Persistent Improvements in Running Economy With Advanced Footwear Technology During Prolonged Running in Trained Male Runners

ABSTRACT

This study evaluated the effects of a top‐tier carbon plated shoe with highly responsive foam (CP‐shoe) on changes in running economy (RE) and running speed at lactate threshold (LT) before, during, and after prolonged running. Ten male distance runners (half‐marathon time: 75 ± 3 min) completed a series of tests. Initially, two graded treadmill tests were undertaken in a well‐rested state while wearing either a CP‐shoe or non‐carbon plated shoe (NCP‐shoe) to determine RE and LT‐speed (LT_CP and LT_NCP). On separate days, participants then performed three 80‐min runs in randomized order: twice at 95% LT_NCP (14.7 ± 0.5 km·h⁻¹, once with each shoe) and once at 95% LT_CP (15.2 ± 0.4 km·h⁻¹, CP‐shoe only). RE, blood lactate concentration ([La⁻]_b), heart rate (HR), and perceived exertion (Borg scale) were recorded throughout. Each 80‐min run was followed by a graded exercise test with assessment of RE and LT‐speed. At matched external workload (95% LT_NCP), the CP‐shoe improved RE and lowered [La⁻]_b, HR, and Borg compared with the NCP‐shoe. At matched internal workload (95% LT specific to each shoe), the CP‐shoe again demonstrated superior RE and lower HR. However, the time course of changes during the 80‐min runs did not differ between shoes (i.e., no shoe × time interactions). Across all 80‐min runs, LT‐speed and RE improved postexercise, with the CP‐shoe yielding higher LT speeds than the NCP‐shoe (+0.5–0.6 km·h⁻¹). In conclusion, the CP‐shoe enhanced RE and reduced perceived exertion and [La⁻]_b compared with the normal running shoe during prolonged running, although the magnitude of changes over time was not different between shoes.

1. Introduction

A seminal review by Sjödin and Svedenhag identified three key physiological determinants of marathon performance: maximal oxygen uptake (V̇O₂max), the fraction of V̇O₂max that can be sustained during long‐distance running (i.e., performance V̇O₂), typically measured as running speed at lactate threshold (LT), and running economy (RE), defined as the oxygen or energetic cost of running at a given speed or distance [1]. These factors have been used in models to estimate running speed and predict marathon performance and have catalyzed scientific and public interest in whether a sub‐2‐h marathon is physiologically attainable [2, 3]. Due to the widespread acceptance of this physiological model, laboratory‐based assessments of V̇O₂max, performance V̇O₂, LT, and RE have become standard practice among competitive long‐distance runners.

Metrics such as RE and LT are typically assessed in a well‐rested, non‐fatigued state to ensure standardized and reliable measurements. While this approach certainly yields valuable insights, it does not consider the progressive physiological and biomechanical changes that occur during prolonged exercise, factors that may partly explain the limited predictive power of these variables for marathon performance among elite runners [3, 4]. The ability to resist deteriorations in, for example, RE and LT over time has been termed “durability” or “resilience”, and has been proposed as a distinct, temporal fourth dimension influencing endurance performance [5, 6]. Most research on durability in athletes has employed cycling as the exercise modality and consistently found that power output eliciting the first ventilatory threshold (VT₁) declines during prolonged exercise [7, 8]. Although durability has been less extensively studied in running, similar patterns have been observed, with running speed at LT declining after prolonged efforts [9, 10]. While the factors underlying the observed declines in running speed at LT are likely multifactorial, the concurrent deterioration of RE appears to play a role. Accordingly, most studies examining exercise‐induced changes in RE have reported clear impairments following prolonged running [9, 10, 11, 12], while others have reported no or less clear changes [13, 14, 15]. Collectively, these observations suggest that impairments in both RE and LT can be key factors contributing to reduced performance during prolonged running.

Over the past decade, distance running has experienced a technological revolution, largely driven by the emergence of advanced footwear technology incorporating carbon fiber midsoles and highly responsive, energy‐returning foam. This footwear type has been demonstrated to enhance RE by ~4% in laboratory settings [16, 17], translating into improvements in distance running performance [18]. While these benefits have been observed across various performance levels of competitive runners, studies have primarily assessed their effects under well‐rested, non‐fatigued conditions. As a result, it remains unclear whether improvements in RE are maintained during later stages of prolonged running. A recent study reported that adding only a carbon plate to running shoes elicited a persistent 1% improvement in RE over the course of a half marathon [15], whereas the effect of incorporating both a carbon plate and highly responsive, energy‐returning foam remains unclear. In this context, Jones and Kirby [6] have proposed that the performance benefits of carbon‐plated shoes with cushioned responsive foam during prolonged running may also be attributed to reduced muscle damage and thereby improved durability [19, 20]. Supporting this idea, running on softer surfaces has been shown to lessen postexercise muscle soreness [21], suggesting that cushioned footwear may confer a similar benefit. Indeed, advanced footwear technology has demonstrated protective benefits by reducing muscle soreness, damage, and inflammation following marathon running [19]. This potential protective effect emphasizes the importance of assessing modern running shoes not only under well‐rested conditions but also during the later stages of prolonged running. Accordingly, the present study was designed to investigate the effects of carbon‐plated shoes with highly responsive foam on changes in RE and running speed at LT during and after prolonged running. We hypothesized that these shoes would attenuate exercise‐induced deteriorations in RE and running speed at LT. Given the relevance of these outcomes for trained runners, the study was conducted specifically in this population.

