Is the healthy respiratory system built just right, overbuilt, or underbuilt to meet the demands imposed by exercise?

Abstract

In the healthy, untrained young adult, a case is made for a respiratory system (airways, pulmonary vasculature, lung parenchyma, respiratory muscles, and neural ventilatory control system) that is near ideally designed to ensure a highly efficient, homeostatic response to exercise of varying intensities and durations. Our aim was then to consider circumstances in which the intra/extrathoracic airways, pulmonary vasculature, respiratory muscles, and/or blood-gas distribution are underbuilt or inadequately regulated relative to the demands imposed by the cardiovascular system. In these instances, the respiratory system presents a significant limitation to O₂ transport and contributes to the occurrence of locomotor muscle fatigue, inhibition of central locomotor output, and exercise performance. Most prominent in these examples of an “underbuilt” respiratory system are highly trained endurance athletes, with additional influences of sex, aging, hypoxic environments, and the highly inbred equine. We summarize by evaluating the relative influences of these respiratory system limitations on exercise performance and their impact on pathophysiology and provide recommendations for future investigation.

INTRODUCTION

Four to five decades ago, when the study of exercise physiology was in its infancy, debate over the “limiting factors” to maximum oxygen transport and exercise performance centered solely on cardiovascular responses and muscle metabolism. The respiratory system, including the airways, lung parenchyma, pulmonary vasculature, and respiratory muscles, was viewed as overbuilt and unmalleable in response to endurance training. In the ensuing decades, we have witnessed a gradual accumulation of evidence that has enriched our knowledge of the remarkable precision and capacity of the healthy respiratory system’s response to the substantial and varied demands of exercise. Additionally, research has also revealed several circumstances in which one or more links in this system was shown to be imperfect, anatomically underbuilt, and/or incurred excessive biological costs. Indeed, select components of the respiratory system structure and/or function have even been shown to be negatively impacted via intense exercise training. Our synthesis aims to provide a comprehensive examination of evidence obtained across this broad spectrum of successes and failures experienced by the healthy respiratory system.

KEY FEATURES OF THE JUST RIGHT/OVERBUILT RESPIRATORY SYSTEM RESPONSE TO EXERCISE IN THE UNTRAINED HEALTHY YOUNG ADULT MALE

Ventilation

A key response is the near isocapnic hyperpnea during mild to moderate exercise intensities (see Fig. 1). Remarkably, this proportional match of increasing alveolar ventilation to increasing CO₂ production occurs even in the face of a falling dead space to tidal volume ratio (Vd/Vt) from rest through increasing exercise intensity. In the late Brian Whipp’s words, “… the system seems to ‘know’ that when Vd/Vt is reduced, V̇e ‘needs’ to increase less per unit V̇co₂ to affect its regulatory function,” i.e., to maintain the ratio of alveolar ventilation to carbon dioxide production (V̇a/V̇co₂) and arterial partial pressure of carbon dioxide ($\text{P}{\text{a}}_{\text{C}{\text{O}}_2}$) near resting values (212). This remarkable tracking of V̇a to V̇co₂ is also evidenced with healthy aging, during which the ratio of alveolar ventilation to perfusion (V̇a/Q̇)·nonuniformity and Vd/Vt increase in an aging lung with loss of elastic recoil and airway narrowing, yet the ratio of minute ventilation to carbon dioxide production (V̇e/V̇co₂) increases during exercise so the $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ remains precisely regulated (92). Thus, the correlative data linking V̇a to V̇co₂ are undeniable, and some ingenious experiments in animals at rest have used manipulation of “CO₂ flow” to the lung to demonstrate a causative proportional response of V̇a to V̇co₂ in the absence of muscle contraction, at least over a limited range of ΔV̇co₂ (152). However, the underlying mechanisms through which V̇co₂ acts as its own controller remain a mystery, i.e., without identifiable stimuli or receptor sites. Theoretically, an argument has been made for an “internal stored memory model” adapted by the respiratory controller over time (perhaps through trial and error) to somehow track metabolic CO₂ production and/or its exchange at the lung, resulting in an energetically cost-effective elimination of CO₂ in the steady state of exercise (158). In addition to this underlying V̇co₂-V̇a link, feedback from group III-IV muscle afferents and feedforward from central command to locomotor muscles are also likely to contribute significantly to both exercise hyperpnea and to the hyperventilation of heavy intensity exercise (53).

Fig. 1.

