Endurance athletes compete against each other in a race to finish first. Independently of the sport performed, these athletes aim to reach the finish line as quickly as possible, by producing the highest average velocity over the course of the event. Endurance competitions typically range from ~3 to 230 min within sports such as running, swimming, speed skating, skiing, and cycling (Riegel, 1981), excluding ultra-endurance competitions. Within such performance duration, world records highlight a distinct linear log-log relationship between finishing times and distance covered for many endurance sports (Lietzke, 1954; Riegel, 1981). The slope of these equations quantifies human endurance, the highest average velocity that is sustained over a given time period or the longest time a given velocity is sustained (Lietzke, 1954; Riegel, 1981; Billat et al., 1999). Later work also assessed the relationships between power and exercise duration, which presumably link more directly to the athlete's underlying physiology (Billat et al., 1999). Such power-duration profiles are widely applied in endurance sports (e.g., cycling) to determine endurance performance and the physical/physiological signature of an athlete (Pinot and Grappe, 2011; Sanders and Erp van, 2020). In a conceptual framework, Joyner and Coyle (2008) already described “potential” physiological determinants of endurance performance. In this review, we advance on this framework, now including a meta-analysis highlighting the (effect size of) interrelationships between endurance performance and key physiological determinants of endurance performance at the macroscopic and microscopic level. Moreover, we discuss how these physiological determinants can be improved by training. Note that this mini review focuses mainly on structural and functional muscle (fiber) characteristics rather than other relevant topics for endurance performance, such as pulmonary or circulatory physiology or nutrition.

Already in 1925, A.V. Hill recognized that fatigue was underlying the decline in velocity with increasing race distance (Hill, 1925). He speculated that performance was determined by physiological characteristics related to energetic costs, oxygen demands, and supply, and oxygen debt (or racing economy). Almost a century later, Joyner and Coyle (2008) highlighted the importance of the maximal oxygen uptake (V˙O_2max), lactate threshold, and efficiency/economy in endurance performance. The V˙O_2max and lactate threshold interact to determine how long a given rate of aerobic and anaerobic metabolism can be sustained (i.e., performance V˙O₂) by an athlete, whereas efficiency determines the velocity or power that can be achieved with a given amount of energy consumption (Joyner and Coyle, 2008).

Elite endurance athletes may score exceptionally high on one (or several) of these three determinants. Upper limits for absolute V˙O_2max have been reported in rowers and cross-country skiers (7.0–7.5 L·min⁻¹ in males and 5.0−5.5 L·min⁻¹ in females) and for relative V˙O_2max in cyclists, runners, and cross-country skiers (up to ~90 mL·kg⁻¹·min⁻¹ in males and ~80 mL·kg⁻¹·min⁻¹ in females) (Haugen et al., 2018; van der Zwaard et al., 2018c). The (first) lactate threshold discriminates moderate- from heavy-intensity exercise and can be determined by gas exchange and/or blood lactate measurements (Poole et al., 2020). Typically, values approximate 75–85% of V˙O_2max in athletes (Coyle et al., 1988; Seiler and Kjerland, 2006; Joyner and Coyle, 2008). Knowing that all Olympic endurance events are decided at intensities above 85% of V˙O_2max (Joyner and Coyle, 2008), athletes benefit from being relatively fatigue resistant at high exercise intensities and in this context a high lactate threshold can be very advantageous. As for gross efficiency, values generally range ~20–23% in elite cyclists (Jeukendrup et al., 2003; Ettema and Lorås, 2009) and ~18–19% in elite rowers (Bourdin et al., 2004). In world-class cyclists, Lucía et al. (2002b) reported efficiency values up to ~24.5–28.1%, although possible methodological issues may be at play (Jeukendrup et al., 2003). Even in elite athletes, scoring high on all three determinants remains unusual, as illustrated by the inverse relationship between V˙O_2max and gross efficiency (Lucía et al., 2002a). However, interestingly, a two-time Tour de France champion was able to display both a very high V˙O_2peak (84 mL·kg⁻¹·min⁻¹) and gross efficiency (23.2%), together with a reasonably high lactate threshold (~81% of V˙O_2peak), which is likely contributing to his athletic success (Bell et al., 2017).