2. Methods

2.1. Subjects, Ethics and Laboratory Settings

Eligibility required participants to: (1) be 18–40 years old; (2) have completed a half‐marathon within the previous year in ≤ 80 min (males) or ≤ 85 min (females); (3) regularly compete in running events. Exclusion criteria included any injury that hindered normal training within four weeks prior to preliminary testing. No specific familiarization with the tested shoes or treadmill running was required. Ten trained and highly trained male runners [22] participated in the study (age 27 ± 3 years, body mass 77.9 ± 6.2 kg, height 186 ± 8 cm, ½ marathon 75.4 ± 3.3 min, weekly training volume 74 ± 21 km). None of the participants regularly used either the NCP‐shoe model or CP‐shoe model during training. Four runners reported using the specific CP‐shoe in competitions, with the remaining six participants using other CP‐shoe models for this purpose. In addition, four participants reported regular treadmill running during the winter. Before inclusion, participants received comprehensive written and oral information about the study. Prior to preliminary testing, all participants provided written informed consent, confirming their understanding of the study's methodology, and potential benefits, and risks. The Regional Committee on Health Research Ethics for Southern Denmark reviewed the study and determined that formal registration was not require (project‐ID: S‐2024200‐121). Utilizing a cross‐over design and assuming that changes in RE would deteriorate 5 ± 5 mL·kg⁻¹·km⁻¹ less during the 80‐min trial in the CP‐shoe with highly responsive foam compared to the NCP‐shoe, a sample size calculation revealed that nine participants would be required to achieve statistical power of 80%. One additional participant was recruited to account for potential drop‐out.

The study was conducted at two locations: The Department of Sports Science and Clinical Biomechanics, University of Southern Denmark, Odense, or The Section for Sports Science, Aarhus University, Aarhus. Identical equipment (i.e., motorized treadmills, metabolic carts and lactate analyzers) was used at both sites (see details later). Recruitment began in August 2024, with test sessions conducted from November 2024 to April 2025. During this period, the runners were in their general preparation phase leading up to the 2025 road racing season.

2.2. Study Design

Using a randomized, crossover and counterbalanced study design, trained to well‐trained male long‐distance runners visited the laboratory on four occasions (Figure 1). The first visit included two graded protocols, while wearing a carbon‐plated running shoe with highly responsive foam (CP‐shoe; Nike Alphafly 3; mass EU 44: 218 g) and a non‐carbon training shoe (NCP‐shoe; Nike Pegasus 41; mass EU 44: 297 g). During these tests, RE and speed at LT were assessed in both the CP‐shoe (LT_CP) and the NCP‐shoe (LT_NCP). The next three visits were designed to compare the physiological responses to prolonged running in the CP‐shoe and the NCP‐shoe at both the same absolute (i.e., external (running speed)) and relative (i.e., internal (% LT1 in the specific shoe)) workload. These sessions were completed in a randomized and counterbalanced order, involving 80 min of running at 95% LT_NCP in both the NCP‐shoe and the CP‐shoe, as well as at 95% LT_CP in the CP‐shoe. Following each run, participants underwent another graded test to assess changes in RE and running speed at LT in response to prolonged running. To minimize bias from factors such as diet and circadian rhythm, all sessions were initiated at the same time of day within participants. Before each of the four laboratory visits, participants were instructed to consume a CHO‐enriched diet, replicating their habitual competition‐day nutrition, and to refrain from caffeine. Participants were instructed to avoid strenuous activities for 24 h prior to testing. Thus, the same individual preparation procedure was repeated prior to all four visits.

FIGURE 1.

Schematic overview of preliminary tests and the three 80‐min running trials. For determination of RE and LT, the preliminary tests included two graded running tests until the RER reached 1.00 ‐ one in the NCP‐shoe and one in the CP‐shoe. These tests were conducted in a randomized, counterbalanced order and separated by 15 min of rest with consumption of 40 g of carbohydrate (well‐rested and fresh condition). The 80‐min running trials were performed on three separate days (5–10 days between) in a randomized, counterbalanced order. These trials were commenced with a 10‐min warm‐up followed by 80 min of running at 95% of LT_NCP on two occasions (i.e., in both the NCP‐ and CP‐shoe) and at 95% of LT_CP on one occasion (i.e., in the CP‐shoe). Throughout the running trials, runners consumed 46 g CHO·h⁻¹ from energy gels. [La⁻]_b was assessed after 14, 34, 54, and 74 min, while respiratory outcomes (V̇O₂ and V̇CO₂) were measured in 5‐min intervals by indirect calorimetry after 15, 35, 55, and 75 min. Following each of the three trials, participants ran for 10 min at 8 km·h⁻¹ before completing a graded test in the assigned shoe for determination of running economy and LT.

2.3. Preliminary Tests (Day 1)

The preliminary test session consisted of two separate submaximal graded tests with progressively increasing running speeds, starting at 13 km/h and increasing by 1 km/h every three minutes (Figure 1). The test continued until the respiratory exchange ratio (RER) remained above 1.00 for 30 s. Between stages, participants briefly stepped onto the treadmill's side rails, allowing for collection of a capillary blood sample from a fingertip for lactate determination. In a randomized and counterbalanced order, participants completed one test in the CP‐shoe and one in the NCP‐shoe, with 15 min of rest and consumption of 46 g of carbohydrate from energy gels and water ad libitum in between. Prior to each test, participants performed 10 min of warm‐up at 10 km·h⁻¹ in the assigned shoe.

During these tests, oxygen uptake (V̇O₂) and carbon dioxide production (V̇CO₂) were continuously measured using an online mixing chamber system and RE was calculated based on these measurements. Also, heart rate (HR) was continuously recorded using a chest strap and the average from the last 15 s of each stage was used to define heart rate at each running speed (Polar H10 sensor, Polar Electro Oy, Kempele, Finland). [La⁻]_b at each stage was determined in duplicates with [La⁻]_b calculated as the average of both values. Thus, two capillary samples were extracted within the 60 s between stages. During the brief rest periods between stages, the rate of perceived exertion was assessed using the Borg scale (range 6–20). Prior to the graded tests, height and body mass were measured while wearing standardized clothing (i.e., tee, shorts, and socks). For further information about equipment and procedures used to measure gas exchange and blood lactate, please see Section 2.5.

