Vascular dysfunction and the age‐related decline in critical power

Abstract Ageing results in lower exercise tolerance, manifested as decreased critical power (CP). We examined whether the age‐related decrease in CP occurs independently of changes in muscle mass and whether it is related to impaired vascular function. Ten older (63.1 ± 2.5 years) and 10 younger (24.4 ± 4.0 years) physically active volunteers participated. Physical activity was measured with accelerometry. Leg muscle mass was quantified with dual X‐ray absorptiometry. The CP and maximum power during a graded exercise test (P GXT) of single‐leg knee‐extension exercise were determined over the course of four visits. During a fifth visit, vascular function of the leg was assessed with passive leg movement (PLM) hyperaemia and leg blood flow and vascular conductance during knee‐extension exercise at 10 W, 20 W, slightly below CP (90% CP) and P GXT. Despite not differing in leg lean mass (P = 0.901) and physical activity (e.g., steps per day, P = 0.735), older subjects had ∼30% lower mass‐specific CP (old = 3.20 ± 0.94 W kg−1 vs. young = 4.60 ± 0.87 W kg−1; P < 0.001). The PLM‐induced hyperaemia and leg blood flow and/or conductance were blunted in the old at 20 W, 90% CP and P GXT (P < 0.05). When normalized for leg muscle mass, CP was strongly correlated with PLM‐induced hyperaemia (R 2 = 0.52; P < 0.001) and vascular conductance during knee‐extension exercise at 20 W (R 2 = 0.34; P = 0.014) and 90% CP (R 2 = 0.39; P = 0.004). In conclusion, the age‐related decline in CP is not only an issue of muscle quantity, but also of impaired muscle quality that corresponds to impaired vascular function.


INTRODUCTION
Exercise tolerance decreases with age, such that activities of daily living (e.g., walking and climbing the stairs) become increasingly difficult (Taylor et al., 2023).Age-related declines in critical power (CP) might contribute to the decline in exercise tolerance and physical function associated with ageing (Gifford & Collins, 2021).
As the boundary that separates steady-state and non-steady-state exercise (Jones et al., 2019), a decrease in CP could cause previously sustainable, submaximal exercise to elicit uncompensable metabolic conditions that inexorably approach maximal levels [e.g., maximal O 2 uptake ( VO 2 max )].Our group and several others have provided evidence indicating that CP decreases with age (Fulton et al., 2023;Gifford & Collins, 2021;Neder et al., 2000;Overend et al., 1992), but the cause of this decrease has not been investigated thoroughly.Development of a greater understanding of which factors are associated with the age-related decrease in CP might lead to more targeted therapies and interventions to treat or prevent exercise intolerance with ageing.
Critical power is strongly influenced by the ability to resynthesize ATP aerobically, and oxygen delivery appears to be key in determining CP (Poole et al., 2021).For example, Kirby et al. (2021) demonstrated that, in contrast to the prompt establishment of steady-state levels of muscle oxygenation that occurs during exercise below CP, exercise above CP results in an inexorable decrease in muscle oxygenation that continuous to decrease as long as the exercise is performed.Given that muscle oxygenation represents the balance between muscle oxygen delivery and muscle oxygen consumption, the continuous decrease in muscle oxygenation observed when exercising above CP is probably attributable to the inability of the cardiovascular system to keep pace with the elevated oxygen demand of the skeletal muscle mitochondria.
Given that there is a finite amount of cardiac output, the proper redistribution of blood flow by the resistance vasculature (e.g., arterioles) strongly influences muscle oxygen delivery and exercise tolerance (Gifford et al., 2020;Hanson et al., 2020;Joyner & Casey, 2015).Poole et al. (2021) hypothesized that impaired vascular function, especially at the level of the resistance vasculature, could conceivably lead to impaired muscle oxygen delivery and reduced CP.As evidenced by the diminished hyperaemic response to passive leg movement (PLM) (Groot et al., 2015;Hydren et al., 2019;Mortensen et al., 2012;Trinity et al., 2015) and intra-arterial infusion of acetylcholine (Seals et al., 2011), ageing is associated with impaired endothelial function in the resistance vasculature.
It seems possible that the age-related decline in CP is associated with concomitant changes in vascular function and exercise blood flow.
Nevertheless, in addition to being sensitive to oxygen delivery, CP is also strongly influenced by muscle mass (Collins et al., 2022).Given that muscle mass often decreases with age (Distefano & Goodpaster, 2018), it is possible that the age-related decline in CP is merely the result of decreased muscle quantity, not muscle quality.Therefore, the purpose of this study was to test the hypothesis that CP is reduced in older

Highlights
• What is the central question of this study?
Does the age-related decrease in critical power occur independently of changes in muscle mass and is it related to impaired vascular function?
• What is the main finding and its importance?
The age-related reduction in exercise tolerance, assessed by critical power, occurs independently of loss of muscle mass and is related to vascular function and exercise blood flow.Blood flow during exercise slightly below critical power reaches very high steady-state levels that do not differ significantly from the maximum blood flow achieved during exercise above critical power.
adults, even when considering any differences in muscle mass, and that the reduced CP of older adults is related to impaired vascular function and reduced exercise blood flow.

