Special Section on Clinical Near-Infrared Spectroscopy and Imaging

Near-infrared spectroscopy and skeletal muscle oxidative function in vivo in health and disease: a review from an exercise physiology perspective

[+] Author Affiliations
Bruno Grassi

University of Udine, Department of Medical and Biological Sciences, Piazzale M. Kolbe 4, Udine, I–33100 Udine, Italy

Institute of Bioimaging and Molecular Physiology, National Research Council, I-20030 Segrate, Milan, Italy

Valentina Quaresima

University of L’Aquila; Department of Life, Health and Environmental Sciences, Via Vetoio, I-67100 L’Aquila, Italy

J. Biomed. Opt. 21(9), 091313 (Jul 21, 2016). doi:10.1117/1.JBO.21.9.091313
History: Received January 6, 2016; Accepted June 28, 2016
Text Size: A A A

Open Access Open Access

Abstract.  In most daily activities related to work or leisure, the energy for muscle work substantially comes from oxidative metabolism. Functional limitations or impairments of this metabolism can significantly affect exercise tolerance and performance. As a method for the functional evaluation of skeletal muscle oxidative metabolism, near-infrared spectroscopy (NIRS) has important strengths but also several limitations, some of which have been overcome by recent technological developments. Skeletal muscle fractional O2 extraction, the main variable which can be noninvasively evaluated by NIRS, is the result of the dynamic balance between O2 utilization and O2 delivery; it can yield relevant information on key physiological and pathophysiological mechanisms, relevant in the evaluation of exercise performance and exercise tolerance in healthy subjects (in normal and in altered environmental conditions) and in patients. In the right hands, NIRS can offer insights into the physiological and pathophysiological adaptations to conditions of increased O2 needs that involve, in an integrated manner, different organs and systems of the body. In terms of patient evaluation, NIRS allows determination of the evolution of the functional impairments, to identify their correlations with clinical symptoms, to evaluate the effects of therapeutic or rehabilitative interventions, and to gain pathophysiological and diagnostic insights.

Near-infrared (NIR) spectroscopy (NIRS) applied to skeletal muscle has often been seen with some skepticism by many exercise physiologists. Once at a scientific meeting, one of the authors of the present review was told by Dr. Bengt Saltin (one of the most influential exercise physiologists of the last decades): “Unless you manage to measure O2 consumption, this technique has not much physiological interest.” We respectfully disagree with Dr. Saltin’s statement. Although we recognize that NIRS has several limitations, some of which have been overcome by recent technological developments, the method also has important strengths and can give valuable (and noninvasive) functional insights into skeletal muscle oxidative metabolism in vivo during exercise, in health and disease.

At this purpose, the present review has been devoted, from an “exercise physiology point of view,” to discuss some of the main issues related with the role of NIRS in the functional assessment of oxidative metabolism in skeletal muscles during exercise, with specific attention to integrative aspects and to the factors limiting exercise tolerance. Attention will also be paid to studies carried out in diseased populations, as well as to some recent and exciting technical and methodological developments.

During supramaximal exercise lasting up to a few seconds, the metabolic power output in humans can increase by about 150 to 200 times compared with the value observed at rest. Among the tissues of the body, only skeletal muscle can sustain such extraordinary increases in metabolic power output, which is made possible by the splitting of high energy phosphates present in muscle as adenosine triphosphate (ATP) and creatine phosphate (PCr). If the exercise is carried out for longer periods of time (up to several minutes, or even hours), the maximal metabolic power output, which can be sustained, decreases hyperbolically as a function of the duration of the exercise14 and oxidative metabolism becomes the prevalent, or substantially the only, mechanism responsible for ATP resynthesis. In other words, in most of our daily activities, either related to work or leisure, the energy for muscle work substantially comes from the oxidation of glucose and lipids in muscle fibers, culminating in oxidative phosphorylation at the mitochondrial respiratory chain.

It is not surprising, then, that the maximal power by oxidative metabolism and the fraction of this power, which can be sustained for relatively long periods of time, are intrinsically related to exercise performance and tolerance. An impairment of oxidative metabolism, which occurs in several types of patients, inevitably leads to a reduced exercise tolerance, which may represent one of the main determinants of the clinical picture and quality of life, as well as an important predictor of mortality.5

The availability of reliable tools to investigate in vivo, with precision and reproducibility, muscle oxidative metabolism, possibly noninvasively and with a reasonably elevated temporal resolution, appears therefore of utmost importance. This applies to athletes (sports activities), healthy subjects (both in normal and in altered environmental conditions), and patients affected by many diseases. In the patients, these tools would allow the follow-up of the impairments with time, the identification of potential correlations with clinical symptoms, the evaluation of the effects of therapeutic or rehabilitative interventions, and could also yield pathophysiological and diagnostic insights.