2.4. Prolonged Running Sessions (Day 2–4)

Upon arrival on day 2–4, participants were weighed on a digital scale. Participants then completed one of three 80‐min running trials: one at 95% LT_NCP in the NCP‐shoe, one at 95% LT_NCP in the CP‐shoe (CP‐NCP), and one at 95% LT_CP in the CP‐shoe (CP‐CP) (Figure 1). This intensity was chosen based on previous studies reporting deteriorations in RE during prolonged running at or slightly below LT [9, 11, 12]. Each trial was followed by 10 min of easy running at 10 km·h⁻¹, after which [La⁻]_b was assessed to monitor the recovery of blood lactate. Participants then completed a graded test, similar to that performed during the preliminary tests, to evaluate changes in RE and LT from well‐rested to post‐run conditions. Room temperature was stable (21.5°C ± 0.5°C), and subjects consumed individual but standardized amounts of fluid throughout all trials to minimize the risk of inconsistent changes in body mass between trials. Before each trial, a 10‐min warm‐up at 10 km·h⁻¹ was performed in the assigned shoe. Trials were separated by 5–10 days.

During each trial, participants briefly stepped onto the treadmill's side rails after 14, 34, 54, and 74 min for duplicate [La⁻]_b determination, with two capillary samples collected within 30 s at these time points. Immediately afterwards, participants were fitted with equipment for respiratory measurements, and V̇O₂ and V̇CO₂ were measured by a mixing chamber system in 5‐min intervals after 15, 35, 55, and 75 min. After each respiratory measurement, participants rated their perceived exertion using the Borg scale, and heart rate data were recorded as an average of 15 s after 19, 39, 59, and 79 min. To prevent low carbohydrate (CHO) availability and hypoglycemia, participants were provided with 34.5 g CHO·h⁻¹ (~11.5 g CHO per 20 min) from energy gels (Figure 1).

2.5. Analytical Procedures

2.5.1. V̇O₂ and Running Economy

At both test sites, participants ran on similar motorized Woodway treadmills (Woodway ELG, Foster CT, Waukesha, Wisconsin). The treadmill was set to 0% incline, as this has been shown to reliably infer overground RE at submaximal speeds [23]. Using a mixing chamber system (Innocor, Innovision, Odense, Denmark) and based on the pulmonary ventilation and expiratory CO₂ and O₂ concentrations, both V̇O₂ and V̇CO₂ values were continuously sampled throughout all graded tests as well as in 5‐min intervals during the prolonged running trials (Figure 1). Before each measurement, gas analyzers were calibrated with two standard gas mixtures containing 21.00 and 14.97% of O₂ and 0.00 and 4.97% of CO₂, respectively. The flowmeter was calibrated manually at increasing flow rates using a 3 L syringe. Moreover, ambient pressure, humidity, and room temperature were monitored prior to all measurements and accounted for during the calibration. The laboratory was well‐ventilated during all tests to maintain stable ambient conditions. The V̇O₂ values representing each running speed in the graded tests were calculated as the mean of measurements during the last minute of that running speed and subsequently used to calculate RE. During the 80‐min trials, the average of V̇O₂ and V̇CO₂ data from the last 2 min of each 5‐min interval was used to calculate RE and RER. RE calculations were based on participants' pre‐trial body mass. The mixing chamber system has previously been validated, demonstrating a CV of 2.4% for V̇O₂ [24].

2.5.2. Lactate Measurements and Lactate Threshold (LT)

At the termination of each stage of the graded tests, and repeatedly during the 80‐min trials, finger capillary blood samples were collected for lactate analysis (Biosen C‐Line, EKF diagnostics, Cardiff, UK). Blood lactate concentrations were measured in duplicates (i.e., two samples), with [La⁻]_b calculated as the average of both values. LT during the graded tests was determined as the running speed at which [La⁻]_b reached 1 mmol·L⁻¹ above baseline, with baseline calculated as the average [La⁻]_b from the first 2–3 stages and before a rise in [La⁻]_b. The exact LT was identified by plotting the running speed—[La⁻]_b relationship and drawing a straight line through the surrounding lactate concentrations. To ensure a valid relationship between running speed and [La⁻]_b, the stages used to determine LT were checked for steady state conditions. For one participant, baseline [La⁻]_b could not be determined (i.e., increased [La⁻]_b from 1st to 2nd measurement), and consequently, this individual was excluded from the comparison of LT speed before and after the prolonged trials.

2.6. Statistical Analysis

Data were tested for normality using the Shapiro–Wilk test. Two‐way ANOVA with repeated measures was used to assess the effects of running shoe model on physiological responses during the 80‐min running trials, as well as on changes in LT and RE from the graded tests performed in a well‐rested state to the tests performed after the 80‐min trials. Fischer's Least Significance Difference (LSD) was applied for post hoc testing. Results are reported as means or mean differences with either standard deviations (SD) or 95% confidence intervals [CI]. p‐values < 0.05 were considered significant; p‐values between 0.05 and 0.10 were considered trends. All statistical analyses and figures were generated using GraphPad Prism version 6.07 (GraphPad Software LLC, San Diego, CA 92108 USA).

3. Results

3.1. Prolonged Running in Carbon Plated vs. Non‐Carbon Plated Shoes at Matched Speed (95% LT_NCP )

3.1.1. Oxygen Uptake, Running Economy, Heart Rate, and Rate of Perceived Exertion

Unexpectedly, V̇O₂ decreased over the course of both 80‐min trials at 95% LT_NCP (i.e., average speed of 14.7 km·h⁻¹; range: 13.5–15.1 km·h⁻¹), following a comparable pattern between shoe conditions (main time effect: p < 0.0001; shoe × time: p = 0.45) (Figure 2A and Table 1). This reduction was due to higher V̇O₂ values at 18–20 min compared to all later time points (p < 0.0001) (Figure 2A). Notably, V̇O₂ remained consistently lower in the CP‐NCP trial compared to NCP (main shoe effect: p < 0.0001). A similar pattern was observed for RE, which improved across both trials (main time effect: p < 0.0001; shoe × time: p = 0.45)—again driven by initially poorer RE at 18–20 min relative to all later time points (p < 0.0001) (Figure 2B and Table 1). RE was 5 ± 3%, 6 ± 3%, 6 ± 4%, and 4 ± 5% better in the CP‐NCP trial compared to NCP at the four time points, respectively (main shoe effect: p < 0.0001) (Figure 2B).

