Whole body hyperthermia, but not skin hyperthermia, accelerates brain and locomotor limb circulatory strain and impairs exercise capacity in humans

Abstract Cardiovascular strain and hyperthermia are thought to be important factors limiting exercise capacity in heat‐stressed humans, however, the contribution of elevations in skin (T sk) versus whole body temperatures on exercise capacity has not been characterized. To ascertain their relationships with exercise capacity, blood temperature (TB), oxygen uptake (V̇O2), brain perfusion (MCA V mean), locomotor limb hemodynamics, and hematological parameters were assessed during incremental cycling exercise with elevated skin (mild hyperthermia; HYP mild), combined core and skin temperatures (moderate hyperthermia; HYP mod), and under control conditions. Both hyperthermic conditions increased T sk versus control (6.2 ± 0.2°C; P < 0.001), however, only HYP mod increased resting TB, leg blood flow and cardiac output (Q̇), but not MCA V mean. Throughout exercise, T sk remained elevated in both hyperthermic conditions, whereas only TB was greater in HYP mod. At exhaustion, oxygen uptake and exercise capacity were reduced in HYP mod in association with lower leg blood flow, MCA V mean and mean arterial pressure (MAP), but similar maximal heart rate and TB. The attenuated brain and leg perfusion with hyperthermia was associated with a plateau in MCA and two‐legged vascular conductance (VC). Mechanistically, the falling MCA VC was coupled to reductions in PaCO 2, whereas the plateau in leg vascular conductance was related to markedly elevated plasma [NA] and a plateau in plasma ATP. These findings reveal that whole‐body hyperthermia, but not skin hyperthermia, compromises exercise capacity in heat‐stressed humans through the early attenuation of brain and active muscle blood flow.


Introduction
Aerobic exercise performance (Dill et al. 1931;Galloway and Maughan 1997;Tatterson et al. 2000;Ely et al. 2010; precise impairment in V O 2max has been controversial, with some observing little (≤3%) or no reduction (Williams et al. 1962;Rowell et al. 1966;Klausen et al. 1967), and others reporting more substantial decrements (~7-30%) (Sawka et al. 1985;Nybo et al. 2001); the largest decline observed when internal and skin temperatures are markedly elevated prior to exercise (Pirnay et al. 1970;Arngr ımsson et al. 2004). Conversely, brief exposure to heat stress, inducing skin hyperthermia in the absence of internal/core hyperthermia, does not appear to substantially reduce V O 2max (Arngr ımsson et al. 2004). The combined thermal stress of high core and skin temperature, therefore, appears to be a prerequisite for a compromised maximal aerobic capacity. There is, however, surprisingly limited information on the precise mechanisms underpinning the hyperthermia-induced suppression of V O 2max and, in particular, the cardiovascular adjustments to incremental exercise with different extents of skin and wholebody hyperthermia.
It has been hypothesized that elevations in skin temperature, by requiring a large proportion of the cardiovascular capacity, are the primary factor leading to a compromised maximal aerobic performance. This concept was substantiated by the classical observation that, despite a small, nonsignificant fall in V O 2max (~3%, N = 6), cardiac output (Q ) is reduced during the high-intensity stages of graded exercise in the heat, compared with a temperate environment, in untrained men (Rowell et al. 1966). This reduction in Q could conceivably impair O 2 delivery to the active skeletal muscle; however, there is some evidence that Q is higher during the early stages of intense constant-load exercise with body hyperthermia in trained individuals (Gonz alez-Alonso and Calbet 2003;Gonz alez-Alonso et al. 2004). Therefore, it is unlikely that high skin blood flow requirements per se compromise systemic perfusion. On the other hand, restrictions in active skeletal muscle perfusion may play an important role in the reduced aerobic capacity in hyperthermic conditions. Under control (normothermic) conditions, skeletal muscle O 2 delivery is tightly coupled to the metabolic demand during submaximal exercise (Andersen and Saltin 1985;Delp and Laughlin 1998;Saltin et al. 1998;Gonz alez-Alonso et al. 2002;Delp and O'Leary 2004); regulation that is lost at high intensities as, prior to volitional exhaustion, systemic and active skeletal muscle (in addition to brain and respiratory muscle) blood flow becomes restricted (Gonz alez-Alonso and Calbet 2003; Mortensen et al. 2005Mortensen et al. , 2008Vogiatzis et al. 2009;Calbet et al. 2015). The attenuated limb blood flow per unit of power when approaching maximal exercise intensities occurs concomitantly with enhanced local vasoconstrictor activity and reductions in stroke volume (Gonz alez-Alonso and Calbet 2003;Calbet et al. 2007;Mortensen et al. 2008;St€ ohr et al. 2011b;Munch et al. 2014). As a consequence, and in contrast to other important regions of the body such as the brain (Nybo et al. 2002;Gonz alez-Alonso et al. 2004;Trangmar et al. 2014), blunted O 2 delivery may compromise local aerobic metabolism, as maximal skeletal muscle O 2 extraction is achieved during exhaustive exercise (Gonz alez-Alonso and Calbet 2003). Whether the hyperthermia-related reduction in V O 2max seen during graded exercise to volitional exhaustion is associated with alterations in regional and systemic hemodynamics has never been systematically tested.
The primary aim of this study was to investigate the effect of heat stress, inducing two different grades of hyperthermia, on brain and active limb blood flow and metabolism during incremental cycling exercise to volitional exhaustion. Regional hemodynamics and metabolism during incremental exercise were assessed; (1) after heat exposure sufficient to elevate internal and skin temperature, (2) after a brief heat exposure sufficient to elevate skin temperature and (3) in control conditions. We hypothesized that combined core and skin hyperthermia, but not skin hyperthermia, would compromise V O 2max and exercise capacity in close association with restrictions in brain and active limb perfusion.