Whole body oxygen consumption (V˙O2) can be measured with relatively good precision (variability of measurements of about 5% in the best laboratories) at the mouth of the subject or at the alveolar level6 (pulmonary V˙O2, V˙O2p) by the classic open circuit method. These measurements have been carried out in innumerable studies and lead to the definition of concepts like the maximal aerobic power (V˙O2p,max), which evaluates the integrated performance of the respiratory, cardiovascular, and skeletal muscle components of the O2 pathway from ambient air to the mitochondria of skeletal muscles. The availability of methods to determine V˙O2p with an elevated temporal resolution (“breath-by-breath” measurements)68 expanded the spectrum of the functional evaluation variables and allowed the introduction of concepts such as the “gas exchange threshold,”9 the “critical power,”1 or the various phases of the V˙O2 “kinetics.”3,4,6,10 The main limitation of the whole body V˙O2p measurements resides in the impossibility to discriminate between the exercising muscles and the rest of the body, as well as between different muscles engaged in the exercise. Moreover, the presence of O2 stores between the site of measurement (the mouth or the alveoli) and the sites of gas exchange at the skeletal muscle level complicates data interpretation during metabolic transitions.6 A further limitation derives from the diffusion of automated and apparently easy-to-use “metabolimeters,” which led to the proliferation of data and studies of dubious quality.

The relationship between V˙O2, blood flow and O2 extraction can be expressed by the Fick equation, based on the principle of mass conservation. If applied “across” a single muscle or a single muscle group, the equation reads as follows: Display Formula

in which V˙O2m represents muscle V˙O2, Q˙m muscle blood flow, and C(a-v)O2m arterial-venous O2 concentration difference across the muscle.

By rearranging Eq. (1): Display Formula

V˙O2m measurements have been carried out with a high temporal resolution also during metabolic transitions in humans11,12 or in isolated in situ animal models.13,14 The main limitations of this approach can be summarized as follows: invasiveness of the measurements; technical problems inherently associated with in vivo blood flow measurements; the substantial impossibility (in human studies) to sample venous blood directly effluent from the exercising muscle, with a resulting “contamination” of venous blood coming from other muscles.

Alternative methods allowing a functional evaluation in vivo of skeletal muscle oxidative metabolism derive from the utilizations of phosphorus (P31) or proton (H1) nuclear magnetic resonance spectroscopy (P-MRS31, and H-MRS1, respectively). As far as P-MRS31, the method has been utilized to determine the kinetics of recovery of muscle [PCr] following exercise. After assuming the equilibrium of the creatine kinase (CK) reaction, the rate of recovery of [PCr] is a function of mitochondrial ATP production, and therefore it can be considered a tool for evaluating the function of skeletal muscle oxidative metabolism.15H-MRS1, on the other hand, detects myoglobin (Mb) desaturation, thereby allowing an estimation of intracellular PO2.16,17 These technologically formidable methods are intrinsically limited by the very high cost of the instrumentation and by the strict constraints on the exercise modality determined by the size of the bore of the magnet.

In short, the above mentioned methods, although widely utilized, have intrinsic limitations. Therefore, noninvasive, precise, reproducible, and relatively low-cost functional evaluation tools to be utilized in vivo, characterized by an elevated temporal resolution, allowing the discrimination between (and possibly within) skeletal muscle groups, would be needed. Does NIRS satisfy, at least in part, these needs? The following sections of the present review will deal with this issue.

Several excellent review articles about the application of NIRS in skeletal muscle have been published.1825 These reviews carry the relevant information on principles, methods, instruments, and so on. Ferrari et al.24 included a useful list of practical recommendations. The present review article has been specifically devoted to discuss, from an “exercise physiology point of view,” some of the main issues related with the role of NIRS in the functional evaluation in vivo of oxidative metabolism in skeletal muscles, within an integrated perspective related to exercise tolerance. A specific attention will be paid to NIRS studies carried out in diseased populations, as well as to some recent and exciting technical and methodological developments. The latter testify how NIRS, as a tool of functional investigation of skeletal muscles, is still in a steep portion of a gain in knowledge versus time curve. The role of NIRS in evaluating oxidative skeletal muscle performance in sport activities and in athletes, which would deserve a dedicated review article, is not covered in the present article. Likewise, the effects of exercise on brain hemodynamics, as measured by NIRS,26,27 as well as the role of NIRS in the evaluation of brain function21,25 are not discussed.

Tissue Interrogated by the Near-Infrared Spectroscopy Probe

When the probe is applied on the skin overlying a muscle of interest, NIRS instruments can interrogate only a relatively small (2 to 6  cm3) and superficial volume of skeletal muscle tissue.28 It is generally accepted that the depth of penetration of the NIR light in tissues roughly corresponds to half of the distance between the light source and the detector (which is usually between 3 and 5 cm). This intrinsic technical limitation makes the measurement of the thickness of the skin and subcutaneous adipose tissue layer, at the site of placement of the NIRS probe, mandatory. The measurement can be carried out by a caliper or, more precisely, by ultrasound.29,30 The recent availability of low-cost hand-held ultrasound devices has significantly facilitated these measurements. Although no cutoff values are recognized as standard, a value greater than 20  mm would presumably make NIRS measurements, as carried out by standard instruments, rather meaningless in terms of investigating skeletal muscle. This precludes the utilization of the method in subjects with a relatively thick layer of subcutaneous fat, such as obese patients or in patients with significant muscle atrophy.

