (National Instruments Corp.) program was developed for photon counting. A software autocorrelator calculates the autocorrelation function and absolute intensity (sum of photon counts) of diffused light (10). A three-dimensional printed probe was used to hold a multi-mode fiber (125 μm in core diameter) from the optical switch and a single-mode fiber (5 μm in core diameter) to the APD detector. The probe was affixed to the forearm of the exercise arm, directly over the belly of the flexor digitorum profundus, using Velcro strips. The source-to-detector distance was 2.5 cm. The data sampling rate was 0.25 Hz.

DCS determines a BFI based on the autocorrelation function of light reflectance, measured at each of the two near-infrared wavelengths (785 and 852 nm) and averaged. Then, relative muscle blood flow from the initial baseline at t₀ (rMBF) was calculated as the following:

$$\text{rMBF}=\frac{\text{BFI}\left(t\right)}{\text{BFI}\left(t_0\right)}\times 100\%$$ (1)

Based on conventional near-infrared spectroscopy (NIRS) methods, the absolute intensity of light reflectance was also used to determine the hemoglobin concentrations. First, the absolute light intensity measured at each wavelength over time, I(t), was converted to the changes in optical density (ΔOD) as the following:

$$\Delta \text{OD}={\text{log}}_{10}\left[\text{I}\left(t_0\right)/\left(t\right)\right]$$ (2)

Then two ΔOD outputs measured, respectively, at 785 and 852 nm were used to quantify the relative changes in oxygenated hemoglobin/myoglobin (HbO₂) and deoxygenated hemoglobin/myoglobin (Hb) concentrations, ΔHbO₂(t) and ΔHb(t), based on the modified Beer-Lambert Law (8). Finally, the change in total hemoglobin concentration [ΔHbT(t)] was derived as the following:

$$\Delta \text{HbT}\left(t\right)=\Delta {\text{HbO}}_2\left(t\right)+\Delta \text{Hb}\left(t\right)$$ (3)

Since conventional NIRS only quantifies the relative changes in hemoglobin concentrations from an initial baseline, the hemoglobin baseline values [HbO₂(t₀), Hb(t₀), and HbT(t₀)] were measured before the experiment with a frequency domain near-infrared tissue oximeter (OxiplexTS; ISS Inc., Champaign, IL), and then the real-time absolute hemoglobin concentrations were derived as the following:

$${\text{HbO}}_2\left(t\right)={\text{HbO}}_2\left(t_0\right)+\Delta {\text{HbO}}_2\left(t\right)$$

$$\text{Hb}\left(t\right)=\text{Hb}\left(t_0\right)+\Delta \text{Hb}\left(t\right)$$

$$\text{HbT}\left(t\right)=\text{HbT}\left(t_0\right)+\Delta \text{HbT}\left(t\right)$$ (4)

Furthermore, the real-time tissue O₂ saturation [StO₂(t)] was derived as the following:

$${\text{StO}}_2\left(t\right)=\frac{{\text{HbO}}_2\left(t\right)}{{\text{HbO}}_2\left(t\right)+\text{Hb}\left(t\right)}100\%$$ (5)

With the combination of the ΔHbO₂ and ΔHb changes from NIRS and the rMBF change from DCS, the relative change of skeletal muscle metabolic rate of O₂ (MRO₂) was then calculated as the following (6):

$${\text{MRO}}_2=\text{rMBF}\left[1+\frac{\Delta \text{Hb}\left(t\right)}{\text{Hb}\left(t_0\right)}\right]{\left[1+\frac{\text{HbT}\left(t\right)}{\text{HbT}\left(t_0\right)}\right]}^{-1}$$ (6)

Data Analysis

Doppler ultrasound, arterial blood pressure, heart rate, and DCS-derived variables were measured throughout rest, exercise, and postexercise recovery. To minimize the muscle-fiber motion artifact during exercise, DCS data were recorded only during the relaxation phase of the handgrip duty cycle, as previously described (2, 14). Accordingly, brachial artery blood flow measurements were also only made during the relaxation phase of the handgrip duty cycle to match DCS. Venous blood sampling was limited to a single resting baseline and a single sample taken during the final 30 s of each incremental workload. With the exception of data presented continuously in figures, resting baseline values represent a 1-min average of the final minute of a 2-min baseline, and exercise values represent the final 30 s of each 3-min exercise stage.

Venous blood constituents.