Ethical approval
All procedures in this study were approved by the Brunel University London Research Ethics Committee (RE54-12) and conformed to the guidelines of the World Medical Association (Declaration of Helsinki). All participants provided their oral and written and informed consent prior to participation.

Participants
Nine healthy male cyclists (mean AE SD; age 26 AE 6 years, stature 181 AE 6 cm, mass 76 AE 9 kg and V O 2max 4.5 AE 0.1 LÁmin À1 ) participated in the study. Participants arrived at the laboratory postprandial with a normal hydration status and were required to abstain from strenuous exercise and alcohol intake for 24 h and caffeine consumption for 12 h.

Experimental design
Participants visited the laboratory on three occasions, comprising of a preliminary trial, a hyperthermia trial, and a control trial, each separated by 1 week. The preliminary trial familiarized participants with the testing methodology, prior to performing an incremental exercise test on a cycle ergometer (Lode Excalibur, Groningen, Netherlands) to establish maximal work rate (W max ), maximal heart rate (HR max ), and V O 2max . The initial work rate was equivalent to 50% of predicted V O 2max , for 2.5 min, followed by increments of 10% predicted every 2.5 min until the limit of tolerance. Participants were instructed to maintain a cadence between 70 and 90 r.p.m. and the test was terminated when cycling speed dropped below 60 r.p.m. for more than 3 sec, despite strong verbal encouragement to continue. After a 1 h recovery period, participants were dressed in a water-perfused suit (covering the arms, legs and torso), and laid in a supine position while hot water (50°C) was circulated through, by a temperature controlled water circulator (Julabo F-34, Seelbach, Germany). A foil blanket, gloves, and hat were worn to minimize heat loss to the environment. After target increases in skin and core temperature (+6 and +1°C, respectively), participants repeated the incremental test to establish HYP mod W max .
On the hyperthermia trial, participants completed three incremental cycling ergometer exercise tests in the upright position with; (1) HYP mod (with moderate T c and high T sk , after 52 AE 3 min of heat exposure), (2) HYP mild (with a high T sk but normal T c , after 13 AE 1 min of heat exposure) and, (3) control conditions (T a 18°C; 36% relative humidity; with fan cooling). On the control trial, the participants completed three incremental cycling ergometer exercise tests in a thermo-neutral environment (20°C; ≤50% relative humidity; with fan cooling). Each of the incremental cycling tests consisted of 59~2.5 min stages at 20, 40, 60, 80, and 100% W max , and cycling pedal cadence was stable between 70 and 90 r.p.m. On both the hyperthermia and control trials, each incremental test was separated by 1 h of passive recovery while hydration was maintained through the regular consumption of water.
On the hyperthermia trial, brain, active limb and systemic hemodynamics and blood samples from the brachial artery and femoral vein were obtained simultaneously at rest and in the final minute of each exercise stage. Skin and femoral venous temperatures and arterial and femoral venous pressures were recorded continuously. The same measures were collected in the control trial, except for the arterio-venous blood sampling, leg blood flow (LBF), and blood pressure measurements, and with the addition of esophageal temperature (T Oes ). Full depiction of the experimental protocol of the study is presented in Figure 1.
During exercise, LBF was determined using the constant-infusion thermodilution method (Andersen and Saltin 1985;Gonz alez-Alonso et al. 2000). Resting blood flow (n = 4) was obtained using duplex Doppler ultrasonography (Vivid 7, Dimension, GE Healthcare, UK), or calculated from the directly obtained a-vO 2diff and estimated leg V O 2 (n = 5) assuming comparable leg V O 2 values than those measured in four participants in this study and previous reports from this laboratory using similar heating protocols (Pearson et al. 2011;Chiesa et al. 2015). Q at rest was estimated using the Modelflow method (Wesseling et al. 1993), from the directly obtained intra-arterial pressure wave forms, corrected for age, height, and weight.