The fact that the NIR light has to cross the skin and subcutaneous fat in order to reach the underlying skeletal muscle tissue inevitably causes a sensitivity problem and represents a “contamination” of the signal of interest, which is the one coming from skeletal muscle. The problem can be particularly significant in conditions in which skin blood flow increases during exercise for thermoregulatory purposes (see also below). Several “correction algorithms” have been proposed (see the discussion in Ferrari et al.24), which, however, are not included in most of the commercial instruments (Tables 1 and 2). Ohmae et al.31 have recently described a simple method, which does not require ultrasound, for the sensitivity correction for the influence of fat thickness on muscle oxygenation, based on the estimation of fat thickness by time-domain NIRS (TD-NIRS) (see below). van Beekvelt et al.32 and Koga et al.29 proposed other correction algorithms based on the assumption that resting V˙O2m [estimated on the basis of the concentration increases in deoxy-hemoglobin (Hb) and deoxy-myoglobin (Mb), [deoxy(Hb+Mb)]) after applying a cuff occluding the arterial circulation, see below] would be inversely correlated with skin + subcutaneous tissue thickness. Another approach, based on [total(Hb+Mb)] values obtained at rest by quantitative multichannel TD-NIRS, has been proposed.33

Table Grahic Jump Location
Table 1Main commercial muscle NIRS oximeters.
Table Footer NoteaBrain oximeter utilized also in muscle studies
Table Footer NotebOximeter with fat-layer compensation
Table Footer NotecCortical microcirculation blood flow measurement using Doppler shift in coherent light signals.
Table Grahic Jump Location
Table 2Main commercial continuous wave portable/wearable NIRS systems for muscle studies with wireless data transmission.
Table Footer NoteaMuscle imager.
Table Footer NotebAccelerometer is available on request.
Table Footer NotecCommercially available only in Japan.
Table Footer NotedSmartphone controllable system.
Table Footer NoteeWater resistant case.

In the quadriceps muscle (often investigated by NIRS studies) and in other muscles as well, deeper muscle regions are characterized by a greater proportion of oxidative fibers compared to superficial regions34 and by a more oxidative energy metabolism. Accordingly, the deeper fibers have a greater sensitivity toward vasodilatory control mechanisms.35 An anatomical “gradient” in fiber type is presumably associated with a gradient also in terms of metabolism and blood perfusion.36 This has been substantially confirmed, in the quadriceps muscle of humans, in studies carried out by utilizing a “high-power” TD-NIRS instrument,37,38 which allows interrogation of slightly deeper portions of muscle. This instrument, which delivers NIR light with a power about 30 times higher than that of traditional NIRS instruments, allows an effective mean penetration depth of about 3 cm and may represent a substantial technical improvement.

Another problem related to the small volume of tissue interrogated by conventional NIRS instruments derives from the fact that muscle activation and metabolism,39 and even more markedly so blood perfusion36,40,41 are heterogeneously distributed within and between exercising muscles. Excellent reviews on metabolic and blood flow heterogeneities in skeletal muscle, in health and disease, have been recently published.42,43 Since skeletal muscle fractional O2 extraction (see below) can increase during exercise only three to four times above the values at rest, whereas V˙O2m can increase 100-fold, directing blood flow according to O2 needs is crucial to allow sustained contractile performance.43

Most studies (but not all44) found good matching between muscle O2 delivery (Q˙O2m=Q˙m*CaO2, in which CaO2 represents the arterial O2 content), or substrates delivery, and V˙O2m between macroscopic (volume of several cm3) regions of a muscle, or between different muscles involved in the exercise.30,45,46 However, when the resolution capacity of the measurements went down to about 1  cm3, the spatial matching between Q˙O2m and V˙O2m (assessed via PCr depletion) was far from being perfect.45 The presence (or absence) of a matching between the two variables could be more functionally relevant at a microscopic level. According to Segal,47 the recruitment of contiguous “microvascular units,” whose volume is orders of magnitude smaller than the volume investigated by NIRS, could be difficult to match with the recruitment of motor units, whose fibers are sparse within the muscle. In other words, blood flow cannot specifically increase to a specific muscle fiber or to the fibers of a motor unit,47 but must rather increase over a relatively wide region of the muscle, leading to the possibility of “over-perfusing” inactive or relatively inactive fibers.

In any case, the apparently good matching between Q˙O2m and V˙O2m in different macroscopic areas of the same muscle, or of different muscles,45,46,48 appears to be difficult to reconcile with the macroscopic heterogeneity of muscle deoxygenation observed by several studies.29,37,49 The issue, in our opinion, is not settled, and further studies with higher spatial resolution devices are needed.

In summary, the area of muscle investigated by the NIRS probe may not represent, in terms of fiber types, fiber activation and the matching of V.O2m and Q˙O2m, a reliable picture of the situation in the whole muscle. In this respect, however, it should be considered that the same limitation, frequently overlooked, is intrinsically associated with other diffusely utilized methods, which investigate only a small and relatively superficial portion of a muscle, such as muscle biopsy.

Where do the Near-Infrared Spectroscopy Signals Come From?