Depiction of typical ventilatory and arterial blood gas responses to steady-state, incremental exercise in untrained young adult healthy subjects [maximal oxygen consumption (V̇o_2max) ~35–50 mL·kg⁻¹·min⁻¹; age 20–35 yr). Note: a) the incremental rise in tidal volume (Vt) and then breathing frequency (f_B) with inspiratory time remaining slightly < expiratory time and the reduction in end-expiratory lung volume; b) the fall in the dead space to tidal volume ratio (Vd/Vt) and ratio of minute ventilation to carbon dioxide production (V̇e/V̇co₂), which maintain the ratio of alveolar ventilation to carbon dioxide production (V̇a/V̇co₂) and partial pressure of carbon dioxide ($\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$) near resting values throughout moderate intensity exercise; and c) the hyperventilatory response to heavy and maximum exercise intensities commensurate with the onset of metabolic acidosis.

Breathing Mechanics

The mechanical efficiency of the “just right” ventilatory response to exercise ensures little or no symptoms of a conscious “awareness” of hyperpnea during exercise. In Fig. 1 note the tradeoff between increasing breathing frequency and tidal volume and the reduction in end-expiratory lung volume with increasing exercise intensity. These highly reproducible responses ensure minimal increases in absolute dead space ventilation because breathing frequency is constrained. Furthermore, the elastic work of breathing is minimized by constraining the increase in Vt to the linear portion of the pressure-volume relationship. This is achieved by both limiting the increase in Vt and reducing end-expiratory lung volume below functional residual capacity via active expiration (70). One potential mechanism, as yet untested in the human for the Vt/breathing frequency (fb) tradeoff with increasing V̇e, is the vagally mediated inhibitory feedback from lung stretch as Vt rises (86). That the resistive work of breathing is also minimized with even maximum exercise intensities is ensured by the structure and precise neural control of both the intra- and extrathoracic airway calibers.

Although the upper extrathoracic airway from nares to glottis may potentially present a major site of resistance to flow during exercise, this circumstance is avoided at least over a range of five- to sevenfold increases in flows experienced in the untrained subject by the precise distribution of efferent respiratory motor output over cranial nerves, which activate pharyngeal and laryngeal airway dilator muscles just milliseconds before phrenic nerve activation. This precise coordination of upper airway versus pump muscle activation with each inspiration ensures that the upper airway is stiffened and widened, thereby resisting progressively rising collapsing transmural pressures generated by the chest wall pump (48). The smaller intrathoracic airways <2 mm diameter and under smooth muscle control also undergo maximum bronchodilation during exercise as regulated by feedback from group III-IV muscle afferents, lung stretch, and increasing circulating catecholamines (96).

Respiratory Muscles

The diaphragm and accessory inspiratory and expiratory respiratory muscles, in contrast to limb locomotor muscles of similar mixed fast-twitch oxidative and glycolytic and slow-twitch oxidative fiber type, also appear to be ideally suited in many ways for the increased ventilatory demands of even heavy intensity endurance exercise: 1) the mitochondrial volume, capillary density, and aerobic capacity of the diaphragm substantially exceed that of the limbs (155); 2) use of supramaximal motor nerve stimulation showed that diaphragm or expiratory muscle fatigue did not occur as a result of either incremental exercise to maximum or endurance exercise to exhaustion at <80% maximal oxygen consumption (V̇o_2max) (91, 176); 3) in vitro evidence has shown rodent diaphragm arterioles to be relatively resistant to norepinephrine-induced vasoconstriction as compared with feed vessels from predominantly red or white limb locomotor muscles (1), and these data imply a relatively protective effect against sympathetically mediated vasoconstriction during whole body exercise, i.e., enhanced sympatholysis, in the diaphragm; and 4) ventilatory requirements of exercise are met by the combined recruitment not only of the diaphragm and of abdominal expiratory muscles but also up to nearly 50–60 “accessory” respiratory muscles of the trunk, neck, and upper torso during heavy exercise (4, 154), i.e., the “load” is widely distributed. In addition to these ideal characteristics and recruitment of respiratory muscles during exercise, the resistive and elastic loads placed on these muscles are minimized by the aforementioned regulation of airway caliber as well as end-expiratory and tidal volumes. Furthermore, the reduced end-expiratory lung volume and increased intra-abdominal pressures ensure optimal lengthening of the inspiratory muscles in meeting rising demands for greater force production (94). Consequently, in untrained subjects exercising at V̇o_2max and generating 100–110 L/min V̇e, respiratory muscle oxygen consumption and blood flow averaged 8–10% of V̇o_2max and cardiac output (2, 67). Furthermore, inspiratory muscles are required to generate only 40–50% of their dynamic capacity, and the maximum flow-volume envelope readily handles the required volumes and flow rates without generating intrathoracic pressures during expiration in excess of those producing “effective” flows (see Fig. 2, values for untrained subjects).