An i-STAT analyzer (Abbott Point of Care, Princeton, NJ) was used to analyze the whole blood sample for hemoglobin and forearm venous O₂ saturation ($\text{S}{\text{V}}_{{\text{O}}_2}$) using CG8+ test cartridges (Abbott Point of Care). All blood samples were analyzed in duplicate and averaged.

Calculated physiological variables.

Doppler ultrasound, arterial blood pressure, heart rate, and $\text{S}{\text{a}}_{{\text{O}}_2}$ data were analyzed offline by a single observer (R.R.), using the LabChart Pro software environment (ADInstruments). For real-time arterial blood pressure waveforms, a peak-detection algorithm was applied to identify the systolic and diastolic components as the peak and valley of each arterial wave. Then, MAP (in millimeters of mercury) was calculated on a beat-to-beat basis. Similarly, mean blood flow velocity (in centimeters per second) was calculated as the beat-to-beat average from the real-time Doppler ultrasound waveforms. Forearm brachial artery blood flow (FBF) was calculated as [mean blood flow velocity × π  (brachial artery diameter/2)²] × 60, where brachial artery diameter (in centimeters) was measured with B-mode ultrasound imaging. FBF was indexed to lean forearm mass and reported as milliliters per minute per 100 grams lean forearm tissue. Forearm vascular conductance was calculated as FBF/MAP × 100 mmHg. Arterial O₂ concentration ($\text{C}{\text{a}}_{\text{O}}{}_{{}_2}$) was calculated as [($\text{S}{\text{a}}_{{\text{O}}_2}$ × Hb × 1.36) + 0.003 × $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$], where $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ is the partial pressure of O₂ in arterial blood and was assumed to be 100 mmHg (4). Venous blood samples were used to calculate $\text{C}{\text{v}}_{\text{O}}{}_{{}_2}$ using the formula [($\text{S}{\text{V}}_{{\text{O}}_2}$ × Hb × 1.36) + 0.003 × ${\text{Pv}}_{{\text{O}}_2}$], where ${\text{Pv}}_{{\text{O}}_2}$ is venous O₂ partial pressure. mVO₂ was calculated using the Fick equation: FBF × (_{$\text{C}{\text{a}}_{\text{O}}{}_{{}_2}$} − $\text{C}{\text{v}}_{\text{O}}{}_{{}_2}$). Arterial and venous O₂ transport were calculated as FBF × _{$\text{C}{\text{a}}_{\text{O}}{}_{{}_2}$} and $\text{C}{\text{v}}_{\text{O}}{}_{{}_2}$, respectively (11).

NIRS and DCS.

All raw DCS and NIRS data were analyzed in MATLAB (Version R2016A; MathWorks Inc., Natick, MA) and exported to Microsoft Excel (Microsoft Corp., Redmond, WA) for subsequent analyses.

Statistics

All statistical analyses were performed with SPSS Software (SPSS 24.0; IBM Corp., Armonk, NY) and Prism, version 8 (GraphPad Software, La Jolla, CA). For all statistical tests, significance was accepted at P < 0.05. All data are presented as means ± SE, unless otherwise noted.

To examine exercise intensity-dependent changes in physiological variables, we used a one-way, repeated-measures ANOVA. As described above, baseline was defined as the 1-min average before exercise, whereas the final 30-s average of each exercise workload was used for comparative purposes. For all repeated-measures ANOVA testing, if the sphericity assumption was violated (Greenhouse-Geisser Ɛ < 0.75), degrees of freedom values were adjusted using the Greenhouse-Geisser correction. When significant overall effects were observed, a Bonferroni correction post hoc analysis was performed to determine where significant differences existed.

To examine the overall relationship between Doppler-derived brachial artery blood flow and DCS-derived BFI, we modeled the kinetic responses illustrated in Fig. 1, A and D, using an exponential growth model with the time constant representing the amount of time required to change 63% of the amplitude or range of the response. In addition, to assess the intra-individual relationship between Doppler-derived brachial artery blood flow and DCS-derived BFI across exercise intensity, we performed intra-individual linear-regression analysis. Similarly, in inter-individual relationships between Doppler-derived blood flow and DCS-derived BFI, deoxyhemaglobin/myoglobin and DCS-derived blood flow index, and DCS-derived relative mVO₂ and conventionally measured mVO₂, Pearson correlations were performed.

Fig. 1.