The NIRS signals are the result of the weighted average of the O2 saturations of the heme groups of Hb in the vascular bed (small arteries, arterioles, capillaries, venules, small veins) and of the Mb heme group in muscle fibers. In terms of Hb, most of the signal comes from small vessels, because larger arteries and veins (greater than 1  mm in diameter) have very high heme concentrations, which absorb all the NIR light. Although it is often considered that venous–venular compartments represent the majority of blood present in skeletal muscle,19,20 direct measurements suggest that this may not be the case. According to Poole and Mathieu-Costello,50 capillaries would contribute to >90% of the total blood volume in muscle. According to these data, the storage of blood in veins would mainly occur externally to the muscle. Thus, the vascular component of the NIRS signals would predominantly come from the capillaries. Since in normal conditions, all regions of the muscle receive nearly-fully oxygenated arterial blood, oxygenation changes detected by NIRS would mainly reflect changes in capillary (Hb-related) and intracellular (Mb-related) O2 levels.

A relative controversy exists about the role played in NIRS signals by the heme group in Mb. It has been traditionally assumed that Hb is responsible for most of the overall NIRS signal.51,52 The concept is supported by the observation of strong correlations, observed in different experimental models, between muscle oxygenation values obtained by NIRS and O2 saturation in venous blood (see below). Studies carried out by NIRS in combination with H1-MRS, aimed at distinguishing between Hb and Mb desaturation during ischemia and exercise, yielded conflicting results.53,54 According to modeling papers,55,56 Mb may contribute to 50% of the total [Hb+Mb] NIRS signal. This concept is supported by a study in which the contribution of Hb and Mb to in vivo optical spectra was determined by wavelength shift analysis.57

The relative role of Hb versus Mb is likely different at rest and during exercise. As discussed above, most of the intravascular NIRS signals likely come from the capillaries. In resting conditions, capillary hematocrit can be significantly lower than that in the systemic circulation.58 During exercise, on the other hand, capillary hematocrit increases, reaching values that are not substantially different from those in larger vessels.58 Thus, the role of Hb (versus that of Mb) in determining the NIRS signals would be higher during exercise compared to rest.

The possibility to separate Hb and Mb desaturation could give valuable insights into the mechanisms regulating peripheral O2 diffusion and V˙O2m. For example, by applying NIRS on a rat model of isolated hindlimb, perfused with Hb-free well-oxygenated Krebs-Henseleit buffer, Takakura et al.59 isolated the signal deriving from Mb desaturation and could calculate the intracellular PO2, making inferences on the role of Mb in O2 supply to mitochondria at exercise onset.

In any case, the whole diffusion pathway, including vascular Hb and intracellular Mb, would desaturate during exercise with a similar time course.56 Therefore, a greater contribution of Mb than Hb to the NIRS [deoxy(Hb+Mb)] signal would not invalidate the interpretation that changes in this signal reflect fractional O2 extraction, the dynamic balance between V˙O2m and Q˙O2m in the volume of tissue under consideration.56 This would apply even if a separation between the Hb and the Mb signal is not feasible or is not carried out. In the next sections of the manuscript, we will discuss why, in our opinion, fractional O2 extraction is by itself of interest.

In terms of the intracellular signals detectable by NIRS, cytochrome oxidase can also absorb NIR light. However, there is substantial agreement that assessment of the redox state of mitochondrial cytochrome oxidase cannot be done in muscle because of Mb interference,60 whereas it is possible in the brain cortex.61

Which Variable Should be Considered?

Upon the premise that NIRS oxygenation signals reflect fractional O2 extraction (see below) in the investigated volume of tissue, the following question can be asked: is there a correlation between NIRS-derived oxygenation signals and PO2 or Hb saturation in venous blood draining from the exercising muscle(s)? The answer to this question may not be straightforward, and conflicting results have been reported in the past. The issue is complicated by the problem of obtaining, particularly in humans, adequate samples of venous blood coming directly from the exercising muscle(s), without a “contamination” from blood coming from other muscles and/or other tissues.

Wilson et al.62 found an excellent correlation between a NIRS oxygenation index and O2 saturation determined in the venous blood draining from a dog gracilis muscle preparation. The data were subsequently confirmed in humans during forearm exercise.53 On the other hand, in two studies63,64 performed during constant work rate exercise on a cycle ergometer, the following phenomena were observed (in normoxia): a transient decrease in oxygenation following the transition to exercise, which was paralleled by a decrease in femoral vein O2 saturation; these decreases were followed by a paradoxical muscle “reoxygenation,” which was not associated with an increased O2 saturation in the femoral vein. The latter, as expected, remained low for the remaining portion of exercise. In other words, following exercise onset an association between the two variables (decreased muscle oxygenation, decreased femoral vein O2 saturation) was observed, whereas dissociation occurred during the remaining portion of the constant work rate exercise. A similar muscle oxygenation pattern has been observed in different muscles14,65,66 for the oxygenated-Hb and -Mb ([oxy(Hb+Mb)]) signal (see below) while not for [deoxy(Hb+Mb)], which increased and then stayed constantly elevated during the remaining portion of the exercise.