Participants
The Brigham Young University (BYU) Institutional Review Board approved this study (IRB2021-108) before recruitment occurred.
Participants were recruited primarily around BYU campus.Each volunteer was informed of the purpose of the study, experimental procedures and potential risks associated with the study before written consent was obtained.Guidelines put forth in the Declaration of Helsinki were followed, except that the study was not pre-registered on clinicaltrials.org.

Procedures
This study involved five visits of ∼90-120 min each.Visit 1 included consenting, measuring body composition with dual emission Xray absorptiometry (DEXA), and familiarizing subjects with the

Assessment of body composition
A DEXA scan was performed according to the manufacturer's recommendations (Lunar iDXA, GE Healthcare, Chicago, IL, USA).
Software on the DEXA machine automatically calculated the percentage of body fat, total lean mass, total fat mass, leg lean mass and leg fat mass.
After the DEXA scan, subjects underwent thigh volume measurements.While lying supine, the circumference of the upper (around inguinal crease), middle and lower (immediately above the top of the patella) thigh was measured with a standard flexible tape measure.Skinfold thickness of the anterior mid-thigh was also measured with standard skinfold calipers.These measurements were used to estimate quadriceps mass size as described by Layec et al. (2014).

Familiarization with knee-extension exercise
Subjects were then familiarized with single-leg knee extension (e.g., kicking) on a custom-made knee-extension ergometer, with a magnetically braked cycle ergometer providing the resistance (Corival CPET; Lode, Groningen, The Netherlands).Following an initial introduction to the movement and several minutes of practice, the subject performed a graded exercise test (GXT), while maintaining kicking frequency at 80 r.p.m.This test consisted of a 3 min warm-up at 10 W, followed by an increase of 3 W every 1 min until task failure.
Task failure was defined as the point when the subject could no longer maintain a kicking frequency of 80 ± 5 r.p.m. for a consecutive 5 s, despite strong verbal encouragement.Once the subject had reached task failure, the wattage was lowered for a cool-down.The greatest power output maintained for 60 s was identified as the GXT maximum power (i.e., P GXT ).After a 20 min break, the subject performed a verification practice test.This test consisted of a 5 min warm-up at 40% of their P GXT and was followed by a 5 min break.Before starting the test, they performed another 3 min warm-up at 40% of P GXT .At the end of the 3 min warm-up, the wattage was increased to the P GXT wattage, and the subject completed the knee-extension exercise until task failure.

Assessment of blood haemoglobin concentration
During the 20 min break between the two familiarization tests, a finger was pricked with a 2.0 mm lancet in order to measure haemoglobin content.Haemoglobin content was determined by using the HemoCue Hb 801 microcuvettes and analyser (HemoCue America, Brea, CA, USA).This was repeated two or three more times for a total of three or four readings from the same finger prick.The average of the readings was recorded as the subject's haemoglobin concentration.

2.3.4
Assessment of physical activity An Actigraph accelerometer (GT9X, ActiLife v.6.13.4;ActiGraph, Pensacola, FL, USA) was given to each subject.The subjects were instructed to wear the device on the non-dominant wrist for 7-10 days.
This allowed us to assess physical activity levels for young and old subjects.
Step count and the number of minutes per day spent in light, moderate and vigorous exercise were calculated to match the activity levels between the younger and older groups.Activity thresholds were based upon the thresholds recommended by Montoye et al. (2020).

Visits 2-4: Determination of CP and P GXT
During visits 2-4, a series of constant-power knee-extension tests were performed until task failure to determine knee-extension CP.
The subjects started each test by warming up at 40% P GXT for 5 min, followed by a 5 min rest period.Then each subject performed another warm-up at 40% P GXT for 3 min.Directly after the 3 min warm up, the resistance of the ergometer was set at 85% of the subject's P GXT .
Subjects exercised at this intensity until task failure.Following a brief cool-down and a 20 min rest period, the subjects then performed a second test to exhaustion at another power output between 80% and 105% of P GXT .Overall, three to five tests to task failure, ranging in duration between 2 and ∼15 min (Figure 1a), were completed over the course of three visits, with no more than two tests per visit.(Broxterman et al., 2017;Gifford & Richardson, 2017).Previous studies have demonstrated that the hyperaemic response to a single-leg movement is nitric oxide dependent (Broxterman et al., 2017) and elicits a minimal chronotropic response (Venturelli et al., 2017).On arrival, subjects were seated in a raised chair, with their legs resting on a table slightly below their hip height.Subjects were connected to the Sun Tech Tango blood pressure machine (SunTech Tango+; SunTech Medical, Morrisville, NC, USA), in addition to a knee brace set to 90 • of flexion and 180 • of extension.
The researcher then used the Doppler ultrasound machine (9L probe; General Electric Medical Systems, Milwaukee, WI, USA) to scan the subject's common femoral artery after the subject had been resting for ≥15 min.For the first 60 s, the blood velocity and artery diameter were recorded for a baseline.Next, a second researcher moved the right leg from 180 • to 90 • and then back a single time in 1 s.For the next 60 s, the researcher continued to record blood flow through the common femoral artery.This process was repeated three or four times, with a 5 min resting period between each test.Blood flow and vascular conductance were calculated as described below.Arterial diameter was calculated as the average of five diameter measurements assessed with software on the ultrasound machine at end diastole during baseline.Blood velocity data were binned into 1 s averages, and a 3 s rolling average was applied to smooth the data.The greatest blood flow achieved over 1 s was identified as the peak flow, and the total response was quantified as the area under the curve, as previously described (Gifford & Richardson, 2017).Resting blood pressure was determined as the average of three measurements performed while subjects sat recumbent during the 5 min break between tests.