A–C: conventional measures of muscle O₂ consumption. A: Doppler-derived brachial artery blood flow. B: arterial-venous O₂ difference. C: skeletal muscle O₂ consumption (V̇o₂). D–F: diffuse correlation spectroscopy (DCS)-derived determinants of relative muscle O₂ consumption.D: DCS-derived blood flow index. E: near-infrared spectroscopy-derived skeletal muscle tissue saturation. F: relative muscle V̇o₂. Rhythmic handgrip began with a fixed workload of 10 kg and progressed by 3 kg every 3 min thereafter. Arrows depict the onset of exercise. Data are reported as means ± SE; n = 8.

RESULTS

Ten subjects volunteered to participate in the study; however, two of these subjects were excluded from the final analysis due to technical difficulties with our DCS device. Subject characteristics for the eight subjects included are presented in Table 1.

Table 1.

Subject characteristics

Variable	Means ± SD
Age, yr	27 ± 4
Height, cm	181.0 ± 6.3
Weight, kg	88.6 ± 16.4
BMI, kg/m2	26.9 ± 4.3
Total body fat, %	23.2 ± 9.4
Left forearm total mass, g	1,689 ± 195
Left forearm lean mass, g	1,551 ± 185
Left forearm adipose tissue thickness, mm	5 ± 1
MVC, kg	53 ± 8

BMI, body mass index; MVC, maximal voluntary contraction; n = 8.

All subjects completed the 19-kg workload; however, only three subjects progressed beyond this workload, with one subject reaching 25 kg and two subjects reaching 22 kg. For comparative purposes, we chose to report data only on the workloads completed by all subjects.

Conventional Determinants of mVO₂

Brachial artery blood flow increased twofold from rest to 10 kg rhythmic handgrip exercise (P < 0.01), followed by a linear increase with each subsequent workload (Fig. 1A). Likewise, the arterial-venous O₂ difference also increased nearly twofold with the onset of exercise (from rest to 10 kg; P < 0.001), however in contrast to brachial artery blood flow, remained unchanged thereafter (Fig. 1B). Of note, the change in the arterial-venous O₂ difference was entirely driven by a decrease in $\text{S}{\text{V}}_{{\text{O}}_2}$ (which changed from 65.3 ± 2.5% at rest to 39.9 ± 3.0% with 10 kg rhythmic handgrip), as we observed no major differences in hemoglobin or arterial saturation. Accordingly, mVO₂ increased fourfold with the onset of exercise (P < 0.001), with subsequent linear increases seemingly driven entirely by brachial artery blood flow (Fig. 1C).

DCS-Derived Determinants of mVO₂

Similar to Doppler-derived brachial artery blood flow, DCS-derived BFI shared a steep rise with the onset of exercise, followed by a linear increase thereafter (Fig. 1D). Likewise, similar to the directly measured $\text{S}{\text{V}}_{{\text{O}}_2}$, DCS-derived tissue saturation significantly decreased with the onset of exercise; however, in contrast to the directly measured $\text{S}{\text{V}}_{{\text{O}}_2}$, tissue saturation continued to decline with each subsequent workload (Fig. 1E). These changes in tissue O₂ saturation were driven almost entirely by deoxyhemoglobin/myoglobin (a surrogate measure of O₂ extraction), which increased rapidly during the onset of exercise, followed by a progressive linear increase thereafter (Fig. 2B). Oxyhemoglobin/myoglobin remained unchanged throughout exercise (Fig. 2A); as a result, total hemoglobin/myoglobin increased throughout the exercise challenge (Fig. 2C; P < 0.01). Once integrated, relative mVO₂ shared a similar increase across the incremental challenge, as conventionally measured mVO₂ (Fig. 1F).

Fig. 2.

Near-infrared-derived oxygenated (Oxy; A), deoxygenated (Deoxy; B), and total (C) hemoglobin (Hb)/myoglobin (Mb) at rest and throughout the incremental handgrip exercise protocol. Rhythmic handgrip began with a fixed workload of 10 kg and progressed by 3 kg every 3 min thereafter. Data are reported as means ± SE; n = 8. Oxygenated hemoglobin/myoglobin remained unchanged from rest throughout exercise, whereas deoxygenated and total hemoglobin/myoglobin increased with graded exercise (P < 0.01 for both).