Grassi et al.65 reasoned that the muscle reoxygenation suggested by the [oxy(Hb+Mb)] time-course could derive from an increased blood flow to the skin, occurring for thermoregulatory purposes, as also suggested by the results obtained by Maehara et al.,67 Chuang et al.,68 and Davis et al.69 An increased skin blood flow for heat dispersion would increase the [oxy(Hb+Mb)] signal, whereas would not substantially affect [deoxy(Hb+Mb)], since the enhanced blood flow would not be associated with an increased gas exchange. This was substantially confirmed in a study carried out by Koga et al.37 These authors observed by continuous-wave NIRS (CW-NIRS) (see below and Ferrari et al.24 for a definition), during whole body heating, a more pronounced increase of [oxy(Hb+Mb)] compared to that of [deoxy(Hb+Mb)]. The apparent increased oxygenation described by CW-NIRS was not observed when utilizing TD-NIRS37 (see below). The confounding effects of an increased skin blood flow on oxygenation variables [in particular [oxy(Hb+Mb)] determined by CW-NIRS] have been confirmed in a study carried out by Messere and Roatta.70

These authors observed that most of the warming-induced increase of the sum between oxy- and deoxy-Hb and -Mb ([total(Hb+Mb)]) resulted from an increase in [oxy(Hb+Mb)], with a relatively small contribution by [deoxy(Hb+Mb)]. They also observed that the “contamination” of the muscle NIRS signal by an increased skin blood flow was substantially reduced by using spatially resolved (SRS) NIRS (see below). No increases in muscle O2 saturation (SO2m, see below) were observed, by utilizing SRS-NIRS, in the vastus lateralis of subjects performing dynamic knee-extension exercise at 20% of MVC after thigh heating at 37°C and 42°C.71 The unchanged SO2m was observed in the presence of a marked increase in cutaneous vascular conductance.54

Several studies have confirmed the presence of a good or very good correlation between the [deoxy(Hb+Mb)] signal, or other tissue oxygenation variables determined by NIRS, and venous O2 saturation, both in an animal model of isolated muscle in situ14,72,73 and in exercising humans.20,30,74,75 In Wüst et al.,72 the correlation was worse during the initial part of contraction period, which in that model is associated with a forceful “muscle pump” effect and a substantial blood squeezing resulting from the sudden onset of tetanic contractions.

The first NIRS instruments widely utilized on skeletal muscle adopted, as a deoxygenation index, the differential absorption of light between 760 and 800 (or 850) nm.52,53,67,7683 The approach was based on the absorption characteristics of NIR light by the chromophores of interest at the two wavelengths. Subsequent instruments allowed the obtainment of relative (with respect to an initial value arbitrarily set equal to zero) or absolute [oxy(Hb+Mb)] and [deoxy(Hb+Mb)] values. The sum of [oxy(Hb+Mb)] and [deoxy(Hb+Mb)] indicates the total Hb + Mb volume ([total(Hb+Mb)]) in the tissue under consideration. Since total [Mb] cannot change acutely during exercise, changes in [total(Hb+Mb)] would reflect a vasodilation or an increased capillary hematocrit in the tissue under consideration. Absolute values of the above-mentioned variables can be obtained by the more technologically sophisticated (and more expensive) SRS, TD, or frequency-domain (FD) instruments, whereas the less technologically sophisticated, less expensive, and more widely utilized CW instruments allow only relative values (for a more detailed discussion on this topic, see below and Ferrari et al.18,21,24). This represents a limitation, which can be at least mitigated by performing a “physiological calibration” at the end of each test, by a transient ischemia of the investigated limb, obtained by applying for a few minutes a markedly suprasystolic pressure by a cuff, “upstream” of the region of investigation (see, e.g., Fig. 1 in Porcelli et al.84). This maneuver is not too uncomfortable for the subject. The range of values of [deoxy(Hb+Mb)] obtained in the muscle between rest and the condition of full extraction (i.e., when the [deoxy(Hb+Mb)] signal reaches a “plateau” after 4 to 5 min of ischemia) should be related to the range of C(a-v)O2m values between rest and full extraction. By adopting this physiological calibration, “semiquantitative” fractional O2 extraction values can be obtained, also by CW-NIRS instruments.

Some controversy has been raised85,86 about which one of the two variables ([oxy(Hb+Mb)] or [deoxy(Hb+Mb)]) should more precisely reflect changes in skeletal muscle fractional O2 extraction. As discussed by Grassi,87 the time-course of [deoxy(Hb+Mb)] (while not that of [oxy(Hb+Mb)]) during a constant work rate exercise65 qualitatively reflects, rather closely, the time courses of variables related to fractional O2 extraction across a wide spectrum of experimental models, such as C(a-v)O2 determined across a contracting isolated in situ muscle preparation88 or an exercising limb,11,20 microvascular PO2 (PO2mv) determined by phosphorescence quenching in rat spinotrapezius muscle,89 intracellular PO2 determined by phosphorescence quenching in isolated amphibian muscle fiber.90 Similar kinetics between PO2mv and [deoxy(Hb+Mb)] during electrically stimulated contractions in the rat gastrocnemius were observed.91 A study described a very similar pattern, also, for deoxy-Mb determined by H1-MRS17. In other words, during constant work rate exercise (or isometric contractions of constant metabolic rate), a series of variables related to O2 extraction, spanning from C(a-v)O2 across a whole limb to Mb saturation, show a very similar time-course: unchanged versus rest for a few seconds (suggesting an adequacy of Q˙O2m with respect to V˙O2m—this information is relevant in terms of the factors determining the V˙O2 kinetics3,4,10), then a rapid exponential increase (the increase V˙O2m exceeds the increase in Q˙O2m) to a new steady state, in which the ratio V˙O2m/Q˙O2m reaches a new equilibrium, although at a higher value compared to that at rest. In other words, [deoxy(Hb+Mb)] (while not [oxy(Hb+Mb)]) belongs to this family of variables related to fractional O2 extraction. This indirectly but strongly supports the role of [deoxy(Hb+Mb)] as an estimate of skeletal muscle fractional O2 extraction. The concept is further strengthened by the issue of [oxy(Hb+Mb)] increases deriving from the enhanced blood flow to the skin for thermoregulatory purposes, as discussed above.