Catheter placement and blood sampling
Participants rested with a slight head-down tilt while catheters for blood sampling, mean arterial pressure (MAP), femoral venous pressure, and blood temperature were inserted after local anesthesia (1% lidocaine) into the brachial artery of the nondominant arm and anterograde into the right common femoral vein (Logicath Quad lumen, 18 gage, 2.3 mm; MXA234X16X85, Smiths Medical International LTD), the latter using the Seldinger technique. Catheters were inserted by an experienced clinician under ultrasound guidance and were regularly flushed with normal saline (0.9% NaCl) to maintain patency. The time from catheterization to the commencement of resting measurements was~1 h to allow time for the restoration of normal hemodynamics.

Blood variables
Arterial and femoral venous blood samples were drawn into preheparinized syringes and analyzed immediately for blood gas variables (ABL 800 FLEX, Radiometer, Copenhagen, Denmark) corrected to blood temperature in the femoral vein. The analyzer was calibrated (one and two-point) at regular intervals in accordance with manufacturer guidelines. Additional arterial blood samples were collected in 2 mL syringes and transferred to EDTA tubes, centrifuged and separated. Plasma noradrenaline was subsequently determined using an enzyme immunoassay kit (DEE6200, Demeditec Diagnostics GmbH, Kiel, Germany).
Heart rate, blood pressure and body temperatures Heart rate was obtained by telemetry (Polar Electro, Kempele, Finland). Arterial and femoral venous pressure waveforms were recorded using pressure transducers (Pressure Monitoring Kit, TruWave, Edwards Lifesciences, Germany) zeroed at the level of the right atrium in the mid-axillary line (arterial) and at the level of the tip of the catheter (femoral venous). Pressure waveforms were amplified (BP amp, ADIstruments) and sampled at 1000 Hz using a data acquisition unit (Powerlab 16/30, ADInstruments, Oxfordshire, UK) for offline analysis. For measurements of femoral venous blood temperature (T B ), a thermistor (T204a, PhysiTemp, Clifton, NJ) was inserted through the femoral venous catheter and connected to a thermocouple meter (TC-2000, Sable Systems, NV) and routed through the data acquisition system. In the control trial, esophageal temperature (T Oes ) was measured using a thermistor (Physitemp, New England), inserted pernasally into the esophagus at a depth of ¼ standing height. Increases in core temperature during cycling exercise reflect the rise in femoral venous blood temperature, as T B and T Oes have been shown to be within~0.1°C (Gonz alez-Alonso et al. 1999). Mean skin temperature (T sk ) from four sites (standard weightings of chest, arm, thigh, and calf; (Ramanathan 1964) was obtained using a wireless monitoring system (iButton â , Maxim Integrated, San Jos e, CA).

Calculations
In the hyperthermia trials, brain and active limb vascular conductance (VC) indices were calculated by dividing MCA V mean and LBF (for two-legged) by perfusion pressure (MAP). Direct measurements of Q were not possible during exercise, however, to provide some insight into these responses, Q was calculated using the Fick principle, by estimation of systemic O 2 extraction from the directly measured limb O 2 extractions (assuming a linear relationship between these variables, reported in similar exercise protocols; Mortensen et al. 2008;Munch et al. 2014; and accounting for the known reduction in systemic O 2 extraction with core hyperthermia; Gonz alez-Alonso et al.  Figure 1. Sequence of the exercise protocols. Participants visited the laboratory on two occasions, with each trial consisting of three incremental cycling exercise tests at intensities relative to V̇O 2max . As HYP mod reduced V̇O 2max (obtained on the preliminary trial), the absolute work rates of each stage were lower than all other incremental tests (321 AE 9 W vs. 371 AE 11 W). This adjustment allowed for comparisons between incremental tests, relative to V̇O 2max , in either HYP mod or HYP mild /control, where the latter two conditions did not reduce V̇O 2max . Passive heating/matched rest durations prior to exercise in HYP mod and HYP mild were 52 AE 3 and 13 AE 1 min, respectively. A minimum of 1 h passive rest separated each incremental exercise bout. 2004). The following equations were used: Y = 1.43XÀ44.7; R 2 = 0.99; P > 0.05 for control and HYP mild and Y = 1.7322X-76.126; R 2 = 0.98; P > 0.05 for HYP mod .
When leg blood flow measurements were not possible, LBF was calculated from the estimated leg V O 2 (assuming that the increase in pulmonary V O 2 from baseline reflected only the increase in leg V O 2 ) (Mortensen et al. 2005(Mortensen et al. , 2008Calbet et al. 2007) and directly measured leg arterial-to-femoral venous O 2 difference.