2.5.2
Determining the cardiovascular response to exercise at 10 and 20 W Femoral artery blood flow and blood pressure were measured as subjects performed several bouts of knee-extension exercise below and above CP.Initially, subjects performed knee-extension exercise at 10 W for 5 min while femoral artery blood flow and brachial artery blood pressure were measured.After performing at 10 W for 5 min, the resistance on the ergometer was increased to 20 W. Subjects exercised at 20 W for 5 min or until task failure (whichever came first).

Determining the cardiovascular response to exercise slightly below CP
We next investigated the blood flow response to exercise as near to CP as possible.Owing to the limitations of 1 W resolution on our cycle ergometer and error in the estimate of CP, 90% CP was the closest to CP that we could have everyone exercise, while still being certain that they did not exceed CP.
After a rest period of ≥10 min, subjects performed a warm-up, exercising at 40% CP for 3 min, then immediately transitioned to an intensity of 90% CP.Subjects sustained this intensity for 10 min.During this test, a researcher scanned the femoral artery to record blood flow slightly below CP.Brachial artery blood pressure was taken every third minute as described above.

2.5.4
Determining the cardiovascular response to exercise at P GXT After another 10 min of rest, subjects exercised at 40% CP for 3 min, then immediately transitioned to an intensity of 100% P GXT (determined during visit 4).Subjects maintained this intensity until task failure.Blood pressures were taken every third minute.

2.6
Data analysis and statistical analysis 2.6.1 Calculation of femoral artery blood flow and conductance Blood flow was measured with the Doppler ultrasound probe on the common femoral artery during the exercise.Specifically, a 9 MHz probe operating at a B-mode frequency of 12 MHz and a Doppler frequency of 4 MHz was used to insonate the artery and measure blood velocity at an insonation angle of 60 • .After data collection, blood flow was calculated with the following equation: Flow = mean blood velocity × πr 2 , where mean blood velocity is the time-averaged mean (TAmean) reported by the ultrasound machine and r is the radius of the artery, calculated as one-half of the measured diameter.Arterial diameter was measured as the average of five measurements during each preceding baseline at a perpendicular angle along the central axis of the artery during end diastole using software on the ultrasound system (Gifford et al., 2016).Mean arterial pressure was calculated as diastolic blood pressure + ([systolic blood pressure − diastolic blood pressure]/3).Vascular conductance was calculated as the quotient of simultaneously measured blood flow and mean arterial pressure.For the purposes of this paper, exercise blood flow and conductance are represented as the average of the final 30 s of each stage, while peak flow and conductance measured during PLM are reported as the greatest 1 s average during the test (see above).

Statistical analysis
The main variables of interest were CP, W′, peak blood flow during PLM, and blood flow, blood pressure and vascular conductance during submaximal exercise (10 W, 20 W and 90% CP) and maximal exercise (100% P GXT ).Two-way ANOVA, with age group (young vs. old) and sex (female vs. male), was performed to determine the difference in variables between young and old subjects.In the event of a significant F-statistic, planned comparisons with Student's t-tests were performed.Pearson correlation between the variables mentioned above was used to explore the relationship between the various indices of blood flow and exercise tolerance.The value of α was set at 0.05.
Data are presented as the mean ± SD unless stated otherwise.Data are available upon request.was observed (P = 0.980).When P GXT was normalized for body mass, a significant main effect of age was detected (old = 0.44 ± 0.10 W kg −1 vs. young = 0.73 ± 0.17 W kg −1 ; P < 0.001).As illustrated in Figure 1b, when normalized for leg lean mass, there was no longer a significant effect of sex (P = 0.443).Importantly, when normalized by leg lean mass, P GXT was significantly lower in the older subjects (old = 4.15 ± 0.90 W kg −1 vs. young = 5.88 ± 0.99 W kg −1 ; P < 0.001), with no significant interaction observed (P = 0.283).