Strictly speaking, an increased [deoxy(Hb+Mb)] can be considered an estimate of an increased fractional O2 extraction only if [total(Hb+Mb)] is constant, which is not always the case in exercising skeletal muscles. However, there is ample evidence14,18,65,86,9294 suggesting that changes in [deoxy(Hb+Mb)] are less influenced by changes in [total(Hb+Mb)] compared to changes in [oxy(Hb+Mb)]. Adami et al.92 observed a good correlation between the increase in [deoxy(Hb+Mb)] and the decrease in femoral blood flow during passive head-up tilting, in the presence of a presumably unchanged V˙O2m. The data confirm the role of changes in [deoxy(Hb+Mb)] in evaluating O2 extraction in the presence of acute and passively induced changes in Q˙O2m. In the study, passive head-up tilt induced also an increased venous blood volume, which complicated the interpretation of [deoxy(Hb+Mb)] as an index of O2 extraction;92 venous blood pooling is, however, unlikely to occur during exercise. In any case, the issue of the influence of blood volume has been substantially overcome by a recent study94 in which the authors calculated, on the basis of the assumption that during an arterial occlusion changes in [oxy(Hb+Mb)] and [deoxy(Hb+Mb)] should occur with a 11 ratio, a useful “blood volume correction factor,” which can be applied to both [oxy(Hb+Mb)] and [deoxy(Hb+Mb)].

As mentioned above, SRS-, TD-, or FD-NIRS instruments allow the measurement of absolute values of [oxy(Hb+Mb)] and [deoxy(Hb+Mb)].24 By utilizing these values, a tissue oxygenation index (TOI) can be calculated, as the ratio [oxy(Hb+Mb)]/[total(Hb+Mb)].95 As an oxygenation variable, TOI is appealing since it is expressed as a percentage, similarly to arterial O2 saturation (SO2a). Since it includes [oxy(Hb+Mb)], however, also TOI may be more significantly influenced by skin blood flow compared to [deoxy(Hb+Mb)].

Physiological Variables of Functional Evaluation

Which of the variables in the Fick equation [Eq. (1)] can be at least estimated by NIRS? The answer is: all of them, although in order to determine V˙O2m or Q˙m some specific (and rather unphysiological) or invasive protocols need to be adopted. When these protocols are not utilized, the oxygenation indices obtained by NIRS evaluate the dynamic balance between V˙O2m and Q˙O2m. In other words, these indices allow the estimation of skeletal muscle fractional O2 extraction, a proxy of C(a-v)O2m (see also below).

Skeletal muscle (or regional) blood flow

De Blasi et al.96 determined forearm Q. by NIRS, at rest and after hand exercise, by inducing a 50-mmHg venous occlusion by a cuff. They reasoned that if venous outflow from a muscle is impeded by the cuff, whereas arterial inflow is not (or at least is only partially impeded), the increase in [total(Hb+Mb)] as a function of time should be related to Q˙. The authors observed indeed an excellent correlation between forearm Q˙, determined by NIRS as described above, and the increase in forearm volume determined by plethysmography. A similar approach was adopted by van Beekvelt et al.32 No comparison with other methods to determine Q˙m was made. Limitations of the approach relate to the fact that pressure by the cuff inevitably limits, although only in part, also arterial inflow, as recently confirmed.97 Moreover, regional Q˙ measurements could be carried out only during the recovery from exercise, and it is well known that the variable may rapidly change in the recovery phase.6

Q˙m in the exercising calf muscles has been determined in humans by utilizing NIRS in association with the intravenous infusion of the NIR tracer indocyanine green (ICG).98 ICG is a fluorescent dye approved by the U.S. Food and Drug Administration and European Medicines Agency for human use. Arterial blood was withdrawn by a pump and [ICG] was detected by photodensitometry, whereas NIRS optodes positioned over the muscle detected ICG transit in the microcirculation at several wavelengths. Q˙m measurements were compared with those obtained by established methods. The main drawback of this elegant approach resides in its invasiveness, related to the need of obtaining arterial blood samples. The method has been utilized to determine Q˙m in respiratory99 and skeletal muscles100. Vogiatzis et al.30 utilized it to evaluate the intramuscular matching between V˙O2m and Q˙O2m.