Discussing the physiology of champions, Joyner and Coyle (2008) already suggest that several anthropometrical, cardiac, and skeletal muscle characteristics underlie endurance performance and its whole-body determinants. In particular, skeletal muscular properties could explain differences in (whole-body determinants of) endurance performance, and have been almost exclusively derived from invasive muscle biopsies (Bergstrom, 1975). Although not commonly practiced with elite athletes, biopsies provide valuable and reproducible data on human muscle physiology using only a few 100 muscle cells (Ekblom, 2017), such as on muscle fiber type, cross-sectional area, mitochondrial volume density or function and capillarization, and can be safely obtained (Tarnopolsky et al., 2011). Yet, the interrelationships and corresponding effect sizes between these physiological determinants and endurance performance remain largely overlooked.

We performed a meta-analysis (described in detail in Supplementary Material 1) to examine how skeletal muscle characteristics may contribute to endurance performance and whole-body determinants of endurance performance, of which the results are summarized in Figure 2. This figure displays reported relationships and their effect size from previous studies. Because no single determinant is able to explain all variance in endurance performance, statistical models have been established to include interactions between multiple macroscopic and microscopic determinants (whole-body and/or skeletal muscle, respectively). In line with the schematic of Joyner and Coyle (2008), the combination of performance V˙O₂ and gross efficiency could explain >90% of the variance in 1-h cycling in trained men (Hopker et al., 2013) and 15-km time trial in elite cyclists (data from van der Zwaard et al., 2018b). Subsequently, performance V˙O₂ was largely explained by a combination of the first lactate threshold and V˙O_2max in these cyclists (86% explained variance; data from van der Zwaard et al., 2018b). Also, the combination of cycling efficiency, V˙O_2max and the lactate threshold could explain >75% of the variance in 25-km cycling performance in trained men (Batterson et al., 2020). In professional cyclists, long-distance time-trial performance was highly correlated to the first lactate threshold (Lucía et al., 2004), and in competitive cyclists, a combination of the first lactate threshold with capillary density explained 92% of the differences in time-to-fatigue (Coyle et al., 1988), stressing the importance of oxidative capacity and oxygen delivery in regulating oxygen flux necessary to sustain performance. This narrative is supported by a study in highly-trained cyclists, demonstrating that 78% of the variance in 26-km time-trial could be explained by oxidative capacity of skeletal muscle, submaximal lactate concentration and ability to extract oxygen in skeletal muscle (Jacobs et al., 2011). Interestingly, recent observations in elite cyclists showed that 67% of the variance in performance V˙O₂ was explained by skeletal muscle determinants, which are a high muscle fiber oxidative capacity, high oxygen supply capacity ([Mb] × capillary-to-fiber ratio) and a small physiological cross-sectional area of the M. vastus lateralis (van der Zwaard et al., 2018b). While confirming the importance of oxygen supply and oxidative capacity, these results also reveal that muscle architecture [e.g., the physiological cross-sectional area (PCSA), the cross-sectional area of a muscle perpendicular to its muscle fibers], which can be obtained using non-invasive techniques, such as 3D ultrasound imaging (Weide et al., 2017), may add novel and valuable insights into endurance performance. At present, however, these whole-muscle characteristics have been largely overlooked as determinants of endurance performance. The finding that PCSA negatively contributed to performance V˙O₂ is in line with previous observations showing that a large FCSA may negatively affect endurance performance (Coyle et al., 1988; Bishop et al., 2000) and the lactate threshold (Bishop et al., 2000), likely because this increases the diffusion distance of oxygen to the mitochondria. Importantly, literature suggests that larger muscle FCSA or PCSA may (partly) be accommodated if the oxygen supply toward the muscle—capillarization, and myoglobin—is well-developed (van der Zwaard et al., 2018a,b). Here, non-invasive measurements with near-infrared spectroscopy may also provide valuable insights into spatial and temporal matching of oxygen supply and demand during exercise, and could explain a large proportion of the differences in endurance performance (Jacobs et al., 2011; van der Zwaard et al., 2016b; Batterson et al., 2020). In summary, integrative assessment of skeletal muscle determinants of endurance performance points toward optimisation of efficient type-I fibers with a high mitochondrial oxidative capacity and strongly encourages well-adjusted capillarization and myoglobin concentrations to accommodate the required oxygen flux during endurance performance, especially in a large muscle FCSA or PCSA with attenuated diffusion of oxygen to the mitochondria.