Fig. 2.

Mean values for tidal flow-volume and pressure-volume loops at rest and with incremental treadmill running in untrained and trained subjects [aged 25 ± 1 yr, maximal oxygen consumption (V̇o_2max) 40 ± 2 (untrained) and 73 ± 1 mL·kg⁻¹·min⁻¹ (trained)]. Both panels are plotted relative to group mean total lung capacity and residual lung volumes. The average maximum effective pleural pressure on expiration (Pmaxe) indicates the pressures at which further increases in expiratory pressure elicit no further increments in flow rate. The maximum dynamic inspiratory pressures available (Pcapi) are also shown across all lung volumes and flow rates achieved during exercise. These values for dynamic inspiratory capacity were determined before exercise via body plethysmography (94). The width of Pmaxe and Pcapi at a given lung volume indicate 95% confidence interval. Note that: a) untrained subjects throughout exercise showed no or little intersection of tidal with maximum flow-volume loops, expiratory pressures remained below Pmaxe, and peak tidal inspiratory pressures remained <50–60% of maximum dynamic pressure; and b) all trained subjects reached Pmaxe at 90–100% of V̇o_2max and 85–100% of Pcapi at V̇o_2max, showed dynamic hyperinflation at the highest exercise intensities and at V̇o_2max, and showed no further increase in minute ventilation (V̇e) with added fraction of inspired carbon dioxide or reduced fraction of inspired oxygen [Adapted from Johnson et al. (94)].

Seven decades ago, Arthur Otis, Wallace Fenn, and Hermann Rahn provided underlying principles governing optimal responses to ensure minimization of the elastic, viscous, and turbulent elements of the work of breathing with changing ventilatory requirements (149). These pioneers of respiratory science surely must have had exercise hyperpnea in mind as the ultimate example of the “just right” response.

Pulmonary Vasculature

The structure of the pulmonary (vs. systemic) vasculature seems to be ideally suited for its role in accommodating the entire cardiac output because it is thin-walled, highly compliant and distensible, and relatively protected from the vasoconstrictive effects of high sympathetic nerve activity (139, 167). At rest, the mean Ppa ~15 mmHg is sufficient to maintain flow through to the left atrium at a modest perfusion pressure of ~7 mmHg (vs. 90 mmHg in the systemic circulation). At 15–20 L/min cardiac output, pulmonary vascular distensibility in vivo equals that determined in vitro in isolated vessels, resistance decreases ~50–60% below rest, and pulmonary arterial pressure increases to 20–25 mmHg, probably due exclusively to increasing left atrial pressure that is transmitted upstream across the pulmonary vasculature (166, 167). This high distensibility of the pulmonary vasculature also allows vessel recruitment to provide a maximum pulmonary capillary blood volume of about three times resting levels at maximum exercise. The expanded capillary volume ensures a sufficiently long average red cell transit time in the pulmonary capillaries, i.e., from ~1.0 s at rest to 0.4–0.6 s at 15–20 L/min cardiac output (CO). Thus, an alveolar to capillary O₂ equilibrium is readily achieved despite a falling mixed venous Po₂ and rising cardiac output (mean red blood cell transit time = pulmonary capillary volume/blood flow). Furthermore, because distensibility is independent of vessel diameter and therefore similar across lung regions, this ensures a fairly homogeneous distribution of blood flow.

Gas Exchange

There are some inefficiencies in gas exchange with exercise, as evidenced by two- to threefold increases in the alveolar to arterial Po₂ difference above resting values at V̇o_2max in the 3–3.5 L/min range. Most of this inefficiency appears to be due to increases in nonuniformity in V̇a/Q̇ distribution within isogravitational lung regions from rest to exercise (see Fig. 3) (59, 204). Despite this increased nonuniformity, during exercise overall V̇a increases out of proportion to cardiac output, thereby ensuring high overall V̇a/Q̇ and avoiding areas of low V̇a/Q̇ during exercise. The hyperventilatory response to heavy exercise raises alveolar Po₂ sufficiently to offset the widened alveolar-arterial partial pressure difference (A-aDO₂) and prevent arterial hypoxemia. Thus, arterial oxyhemoglobin (HbO₂) saturation is typically reduced 2–4% below rest during heavy intensity exercise due solely to a reduced HbO₂ affinity induced via metabolic acidosis and increased temperature (Bohr effect). Finally, plasma water turnover is increased into the lung’s interstitial fluid space during exercise, but extraordinary drainage capacity of the interstitial fluid via the thoracic lymphatic system provides protection against movement from the interstitial into the alveolar spaces (100).