Skeletal muscle O2 consumption

Hamaoka et al.101 determined by NIRS the initial rate of Hb and Mb deoxygenation in finger flexors of humans during ischemia performed at rest and immediately after submaximal exercise of different intensities, and considered this variable a reflection of V˙O2m. These authors observed a significant correlation between this rate and the concentrations of intramuscular metabolites (PCr and adenosine diphosphate), determined by P31-MRS, considered to be regulators of oxidative phosphorylation. They concluded that NIRS can quantify the rate of oxidative metabolism at rest and after exercise. The validity of the approach was subsequently confirmed.102 The limitations are represented by the fact that the NIRS measurements need to be carried out in ischemic conditions and that the measurements are possible only in the recovery phase following exercise, during which V˙O2 rapidly decreases.6

While ischemic contractions do not represent a physiological condition, a “less unphysiological” experimental model would be one in which blood flow is held constant. Also in this case, according to the rearranged Fick equation [Eq. (2)], and after considering [deoxy(Hb+Mb)] a proxy of C(a-v)O2m, the increase in [deoxy(Hb+Mb)] should represent an estimate of V˙O2m. This has indeed been experimentally confirmed in the study by Wüst et al.,72 which carried out in an isolated dog gastrocnemius in situ preparation. In the study, Q˙m was held constant by a pump, V˙O2m was calculated by the Fick equation across the muscle and NIRS optodes were directly applied on the surface of the exposed muscle. During metabolic transitions, the kinetics of adjustment of [deoxy(Hb+Mb)] was very similar to the kinetics of adjustment of V˙O2m.

Brief sequential ischemic periods carried out during the recovery from exercise/contractions have been utilized in order to determine, by NIRS, the kinetics of recovery of V˙O2m following exercise.94,103106 Also in this case, the rate of increase in [deoxy(Hb+Mb)] (and/or the rate of decrease in [oxy(Hb+Mb)]) during each of the brief ischemic periods is considered an estimate of V˙O2m. The kinetics of recovery of V˙O2m, as determined by the abovementioned protocol, has been shown to be closely correlated with two well-established variables of functional evaluation of skeletal muscle oxidative metabolism (frequently and incorrectly termed “oxidative capacity” by some of these authors): (1) the kinetics of recovery of [PCr], as determined by P31-MRS following exercise104 and (2) maximal adenosine diphosphate-stimulated mitochondrial respiration, as determined by high-resolution respirometry in isolated and permeabilized skeletal muscle fibers.106

Skeletal muscle fractional O2 extraction

As discussed above, in “normal” experimental conditions, oxygenation variables obtained by NIRS allow the evaluation of the dynamic balance between V˙O2m and Q˙O2m, in other words of skeletal muscle fractional O2 extraction, a proxy of C(a-v)O2m. NIRS oxygenation variables can represent only a proxy of C(a-v)O2m since they are usually expressed in relative terms, and the proportional contributions of arterial and venous blood to the overall signal are unknown. In the next paragraphs, we will provide examples of conditions in which analysis of skeletal muscle fractional O2 extraction appears relevant and of physiological interest.

The capacity for prolonged exercise depends upon the capacity to supply the exercising muscles with O2 in order to satisfy the metabolic requirements. During exercise, however, fractional O2 extraction increases. In other words, compared to rest, the muscles (as other organs) utilize a greater percentage of the O2 they receive by the cardiovascular system. We will discuss, later in this section, how this increase in fractional O2 extraction could impair peripheral O2 diffusion and oxidative metabolism through the resulting decrease in microvascular PO2 (PO2mv). Textbook physiology says that at the systemic level, in humans at rest, fractional O2 extraction is 25%, whereas during maximal exercise, the variable increases to 75%. In steady-state conditions during exercise, the relationship between Q˙ and V˙O2p is linear, with a slope of 5 (1  Lmin1 increase in V˙O2p is accompanied by a 5  Lmin1 increase in Q˙). Considering that 1/5 of the volume of arterial blood is represented by O2 (CaO2200  mLL1 of blood), the linear relationship between Q˙O2 and V˙O2p has a slope of 1. As a consequence of the linear Q˙ (or Q˙O2) versus V˙O2p relationship, systemic C(a-v)O2 would increase hyperbolically as a function of work rate. This has indeed been observed during ramp incremental cycling exercise.107

The relationship between Q˙mO2 and V˙O2m is somehow different when determined at the microvascular level, which is the domain investigated by NIRS. During an incremental or ramp exercise, the NIRS-derived [deoxy(Hb+Mb)] follows a sigmoid pattern,62,84,107113 with a shallow increase at low work rates (suggesting a tighter matching between muscle V˙O2m and Q˙O2m, presumably associated with the recruitment of more oxidative fibers), a linear intermediate portion (suggesting greater O2 extraction and a progressively higher V˙O2m/Q˙O2m, presumably associated with the recruitment of less oxidative fibers), followed by a plateau at work rates approaching 80% of V˙O2m,max.