Fick Principle and Fick’s Law of Diffusion

To help interpret the apparent discrepancy between the determinants of mVO₂ when assessed by the conventional approach versus DCS-derived MRO₂, we plotted the mean convective and diffusive components of O₂ transport between two distinct phases in the incremental exercise test: 1) rest to 10 kg of force (Fig. 3A) and 2) 10–19 kg of force (Fig. 3B). The Fick principle curve depicts mVO₂ as a function of convective O₂ delivery versus venous partial pressure of O₂ (Po₂. Fick’s law of diffusion depicts O₂ utilization (V̇o₂) as a function of venous Po₂ with the slope representative of muscle O₂ diffusive conductance (20). The intersection of these lines determines the V̇o₂ achieved. As illustrated, the transition from rest to 10 kg rhythmic handgrip exercise was achieved by a greater increase in diffusive O₂ conductance (10-fold increase) compared with convective O₂ delivery (fourfold increase) that resulted in increased O₂ extraction. The increase in mVO₂, from 10 to 19 kg, was achieved by a 1.8-fold increase in both convective O₂ delivery and diffusive O₂ conductance that resulted in minimal change in the arterial-venous O₂ difference. These models are supported by the expression of the proportionate changes in both arterial and venous O₂ transport across each exercise workload and the calculation of their respective difference [i.e., V̇o₂; Fig. 3C]. Indeed, the greatest increase in V̇o₂ occurs during the transition from rest to 10 kg of rhythmic handgrip. During this transition, because diffusive O₂ conductance increases to a greater extent than convective O₂ delivery, an appreciable increase in O₂ extraction (i.e., arterial-venous O₂ difference) is observed. However, after this first 10-kg workload, when diffusive and convective O₂ conductance are matched, no further changes in O₂ extraction (i.e., arterial-venous O₂ difference) are observed. This is supported at the microvascular level using DCS. Indeed, during the transition from 10 to 19 kg, absolute O₂ extraction increases with each transition (defined by the rise in deoxyhemoglobin/myoglobin; Fig. 2B) but does so at a similar rate as O₂ delivery (i.e., DCS-derived BFI; Fig. 1D), thus preserving the arterial-venous O₂ difference.

Fig. 3.

The Fick principle curve depicts muscle O₂ consumption (mVO₂) as a function of convective O₂ delivery (curved line) vs. venous partial pressure of O₂ (Po₂). Fick’s law of diffusion depicts mVO₂ as a function of venous Po₂, with the slope (straight line) representative of muscle O₂ diffusive conductance. The intersection of these lines determines the mVO₂ achieved. A: the transition from rest to 10 kg of rhythmic handgrip exercise was achieved by a greater increase in diffusive O₂ conductance compared with convective O₂ delivery. B: the increase in mVO₂ from 10 to 19 kg was achieved by an increase in both convective O₂ delivery and diffuse O₂ conductance that resulted in a minimal change in venous Po₂. C: the arterial (open bars) and venous (closed bars) O₂ transport at rest and at the end of each respective exercise workload. O₂ utilization (O₂ Util.) was calculated as the differences between arterial and venous O₂ transport, expressed in milliliters of O₂ per minute. Data are reported as means ± SE; n = 8.

Relationships Between Conventional and DCS-Derived Determinants of mVO₂

As mentioned, brachial artery blood flow and DCS-derived BFI shared a similar response, increasing sharply with the onset of exercise, followed by a linear increase thereafter (Fig. 1, A and D). To assess the agreement between these two variables, we performed nonlinear-regression modeling. Consistent with our prior reports (2, 36), Doppler-derived brachial blood flow and DCS-derived BFI were very closely related, sharing very similar time constants (264 vs. 270 s, respectively), along with large overlap between their respective 95% confidence intervals (216–324 vs. 192–414 s, Doppler vs. DCS, respectively). In addition to the evaluation of the goodness of fit, we evaluated the within-subject agreement between the two approaches. As illustrated in Fig. 4A, we observed excellent within-subject agreement between Doppler-derived brachial blood flow and DCS-derived BFI (correlation coefficients ranging from 0.79 to 0.99, median = 0.97). We also observed good between-subject agreement between Doppler-derived brachial blood flow and DCS-derived BFI (r² = 0.57; Fig. 4A).

Fig. 4.