Effect of age on critical power
When examining differences in CP with a two-way ANOVA (age and sex), a significant main effect of age was observed, such that older individuals had a 32% lower absolute CP than younger individuals (Figure 1a; old = 28.5 ± W 10.4 vs. young = 41.8 ± 8.3 W; P = 0.007).A significant main effect of sex was also observed, such that females tended to have a lower absolute CP than males (females = 28.5 ± 10.4 W vs. males = 41.8 ± 8.3 W; P = 0.001).No interaction between age and sex was observed (P = 0.273).When normalized by total body mass, the main effect of age on CP was widened to a 40% difference between young (0.57 ± 0.13 W kg −1 ) and old (0.34 ± 0.10 W kg −1 ; P < 0.001).As illustrated in Figure 1c, the main effect of age persisted when normalized for leg lean mass, assessed by DEXA (old = 3.20 ± 0.94 W kg −1 vs. young = 4.60 ± 0.87 W kg −1 ; P < 0.001), and no sex differences (P = 0.486) or interactions (P = 0.940) were apparent.

Effect of age on W′
When examining differences in CP with a two-way ANOVA (age and sex), a significant interaction between age and sex for W′ existed (P = 0.038).Post hoc analysis revealed that young males exhibited a greater W′ than all other groups, such that there was a significant difference in W′ between young and old males (old males = 1897 ± 254 J vs. young males = 3535 ± 1445 J; P = 0.013), but not young and old females (old females = 1713 ± 774 J vs. young females = 1474 ± 294 J; P = 0.690).A significant main effect of sex was also observed (P = 0.016), such that, on average, males had a significantly greater W′ than females.As indicated in the interaction, this was primarily driven by the greater W′ of the young males.
When W′ was normalized by total body mass, significant main effects of age (P = 0.029), sex (P = 0.044) and an age-by-sex interaction (P = 0.018) were detected.This interaction was driven primarily by a large decrease in W′ in old men (20.06 ± 4.37 J kg −1 ) compared with young men (46.97 ± 16.34 J kg −1 ; P = 0.016).
As illustrated in Figure 1d, the significant interaction between age and sex persisted when W′ was normalized for leg lean mass (Figure 1d; P = 0.012), such that young males exhibited a greater mass-specific W′ than all other groups.Besides the interaction, no significant main effect of age (P = 0.127) or sex (P = 0.221) was observed for W′ when normalized by leg lean mass.
As illustrated in Figure 1e, a significant interaction was observed when considering CP as a percentage of P GXT (P = 0.032), with the CP tending to occur at a higher percentage of P GXT in older males than younger males (82.35 ± 1.23% vs. 75.03± 2.29%, respectively; P = 0.083), and at a lower percentage of P GXT in older females than younger females (79.63 ± 3.96% vs. 81.72 ± 1.17%, respectively; P = 0.089).Notably, the percentage of P GXT at which CP occurred was inversely related to mass-specific W′ (R 2 = 0.52, P < 0.001).
Figure 2a illustrates leg blood flow normalized by leg lean mass when exercising at 90% CP.When normalized by leg lean mass, sex differences in blood flow at 90% CP disappeared (P = 0.251), while the main effect of age persisted (young = 522.08 ± 123.47 mL min −1 kg −1 vs. old = 371.04± 115.84 mL min −1 kg −1 ; P = 0.014).No age-bysex interaction was observed (P = 0.711).Blood flow achieved during the 10th minute of exercise at 90% CP did not differ significantly from maximum blood flow achieved when exercising at P GXT , averaging 93.56 ± 16.27% of maximum flow achieved during P GXT (P = 0.122).
As illustrated in Figure 2b, neither age (P = 0.815) nor sex (P = 0.579) affected the relationship between blood flow at 90% CP and P GXT .
As illustrated in Figure 2c, older individuals exhibited significantly lower vascular conductance per kilogram of leg lean mass than their young counterparts when exercising at 90% CP (P = 0.008).As illustrated in Figure 2d, neither age nor sex affected the relationship between conductance at 90% CP and P GXT .Notably, the vascular conductance achieved at 90% CP did not differ from maximum conductance (P = 0.897), averaging 99.38 ± 18.15% of the maximum vascular conductance achieved during P GXT .Absolute values for blood flow (i.e., not normalized by leg lean mass) are presented in Table 2.

3.3
Effect of age on vascular function

Effect of age on vascular endothelial function assessed during PLM
When considered in absolute terms, older adults exhibited significantly lower peak responses (P = 0.015) and total hyperaemic responses (P = 0.008) to PLM.A main effect of sex was observed for the peak response to PLM, such that females exhibited a significantly lower absolute response to PLM (P = 0.039).
Given that differences in muscle mass could mediate the observed differences in vascular function, we normalized the response to PLM by leg lean mass.As illustrated in Figure 3b, c, all sex differences disappeared, whereas all age differences persisted when normalizing for leg lean mass.

Relationship between vascular function, blood flow and exercise tolerance in young and old adults
Absolute CP was correlated with peak leg blood flow in response to PLM (R 2 = 0.40; P = 0.004) and correlated with leg blood flow during knee-extension exercise at 90% CP (R 2 = 0.30; P = 0.015).As illustrated in Figure 4, when normalizing for leg muscle mass, CP is strongly correlated with leg blood flow (R 2 = 0.44; P = 0.002; Figure 4a) and vascular conductance (R 2 = 0.39; P = 0.004; Figure 4b) during kneeextension exercise at 90% CP and the peak hyperaemic response to PLM (R 2 = 0.52; P < 0.001; Figure 4c).As illustrated in Figure 5, the blood flow and vascular conductance responses to various intensities of knee-extension exercise (e.g., 20 W, 90% CP and P GXT ) were related to vascular endothelial function assessed by PLM.