Statistics
Differences between exercise conditions were assessed using a two-way repeated-measures ANOVA in which condition (Moderate heat stress, mild heat stress, and control) and exercise phase (Rest, 20, 40, 60, 80, and 100%) were the main factors. Where a significant main effect was found, pairwise comparisons were made using the Holm-Bonferroni procedure. Statistical significance was set at P < 0.05 and all analyses were made using IBM SPSS Statistics (Version 20, IBM Corporation, Armonk, NY, USA).

Impact of heat stress and repeated incremental exercise on exercise capacity
On the preliminary visit, heat stress exposure sufficient to induce HYP mod resulted in a reduction in W max bỹ 13 AE 1% (range: 11-17%) and a fall in V O 2max by 8 AE 3% (range 5-12%), despite a similar HR max compared to control. To ensure a comparable percentage of W max across experimental conditions in the subsequent hyperthermia and control trials, the absolute work rates for the incremental stages in HYP mod were reduced by 13 AE 1% (64 AE 2, 128 AE 4, 193 AE 5, 257 AE 7 and 321 AE 9 W) compared to all other incremental tests (74 AE 2, 148 AE 4, 223 AE 7, 297 AE 9 and 371 AE 11 W; Fig. 1).
During the control trial, where exercise capacity across the three incremental tests was the same, V O 2max (4.4 AE 0.1, 4.5 AE 0.2 and 4.5 AE 0.1 LÁmin À1 ), HR max (177 AE 3, 181 AE 3, and 182 AE 3 beatsÁmin À1 ), T Oes (38.2 AE 0.1, 38.6 AE 0.1, and 38.7 AE 0.1°C) and end-exercise MCA V mean (68 AE 5, 66 AE 5, and 68 AE 3 cmÁs À1 ) were not significantly different (Fig. 2). Moreover, the increase in V O 2 per unit of power was linear from low to maximal exercise intensities in all three tests (9.2 AE 0.3, 9.5 AE 0.3 and 9.1 AE 0.3 mL.min À1 .W À1 ; R 2 = 0.99; P < 0.001). Given the similar exercise capacity, body temperatures and cardio-respiratory responses to exercise in the control trial, the following sections focus on the effects of temperature manipulation on wholebody hemodynamics in the hyperthermia trial only.

Temperature and cardiorespiratory responses to heat stress
Resting T B was elevated in HYP mod compared to HYP mild and control exercise conditions (37.5 AE 0.1 vs. 36.7 AE 0.1 and 37.0 AE 0.1°C; P = 0.03), whereas T sk was elevated in both heat stress conditions compared to control (~38.2 AE 0.3 vs. 32.3 AE 0.4°C; P < 0.001: Fig. 3A and B). During incremental exercise in HYP mod , T B was initially unchanged before increasing to a peak of 39.3 AE 0.1°C (P < 0.01 vs. rest), whereas in HYP mild and control, T B increased from rest to W max (39.1 AE 0.1°C; P < 0.001) and was lower overall compare to HYP mod . T sk was maintained elevated in both heat stress conditions (~36.9 AE 0.4 vs. 32.0 AE 0.4°C; P < 0.001) and was maintained stable throughout exercise.
Cardiorespiratory variables are presented in Table 1. Briefly, systolic and diastolic blood pressures were lower in HYP mod compared to HYP mild and control (P < 0.001). Respiratory frequency, CO 2 production (V CO 2 ), and minute ventilation (V E ) increased with exercise intensity and were lower in HYP mod compared to HYP mild and control (both P < 0.001). End-tidal PO 2 initially declined before increasing at W max, with the reverse response observed for PCO 2 ; however, there were no differences between the exercise test conditions (P = 0.492).

