Establishing the contribution of the muscle-tendon interaction to the metabolic cost of running using a blended experimental and computational approach

Although humans have been running bipedally for thousands of years, there is still substantial variability in the way we run, i.e. our preferred running pattern. Moreover, also the metabolic energy we consume to run at a certain, submaximal velocity varies between runners. Despite the large body of running related research, our understanding on the underlying mechanisms explaining the variability in preferred running pattern and/or the whole-body metabolic rate during running is limited. Over the last decades, researchers have tried to describe the mechanics of running and have attributed the whole-body metabolic rate during running to specific biomechanical tasks such as body weight support, forward propulsion, and leg swing. While these biomechanical tasks provide valuable insight in why running requires metabolic energy, it only suggests that producing high forces over a short time period is energy expensive and as such gives little insights into what makes running require such a substantial amount of energy. Therefore, to further enhance our understanding on the whole-body metabolic rate during running we have to consider the tissues consuming the energy. Skeletal muscles are the primary metabolic energy consumers during running and their energy consumption is determined by the muscle force, duration of activation, length, velocity and frequency of activation/deactivation.

Earlier research has already demonstrated that the mechanical efficiency of running exceeds the maximal mechanical efficiency of concentrically contracting skeletal muscle, highlighting that other elastic tissues, such as tendons, enhance the mechanical efficiency of skeletal muscles and therefore aid in reducing the metabolic cost of running. As the ankle joint provides approximately two thirds of the total positive lower limb work during stance, the plantar flexor muscles can be considered among the most important muscles during running. The triceps surae muscle, the major ankle plantar flexor consisting of the gastrocnemius medialis, gastrocnemius lateralis, and soleus muscles, is connected to the calcaneal bone through the long and highly compliant Achilles tendon (i.e. in series connected tendinous tissue). This in series connected tendinous tissue stores and returns mechanical energy and interacts with the triceps surae muscle, thereby decoupling muscle fascicle length changes from length changes of the entire triceps surae muscle-tendon unit. Subsequently, during running the triceps surae can operate at favorable operating conditions, i.e. close to optimal muscle fascicle length and at relatively slow muscle contraction velocities, while ankle motion is, and thus muscle-tendon length changes are, considerable. Although it is clear that the triceps surae muscle-tendon unit is likely to affect the whole-body metabolic rate during running, we still do not know whether a runner’s preferred running pattern allows for optimal triceps surae muscle-tendon interaction or if triceps surae muscle kinematics can help explain a runner’s optimal running pattern.

The overarching aim of this PhD was to study how triceps surae muscle-tendon interactions affect whole-body metabolic rate and preferred movement patterns during running. After assessing the effect of triceps surae contraction velocity on whole-body metabolic rate, we investigated how two different running patterns (e.g. different habitual foot strike patterns and different imposed stride frequencies) affect triceps surae muscle kinematics and force production using a combined approach of experiments and computational simulations.