F I G U R E 2
The effect of age on blood flow when exercising slightly below (90%) critical power (CP).(a) Leg blood flow, normalized by leg lean mass, during the 10th minute of exercise at 90% CP.(b) Blood flow during the 10th minute of exercise at 90% CP expressed as a percentage of maximum blood flow achieved when exercising at the maximum power achieved during a graded exercise test (P GXT ).(c) Leg vascular conductance, normalized by leg lean mass, during the 10th minute of exercise at 90% CP.(d) Vascular conductance during the 10th minute of exercise at 90% CP expressed as a percentage of maximum conductance achieved when exercising at P GXT .*Significant effect of age.Small symbols represent individual data points, with triangles representing females and circles representing males.Statistics were determined with a 2 × 2 ANOVA, with 10 subjects (five females and five male) in each group for (a) and (c).Owing to difficulties in measuring blood flow/pressure during maximum exercise, only eight young (three female and five male) and seven old (five female and two male) subjects were included in the analysis for (b) and (d).

DISCUSSION
The purpose of this study was to determine whether the age-related decrease in CP occurs independently of changes in body composition and whether it is related to impaired muscle blood flow and vascular function.We observed that CP during knee-extension exercise was lower in active older adults than in active young adults, even after controlling for muscle mass.We also observed that the age-associated decrease in CP was strongly related to impaired blood flow and vascular conductance during exercise and to decreased endothelial function of the resistance vasculature.The context and implications of these findings is discussed below.

Does decreased muscle quality contribute to the age-related reduction in CP?
As the boundary between steady-state and non-steady-state exercise (Jones et al., 2019;Poole et al., 2016), a person's CP strongly influences the metabolic conditions, and consequently, the fatigue and sustainability that will be experienced at a given power output (Jones TA B L E 2 Absolute leg blood flow and vascular conductance during single-leg knee-extension exercise.Data presented here are absolute values, not normalized for leg lean mass.Abbreviations: CP, critical power; P GXT , maximum power achieved during a graded exercise test.Statistics were determined with Student's unpaired t-tests, with 10 subjects (five females and five male) in each group.Data are means ± SD.   et al., 2008;Vanhatalo et al., 2016).Neder et al. (2000) previously showed that sedentary males >60 years old exhibited significantly lower absolute CP during cycling than sedentary males <30 years old.Gifford and Collins (2021) and Fulton et al. (2023) demonstrated that critical speed, the running surrogate of CP, decreases with age even in very active, world-class masters' athletes.Unfortunately, body composition was not assessed in these previous studies, making it unclear whether the reported decrease was merely attributable to age-associated decreases in muscle mass or to increases in total body mass (Distefano & Goodpaster, 2018;Fleg et al., 2005).

Parameter
In our study, we measured CP during single-leg knee-extension exercise.Consistent with previous studies (Fulton et al., 2023;Gifford & Collins, 2021;Neder et al., 2000), we observed that absolute CP during knee-extension exercise was 32% lower in older adults than in young adults (Figure 1a).Importantly, CP was still 30% lower in the old adults when normalized for leg lean mass (Figure 1c).With The relationship between leg vascular conductance during 90% CP knee-extension exercise when normalized for muscle mass and peak leg blood flow in response to PLM when normalized for muscle mass.(f) The relationship between leg vascular conductance during P GXT when normalized for muscle mass and peak leg blood flow in response to PLM when normalized for muscle mass.Statistics were determined by linear regression, with five young females, five young males, five old females and five old males included in the analysis, except for (c) and (f), which had only eight young (three female and five male) and seven old (five female and two male) subjects included in the analysis owing to difficulties in measuring blood flow/pressure during maximal exercise.
age differences persisting after normalizing for muscle mass, it is clear that the age-associated decrease in CP involves impairments in muscle quality, not only muscle quantity (i.e., mass).Importantly, typical age-associated changes in muscle mass (e.g., sarcopenia) and body composition (e.g., increased body mass) are likely to compound the impact of the reduced muscle quality by requiring less muscle (of a lower quality) to lift more weight with each step (Distefano & Goodpaster, 2018;Fleg et al., 2005).Although indices of leg lean mass did not differ between groups in the present study, total body mass tended to be greater in the old adults (Table 1).Consequently, the disparity in CP between young and old adults grows to 40% when considering CP relative to total body mass (young = 0.57 ± 0.13 W kg −1 vs. old = 0.34 ± 0.10 W kg −1 ; P < 0.001).Thus, older adults, especially those who are overweight, are likely to be even more susceptible to fatigue during weight-bearing exercises, such as walking, than non-weight-bearing exercises, such as knee extension or stationary cycling.