FIGURE 2.

Effects of 80 min of running at matched external workload (95% LT_NCP) in the NCP‐shoe and the CP‐shoe on (A) V̇O₂, (B) running economy, (C) heart rate and (D) rating of perceived exertion. V̇O₂ and running economy were obtained from indirect calorimetry measured from 18 to 20 min, 38 to 40 min, 58 to 60 min, and 78 to 80 min. Individual data is indicated by filled circles (NCP‐shoe) and open circles (CP‐shoe) and connected by black lines. Bars indicate means at each time‐point. ^#Main time effect, p < 0.05. ^§Main shoe effect, p < 0.05. ^aDifference between shoes, p < 0.05. ^(a)Trend towards difference between shoes, p = 0.05–0.10. ^bDifferent from 18th to 20th min across shoes, p < 0.0001. ^cDifferent from 38th to 40th min across shoes, p < 0.01.

TABLE 1.

Physiological and perceived responses to 80 min of running in carbon‐plated (CP) vs. non‐carbon‐plated (N‐CP) shoes at matched running speed (95% LT_NCP).

	[La−]b (mmol·L−1)		V̇O2 (L·min−1)		Running economy (mL·kg−1·km−1)		Heart rate (bpm)		Perceived exertion, Borg
	Mean diff., CP‐NCP	p	Mean diff., CP‐NCP	p	Mean diff., CP‐NCP	p	Mean diff., CP‐NCP	p	Mean diff., CP‐NCP	p
Shoe × time		0.63		0.45		0.45		0.79		0.40
Time effect		0.0008		< 0.0001		< 0.0001		< 0.0001		< 0.0001
Shoe effect		< 0.0001		< 0.0001		< 0.0001		< 0.0001		< 0.0001
15–20 min	−0.3 [−0.6; 0.0]	0.06	−0.17 [−0.26; −0.07]	0.001	−9.8 [−15.0; −4.7]	0.0004	−6 [−9; −3]	0.0006	−1 [−1; 0]	0.06
35–40 min	−0.4 [−0.7; −0.1]	0.008	−0.20 [−0.30; −0.11]	< 0.0001	−11.6 [−16.7; −6.5]	< 0.0001	−7 [−10; −4]	< 0.0001	−1 [−2; −1]	0.0002
55–60 min	−0.5 [−0.8; −0.2]	0.001	−0.18 [−0.28; −0.09]	0.0003	−11.0 [−16.1; −5.9]	< 0.0001	−8 [−11; −5]	< 0.0001	−1 [−2; 0]	0.003
75–80 min	−0.3 [−0.6; 0.0]	0.06	−0.10 [−0.20; −0.01]	0.03	−7.0 [−12.1; −1.9]	0.02	−7 [−10; −4]	< 0.0001	−1 [−2; −1]	0.0004

Note: Data is presented as mean differences between shoes and [95% CI]. See text for exact times of measurements within the given periods.

Abbreviations: [La⁻]_b, blood lactate concentration; V̇O₂, oxygen uptake.

HR increased progressively during both trials (main time effect: p < 0.0001; shoe × time: p = 0.79) but was 4 ± 1%, 4 ± 2%, 5 ± 4%, and 4 ± 4% lower in the CP‐NCP trial at the four time points (main shoe effect: p < 0.0001) (Figure 2C and Table 1). Rating of perceived exertion increased similarly across both trials (main time effect: p < 0.0001; shoe × time: p = 0.40), though participants consistently reported lower RPE in the CP‐NCP trial compared to NCP (main shoe effect: p < 0.0001) (Figure 2D and Table 1). Regarding substrate utilization, RER was slightly decreased throughout both trials (CP‐NCP: 0.94 ± 0.03 to 0.91 ± 0.04; NCP: 0.93 ± 0.02 to 0.91 ± 0.03, main effect of time: p < 0.0001), with no shoe × time interaction (p = 0.92) or main effect of shoe (p = 0.36).

3.1.2. Blood Lactate Concentrations

During the two 80‐min trials at 95% LT_NCP, [La⁻]_b increased from 14 to 74 min (NCP 1.7 ± 0.5 to 2.0 ± 0.6 mmol·L⁻¹; CP 1.4 ± 0.2 to 1.8 ± 0.4 mmol·L⁻¹, main time effect p = 0.0008), with no difference between shoes (shoe × time, p = 0.63) (Figure 3 and Table 1). Across the different time points, [La⁻]_b was 0.3–0.5 mmol·L⁻¹ lower in the CP‐NCP trial compared to the NCP trial (main shoe effect: p < 0.0001) (Table 1).

FIGURE 3.

Effects of 80 min of running at matched external workload (95% LT_NCP) on [La⁻]_b in the NCP‐shoe and the CP‐shoe. [La⁻]_b was measured after 14, 34, 54, and 74 min. Individual data is indicated by filled circles (NCP‐shoe) and open circles (CP‐shoe) and connected by black lines. Bars indicate mean [La⁻]_b at each time‐point. ^#Main time effect, p < 0.05. ^§Main shoe effect, p < 0.05. ^aDifference between shoes, p < 0.05. ^(a)Trend towards a difference between shoes, p = 0.05–0.10 (see Table 1 for exact p‐values). ^bDifferent from 14th, 34th, and 54th min across shoes, p < 0.01.

3.2. Prolonged Running in Carbon‐Plated vs. Non‐Carbon‐Plated Shoes at Matched Internal Workload

To compare physiological responses between shoes under matched internal workload, a third trial was conducted using the CP‐shoe, with running speed set at 95% LT_CP. In this comparison, the average running speeds were 14.7 km·h⁻¹ in the NCP‐shoe trial and 15.2 km·h⁻¹ (range: 14.5–15.7 km·h⁻¹) in the CP‐shoe trial.

