Local oxygenation profiles were monitored simultaneously from leg muscle, arm muscle, and cerebral tissue using a continuous wave NIRS system (Oxymon Mk III Near-Infrared Spectrophotometer, Artinis Medical Systems, Zetten, The Netherlands). For local leg muscle oxygenation profiles, the transmitting and receiving optodes were placed over the vastus lateralis muscle of the right leg at mid-thigh level and parallel to the long axis of the muscle. Since the activity of m. vastus lateralis increases during running (Guidetti et al., 1996), this muscle is suitable for examining exercise-induced changes in leg muscle oxygenation. To monitor less active arm muscle during running, the optodes were positioned longitudinally on the biceps brachii muscle of the right arm above the elbow joint and lateral to middle line. For cerebral monitoring, the optodes were placed over the right frontal cortex, about 2 cm above the right eyebrow and as laterally as possible to the longitudinal fissure of cerebrum. The prefrontal association cortex occupies most of the rostral part of the frontal cerebral lobe. This site has been linked to the planning of voluntary movement (Sahyoun et al., 2004), and it may contribute to an integrative decision to stop exercising. For both muscle and cerebral measurements, the optodes were housed in an optically dense plastic holder which was attached to the skin using double-sided adhesive tape. For the cerebral optode holder, a special headband was used. The setup helped to minimize the intrusion of extra outside light and the loss of transmitted near-infrared light from the field of interrogation and to ensure the fixed positions of the optodes. The inter-optode distances were chosen to be between 35 and 50 mm so that a good signal quality was reached before starting the measurements. A sampling frequency of 50 Hz was used for collecting NIRS data.

Each NIRS probe consisted of one receiver and three transmitters operating at wavelengths of 765 and 860 nm. The theory of NIRS and its use in exercise measurements have been described in detail elsewhere (Boushel et al., 2001). Briefly, the intensity of incident and transmitted light was recorded continuously and, along with the specific optical pathlength and extinction coefficients, used for on-line estimation and display of concentration changes (ΔμM) from the resting baseline of oxyhemoglobin (Δ[O₂Hb]), deoxyhemoglobin (Δ[HHb]), and total hemoglobin (Δ[tHb] = Δ[O₂Hb] + Δ[HHb]). The values used for the differential pathlength factor (DPF) were 5.51 for the leg, and 4.16 for the arm (van der Zee et al., 1992; Duncan et al., 1995). DPF for cerebral tissue was calculated (DPF = 4.99 + 0.067 × Age^0.814) according to the manufacturer's guidelines. The tissue saturation index (TSI = Δ[O₂Hb]/Δ[tHb] × 100%) was calculated from the light attenuation slope along the distance from the three emitting points as detected by the sensor of the receiving optode. Obtained NIRS data were averaged to give values in 1 s intervals, and time-aligned with gas exchange, heart rate and SpO₂% data. The Δ[O₂Hb], Δ[HHb] and TSI recordings were presumed to be reliable estimators of changes in tissue deoxygenation status representing regional imbalances between O₂ delivery and O₂ utilization in the field of interrogation (Boushel et al., 2001; DeLorey et al., 2003). Total hemoglobin (Δ[tHb]) was analyzed since it reflects changes in tissue blood volume (Boushel et al., 2001; DeLorey et al., 2003).

Values of TSI and Δ[HHb] at peak exercise are referred as TSI_peak and Δ_peak[HHb], respectively. The absolute exercise-induced changes (Δ_t) in TSI and Δ[HHb] were determined as a subtraction [the value at peak exercise minus the value at baseline walking (5 km · h⁻¹)] and are referred as Δ_tTSI and Δ_t[HHb], respectively.

To determine tissue-specific NIRS inflection points (NIP) reflecting local oxygenation changes in leg muscle, arm muscle, and cerebral tissue, the following procedures were performed:

Second by second data of Δ[O₂Hb], TSI, and Δ[HHb] were plotted against time starting from locomotion at 8 km · h⁻¹ until the end of exercise. The plottings were performed to all the three tissues of interest (m. vastus lateralis, m. biceps brachii, frontal cerebral cortex) of every subject.