At submaximal work rates, before the “plateau” is reached, the observation of higher fractional O2 extraction for the same V˙O2m (or work rate), in other words, a left-shifted [deoxy(Hb+Mb)] versus work rate curve would suggest an impaired Q˙O2m. On the other hand, a lower fractional O2 extraction (right-shifted [deoxy(Hb+Mb)] versus work rate) would suggest an excess Q˙O2m with respect to V˙O2m. This information may have profound physiological and pathophysiological significance (see below). The sigmoid shape of the [deoxy(Hb+Mb)] versus work rate curve has been shown to vary in muscles characterized by different fiber type compositions49 and has been confirmed (with minor differences) also during supine exercise114 as well as during incremental exercise in acute hypoxia.115 A right-shifted [deoxy(Hb+Mb)] versus work rate curve was described in trained cyclists.116

The [deoxy(Hb+Mb)] kinetics during constant work rate exercise has been utilized to evaluate the effects on the rate of adjustment of oxidative metabolism by a “prior” or “warm up” or “increased baseline” exercise,72,117119 or to investigate the effects on microvascular fractional O2 extraction by interventions such as dietary nitrate supplementation, aimed at increasing microvascular nitric oxide (NO) availability.120 For example, the increased muscle oxygenation observed by Gurd et al.118 following a prior heavy intensity exercise suggests an enhanced O2 availability, which should have contributed to the observed faster pulmonary V˙O2 kinetics. In other studies, the results were not conclusive. Bailey et al.,117 for example, observed after oral nitrate supplementation, a decreased muscle O2 extraction, which, however, could not be simply attributed to enhanced O2 delivery since it was associated with a decreased V˙O2.

Skeletal muscle fractional O2 extraction determined by NIRS can also give valuable physiological and pathophysiological insights into the integrated responses to exercise (and to other conditions) involving different muscle groups or skeletal muscles and other organs, as well as into the physiological mechanisms of Q˙m regulation.

For example, arm Q˙ (by thermodilution) and muscle oxygenation (by NIRS) were measured at the transition from arm to arm + leg exercise.66 During maximal exercise, arm Q˙ decreased at the transition, to a degree that V˙O2m,max and muscle oxygenation were compromised. Thus, NIRS can give valuable insights into the coordination of Q˙m and V˙O2m adjustments to exercise between different muscle groups, in conditions of increased O2 demands and/or limited Q˙O2m (see also below).

Turner et al.121 observed [deoxy(Hb+Mb)] increases both in limb locomotor muscles and in respiratory muscles after interventions, such as inspiratory muscle loading performed during cycling exercise at 80% of V˙O2p,max, or after increasing work rate from 80% to 100% of V˙O2p,max. The results suggest a limited Q˙O2m, both in locomotor and in respiratory muscles, which may impair exercise tolerance during maximal exercise or in conditions of increased respiratory muscles work.

By applying lower body negative pressure, Soller et al.122 induced a progressive decrease of central blood volume and a decreased stroke volume, which were paralleled by a decreased muscle oxygenation from vasoconstriction. According to these authors, skeletal muscle deoxygenation determined by NIRS could be utilized as an early indicator of central hypovolemia and impending cardiovascular collapse.

NIRS has also been utilized to elucidate basic physiological mechanisms of Q˙m regulation, such as the metabolic modulation of sympathetic vasoconstriction (functional sympatholysis) by exercise123 or by tissue hypoxia.124 By simultaneously measuring sympathetic nerve activity by microelectrodes and forearm oxygenation by NIRS, Hansen et al.123 provided evidence in favor of functional sympatholysis. Reflex sympathetic activation, which consistently decreased oxygenation in resting forearm muscle, had no effect on oxygenation when the muscles were exercised.

The [deoxy(Hb+Mb)] plateau during an incremental exercise can be considered an estimate of the maximal capacity of O2 extraction during exercise and has been utilized in several studies84,109113,125127 as a variable of functional evaluation of skeletal muscle oxidative metabolism.

Venous blood draining from exercising muscles still contains small but significant amounts of O2; in other words, fractional O2 extraction is not 100%.128130 Which are the normal values of fractional O2 extraction across skeletal muscle during maximal exercise? According to Knight et al.,128 during cycling, the value can go up to 90%, as determined by invasive measurements carried out across an exercising limb. A fractional O2 extraction value of 85%, also in this case determined invasively across an exercising limb, was observed during maximal knee-extensor exercise,130 which is an exercise paradigm (involvement of a relatively low small muscle mass) in which cardiovascular constraints to O2 delivery are not present or are significantly reduced, and skeletal muscle blood flow can be three times higher than during conventional cycle ergometer exercise. These values are very similar to those obtained in normal subjects by NIRS for [deoxy(Hb+Mb)] at peak exercise, expressed as a percentage of the maximal [deoxy(Hb+Mb)] value observed during a transient limb ischemia.84

Although, in normal subjects, V˙O2p,max is traditionally considered to be mainly limited by maximal cardiac output (Q˙max) and maximal cardiovascular O2 delivery (Q˙O2max),131 the role of peripheral factors cannot be overlooked.132,133 If one considers the Fick equation [see Eq. (1)] applied to conditions of maximal exercise, it is obvious that the main difference between subjects with high V˙O2p,max and subjects with low V˙O2p,max resides in Q˙O2max, and not in C(a-v)O2max. However, the concept that