3.2.1. Oxygen Uptake, Running Economy, Heart Rate, and Rate of Perceived Exertion

V̇O₂ decreased similarly during both 80‐min trials (main time effect: p < 0.0001), driven by the initial higher values after 18–20 min compared to all later time points (p < 0.001) (Figure 4A and Table 2). Despite the ~0.5 km·h⁻¹ difference in running speed, no overall difference in V̇O₂ was detected between trials (main shoe effect: p = 0.44) (Table 2). Also, changes in RE showed a comparable improvement across both trials (main time effect: p = 0.0001; shoe × time: p = 0.84), with RE better overall in the CP‐shoe trial (+4 ± 3%, +3 ± 5%, +4 ± 7%, and +2 ± 6%; main shoe effect: p = 0.002) (Figure 4B and Table 2).

FIGURE 4.

Effects of 80 min of running at matched internal workload (95% LT specific to each shoe) in the NCP‐shoe and the CP‐shoe on (A) V̇O₂, (B) running economy, (C) heart rate, and (D) rating of perceived exertion (Borg scale). V̇O₂ and running economy were obtained from indirect calorimetry measured after 18–20, 38–40, 58–60, and 78–80 min. Individual data is indicated by filled circles (NCP‐shoe) and open circles (CP‐shoe) and connected by black lines. Bars indicate means at each time‐point. ^#Main time effect, p < 0.05. ^§Main shoe effect, p < 0.05. ^aDifference between shoes, p < 0.05. ^(a)Trend towards a difference between shoes, p = 0.05–0.10 (see Table 2 for exact p‐values). ^bDifferent from 18th to 20th min across shoes, p < 0.01. ^cDifferent from 38th to 40th min across shoes, p < 0.05. ^dDifferent from 58th to 60th min across shoes, p < 0.05.

TABLE 2.

Physiological and perceived responses to 80 min of running in carbon‐plated (CP) vs. non‐carbon‐plated (N‐CP) shoes at matched internal workload (95% LT in each shoe).

	[La−]b (mmol·L−1)		V̇O2 (L·min−1)		Running economy (mL·kg−1·km−1)		Heart rate (bpm)		Perceived exertion (Borg)
	Mean diff. (CP‐NCP)	p	Mean diff. (CP‐NCP)	p	Mean diff. (CP‐NCP)	p	Mean diff. (CP‐NCP)	p	Mean diff. (CP‐NCP)	p
Shoe × time		0.13		0.82		0.84		0.85		0.57
Time effect		0.48		< 0.0001		0.0001		< 0.0001		< 0.0001
Shoe effect		0.38		0.44		0.002		0.01		0.27
15–20 min	0.1 [−0.3; 0.5]	0.59	0.00 [−0.15; 0.13]	0.90	−7.1 [−14.6; 0.4]	0.06	−2 [−5; 1]	0.18	0 [−1; 1]	0.54
35–40 min	0.0 [−0.4; 0.4]	0.98	0.01 [−0.13; 0.15]	0.86	−5.2 [−12.7; 2.3]	0.17	−2 [−5; 2]	0.32	0 [−1; 1]	0.54
55–60 min	−0.2 [−0.5; 0.3]	0.37	0.01 [−0.13; 0.15]	0.82	−8.1 [−15.6; −0.6]	0.03	−3 [−6; 0]	0.05	−1 [−2; 0]	0.23
75–80 min	−0.3 [−0.7; 0.1]	0.16	0.07 [−0.06; 0.22]	0.26	−3.8 [−11.3; 3.7]	0.31	−1 [−4; 0]	0.39	−1 [−1; 0]	0.27

Note: Data is presented as mean differences between shoes and [95% CI]. See text for exact times of measurements within the given periods.

Abbreviations: [La⁻]_b, blood lactate concentration; V̇O₂, oxygen uptake.

HR increased similarly across both trials (main time effect: p < 0.0001; shoe × time: 0.85) but was lower overall in the CP‐shoe trial (main shoe effect: p = 0.01) (Figure 4C and Table 2). Also, the rating of perceived exertion increased similarly during both 80‐min trials (main time effect: p < 0.0001; shoe × time: p = 0.57), but with no differences in Borg ratings between shoes (main shoe effect: p = 0.27) (Figure 4D and Table 2). Again, RER decreased in a comparable manner throughout both trials (NCP: 0.93 ± 0.02 to 0.91 ± 0.03; CP‐CP: 0.94 ± 0.03 to 0.91 ± 0.02, main effect of time: p < 0.0001) with no shoe × time interaction (p = 0.63) or main effect of shoe (p = 0.13).

3.2.2. Blood Lactate Concentration

During running at matched internal workloads, [La⁻]_b remained stable throughout the trial in both conditions (main time effect: p = 0.48; shoe × time: 0.13), with no main effect of shoe type (p = 0.37) (Figure 5 and Table 2).

FIGURE 5.

Effects of 80 min of running at matched internal (95% LT specific to each shoe) on [La⁻]_b in the NCP‐shoe and the CP‐shoe. [La⁻]_b was measured after 14, 34, 54 and 74 min. Individual data is indicated by filled circles (NCP‐shoe) and open circles (CP‐shoe) and connected by black lines. Bars indicate mean [La⁻]_b at each time‐point.

3.3. Changes in LT and RE After Prolonged Running at Matched Speed (95% LT_NCP )

Changes in LT after running at 95% LT_NCP were not different between shoes (shoe × time, p = 0.54) (Figure 6). Overall, LT‐speed increased (NCP: +0.4 km·h⁻¹ [0.1; 0.8]; CP‐NCP: +0.6 km·h⁻¹ [0.3; 0.9], main time effect: p < 0.0001). Accordingly, [La⁻]_b was lower at 15 km·h⁻¹ and 16 km·h⁻¹ following these trials compared with the well‐rested conditions (15 km·h⁻¹: NCP: −0.4 mmol·L⁻¹ [−0.8; −0.1], CP: −0.2 mmol·L⁻¹ [−0.5; 0.2], main time effect: p = 0.01; 16 km·h⁻¹: NCP: −0.6 mmol·L⁻¹ [−1.0; −0.2], CP: −0.5 mmol·L⁻¹ [−0.9; −0.0], main time effect: p = 0.001). Moreover, LT was consistently attained at higher speeds in the CP‐shoe compared with the NCP‐shoe (main shoe effect: Pre: +0.5 km·h⁻¹ [0.2; 0.7]; Post: +0.6 km·h⁻¹ [0.4; 0.9], p < 0.0001). These exercise‐induced improvements were accompanied by improvements in RE from the well‐rested state to postexercise (i.e., after the 80‐min trials) (NCP: 14 km·h⁻¹: 194 ± 11 mL·kg⁻¹·km⁻¹ to 184 ± 21 mL·kg⁻¹·km⁻¹; CP: 185 ± 11 mL·kg⁻¹·km⁻¹ to 174 ± 21 mL·kg⁻¹·km⁻¹; main time effect: p = 0.04; main shoe effect: p = 0.002).