Endurance training shifts velocity/power-duration relationships to the right, allowing athletes to perform a given distance faster or to sustain a given velocity longer (Jones and Carter, 2000). As such, endurance training plays a central role within exercise physiology (van der Zwaard et al., 2020), with training methods facilitating adaptations in macroscopic and microscopic physiological determinants in response to training intensity, duration, frequency, and periodization (Seiler, 2010; Hawley et al., 2018). For example, polarized training consisting predominantly of low-intensity, long-duration training along with some high-intensity training, may be optimal to enhance (whole-body determinants of) endurance performance (Stöggl and Sperlich, 2014; Tønnessen et al., 2014), through higher adaptive cellular signaling, gene expression, and stress responses (Seiler, 2010). Linking training behavior to molecular adaptations, Seiler (2010) previously suggested that mitochondrial biosynthesis might be driven through different pathways depending on exercise duration and intensity. These pathways include, but are not limited to, sustained contractile activity during long exercise bouts that induces chronic release of calcium, low energy status (low AMP:ATP ratio) with high-intensity exercise that increases AMPK and PGC-1α phosphorylation, and training-induced oxidative stress and altered redox states that affect PGC-1α, which trigger mitochondrial adaptations (Hood, 2001; Egan and Zierath, 2013). However, whether training intensity or volume is the most important to increase mitochondrial oxidative capacity remains highly debated (Bishop et al., 2019; MacInnis et al., 2019). Recent evidence suggests that too much high-intensity training may also adversely affect mitochondrial adaptations (Granata et al., 2020; Cardinale et al., 2021; Flockhart et al., 2021). Polarized training may therefore have an advantageous mix of low- and high-intensity exercise for maximizing mitochondrial adaptations.

To benefit from this enhanced oxidative capacity, oxygen supply needs to match oxygen demand (Hepple, 2000), which requires adaptations in capillary density and/or Mb content. High-intensity or (repeated-)sprint exercise may induce local tissue hypoxia (with reduced PO₂) leading to stabilization of the transcription factor HIF-1α, which increases transcription of genes for capillary growth (Lundby et al., 2009; Egan and Zierath, 2013). Although controversies may exist (e.g., Millet and Brocherie, 2020; Siebenmann and Dempsey, 2020), when performed in hypoxia, high-intensity training also seems effective in increasing transcription of VEGF and Mb (Vogt et al., 2001; Kanatous et al., 2009; Faiss et al., 2013; Brocherie et al., 2018). Importantly, the prolonged low-intensity exercise in polarized training could stimulate muscle protein breakdown via enhanced expression of E3 ligases (Stefanetti et al., 2015), resulting in small muscle fibers that facilitate oxygen diffusion to the mitochondria. Growing evidence supports favorable adaptations (i.e., shift in muscle fiber phenotype (from less efficient myosin heavy chain IIX to more oxidative type myosins I and IIA), improvements in motor unit recruitment and firing frequency, greater muscle activation and intra- and inter-muscular coordination, greater muscle–tendon unit stiffness and biomechanical efficiency, changes in substrate utilization, metabolites accumulation, and important signaling cascades related to PGC-1α) following sprint interval training (Sloth et al., 2013), repeated-sprint training (Bishop et al., 2011; Girard et al., 2011; Taylor et al., 2015) or even resistance training (Blagrove et al., 2018) that are more and more implemented among (elite) endurance athletes. Note that training interventions require optimal periodisation, taking into account time windows of adaptation and decay of the underlying skeletal muscle determinants (Hickson and Rosenkoetter, 1981). Considering periodisation, dose-response relationships, and individual trainability, discovering what type of training is the most effective remains an unsolved question that nourish the endless debate among practitioners and researchers.

Determinants of endurance performance have previously been established in the conceptual framework by Joyner and Coyle (2008). However, less is known on how the differences in endurance performance are explained by skeletal muscle characteristics and whole-body determinants. This review includes a meta-analysis that highlights the interrelationships and corresponding effect sizes between endurance performance, V˙O_2max, lactate threshold and efficiency/economy as well as skeletal muscle fiber type, fiber size, capillarization, myoglobin, mitochondrial oxidative capacity, and whole-muscle morphology. Integrative assessment of these determinants is encouraged, stressing the importance of capillarization and myoglobin, and training strategies to improve the physiological determinants have been mentioned, taking into account the different time windows of their adaptation and decay. In summary, optimisation of endurance performance requires careful design of training interventions that finetune modulation of exercise intensity, frequency and duration, and particularly periodisation with respect to the skeletal muscle determinants.

SZ, FB, and RJ conceived, designed the work, drafted, and revised the manuscript. SZ acquired, analyzed, and interpreted the data (from the meta-analysis). All authors read and approved the final version of the manuscript for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.