Brain, active limb, and systemic hemodynamics
At baseline, HR was 57 AE 3 beatsÁmin À1 , two-legged blood flow 0.8 AE 0.1 LÁmin À1 , Q 5.5 AE 0.4 LÁmin À1 , and MCA V mean 64 AE 1 cmÁs À1 (Fig. 4). At rest following passive heat stress or control, HR (88 AE 3 vs. 76 AE 5 bpm), two-legged blood flow (1.9 AE 0.1 vs. 1.0 AE 0.1 LÁmin À1 ), and Q (8.9 AE 0.7 vs.~6.9 AE 0.8 LÁmin À1 ) were elevated in HYP mod compared to HYP mild and control (all P < 0.05), whereas MCA V mean was not different (~63 AE 2 cmÁs À1 ). From rest to submaximal exercise, HR and two-legged blood flow increased with exercise intensity in all conditions (P < 0.05 vs. rest) and MCA V mean was elevated ( Fig. 4C; P < 0.05). However, overall, two-legged blood flow was lower ( Fig. 4A; P < 0.05) and HR higher, in HYP mod exercise compared to control exercise. At exhaustion, HR increased to similar peak values in HYP mod, HYP mild and control, respectively (189 AE 4, 187 AE 3, and 184 AE 3 beatsÁmin À1 ). In all conditions, the rate of rise in two-legged blood flow was attenuated, and MCA V mean was reduced in all exercise conditions. Final twolegged blood flow (16.2 AE 1.3, 18.4 AE 1.1, and 18.9 AE 1.1 LÁmin À1 ) and MCA V mean (57 AE 1 vs. 66 AE 3 cmÁs À1 ) were lower in lower in HYP mod than in HYP mild and control conditions.
On the transition from rest to submaximal exercise, estimated Q increased at a similar rate among conditions (~0.04 LÁmin À1 ÁW À1 ). Prior to exhaustion, Q paralleled the attenuation in two-legged blood flow, to a greater extent in HYP mod versus HYP mild and control conditions (Gradient = 0.007 vs. 0.017 LÁmin À1 ÁW À1 ), at a lower absolute work rate, and was similar at end-exercise (26.6 AE 2 LÁmin À1 ).
Blood pressure, oxygen uptake, and brain oxygenation At rest, MAP and FVP were not different among conditions (Fig. 4D). From rest to maximal exercise, MAP increased in all conditions, but was reduced in HYP mod compared to HYP mild and control, respectively (124 AE 7, 139 AE 7, and 153 AE 7 mmHg; P < 0.05). Femoral venous pressure increased with exercise intensity but was not different among exercise conditions. At rest, leg a-v O 2diff was lower in HYP mod compared to HYP mild (24 AE 3 vs.~56 AE 7 mLÁL À1 ; P < 0.05: At exhaustion, leg a-v O 2diff was not different among conditions; however, systemic V O 2max was reduced in HYP mod compared to HYP mild and control exercise conditions (3.94 AE 0.11 vs. 4.23 AE 0.13 and 4.23 AE 0.14 LÁmin À1 , respectively; P < 0.05). Compared to the three maximal incremental tests in the control trial, the rise in systemic V O 2 per unit of power was identical from 20 to 70-80%W max (9.6 AE 0.3 mLÁmin À1 ÁW À1 ), but became attenuated thereafter (8.2 AE 0.6 mLÁmin À1 ÁW À1 ). At rest, NIRS-derived rSO 2 % was elevated in HYP mod and HYP mild versus control conditions (77 AE 2 & 75 AE 3 vs. 67 AE 3%; P < 0.05) and remained unchanged across all conditions during incremental exercise, but declined before exhaustion (rSO 2~6 4%; P < 0.05).
Brain and active limb conductance, blood gasses, plasma catecholamines and ATP Arterial and venous hemoglobin [Hb] and arterial oxygen content increased with incremental exercise in all conditions, despite a reduction in arterial oxygen saturation (all  (Table 3). However, arterial and venous glucose concentrations were elevated at exercise intensities ≥ 60% W max in the HYP mod compared to control and HYP mild . At rest, MCA vascular conductance was not different among conditions (Fig. 5A). The elevations in limb and systemic perfusion in HYP mod were coupled to an enhanced limb vascular conductance ( Fig. 5B; P < 0.05). MCA vascular conductance declined with exercise intensity. Contrastingly, limb vascular conductance increased with exercise intensity (P < 0.05), but was not different among conditions.
At rest, arterial [NA] (Fig. 5C) was augmented in HYP mod versus HYP mild and control conditions (3.7 AE 0.8 vs.~1.9 AE 0.5 nmolÁL À1 ; P < 0.05), whereas venous [NA] was not different (data not shown). Thereafter, both arterial and venous [NA] increased with exercise intensity to a similar peak value   (~48 AE 5 nmolÁL À1 ). The rise in arterial [NA] was coupled to a blunted two-legged vascular conductance at maximal exercise intensities (R 2 = 0.64; P < 0.01). At rest and during submaximal exercise, P a CO 2 was maintained stable (Fig. 5D. However, beyond submaximal intensities, and in association with a marked increase in V E (Table 1), P a CO 2 declined to a similar end-exercise value across conditions (P < 0.05). The decline in P a CO 2 was moderately related to the fall in MCA vascular conductance (R 2 = 0.29; P < 0.01). Lastly, femoral venous (ATP fv ) and arterial (ATP a ) ATP concentrations were not different at rest, but increased up to submaximal exercise intensities in all conditions ( Fig. 5E and F). Beyond 80% W max , both ATP fv , in association with an attenuation in two-legged vascular conductance (R 2 = 0.46; P < 0.01), and ATP a plateaued.