Our first goal was to quantify the effect of increasing triceps surae muscle fascicle contraction velocity on whole-body metabolic rate and triceps surae muscle activations. Therefore, in study 1 we adopted a new application of an already existing approach to specifically target the triceps surae muscle, allowing for tight control of muscle fascicle contraction velocity while keeping other factors related to a muscle’s metabolic rate constant. Fifteen male participants were positioned prone on a dynamometer and performed cyclic plantar flexion contractions (~1/3s active and ~2/3s passive) for six minutes at three different ankle angular velocities: 1) isometric, i.e. no angular rotation, with the ankle fixed at 10° plantarflexion (PF), 2) isokinetic at a relatively slow angular velocity (30°/s) during which the ankle rotated from 5° PF to 15° PF, and 3) isokinetic at a relatively fast angular velocity (60°/s) during which the ankle rotated from 0° PF tot 20° PF. Target torque (25% of maximal voluntary contraction) and frequency (1 Hz) were held constant across conditions. Conditions were randomized and duplicated in a mirrored design resulting in six trials in total. During each trial, we collected whole-body metabolic rate, triceps surae muscle activity and gastrocnemius medialis muscle fascicle kinematics. As expected, increasing ankle angular velocity increased gastrocnemius medialis muscle fascicle velocity and mechanical work (p < 0.01). Despite similar mean and peak torques and no difference in cycle frequency nor duty factor (time of active force production over cycle time), increasing ankle angular velocity increased mean and peak triceps surae muscle activity and substantially increased net whole-body metabolic rate (p < 0.01). Furthermore, when computing the relative increase in muscle activations with faster shortening velocities, we found a significantly greater increase in gastrocnemius lateralis activation compared to gastrocnemius medialis and soleus activation (p < 0.05). This first study provides strong evidence that increasing triceps surae contraction velocity and positive mechanical work increases triceps surae muscle activation and causes a substantial increase in whole-body metabolic rate. In addition, it demonstrated that the triceps surae activation strategy depends on the mechanical demands of the task. When translating these results to locomotion, it is suggested that a fine-tuned triceps surae muscle-tendon interaction facilitating slow or - ideally - isometric triceps surae muscle contractions and thus performing little muscle mechanical work will aid in reducing the metabolic cost during locomotion.

Next, in study 2 we investigated whether distinct habitual foot strike running patterns - previously demonstrated to alter ankle and knee joint kinetics and kinematics - affect gastrocnemius medialis muscle kinematics and force production. Nineteen experienced runners, of which nine habitual rearfoot strikers and ten habitual mid-/forefoot strikers participated in the study. We collected lower limb joint kinematics, kinetics, muscle activation data and in vivo gastrocnemius medialis ultrasound images while running at 10 and 14 km/h. To obtain triceps surae muscle forces we used ultrasound informed dynamic optimization. During stance, gastrocnemius medialis muscle fascicle shortening was 40 to 45% greater (p = 0.02) and stance averaged shortening velocities were greater (p = 0.01) in rearfoot than in forefoot strikers, with these differences especially prominent during early stance. During a similar period in early stance, gastrocnemius medialis muscle force appeared smaller in rearfoot strike runners compared to mid-/forefoot strike runners. In contrast to speculations in literature, stretch nor recoil of the triceps surae in series connected elastic tissue was different between foot strike patterns. These results suggest that a potential increase in gastrocnemius medialis muscle metabolic rate due to greater muscle contraction velocity in rearfoot strikers may be counteracted by a lower gastrocnemius medialis muscle force reducing the muscle’s metabolic rate.

From the gastrocnemius medialis muscle kinematics and force production, we speculated that metabolic rate in this triceps surae muscle was not different between habitual foot strike patterns. Since it is impossible to directly measure the metabolic rate of individual muscles, we used a model-based approach to estimate the metabolic rate of the gastrocnemius medialis muscle, the other triceps surae muscles and the whole-body metabolic rate and assessed whether triceps surae or whole-body metabolic rate differed between habitual foot strike patterns. From the muscle states (force, length, velocity, activation and excitation) acquired from dynamic optimization, we estimated muscle metabolic rates using four different Hill-type based muscle metabolic energy models. Neither gastrocnemius medialis metabolic rate nor triceps surae metabolic rate (p > 0.35) was different between habitual foot strike patterns regardless of the muscle metabolic energy model used or running speed adopted. Additionally, also estimated whole-body metabolic rate was not different between habitual foot strike patterns (p > 0.14). Overall, this second study demonstrated that, although gastrocnemius medialis muscle kinematics and force demand are different in habitual mid-/forefoot strike runners compared to rearfoot strike runners, triceps surae metabolic rate is similar. Hence, we further support the already existing evidence that different habitual foot strike patterns are metabolically equivalent.