Does W′ vary with age?
The W′ can be defined as the amount of work and accompanying metabolic disturbance that can be tolerated when working above CP (Poole et al., 2016).Neder et al. (2000) showed that W′ was lower in sedentary males, as was CP.Likewise, Gifford and Collins (2021) reported that distance prime (D′; a surrogate of W′) was lower in 60-year-old athletes than in 35-year-old athletes.We observed a significant interaction between age and sex, such that mass-specific W′ was lower in older males than young males, but there was no significant difference in W′ between young and older females, as illustrated in Figure 1d.In other words, male subjects exhibited a greater impact of age on W′ than female subjects.Additional studies designed to investigate sex differences are needed to confirm or reject these findings in our small sample; nevertheless, the differences in W′ between young males and young females are consistent with previous studies showing sex differences in W′ or D′ during cycling exercise (Collins et al., 2022;James et al., 2023) and running exercise (Gifford & Collins, 2021).Like CP, the deficit in W′ is exacerbated when considering total body mass, to the point that the younger subjects were able to perform ∼35% more work per kilogram than the older subjects (P = 0.028).Thus, even if young and old subjects had the same CP, the older subjects would reach exhaustion sooner than the younger subjects when performing weight-bearing activities above CP.
The W′ appears to reflect the relationship between CP and P GXT , with a higher mass-specific W′ being related to CP occurring at a lower percentage of P GXT (R 2 = 0.52, P < 0.001).On average, young and old adults reached CP at the same percentage of P GXT (Figure 1e); however, there was substantial variability from person to person, with CP occurring as low as 57% P GXT and as high as 89% P GXT .This is in agreement with previous studies (Collins et al., 2022;James et al., 2023;Poole et al., 2016), demonstrating that CP cannot be estimated accurately as a fixed percentage (e.g., 80%) of P GXT or VO 2 max .
Such variability in how CP relates to P GXT might explain the inconsistency of responses to exercise between individuals, even when the exercise is performed at a given percentage of P GXT or VO 2 max (Collins et al., 2022(Collins et al., , 2023;;Meyler et al., 2021).For example, if a study tried to determine the impact of age on fatigability in our subjects by having them exercise at 80% P GXT , most of the young men would be exercising in the uncompensable, severe-intensity domain, with its unique physiology, whereas most of the older men would be exercising in the compensable, heavy-intensity domain, with its unique physiology (Poole et al., 2016).Thus, it might be more effective to describe and prescribe exercise in relationship to CP, not P GXT or VO 2 max power (Collins et al., 2022;Meyler et al., 2023;Poole & Jones, 2023).

Is leg blood flow reduced in older adults when exercising near CP?
Having observed that a decrease in muscle quality contributes to the age-related decrease in CP, we next sought to determine whether decreases in the function of the vascular system might play a role.
As illustrated in Figure 2, we measured leg blood flow and vascular conductance while young and old subjects exercised slightly below CP and again while they exercised to exhaustion at the maximum power previously achieved during a graded exercise test.Older subjects exhibited substantially lower blood flow (Figure 2a) and conductance (Figure 2c) per kilogram of lean mass than young subjects.Given that CP is considered a submaximal exercise, we were impressed at the high rates of leg blood flow achieved when exercising slightly below CP.Despite exercising at ∼30% lower power output (90% CP = 31.64± 2.39 W vs. 100% P GXT = 45.15 ± 3.08 W; P < 0.001), the blood flow and conductance reached at 90% CP did not differ significantly from the maximum blood flow and conductance achieved during the P GXT trial (Figure 2b, d).As suggested by Hammer et al. (2020b), CP appears to require maximal or nearly maximal rates of blood flow.This raises the possibility that, like VO 2 max (Gifford et al., 2016;Levine, 2008), CP is strongly influenced, and potentially limited, by maximum blood flow.
Our observation that blood flow at 90% CP reached near-maximal levels might seem contradictory to observations by Copp et al. (2010), who reported substantially lower blood flow when rodents ran at ∼88% critical speed compared with ∼119% critical speed.We measured blood flow during 10 min of exercise at 90% CP, whereas Copp et al. (2010) measured flow ∼3.5 min into each exercise bout, probably before the slow component increase in flow (Figure 6) and O 2 uptake fully matured.Therefore, it is very likely that flow and O 2 uptake reached higher levels once a steady state was reached when exercising below critical speed.