A: relationship between Doppler-derived brachial artery blood flow and diffuse correlation spectroscopy (DCS)-derived blood flow index (BFI) across the incremental exercise test. B: relationship between DCS-derived BFI and near-infrared spectroscopy-derived deoxy hemoglobin (Hb) + myoglobin (Mb; an indicator fractional O₂ extraction). C: relationship between conventional muscle O₂ utilization (mVO₂) and DCS-derived metabolic rate of O₂ (MRO₂). Individuals are color coded across each panel to highlight intra-individual relationships, which ranged between 0.62 and 0.99 for A, 0.76 and 0.98 for B, and 0.54 and 0.98 for C.

Similar to the comparisons made between Doppler-derived and DCS-derived blood flow, we also assessed relationships between DCS-derived BFI and deoxyhemoglobin/myoglobin [a measure of tissue O₂ extraction (9, 13); Fig. 4B]. The tight relationship observed supports the conventionally derived data described immediately above. Together, the data show that when diffusive O₂ conductance is greater than convective O₂ delivery (i.e., rest to 10 kg), the change in mVO₂ is driven largely by O₂ extraction (i.e., increased deoxyhemoglobin/myoglobin), whereas when diffusive and convective O₂ conductance is similar (i.e., 10–19 kg), the change in mVO₂ is driven by seemingly proportional changes in O₂ delivery (DCS-derived BFI) and O₂ extraction.

Consistent with the close linear relationships already mentioned, we also observed good agreement between conventionally derived mVO₂ and DCS-derived MRO₂. Indeed, within-subject correlation coefficients ranged from 0.54 and 0.98, with a between-subject agreement of 0.68 (Fig. 4C).

DISCUSSION

DCS is emerging as a noninvasive optical imaging technique for quantifying skeletal muscle O₂ delivery and mVO₂ at the microvascular level. With the comparison of DCS with conventional measures of convective O₂ delivery and mVO₂, we show that the two approaches are far more similar than they are different. The data also highlight a common misconception of the Fick principle (25, 28), namely, that increases in skeletal mVO₂ are almost entirely dependent on convective O₂ delivery, and show instead an equal contribution from diffusive O₂ conductance. Taken together, these data highlight the potential role that DCS can play, both clinically and as a research tool, for extending our pathophysiologic understanding of exercise (in)tolerance across the spectrum of health and disease noninvasively.

The observed plateau in $\text{S}{\text{V}}_{{\text{O}}_2}$ and the arterial-venous O₂ difference are in line with several recent reports involving handgrip exercise (5, 23, 24). Yet, this observed plateau is in direct contrast with the DCS-derived tissue-saturation measurement, which showed a linear decline throughout the incremental exercise challenge. Each technique therefore provided—at least at first glance—seemingly different interpretations of the determinants of mVO₂. That the arterial-venous O₂ difference was unchanged during exercise performed at workloads >10 kg suggested that the increase in mVO₂ was entirely driven by proportional increases in convective O₂ delivery. However, after more detailed analysis incorporating Fick’s law of diffusion, one can appreciate that the increase in mVO₂ is also driven by an increase in muscle O₂ diffusive conductance (Fig. 3, A and B). Whereas the exact mechanism driving these changes is beyond the scope of the present investigation, the initial increase in diffusive O₂ conductance is likely attributable to a rapid decrease in intramuscular Po₂ at the onset of exercise, which has been shown to reach its nadir even at submaximal workloads (22, 31, 32). Increases in the diffusive O₂ conductance are also likely to be attributable to longitudinal capillary recruitment and an increased capillary hematocrit (26, 27). Given the observed increase in NIRS-derived total hemoglobin in our participants, it is interesting to speculate that capillary hematocrit concentration likely played an important role (3).

Importantly, if diffusive O₂ conductance was not matched with the marked linear increase in O₂ delivery observed, then the arterial-venous O₂ difference would have declined during the transitions from 10 to 19 kg (i.e., an effective shunt). Alternatively, had diffuse O₂ conductance increased beyond the rate of O₂ delivery (as was the case during the transition from rest to 10 kg), then the arterial-venous O₂ difference would have continued to increase throughout exercise. That the arterial-venous O₂ difference did not increase with each subsequent workload suggests that muscle O₂ extraction was matched with muscle O₂ delivery. Indeed, the close linear relationship between DCS-derived BFI and NIRS-derived deoxyhemoglobin/myoglobin [an indirect measure of muscle O₂ extraction (9, 13)] supports this interpretation. Because the arterial-venous O₂ difference is simply a fractional representation of O₂ extraction, it does not provide insight into the absolute amount of O₂ being extracted. To determine this, one needs to account for absolute O₂ delivery (as illustrated in Fig. 3C). As such, when both convective O₂ delivery and diffusive O₂ conductance increase proportionately (as was the case in our study between 10 and 19 kg), the arterial-venous O₂ difference remains constant. Accordingly, both approaches (conventional vs. DCS) seem to provide complementary insight into the determinants of V̇o₂.