In study 3 we aimed to pinpoint the drivers of the textbook example of self-optimization: preferred stride frequency. Despite the considerable variation in runners’ preferred stride frequency, it is well recognized that an experienced runner’s preferred stride frequency closely matches with their optimal stride frequency, i.e. the frequency associated with lowest whole-body metabolic rate. However, a clear mechanistic explanation of why a certain stride frequency corresponds with minimal metabolic rate is still lacking. Altering one’s stride frequency changes lower limb joint kinematics and kinetics. Yet, each joint is affected differently, potentially with counteracting effects during the stance and leg swing phases. First, we conducted a joint-level analysis on the effects of changing stride frequency specifically focusing on stance and leg swing separately. We collected whole-body metabolic rate, triceps surae muscle activity, and lower limb joint kinematics and kinetics of seventeen experienced runners while they adopted five different stride frequencies (preferred frequency, preferred ± 8% and preferred ± 15%) at 12 km/h. In line with the self-optimization hypothesis, the stride frequency – whole-body metabolic rate relationship followed a U-shaped relationship with the preferred stride frequency corresponding with the lowest metabolic rate (p < 0.01). Decreasing stride frequency caused an increase in average positive ankle and knee joint power during stance (p < 0.01), but decreased average positive hip joint power during leg swing remarkably (p < 0.01). The sharp increase in average positive hip joint power during leg swing with increasing stride frequency, suggests that the metabolic cost associated with leg swing may increase substantially. Consistent with the increased average positive ankle joint power, average soleus activation during stance also increased with decreasing stride frequency (p = 0.01). Furthermore, increasing stride frequency increased duty factor emphasizing that, although stance time shortened per stride, runners adapted their stride kinetics to allow relatively more time on the ground to produce the required forces. Overall, this first part of the study primarily focusing on a joint-level analysis suggests that the optimal stride frequency represents a trade-off between minimizing the cost of stance, which likely increases with decreasing stride frequency, without excessively increasing the cost of leg swing or reducing the available time to produce the necessary forces.

The altered average positive ankle and knee joint power with changing stride frequencies likely induce changes in the underlying triceps surae muscle-tendon unit. Especially the increased average positive ankle and knee joint power at lower than preferred stride frequencies are likely associated with greater triceps surae energy consumption during stance. To investigate the effect of changing stride frequency from the triceps surae muscle’s perspective we collected soleus and gastrocnemius medialis kinematics in twelve runners, a subset of the participants used in the joint level analysis. Remarkably, mean soleus and gastrocnemius medialis muscle fascicle lengths during stance followed an inverted U-shape relationship, with the longest fascicle lengths corresponding to the preferred stride frequency. Additionally, soleus muscle force demand was significantly higher when running at the slowest stride frequency (preferred-15%) compared to the preferred stride frequency. These results suggest that the optimal stride frequency may be (partially) explained by more favorable operating conditions for the triceps surae muscle at the preferred stride frequency. Particularly at stride frequencies lower than the optimal frequency, the shorter muscle fascicle operating lengths and increased soleus force demand suggest an elevated triceps surae muscle metabolic rate.

The blended experimental and computational approach adopted during this PhD project highlighted the importance of the triceps surae muscle for economical endurance running. We demonstrated that running pattern (habitual foot strike pattern and stride frequency) affect the triceps surae muscle kinematics and force production. More specifically, in rearfoot striking we found more and faster gastrocnemius medialis fascicle shortening, but lower forces explaining the equivalent (whole-body) metabolic rate across habitual foot strike patterns. Next, we found that specific biomechanical changes can explain the existence of the optimal stride frequency. Deviating from the preferred – and optimal – stride frequency induced shorter and less force-efficient triceps surae operating lengths. Moreover, at lower than preferred stride frequencies soleus muscle force was increased whereas at higher than preferred stride frequency positive average hip joint power substantially increased, both explaining (some) of the increased whole-body metabolic rate. Lastly, differences in triceps surae muscle metabolic rate – assessed indirectly through its force, length and velocity or estimated by computer simulations – were in line with differences in whole-body metabolic rate, suggesting that triceps surae muscle metabolic rate might be a driver behind selecting certain preferred running pattern.