After determining them, the tissue-specific NIP were chosen for further analyses on the following condition; an angle of ≥ 15° between consecutive trend lines had to be observed at the same time point in all the three parameters (Δ[O₂Hb], TSI, and Δ[HHb]). The change of this magnitude was expected to reflect permanent changes in tissue oxygenation trend. In addition, for cerebral tissue, inflection points with a rise only in cerebral Δ[HHb], were included in further analyses. The tissue-specific inflection points were named as follows: NIP_LegAT (the NIP that was observed in the leg muscle and was the closest to AT), NIP_LegRC, NIP_ArmAT, NIP_ArmRC, NIP_CerAT, and NIP_CerRC, respectively.

Regional muscle and cerebral oxygenation patterns are shown in Figures 2 (leg muscle), 3 (arm muscle), and 4 (cerebral tissue). In the leg muscle (m. vastus lateralis), Δ[O₂Hb] and TSI decreased (−8.7 ± 5.1 μM; −28.7 ± 10.9%, respectively) from baseline walking to peak exercise, while Δ[HHb] and Δ[tHb] increased (12.2 ± 5.8 μM; 3.5 ± 3.8 μM, respectively). In the arm muscle (m. biceps brachii), Δ[O₂Hb] and TSI decreased (−19.1 ± 9.0 μM; −42.0 ± 20.7%, respectively), while Δ[HHb] and Δ[tHb] increased (21.9 ± 12.7 μM; 2.8 ± 13.8 μM, respectively). In the cerebral cortex, TSI decreased (−9.2 ± 7.6%), while Δ[O₂Hb], Δ[HHb], and Δ[tHb] increased (3.5 ± 4.7 μM; 6.2 ± 2.4 μM; 9.8 ± 4.9 μM, respectively). Deoxygenation at peak exercise was greatest in the arm muscle, then in the leg muscle, and least in cerebral tissue as detected by TSI_peak (arm vs. leg; arm vs. cerebral; leg vs. cerebral, all p < 0.001) and Δ_peak[HHb] (arm vs. leg, p < 0.001; arm vs. cerebral, p < 0.01; leg vs. cerebral, n.s.).

In the leg muscle, Δ_t[HHb] had a positive association with V˙O2peak (r = 0.48, p < 0.05). No other relationships between the extent of leg muscle deoxygenation and V˙O2peak were observed at any exercise intensities. In arm muscle- and cerebral tissue, the extent of deoxygenation did not associate with V˙O2peak.

An example of NIP determination of the cerebral tissue is presented in Figure 5. For the 22 subjects, a total of 24 NIPs were observed in the leg muscle, 36 in the arm muscle, and 64 in the cerebral tissue. That is to say, a per-person average of 1.1 NIPs were observed in the leg muscle, 1.6 in the arm muscle, and 2.9 in the cerebral tissue. Of the 24 NIPs observed in the leg muscle, 16 reflected deoxygenation (Δ[O₂Hb]↓, TSI↓ and Δ[HHb]↑), 1 was indefinable (e.g., Δ[O₂Hb]↓, Δ[HHb]↓), and 7 reflected oxygenation (Δ[O₂Hb]↑, TSI↑ and Δ[HHb]↓). In the arm muscle, 30 NIPs reflected deoxygenation and 6 reflected oxygenation. In cerebral tissue, 63 NIPs reflected deoxygenation, while 1 was indefinable. The data of six tissue-specific NIPs being the closest to the ventilatory thresholds are presented in Table 3. There were no differences between the NIPs closest to AT and AT (NIP_LegAT vs. AT, p = 0.587; NIP_ArmAT vs. AT, p = 0.356; NIP_CerAT vs. AT, p = 0.332). Concerning the NIPs closest to RC, NIP_LegRC did not differ from RC (NIP_LegRC vs. RC, p = 0.058), while NIP_ArmRC (vs. RC, p < 0.001) and NIP_CerRC (vs. RC, p < 0.05) did.