4.4
Is the age-associated reduction in CP related to vascular dysfunction?
Blood flow is determined primarily by the perfusion pressure across a vascular bed and the accumulative radius or size of the vascular F I G U R E 6 Leg blood flow response to knee-extension (KE) exercise at four different intensities.All subjects performed 10 min of KE at 10 W, 20 W and 90% of their personal critical power (CP).Subjects also performed exercise at their previously determined maximum power (P GXT ) until task failure.Leg blood flow is expressed as a percentage of maximum blood flow achieved when exercising at P GXT .Owing to missing time points (e.g., poor blood flow measurements at the onset of an exercise bout), only 15 subjects (nine young and six old) were included in this analysis.Data are not separated by age.Significant main effects of time (P < 0.001) were observed for each intensity.*Blood flow at that time point is significantly different from the final time point for that intensity.Data are presented as means ± SEM.
bed (Joyner & Casey, 2015).The size of the vascular bed can be manipulated structurally, with angiogenesis and rarefaction, or it can be manipulated functionally, with vasodilatation and vasoconstriction (Joyner & Casey, 2015).Unfortunately, we were unable to measure capillary density in these subjects, and existing data are equivocal on whether capillary density decreases in active older adults (Landers-Ramos & Prior, 2018).However, we did measure vasodilatory function of the leg with vascular conductance during KE exercise and with PLM-induced hyperaemia (Figure 3).
As the quotient of flow and perfusion pressure, vascular conductance provides insight into the size and vasodilatory ability of the vascular bed.In addition to lower vascular conductance at 90% CP, older adults also exhibited significantly blunted conductance when exercising at the same absolute power output (20 W) as the young (Figure 3a).Importantly, mass-specific conductance at 20 W (R 2 = 0.24, P = 0.039) and at 90% CP (R 2 = 0.39, P = 0.004; Figure 4a) were both significantly related to mass-specific CP across all subjects, supporting the notion that CP might be related to vascular (dys)function (Poole et al., 2021).
Like others (Groot et al., 2015;Hydren et al., 2019;Mortensen et al., 2012;Trinity et al., 2015), we also observed a blunted hyperaemic response to a single passive movement of the leg.However, for the first time, we demonstrated that the age-related difference in PLM-induced hyperaemia persists even when normalizing for leg lean mass (Figure 3b, c), thereby providing further evidence of impaired muscle quality with ageing.Given that the hyperaemic response to PLM is strongly related to acetylcholine-induced hyperaemia (Mortensen et al., 2012) and is mostly blunted with the inhibition of nitric oxide synthase (Broxterman et al., 2017;Mortensen et al., 2012), the blunted hyperaemic response of the old is suggestive of impaired endotheliumdependent dilatation and reduced nitric oxide availability at the level of the resistance arteries that regulate blood flow (Broxterman et al., 2017;Gifford & Richardson, 2017;Limberg et al., 2020).Such impairments in dilatory function could conceivably reduce maximal muscle blood flow (Hanson et al., 2020) and contribute to decreased CP.In fact, PLM-induced hyperaemia was strongly related to CP (Figure 4c) and the vascular response to exercise at 20 W, 90% CP and P GXT (Figure 5).
In addition to causing limitations to oxygen delivery (Gifford et al., 2016)

Experimental considerations
This study focused solely on healthy, unmedicated active adults.
Although it seems likely that inactivity and diseases typical of ageing would exacerbate the discrepancies between young and old seen here, specific research on sedentary, diseased and medicated populations is needed to see whether this is the case.We also included equal representation of male and female subjects in this study.With our small sample size, sex-difference data should be viewed as exploratory and interpreted with caution.Future studies are required to confirm or reject our exploratory findings.
The leg muscle mass used for normalization purposes was estimated with DEXA, which includes muscles other than the quadriceps femoris in its estimate.We also estimated quadriceps muscle mass with validated anthropometric measures (Layec et al., 2014).Like leg lean mass estimated by DEXA, estimated quadriceps mass did not differ between groups (Table 1).Importantly, leg lean mass obtained with DEXA and quadriceps mass obtained with anthropometry were significantly correlated (r = 0.66, P = 0.001).Given that normalizing the data by either estimate resulted in the same conclusions, we present only the data normalized by DEXA-estimated leg lean mass.
We determined CP with two bouts of exhaustive exercise with 30 min of cool-down, rest and warm-up between them.We have found that this approach yields very reproducible measures of CP (coefficient of variation ≤ 3% across three separate sets of tests across a month) (James et al., 2023).Although previous data suggest that 30 min of recovery between trials results in valid estimates of CP (Karsten et al., 2017;Triska et al., 2021), it is possible that fatigue from the first bout affected the second bout, which might have caused us to overestimate or underestimate CP or W′.Exercise below CP is typified by steadystate conditions (Jones et al., 2019;Poole et al., 2016).In Figure 6, we illustrate the leg blood flow response to the various exercise intensities for young and old adults combined.We tested for the existence of a steady state with post-hoc analyses comparing blood flow at each time point with the final blood flow measured for that intensity.If we overestimated CP, a steady state would not be observed.Notably, exercise at 90% CP resulted in steady-state blood flow after only 5 min, supporting the notion that the exercise was performed in the heavy domain.If we underestimated CP systematically, blood flow would probably be much lower than the maximum blood flow achieved during P GXT .Impressively, the data presented in Figures 2b and 6 indicate that steady-state blood flow was reached at a very high percentage (∼94%) of the maximum flow observed during P GXT .With blood flow reaching such a high steady state that did not differ statistically from maximum blood flow (Figure 2b), it seems likely that, despite performing two exhaustive bouts in a day to determine CP, our estimates of CP were accurate.Nevertheless, it is still possible that the estimates of W′, which is often less reproducible (James et al., 2023;Karsten et al., 2017), were affected by our methodology and should be interpreted with caution.
Vascular function and exercise tolerance were assessed in a relatively small percentage of muscle mass that is not typically limited by cardiac output.Although the quadriceps are very relevant to locomotion, it is unclear how well the results of the present study apply to whole-body exercise, such as walking, running or cycling.Given that blood flow at CP was not significantly different from maximum blood flow, it seems likely that CP could be sensitive to changes in maximum cardiac output, as was the case in a rodent model of heart failure (Craig et al., 2019).