Discussion
To the best of our knowledge, this is the first study to separate the effects of skin hyperthermia from the combined effects of skin and internal hyperthermia on the , P a CO 2 and femoral venous and arterial plasma ATP in response to incremental exercise with different grades of hyperthermia. Values are meansAESEM for seven participants. ‡ different versus mild hyperthermia, † different versus control. Presented symbols denote differences between conditions, when compared at a relative % of maximal work rate in hyperthermia or normothermia (100% = 321 W in HYP mod vs. 371 W in HYP mild and control, respectively).
brain and locomotor limb perfusion, aerobic metabolism and exercise capacity. The major novel finding was that marked skin hyperthermia was insufficient to compromise V O 2max and incremental exercise capacity. On the other hand, superimposed internal and skin hyperthermia reduced maximal exercise capacity and brain and locomotor limb perfusion, which were mechanistically coupled to a plateau or decline in regional vascular conductance and a reduced arterial pressure. Finally, the attenuation in the brain and exercising limb blood flow appears to be an important mechanism by which combined skin and internal hyperthermia reduces aerobic metabolism and exercise capacity. Together, these findings demonstrate that the combination of skin and internal body hyperthermia is a critical factor in whether, or not, brain and active muscle perfusion and aerobic metabolism is compromised during incremental exercise to volitional exhaustion in hot environments.
Skin hyperthermia does not independently compromise cardiovascular capacity or aerobic exercise performance In this study, we clamped skin temperature at a high level (i.e.,~37°C vs.~32°C), without increasing internal temperature, prior to and during incremental exercise (HYP mild ). To achieve this, the participants were first exposed to passive whole body heat stress for~13 min and then combined whole body heat stress and exercise for an additional~12.5 min. An important finding under these conditions was that V O 2max and exercise capacity was the same compared to control exercise. Narrow coreto-skin temperature gradients, as seen in both HYP conditions in this study (range: 0-2.6°C), are purported to place a significant burden on cardiovascular capacity owing to the increased demand for skin blood flow (Rowell 1986;Sawka et al. 2012). This theory has been taken to mean that high skin temperatures play a dominant role in reduced exercise capacity in the heat, by promoting the displacement of blood volume and flow to the skin thereby compromising active muscle perfusion (Tatterson et al. 2000;Ely et al. 2009;Kenefick et al. 2010;Lorenzo et al. 2010;Cheuvront et al. 2010;Sawka et al. 2012). However, we show that brain, active limb, and systemic blood flow during HYP mild is not reduced compared to that observed during control conditions. Moreover, the results on the experimental trial are supported by those on the control trial (Fig. 2) where despite some differences in exercise internal temperature (~0.5°C at exhaustion), exercise capacity was not different during repeated incremental exercise with normal skin temperature (~32°C). Our findings collectively suggest that skin hyperthermia or small elevations in internal temperature alone do not compromise aerobic power or exercise capacity in trained individuals. In contrast, when combined internal and skin hyperthermia was present (i.e., achieved by extending the exposure to passive whole body heat stress to~52 min, while the exercise duration was not different), V O 2max was reduced by~8%; a decline similar to that previously reported (Klausen et al. 1967;Pirnay et al. 1970;Sawka et al. 1985;Nybo et al. 2001;Arngr ımsson et al. 2004). The reduced aerobic power and work capacity were associated with a diminished arterial pressure, an attenuation in active limb (and systemic perfusion), a reduced brain blood flow, and high internal and skin temperatures (39.3 and 37°C, respectively) (Fig. 4). Restricted LBF, via a plateau in local vascular conductance, precedes fatigue during incremental (Mortensen et al. 2008) and constant load maximal exercise (Gonz alez-Alonso and Calbet 2003); whole body hyperthermia advances this cardiovascular instability and may explain the reduced maximal aerobic power (Gonz alez-Alonso and Calbet 2003). Our data demonstrate that the duration of heat exposure is critical to whether or not cardiovascular function is impaired during strenuous exercise in the heat-stressed human.