NIPs that reflected deoxygenation (Δ[O₂Hb]↓, TSI↓ and Δ[HHb]↑) had consistently positive associations with V˙O2peak when examining absolute V˙O2 at NIPs (e.g., V˙O2 at the second NIP reflecting deoxygenation in m. vastus lateralis vs. V˙O2peak: r = 0.85, p < 0.05). However, when examining %V˙O2R at NIPs, no associations between NIPs and V˙O2peak were observed.

The NIP reflecting acceleration of arm deoxygenation was observed in 10 subjects during last two minutes of exercise (63 ± 42 s before exhaustion). In nine of those 10 subjects, an accelerated rise of V˙E occurring 60 ± 72 s before the acceleration of arm deoxygenation was also observed.

In the leg muscle (m. vastus lateralis), the oxygenation pattern is in accordance with other studies that have examined incremental treadmill exercise (Hiroyuki et al., 2002; Lee et al., 2011). At the onset of baseline walking, TSI immediately increased above standing level, which indicates increased leg muscle oxygenation. The rapid increase of TSI resulted from activation of the muscle pump, that expelled pooling venous blood toward the heart (DeLorey et al., 2003), and thus, led to decreased Δ[tHb]. At speeds of greater than or equal to 8 km · h⁻¹, subjects began running and leg muscle oxygenation decreased (TSI↓, Δ[HHb]↑) with the increase in running speed and O₂ demand.

At the speeds of greater than or equal to 8 km · h⁻¹, Δ[tHb] increased, reflecting an increase in leg muscle blood volume. However, in agreement with previous findings (Legrand et al., 2007), Δ[tHb] decreased at high intensities (≥ 12 km · h⁻¹) but remained significantly higher than values observed during baseline walking. The decrease primarily resulted from local vasoconstriction (Secher and Volianitis, 2006) and possibly from mechanical constraints (Kowalchuk et al., 2002). Moreover, blood flow distribution changes (i.e., blood flow heterogeneity decreases) within the active muscle mass with increasing speed (Heinonen et al., 2007). Further investigations are needed to determine the combined contribution of these factors to changes in muscle blood volume during high intensities.

In this study, we employed a novel method to determine NIP. In the leg muscle, an average of 1.1 NIPs per subject were observed; however, there was inter-individual variability. The main findings concerning NIP determination are: (1) NIPs observed coincided with AT and/or RC; (2) both NIP_LegAT and NIP_LegRC were manifested in nine subjects; (3) NIPs reflecting deoxygenation had positive associations with V˙O2peak when examining absolute V˙O2 at NIPs, but when examining relative values (%V˙O2R), no associations were observed. The 1st and 2nd findings imply that metabolic changes within the muscle tissue affect oxygenation changes, and are one essential part of ventilatory response control during incremental exercise (Turner, 1991; Bhambhani, 2004), but there likely is more inter-individual variability during treadmill exercise than during cycling. It should be noted that NIP_LegRC was observed slightly later than RC, although the difference was not statistically significant. The 3rd finding is congruent with the previously reported rightward shift of the Δ[HHb] pattern relative to absolute exercise capacity in individuals of higher aerobic capacity (Boone et al., 2009). However, it is incongruent with a rightward shift of the Δ[HHb] pattern relative to percentage of exercise capacity in individuals with higher V˙O2peak, also reported by Boone et al. (2009). These data suggest that matching local muscle perfusion to V˙O2 at low exercise intensities is likely to be better in subjects with higher aerobic capacity. Similar observations have been made during static knee-extensor exercise (Kalliokoski et al., 2005). Thus, subjects with lower V˙O2peak place greater reliance on O₂ extraction in providing adequate V˙O2, which highlights their lower O₂ delivery (Bassett and Howley, 2000).