CONCLUSIONS
Even when accounting for differences in muscle mass, physically active older adults exhibit reduced CP, indicating that changes in muscle quality contribute to the age-related decline in exercise tolerance.

F
I G U R E 1 The effect of age on mass-specific exercise tolerance during single-leg knee-extension exercise.(a) Illustration of the relationship between exercise power and time to task failure for young and older adults (data are means ± SEM).(b) Maximum power output achieved during a graded exercise test (P GXT ) normalized by leg lean mass.(c) Critical power (CP) normalized by leg lean mass.(d) Work prime (W′) normalized by leg lean mass.(e) Critical power expressed as a percentage of P GXT .*Significant effect of age.Small symbols in (b-e) represent individual data points, with triangles representing females and circles representing males.Statistics were determined with a 2 × 2 ANOVA, with 10 subjects (five females and five male) in each group.

E 3
Effect of age on vascular function.(a) Leg vascular conductance when performing 20 W of knee-extension (KE) exercise.(b) Total leg blood flow response to passive leg movement (PLM), normalized by leg lean mass, with statistics performed on the area under the curve.(c) Peak blood flow response to PLM. *Significant effect of age.Small symbols in (a) and (c) represent individual data points, with triangles representing females and circles representing males.Statistics were determined with a 2 × 2 ANOVA, with 10 subjects (five females and five male) in each group.

F
Relationship between critical power (CP) and vascular function.(a) The relationship between mass-specific CP and mass-specific leg blood flow during 90% CP knee-extension exercise.(b) The relationship between mass-specific CP and mass-specific vascular conductance at 90% CP.(c) The relationship between mass-specific CP and vascular function assessed by passive leg movement (PLM).Statistics were determined by linear regression, with five young females, five young males, five old females and five old males included in the analysis.F I G U R E 5 Relationship between resistance artery function and exercise blood flow and conductance.(a) The relationship between leg blood flow during 20 W knee-extension exercise when normalized for muscle mass and peak leg blood flow in response to passive leg movement (PLM) when normalized for muscle mass.(b) The relationship between leg blood flow during 90% critical power (CP) knee-extension exercise when normalized for muscle mass and peak leg blood flow in response to PLM when normalized for muscle mass.(c) The relationship between leg blood flow during maximum power achieved during a graded exercise test (P GXT ) knee-extension exercise when normalized for muscle mass and peak leg blood flow in response to PLM when normalized for muscle mass.(d) The relationship between leg vascular conductance during 20 W knee-extension exercise when normalized for muscle mass and peak leg blood flow in response to PLM when normalized for muscle mass.(e) , Poole et al. (2021) recently offered an additional hypothesis for how vascular dysfunction might impair CP.They suggested that impaired vascular function could result in an imbalance in oxygen delivery and consumption at the start of exercise, which would lower the intramuscular pressure of oxygen, leading to exaggerated metabolic stress and exercise intolerance.The strong relationship between PLM-induced hyperaemia and CP (Figure 4c) supports this hypothesis put forward by Poole et al. (2021).Although these data do not establish a causal relationship between vascular function and CP, they do justify further investigation of the impact of vascular function on exercise tolerance.Research examining the impact of vasoactive interventions (Poole et al., 2021) on exercise tolerance seem especially justified by these data.
Typical age-related changes in body composition (e.g., decreased muscle mass and increased total body mass) are likely to magnify the exercise intolerance caused by impaired muscle quality, making previously sustainable activities of daily living very fatiguing.Differences in exercise blood flow and vascular function are strongly related to CP in young and old adults.Future research should determine whether improvements to vascular function have a meaningful impact on CP and exercise tolerance.AUTHOR CONTRIBUTIONSThis study was performed in the Human Performance Research Laboratory at Brigham Young University.Abigail Dorff conceived the work, acquired and analysed the data and drafted the manuscript.Christy Bradford acquired and analysed the data and critically revised the manuscript.Ashley Hunsaker acquired and analysed the data and critically revised the manuscript.Jake Atkinson acquired and analysed the data and critically revised the manuscript.Joshua Rhees acquired and analysed the data and critically revised the manuscript.Olivia Leach acquired and analysed the data and critically revised the manuscript.Jayson Gifford conceived the work, acquired and analysed the data and drafted the manuscript.All authors approved the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Table 1
Subject characteristics.The statistics were determined with Student's unpaired t-tests, with 10 subjects (five females and five male) in each group.Data are means ± SD.
TA B L E 1 knee-extension exercise.Visits 2-4 involved further familiarization with the exercise and a series of tests to task failure to determine each subject's CP, work prime (W′) and maximum power reached during a graded exercise test (P GXT ).Visit 5 involved measuring vascular function with PLM-induced hyperaemia, and determining blood flow and blood pressure at various intensities of exercise below and above CP.