FIGURE 6.

Individual running speeds at LT in the NCP‐shoe (left) and the CP‐shoe (right) in the well‐rested condition (filled black circles), after running for 80 min at 95% LT_NCP (open circles), and after running for 80 min 95% LT_CP (filled gray circles—only conducted in the CP‐shoe). Bars indicate mean LT‐speed in each condition. ^#Main time effect, p < 0.05. ^§Main shoe effect, p < 0.05. ^aDifferent from well‐rested and fresh condition in the same shoe, p < 0.01. ^bDifferent from NCP condition at same time point, p < 0.001.

3.4. Changes in LT and RE After Prolonged Running at Matched Internal Workload

At matched internal workload (i.e., at 95% of LT specific to each shoe), changes in LT‐speed after 80 min of running did not differ between shoes (shoe × time: p = 0.69) (Figure 6). Overall, however, LT‐speed increased over time (main time effect: NCP: 0.4 km·h⁻¹ [0.2; 0.7]; CP‐CP: 0.5 km·h⁻¹ [0.3; 0.8], p < 0.0001). Accordingly, [La⁻]_b was reduced after the 80‐min trials at 15 km·h⁻¹ and 16 km·h⁻¹ compared to the fresh condition (main time effects: 15 km·h⁻¹: NCP: −0.4 mmol·L⁻¹ [−0.8; −0.1], CP: −0.3 mmol·L⁻¹ [−0.7; −0.0], p = 0.003; 16 km·h⁻¹: NCP: −0.6 mmol·L⁻¹ [−1.0; −0.2], CP: −0.5 mmol·L⁻¹ [−0.9; −0.1], p = 0.0006). Additionally, a clear main effect of shoe was observed, with LT‐speeds consistently higher in the CP‐shoe condition (Pre: +0.5 km·h⁻¹ [0.2; 0.7]; Post: +0.5 km·h⁻¹ [0.3; 0.8], p < 0.0001). The improvements in LT‐speed were accompanied by enhanced RE after the 80‐min trials (14 km·h⁻¹: NCP: 193 ± 13 mL·kg⁻¹·km⁻¹ to 184 ± 20 mL·kg⁻¹·km⁻¹; CP: 185 ± 15 mL·kg⁻¹·km⁻¹ to 174 ± 21 mL·kg⁻¹·km⁻¹; main time effect: p = 0.03; main shoe effect: p = 0.03).

4. Discussion

We wish to highlight three key findings of the present study. (1) At fixed running speed, the CP‐shoe featuring highly responsive foam consistently outperformed the NCP‐shoe across all measured physiological and perceptual measures. Specifically, running in the CP‐shoe was associated with improved RE, lower V̇O₂, reduced [La⁻]_b, and lower ratings of perceived exertion. These benefits were evident both in a well‐rested state and throughout the 80‐min runs. Even when trials were matched for internal workload (i.e., at 95% LT specific to each shoe), the CP‐shoe maintained superior RE and a lower heart rate, despite faster running speeds. (2) Contrary to our hypothesis and prior findings with conventional running shoes [9, 11], no worsening in RE or increase in [La⁻]_b occurred over time in either shoe. As a result, the magnitude of running‐induced changes in these outcomes did not differ between shoes. (3) Surprisingly, both RE and LT‐speed improved after the 80‐min runs in both shoe conditions. Notably, running in the CP‐shoe consistently elicited higher LT‐speeds by ~0.6 km·h⁻¹ compared to the NCP‐shoe.

4.1. Advanced Footwear Technologies Enhance Running Speed via Improved Running Economy and Lactate Threshold

Previous studies involving trained runners have reported improvements in RE of ~3%–4% when using the CP‐shoe model featured in the current study, compared to NCP‐shoes and other carbon plated shoe models [16, 25]. These findings align closely with the present results, which demonstrated a 4.5% improvement in RE (range: 1.6%–8.0%) at 14 km·h⁻¹ during the graded test conducted in the well‐rested state. Also, 2%–6% improvements in RE were observed during the 80‐min trials by running in the CP‐shoe compared to the NCP‐shoe, with lower improvements observed when trials were matched for internal workload (Figures 3 and 5). This improvement in RE is expected to translate into faster distance running, as supported by recent findings from “real world” performances at the highest level [18]. The general benefits of the CP‐shoes have been largely attributed to their increased longitudinal bending stiffness, provided by the embedded carbon fiber plate, combined with a highly compliant and responsive midsole foam that returns more mechanical energy upon compression during ground contact than the materials used in other shoes [16, 25]. In the present study, the CP‐shoe differed from the NCP‐shoe by incorporating improved foam properties, the integration of a carbon plate, and by offering a lower mass (218 vs. 297 g). As such, we can only speculate on the relative contribution of each of these factors to the enhanced RE. Regarding shoe mass, we chose not to control for this in order to preserve high ecological validity. Previous findings indicate that adding 100 g to a shoe increases the cost of running by ~1% [26] (Frederick et al. 1984), and controlling for shoe mass would therefore likely have reduced the observed difference in RE. Nevertheless, the effect would likely still be of sufficient magnitude to confer meaningful running performance benefits. Finally, it is worth noting that the NCP‐shoe model used in this study was deliberately selected because, according to the manufacturer, this shoe also features responsive foam with higher energy return than conventional running shoes, thereby providing a meaningful benchmark against the top‐tier CP‐shoe [27].