Impact of hyperthermia on blood flow and pressure at rest and during incremental exercise
To understand the responses to regional hyperthermia during exercise and the potential underlying mechanisms, we need to first scrutinize the resting responses. At rest, combined internal and skin hyperthermia led to elevations in LBF and Q , accompanying a fall in limb a-vO 2diff and a lower MAP, in close agreement with the responses to passive heat stress (Barcroft et al. 1947;Rowell et al. 1969;Rowell 1974;Minson et al. 1998;Crandall et al. 2008;St€ ohr et al. 2011a;Pearson et al. 2011;Heinonen et al. 2011;Chiesa et al. 2016). Interestingly, brief heat exposure, sufficient to raise T sk to that experienced during combined internal and skin hyperthermia (HYP mild ), but without elevations in T c , led to a small increase in systemic (+1.3 LÁmin À1 ) and limb blood flow (+0.25 LÁmin À1 ) compared to baseline values. During passive whole body heat stress, interspersed by single leg exercise, elevations in whole-body perfusion (e.g., Q ; 1.1-1.8 AE 0.3 LÁmin À1 , LBF; 0.5 AE 0.1 LÁmin À1 ) and small but significant reductions in MAP have been observed with skin hyperthermia at rest without increases in T c (Pearson et al. 2011;St€ ohr et al. 2011a). In a recent study from this laboratory, mild heat stress was also shown to induce small but significant increases in systemic and leg perfusion and HR (e.g., Q ; 0.9 LÁmin À1 , LBF; 0.2 LÁmin À1 ; 12 beatsÁmin À1 ), although these alterations occurred concomitant to small increases in T c (~0.4°C) (Chiesa et al. 2016). It is therefore likely that any increased demand for skin and deep limb tissue blood flow, during passive mild hyperthermia, is met by blood flow redistribution from splanchnic vascular beds and a small increase in Q and small reduction in MAP (Rowell et al. 1968;Crandall et al. 2008).
A key question is whether passive hyperthermia-induced hyperperfusion and hypotension alters cardiovascular dynamics during incremental exercise. There were no clear differences in hemodynamic responses, among conditions, during submaximal exercise; in agreement with previous blood flow observations in trained individuals (Savard et al. 1988;Nielsen et al. 1993Nielsen et al. , 1997. However, at the maximal attainable aerobic work rate in HYP mod , blood pressure, brain perfusion, and two-legged blood flow were reduced compared to control exercise (Fig. 4), despite an estimated~1.8 LÁmin À1 elevation in Q and a similar HR max and limb O 2 extraction. The reduced regional blood flow was coupled to an attenuation of vascular conductance, and enhanced vasoconstrictor activity (Fig. 5). Attenuation in blood flow and vascular conductance at maximal exercise may involve the interaction of various reflex, chemical, and thermal mechanisms, in different tissues of the body, responsible for regulating local vascular tone (Rowell 1974;Gonz alez-Alonso et al. 2004;Gonz alez-Alonso 2012;Mortensen et al. 2008;Mortensen and Saltin 2014). To provide mechanistic insight into these circulatory alterations during incremental exercise, with differing combinations of body temperatures, we assessed several vasoactive substances implicated in the regulation of the brain and muscle blood flow. Irrespective of the temperature manipulation, the suppressed (brain) or nonlinear rise in (two-legged) blood flow prior to volitional exhaustion was coupled to a similar fall or plateau in regional vascular conductance (Fig. 5), indicative of vasoconstriction in the active brain and muscle vascular beds. Mechanistically, the brain blood flow velocity decline toward baseline values was associated with a hyperventilationinduced fall in P a CO 2 (r = 0.54; P < 0.05); a potent vasoactive substance affecting cerebrovascular tone (Willie et al. 2012). These dynamics during graded exercise are supported by the literature (Hellstrom et al. 1996;Sato et al. 2011;Trangmar et al. 2014). On the other hand, the restriction in two-legged conductance, prior to exhaustion in all conditions, was related to a plateau in plasma ATP and an exponential rise in sympathetic vasoconstrictor activity even when leg vascular conductance and plasma ATP and [NA] were or tended to be higher in the hyperthermic trials (Fig. 5). It has previously been postulated that the influence of sympathetic vasoconstriction on vascular conductance can be "overridden" by metabolic vasodilation (Remensnyder et al. 1962;Rosenmeier et al. 2004). This theory can explain the regulation of muscle perfusion when exercising limb blood flow, and the intravascular vasodilator milieu including ATP, increase progressively during exercise against a background of relatively low sympathetic drive (Gonz alez-Alonso et al. 2002;Rosenmeier et al. 2004;Mortensen et al. 2011). However, the present findings together with those during maximal and supramaximal exercise (Mortensen et al. 2008), indicate that local vasoconstriction prevails during whole body, intense exercise in association with marked increases in sympathetic nerve activity (Saito et al. 1993;Ichinose et al. 2008) and a blunted rise in plasma ATP concentration. Thus, functional sympatholysis does not prevail at the maximal and supramaximal exercise domain with normal or elevated levels of local hyperthermia.