This is the first study to examine arm muscle oxygenation during incremental treadmill exercise. The shape of the arm muscle (m. biceps brachii) oxygenation pattern resembles that of the leg muscle. At the onset of baseline walking, arm muscle oxygenation rapidly increased and then remained relatively constant. Thereafter at the speeds of greater than or equal to 8 km · h⁻¹, an initial moderate decrease of oxygenation level was observed, followed by a rapid decrease during more severe exercise. At peak exercise, deoxygenation was greatest in the arm muscle in comparison with the leg muscle and cerebral tissue. These findings are consistent with previous findings made during incremental cycling (Ogata et al., 2007; Peltonen et al., 2012).

An average of 1.6 NIPs per subject with inter-individual variability were observed in the arm muscle. Concerning the timings of the NIPs, NIP_ArmAT coincided with AT but was manifested only in six subjects, suggesting that arm muscle deoxygenation hardly couples with ventilatory responses at low intensities during incremental treadmill exercise. However, NIP_ArmRC did not coincide with RC but was observed later than RC. In addition, a NIP reflecting acceleration of arm deoxygenation was observed in 10 subjects during the last two minutes of exercise, and an accelerated rise of V˙E occurred in nine of those 10 subjects approximately 1 min before the deoxygenation acceleration. Ogata et al. (2007) and Peltonen et al. (2012) have similarly reported that the magnitude of decrease in O₂ delivery to less active muscle is coupled to the magnitude of increase in the amount of hyperventilation during incremental cycling. Ogata et al. (2007) suggested that the association would result from metabolite accumulation and the concomitant metabolic acidosis, which concurrently affect both hyperventilation and O₂ delivery to inactive muscle. The acceleration of arm deoxygenation might also be explained by exercise-induced sympathetic flow and the consequential vasoconstriction to support the prevailing blood pressure, and thus favoring blood flow to working muscles (Secher and Volianitis, 2006).

Rooks et al. (2010) recently described that cerebral tissue oxygenation has a quadratic response to incremental exercise: it increases from low-to-hard intensities, which is followed by a plateau or decline toward baseline at very hard, maximal intensities. They specified that in aerobically trained people, the plateau or a slight decline in cerebral Δ[O₂Hb] concurrent with increasing Δ[HHb] was observed at exhaustive, maximal intensities. Our findings are similar: Δ[O₂Hb] rose from low-to-hard intensities and fell slightly at the speeds of greater than 12 km · h⁻¹, whereas Δ[HHb] increased along with running speed.

Thus, cerebral deoxygenation was manifested, and it accelerated after RC, because RC was followed by NIP_CerRC. We measured PETCO₂ [an indirect estimate of partial pressure of arterial CO₂ (PaCO₂)], which began to decrease systematically at RC due to hyperventilation as reported previously (Beaver et al., 1986). The decrease in PaCO₂ is known to lead to cerebral vasoconstriction (Ogoh and Ainslie, 2009). However, no significant cerebral vasoconstriction was observed in our study, demonstrated by the plateau of Δ[tHb] level at high intensities. Previous studies (Subudhi et al., 2007; Rupp and Perrey, 2008) have also reported cerebral Δ[tHb] level remaining elevated in trained individuals at peak exercise, whereas Bhambhani et al. (2007) reported a post-RC decrease in cerebral blood volume (Δ[tHb]) in individuals with a mean V˙O2peak of <40 ml · kg⁻¹ · min⁻¹. One reason for the variability of Δ[tHb] levels at peak exercise may be lower peripheral chemosensitivity and lower submaximal V˙E of trained individuals, which attenuate PaCO₂ reduction at high intensities and result in less cerebral vasoconstriction (Weil and Swanson, 1991).

In cerebral tissue, an average of 2.9 NIPs per subject were observed with inter-individual variability; the number is higher than in muscle tissues because NIPs reflecting the acceleration of cerebral deoxygenation as a rise in cerebral Δ[HHb] only were also included in analyses. Oxygen uptake at NIPs reflecting cerebral deoxygenation correlated positively with V˙O2peak, implying that cerebral O₂ delivery, in relation to O₂ utilization, decreases later in trained individuals throughout incremental exercise. However, cerebral O₂ delivery stays adequate until exhaustion and does not limit exercise performance during normoxia (Subudhi et al., 2007).