4.2. Impact of Accumulated Work on Physiological Responses During Running

While assessing RE in a well‐rested state provides methodological standardization, growing evidence suggests that key physiological determinants of endurance performance are dynamic and evolve in response to accumulated exercise stress. In this context, it has been hypothesized that CP‐shoes may not only enhance RE in the well‐rested state but also prevent exercise‐induced deteriorations in RE by attenuating muscle damage, soreness, and perceived fatigue [6, 19]. Given the well‐established effect of CP‐shoes on RE, one might argue that any beneficial effect of CP‐shoes on exercise‐induced changes at matched external workloads (i.e., fixed running speed) could stem from reduced physiological and metabolic strain. This would in turn minimize metabolic perturbations, cardiac drift, and biomechanical deteriorations. Consequently, comparisons based solely on matched external workload would complicate the assessment of durability effects of the CP‐shoe per se. To address this, we implemented two CP‐shoe trials for comparison with the NCP‐shoe: one in which running speed (external workload) was matched (i.e., 95% LT_NCP) and another in which the intensity was 95% LT specific to each shoe, representing equal internal workload. Under conditions with matched internal workload, RE remained ~2–4% better with the CP‐shoe throughout the trials, yet the magnitude of changes in [La⁻]_b, RE, heart rate, and rate of perceived exertion were similar between shoe conditions. A comparable pattern was observed in the speed‐matched trials, where the CP‐shoe consistently demonstrated ~4–6% greater RE.

During the final hour of all trials, V̇O₂ and RE remained stable with only small drifts in heart rate (i.e., ~5–6 bpm), probably due to increases in body temperature and a gradual loss of plasma volume through sweating [28]. This contrasts with some previous studies in trained runners wearing conventional running shoes, which have reported small increases in V̇O₂ (~50–80 mL·min^‐1) during 60 min of running at 70% V̇O₂max (equivalent to 111% of the gas exchange threshold) [29] and during 90 min at 65% V̇O₂max [30]. Similarly, recent studies involving trained runners of various performance levels observed ~2–4% impairments in RE following 90 min of running at LT [11] and ~4–6% impairments after 90–120 min slightly above LT [10]. On the other hand, a study in trained runners has reported no change in RE during 60 km of outdoor running at 65%–70% V̇O₂max [13], while Brueckner et al. [14] observed that RE remained stable after 15 km but became worse after 32 and 42 km of running in trained marathon runners. These discrepancies could be attributable to differences in training status among the participants as well as differences in running intensity (i.e., below, at, or above LT), running duration, running shoe model, as well as nutritional strategies. Thus, although previous studies have reported deteriorations in RE at comparable intensities [9, 11], it is likely that the present protocol (i.e., 95% LT) was not sufficiently demanding to induce noticeable fatigue during the 80‐min trials in the cohort of trained runners. Also, substrate availability and utilization are important factors influencing RE. Increased fat oxidation during exercise elevates V̇O₂ at a given running speed, as glucose yields more energy per unit of oxygen consumed [5, 31]. Notably, most previous studies reporting exercise‐induced impairments in RE have been conducted following overnight fasting and/or without carbohydrate intake during running [9, 11, 29]. In contrast, carbohydrate was consumed before and during running in the present study, likely preserving a high contribution from carbohydrate oxidation throughout the trials, thereby supporting a stable RE. This notion was supported by only small reductions in RER values during all trials (i.e., 0.02–0.03), suggesting that substrate utilization only changed to a minor extent. Taken together, future studies should employ longer and/or more intense test protocols, likely at or above LT, to better assess the effect of CP‐shoes on exercise‐induced physiological changes and durability.

4.3. Sub‐Optimal Running Economy at the Onset of Prolonged Running in Well‐Trained Runners

V̇O₂ was consistently high and RE poor during the initial 25–30 min of running (i.e., following a 10‐min warm‐up at 10 km·h⁻¹ and the first 18 min of the trials) compared to later time points in both shoe conditions, a pattern observed in 7–9 of the 10 runners depending on the trial. Thereafter, V̇O₂ declined (i.e., improved RE) and plateaued for the remainder of the trials (Figures 1 and 2). The underlying cause of the diminished RE during the initial periods of the study protocol remains speculative, but since we have not previously observed a similar pattern during prolonged cycling trials in elite cyclists [5], it may be related to the use of running as the exercise mode. In this context, a previous study in trained marathon runners showed that the cost of running decreases across repeated trials as runners become familiar with treadmill running [14]. As most participants in the present study did not regularly train on a treadmill, it cannot be ruled out that this factor could have contributed to the initial inefficiency. Moreover, given the participants' unfamiliarity with the tested footwear (i.e., both the NCP‐ and CP‐shoe), it is plausible that a period of familiarization is required to achieve optimal RE.

The role of warm‐up strategies intending to optimize RE among trained runners remains, to the best of our knowledge, largely unexplored, although the warm‐up practice itself is obviously a widely implemented strategy. Studies utilizing plyometrics and resisted runs during the warm‐up have been shown to improve RE [32, 33], emphasizing that RE can be modulated by different warm‐up approaches. Moreover, during the general preparation phase, when the present study was conducted, training loads are typically high in competitive runners. It is therefore possible that the transition from rest to exercise was hindered, requiring additional warm‐up to optimize physiological processes and consequently RE. Finally, shoe temperature may also have affected shoe properties, and potentially RE, as the foam in the midsole and insole may soften with increased temperature, altering cushioning properties [34, 35]. A previous study has demonstrated that the midsole temperature increased by 8°C within the first 15–20 min of running [34], driven in part by body heat dissipation and repetitive friction. As changes in sole temperature likely vary with foam type, studies are needed to understand how sole temperature evolves during overground running and treadmill running in modern shoes, and how these changes affect foam responsiveness, running biomechanics, and RE.