Does cardiovascular strain contribute to hyperthermia-induced fatigue?
An important question from this study is which cardiovascular process underpins the reduced exercise capacity under physiologically stressful environments. Prevailing theory suggests that reduced aerobic capacity during exercise in the heat is due to reductions in active muscle blood flow, secondary to a substantial increase in skin perfusion, and despite active redistribution of blood flow from nonactive tissues (Rowell 1974(Rowell , 1986. This theory was based on observations that body hyperthermia suppressed Q during treadmill running, in untrained and unacclimatized individuals, compared to control conditions (Rowell et al. 1966); thus giving rise to the premise that the limited cardiovascular capacity is insufficient to meet the combined demands of heat dissipation (skin perfusion) and active muscle perfusion. Our findings demonstrate that the attenuated rise in systemic V O 2 (from 9.6 to 8.2 mLÁmin À1 ÁW À1 ) and reduced exercise capacity with combined internal and skin hyperthermia were coupled to an advanced fall in brain blood flow, and an early attenuation in LBF (that is, occurring at a lower absolute work rate); temporal responses that could feasibly result in a compromised local tissue aerobic metabolism when oxygen extraction reaches its upper limits (~90% in the three conditions of this study) (Gonz alez-Alonso and Calbet 2003; Mortensen et al. 2005Mortensen et al. , 2008Calbet et al. 2007). In addition, our estimates of Q suggest that systemic blood flow is similar at exhaustion among temperature manipulations; a conclusion supported by findings in trained participants, during constant-load cycling to volitional exhaustion, with combined internal and skin hyperthermia (Gonz alez-Alonso and Calbet 2003). It is therefore unlikely that the absolute values of Q and high skin blood flow explain early fatigue during incremental exercise. Reductions in cerebral O 2 delivery (and oxygenation) might contribute to fatigue processes when hyperthermic (Nielsen and Nybo 2003;Nybo and Secher 2004;Todd et al. 2005;Rasmussen et al. 2010;Ross et al. 2012). However, it is unlikely that the moderate reductions in cerebral perfusion, seen here, and in previous studies Trangmar et al. 2014Trangmar et al. , 2015, can compromise cerebral metabolism to the extent that can explain the reduced aerobic power with a moderate hyperthermia. Rather, the advanced fall in cerebral perfusion, at lower absolute exercise intensities, is likely a consequence of the overall cardiovascular strain induced by strenuous exercise in the heat and the concomitant respiratory alkalosis. This is supported by similar findings in hypoxia where cerebral O 2 delivery is markedly attenuated, despite elevated systemic blood flow and perfusion pressure (Subudhi et al. 2009;Vogiatzis et al. 2011). Restoring reductions in cerebral O 2 delivery, during exercise in hypoxia and with body hyperthermia, does not improve maximal aerobic power (Subudhi et al. 2011;Keiser et al. 2015), indicating that processes other than a suppressed cerebral O 2 metabolism explain the early fatigue under physiological stressful environments.
Our present findings highlight that combined skin and internal hyperthermia compromises regional and systemic perfusion at the maximal attainable aerobic work rate. Blunted skeletal muscle and systemic blood flow and O 2 delivery, with and without body hyperthermia, appear to be an important factor limiting aerobic capacity (Gonz alez-Alonso and Calbet 2003; Mortensen et al. 2005Mortensen et al. , 2015. We recognize that many interrelating factors likely contribute to the development of fatigue during exercise. In this context, exhaustion in the present experimental conditions may have resulted from the interaction of multiple inhibitory and excitatory regulatory processes in response to reduced O 2 delivery, modified locomotor muscle and brain metabolism, hyperthermia, altered central motor output, changed central nervous system neurotransmitter activity, and stimulation of muscle feedback mechanisms sensing local metabolic milieu (as widely reviewed; Gonz alez- Amann and Calbet 2008;Meeusen and Roelands 2010;Amann et al. 2011;Noakes 2012;Sawka 2012;Nybo et al. 2014;Morales-Alamo et al. 2015;Blain et al. 2016). Supporting the idea that the etiology of fatigue during exercise is multifactorial and typified by cardiovascular strain and disturbed physiological homeostasis, we found that the single stressor skin hyperthermia was apparently met by compensatory physiological adjustments such that muscle and whole body aerobic energy provision was not compromised compared to control. The combination of multiple stressors triggered by whole body hyperthermia, however, resulted in a compromised aerobic capacity, associated with a blunted active muscle and systemic perfusion.

Methodological considerations
Resting blood flow measurements were made using Doppler ultrasonography, rather than thermodilution, as less blood flow variability is seen with ultrasonography in resting conditions. We were unable to obtain direct measures of Q during exercise; on this basis, our conclusions based on estimated Q are purposefully tempered. On the other hand, it is established that systemic O 2 difference shares a strong linear relationship with leg O 2 extraction during incremental exercise (Mortensen et al. 2008;Munch et al. 2014). Moreover, the adjustment to this relationship in HYP mod is in accordance with previous literature demonstrating a reduced systemic O 2 extraction, per unit of leg O 2 extraction with body hyperthermia (Gonz alez- . Finally, our estimations on Q dynamics during exercise were supported by those obtained with the Modelflow method. Nevertheless, future studies measuring central hemodynamics with different manipulations of internal and skin temperature are required to confirm the present observations.

Conclusion
The present findings show that skin hyperthermia, in the absence of high internal temperatures, does not compromise cardiovascular capacity, maximal oxygen uptake, or exercise performance during strenuous whole-body dynamic exercise. The fall in maximal aerobic power with combined internal and skin hyperthermia was associated with compromised active muscle metabolism due to reduced oxygen delivery. Taken together, these observations explain why aerobic exercise performance in hot environments is not universally impaired across all exercise modalities, as the deleterious effects of environmental heat stress are directly dependent upon heat exposure inducing whole-body hyperthermia and uncompensable physiological strain.