That we observed a close relationship and excellent within-subject agreement between DCS-derived muscle BFI and Doppler-derived forearm blood flow is entirely consistent with a growing number of publications (2, 36). Whereas these data support DCS-derived BFI as a viable approach for the measurement of microvascular skeletal muscle O₂ delivery, we recognize that others have not found the same tight relationship. For example, Hammer et al. (15) found a clear dissociation between DCS-derived skeletal muscle perfusion and Doppler-derived forearm blood flow during incremental handgrip exercise. However, in the present investigation, the highest workload achieved by all of the subjects was 19 kg, which was ~40% of their MVC. In contrast, Hammer et al. (15) consistently achieved intensities above 50% of MVC. In addition, in the present investigation, data were collected continuously at 0.25 Hz throughout rhythmic handgrip exercise. In contrast, Hammer et al. (15) used a unique postexercise data collection approach, which may have contributed to the differences between these two studies. Caution is therefore warranted with the interpretation of the tight relationship between Doppler-derived blood flow and DCS-derived BFI presented herein, as differences may exist between conduit artery and microcirculatory blood flow kinetics and may differ depending on muscle mass and exercise intensity. It is also interesting to note that differences were observed in the magnitude of change between DCS-derived BFI (~2.5- to 3-fold change at 19 kg) and Doppler-derived brachial artery blood flow (~7- to 10-fold change at 19 kg). We interpret this difference to reflect the indiscriminant nature of conduit blood vessels, which supply blood to the entire limb, compared with the muscle-specific increases reflected by DCS.

Experimental Considerations

This study is not without limitation. We cannot completely rule out the contribution of oxy- and deoxymyoglobin to the changes in the NIRS signal observed throughout exercise. Indeed, NIRS is incapable of differentiating between hemoglobin and myoglobin, and myoglobin has indeed previously been shown to deoxygenate during heavy-intensity rhythmic exercise (33). However, because oxymyoglobin decreases rapidly during the onset of exercise, before reaching a steady plateau (despite greater intensity exercise) (33), we contend that the NIRS changes reported herein were likely driven by deoxy- and oxyhemoglobin; however, since we cannot be sure, we have reported our NIRS measurements as a grouped variable (i.e., Hb + Mb).

Additionally, we only report data for the workloads for which we had a complete data set. In the majority of cases, the primary reason for termination of the test was poor Doppler data quality secondary to excessive upper-arm movement (alternative muscle recruitment). It is possible that we have omitted important micro- versus macrovascular differences in O₂ delivery, as previously highlighted by Hammer et al. (15).

Despite efforts to optimize our venous sampling—the selection of a deep-vein draining from the most active region of the forearm during exercise and the insertion of the catheter in a retrograde fashion—we cannot rule out that a portion of the venous sample originated from muscles other than the flexor digitorum profundus. Consequently, the venous blood samples may partly reflect forearm mVO₂ from nonexercising tissue. This is in contrast to the DCS-derived measures, which reflect a volume-weighted measure of perfusion and extraction. Attention is also warranted with the consideration of the sample time between our DCS measures and the venous sampling. As stated, DCS data were only taken during the 2-s relaxation period, whereas venous sampling was not limited to a specific phase of the handgrip duty cycle. Indeed, the rhythmic nature of handgrip exercise inevitably imposes a pulsatile pattern on both the blood flow and O₂ extraction characteristics, which cannot be accounted for by the approach used herein. Despite this potential consideration, however, the strong relationships between DCS-derived BFI and NIRS-derived deoxyhemoglobin/myoglobin, illustrated in Fig. 4B, and its close approximation of the arterial-venous O₂ difference measure by conventional approaches suggest that this confounding factor likely had a minimal impact.

Previous work by McDonough et al. (21) showed that to meet exercise-dependent increases in O₂ demand, fast-twitch muscle fibers rely more heavily on increases in fractional O₂ extraction, whereas slow-twitch fibers exhibit greater proportional increases in convective O₂ delivery. It is important to consider these findings when comparing the forearm model used herein with future DCS-based investigations targeting lower-limb muscles, as the proportional differences in fiber type may influence the results and subsequent interpretation.