4.4. Improved LT‐Speed and Running Economy Following Prolonged Running

LT and the first ventilatory threshold (VT1) have been used to demarcate the transition between moderate and heavy exercise intensity domains, and previous studies have reported declines in running speed and power output at LT and VT1 following prolonged running and cycling [7, 8, 9, 36]. For example, Nuuttila et al. [9] observed a ~5–6% reduction in LT‐speed accompanied by reduced RE in recreational runners after 90 min of treadmill running at 90% LT (LT defined as speed corresponding to baseline lactate +0.3 mmol·L⁻¹). In contrast, the present study found ~3% improvements in LT‐speed along with enhanced RE after 80 min of running, irrespective of shoe model and exercise intensity. The reasons for these contrasting findings remain speculative, but the observed reductions in [La⁻]_b at the given running speeds after the prolonged trials and the resulting increase in LT‐speeds suggest that lactate production was either reduced or that lactate oxidation was increased. The capacity to produce lactate has been shown to be gradually reduced during prolonged, energy‐restricted, and glycogen‐depleting exercise [37]. While it cannot be ruled out that the running trials in the present study induced a reduced ability to produce lactate, the primary determinant of this is carbohydrate availability [37]. In this regard, given the ingestion of carbohydrate before and during running, and the relatively small reductions in carbohydrate oxidation and high RER values, carbohydrate availability was likely not critically reduced in the present study.

Rather than reflecting a diminished capacity to produce lactate, the findings more likely indicate a reduced reliance on anaerobic energy contribution, potentially accompanied by increased lactate oxidation during the prolonged running trials. Thus, warm‐up or “priming” prior to the assessments of RE and LT in the well‐rested state may have been suboptimal (i.e., 10 min at 10 km·h⁻¹), leading to lower physiological readiness with respect to muscle temperature, blood flow [38], V̇O₂ kinetics [39], and lactate production and oxidation [39, 40]. Together, such factors may have diminished both RE and the efficiency of lactate shuttling and oxidation processes in the well‐rested condition [41]. Moreover, the improved LT‐speed and RE may be related to the extent of recovery before the posttrial LT assessments. In the present study, the use of a 10‐min active recovery period at 10 km·h⁻¹ prior to LT testing allowed for a complete recovery of [La⁻]_b, and [La⁻]_b at the initial stage of the postexercise progressive test (13 km·h⁻¹) was similar to that observed in the well‐rested condition in all trials (~1.1–1.2 mmol·L⁻¹). In contrast, previous studies in runners and cyclists have employed shorter recovery periods (i.e., 4 min) [7, 8, 9] and it remains uncertain whether these differing procedures influenced recovery of [La⁻]_b. Lastly, and as previously noted, insufficient familiarization with the running shoes, as well as treadmill running while wearing a face mask, may have negatively affected RE measurements and potentially LT during the tests in the well‐rested state [14]. Thus, it cannot be ruled out that the observed improvements in RE, and consequently LT, may be at least partly attributable to a learning effect.

In conclusion, the CP‐shoe with highly responsive foam consistently outperformed the NCP‐shoe across all measured variables, both in the well‐rested state and during prolonged running at matched running speeds. When internal workload was controlled, the CP‐shoe continued to show advantages in RE together with lower heart rate. However, the magnitude of differences between shoe‐conditions remained stable throughout the 80‐min trials. Notably, LT‐speed increased following prolonged running regardless of shoe model.

4.5. Limitations

The relatively small sample size may limit the generalizability of our findings; however, the consistency of responses suggests that a larger sample would not substantially alter the overall interpretations. Participants' blinding would have been preferable but was not feasible given the distinctly different sensation of running in the CP‐shoe. As the CP‐shoe has been reported to enhance performance, participants may have anticipated improved outcomes while wearing it. Similarly, blinding the laboratory technicians conducting the measurements was not practically possible. For the assessment of RE, a single measurement was performed with each shoe model at each time point. Although at least one additional measurement would be ideal to reduce potential measurement error [42], this was not practically or methodologically feasible in the present study. Changes in body mass during the trials were also not accounted for in the RE calculations, as participants were not reweighed every 20 min. Nevertheless, fluid intake was standardized within participants, meaning that changes in body mass after 80 min primarily reflected individual sweat rates. Given the cross‐over design, standardized dietary preparation, and controlled laboratory temperature, body mass changes during running were assumed to be comparable between trials. For illustration, in a 78‐kg runner with a V̇O₂ of 3.5 L·min⁻¹ at 14.7 km·h⁻¹, a 0.5‐kg reduction in body mass would impair RE only marginally from 183 to 184 mL·kg⁻¹·min⁻¹. Therefore, normalizing RE at each time point would not have substantially altered the results or conclusions. Finally, because the study was conducted on a motorized treadmill, the transferability of the findings to other surfaces remains unclear.

5. Perspective

Advanced footwear technology has consistently been shown to improve RE under well‐rested conditions. To our knowledge, this is the first study to examine how a top‐tier CP‐shoe influences physiological responses throughout prolonged running in trained runners. Throughout the 80 min of running at speeds equivalent to 95% LT in the NCP‐shoe, the CP‐shoe consistently demonstrated superior RE, along with lower HR, [La⁻]_b, and V̇O₂. However, the changes in these physiological markers over time did not differ between shoes. When internal workloads were matched (i.e., running at 95% LT specific to each shoe), the CP‐shoe continued to yield better RE and lower HR compared to the NCP‐shoe. Yet again, the changes in physiological responses remained comparable between shoes. Along with previous studies, the present study shows that the effect of advanced footwear technology on RE varies considerably between individuals during non‐fatigued conditions. In this context, as the performance‐enhancing effect may be mediated by spatiotemporal factors such as contact time and flight time in well‐rested conditions [17, 43], it would be relevant for future research to examine potential mediating factors as fatigue develops during prolonged running.

Conflicts of Interest

The authors declare no conflicts of interest.

Madsen L. L., Abel K., Hansen A. A., et al., “Persistent Improvements in Running Economy With Advanced Footwear Technology During Prolonged Running in Trained Male Runners,” Scandinavian Journal of Medicine & Science in Sports 35, no. 10 (2025): e70139, 10.1111/sms.70139.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.