NIRS measures change in tissue oxygenation in small arterioles, capillaries, and venules within a specific region of interest. Therefore, the relative contributions of each of these compartments to the optical signal during handgrip exercise are unclear. However, it has been shown that the majority (~84%) of the skeletal muscle microvascular volume comprises capillaries, with the remainder split between the arterioles and venules (4, 31). As such, changes in BFI and/or oxygenated and deoxygenated hemoglobin/myoglobin likely represent changes predominantly in the microcirculation (28). Moreover, DCS reflects only a small fraction of the limb being interrogated (in this case, forearm) and thus may not be representative of changes across the entire limb. In contrast, brachial artery flow measurements and venous sampling yield a much more global view.

We plotted the mean convective and diffusive components of O₂ transport between two distinct phases of our incremental exercise test to help explain apparent differences between the conventional and DCS approaches, particularly as they relate to tissue versus $\text{S}{\text{V}}_{{\text{O}}_2}$. Whereas this approach helped our overall interpretation of the data, it requires that mitochondrial Po₂ be negligible. Because we cannot be sure that this was the case, caution is warranted when interpreting these results.

Finally, in addition to our relatively small sample size, we only included male subjects with a relatively narrow range in grip strength. Both exclusion criteria were designed to limit inter-individual variability with respect to the relative exercise intensity performed throughout the incremental test. Future investigations are indeed warranted to assess potential sex differences, as well as to explore unique individual differences across a broader population.

Clinical Implications

Exercise intolerance is a hallmark feature in many diseases and is associated with decreased quality of life and functional performance. The understanding of the mechanisms contributing to exercise intolerance is therefore therapeutically important (20). Indeed, prior work has shown that exercise intolerance may be due to convective or diffusive O₂ transport limitations or both (12, 18, 29, 34). Importantly, these works also bring attention to how pharmacological treatment or exercise prescription may or may not be effective in addressing the pathophysiologic basis for the limited ability to exercise.

Exercise intolerance can be objectively measured by O₂ consumption, either at the whole-body level or at the level of an individual muscle. The latter case is both technically challenging and requires multiple venous measurements from an indwelling catheter. Moreover, as evidenced by the present data, it may not provide a readily apparent understanding of the determinants of mVO₂. DCS offers a noninvasive alternative, with the added benefit of evaluating the microvascular environment in question. This approach could assist clinicians in identifying patients who are more convectively or diffusively limited, which will better define their treatment plan. Application of this technology could therefore have a major impact in the era of precision medicine and help to drive therapeutic discovery.

In conclusion, the data herein provide novel insight into the potential use of DCS for simultaneous, noninvasive evaluation of the microvascular determinants of mVO₂. Moreover, DCS appears to provide a more holistic representation of the determinants of mVO₂, particularly with regard to O₂ extraction, highlighting its added value over traditional approaches measured at the macrovascular level.

GRANTS

Support for this research was funded by a University of Texas at Arlington Nagy Family Endowment Award (to M. D. Nelson) and National Heart, Lung, and Blood Institute Grant R15 HL140989-01 (to M. D. Nelson). Financial support for W. J. Tucker was provided by the American Heart Association (AHA) Postdoctoral Fellowship Grant (AHA Award Number 18POST33990210) and for M. J. Haykowsky by the Moritz Chair in Geriatrics at the University of Texas at Arlington.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.R., W.J.T., J.M.K., F.T., and M.D.N. conceived and designed research; R.R., W.J.T., D.T., H.H.C., C.A.A.-M., Y.Z., and M.D.N. performed experiments; R.R., W.J.T., M.J.H., D.T., H.H.C., Y.Z., J.W., F.T., and M.D.N. analyzed data; R.R., W.J.T., M.J.H., D.T., C.A.A.-M., Y.Z., J.W., J.M.K., F.T., and M.D.N. interpreted results of experiments; R.R., M.J.H., J.W., and M.D.N. prepared figures; R.R., M.J.H., H.H.C., J.W., and M.D.N. drafted manuscript; R.R., W.J.T., M.J.H., D.T., H.H.C., C.A.A.-M., Y.Z., J.W., J.M.K., F.T., and M.D.N. edited and revised manuscript; R.R., W.J.T., M.J.H., D.T., H.H.C., C.A.A.-M., Y.Z., J.W., J.M.K., F.T., and M.D.N. approved final version of manuscript.