Fig. 3.

Ratio of ventilation to perfusion (V̇a/Q̇) at rest and during moderate-intensity steady-state exercise [oxygen consumption (V̇o₂) 2.2 L/min, cardiac output (CO) 18 L/min, alveolar-arterial partial pressure difference (A-aDO₂) 22 mmHg] in healthy male subjects, as determined by the multiple inert gas elimination technique (MIGET). MIGET estimates the “continuous” distribution of V̇a/Q̇·based on the retention and elimination of 6 infused inert gases of varying solubility (204, 205). Note that V̇a/Q̇ distribution (log scale x-axis) is very narrow at rest, centered around an overall V̇a/Q̇ of ~0.9 and spanning units with V̇a/Q̇ ~0.5 to 3.0. During exercise, the nonuniformity of V̇a/Q̇ increases > rest, but overall V̇a increases out of proportion to Q̇ so that the mean V̇a/Q̇ has more than doubled from rest to exercise. Thus, the lowest V̇a/Q̇ ratios at rest no longer exist during exercise, and pulmonary capillary blood flow is now exposed to higher alveolar (and therefore end-capillary) Po₂. Given the exercise-induced reduction in mixed venous O₂ content, the widened V̇a/Q̇ distribution accounted for about one-half the widened A-aDO₂ during exercise. The authors could attribute the remainder of the increase in A-aDO₂ during this moderate-intensity exercise to an extrapulmonary shunt of deoxygenated mixed venous blood (Thebesian and bronchial venous drainage) amounting to ~1% of the prevailing CO, which would result in a significant end-capillary to arterial Po₂ difference. An alternate explanation attributes the extra widening of the A-aDO₂ (beyond that accounted for by V̇a/Q̇ nonuniformity) to alveolar-capillary diffusion limitation (204). Given the more uniform topographical distribution of V̇a/Q̇ reported from rest to exercise (13, 20), the increased dispersion of V̇a/Q̇ quantified via MIGET in this study was attributed to an increased nonuniformity of V̇a and Q̇ distribution within isogravitational lung regions (59) [Based on findings reported in Gledhill et al. (58)].

Summary: Just Right/Overbuilt: Optimal Design?

Ewald Weibel (1929–2019), the late, pioneering lung morphologist envisioned “what makes a good lung,” fit to provide our organs with the O₂ they need at rest and exercise, according to the principles of optimal design (209). He emphasized the very large yet extraordinarily thin and delicate surface area of contact between air and blood, maintained by an exceedingly small amount of supportive tissue arranged in a hierarchical design of elastin and collagen fibers woven into the capillaries together with a fine layer of stabilizing surfactant at the air-liquid interface. This ingenious architectural design observed rules of fractal geometry in packing this huge surface area into the limited space of the chest cavity as well as providing a system of airways and blood vessels reaching all points evenly on the gas exchange surface.

As discussed in the preceding sections, optimal design principles also appear to extend to the entire healthy respiratory system when faced with the enhanced demand for gas transport during exercise. Thus, the remarkable architecture of the “good lung” and its vasculature is complemented by the design of the highly aerobic capacity of the respiratory musculature and the precision and efficiency with which the feedback and feedforward components of the neural respiratory control network mysteriously controls its activity and coordination to drive alveolar ventilation precisely in proportion to tissue CO₂ production while minimizing elastic and resistive loads on the respiratory muscles and preserving adequate O₂ transport. Accordingly, it seems highly unlikely that any aspect of the ventilation, gas exchange, pulmonary vasculature, or respiratory muscle response to progressive or sustained exercise presents a significant limitation to exercise performance, at least in the young untrained male adult exercising in a normoxic environment. We now address whether this optimal design also applies to highly trained individuals, the elderly, females as well as males, in hypoxic environments, and in non-human mammalian species selected for their extraordinary exercise capability.

DEMAND VERSUS CAPACITY: RELATIVE MALLEABILITY IN O₂ TRANSPORT SYSTEMS

Many of the physiological variables contributing to maximum O₂ transport undergo substantial levels of phenotypic plasticity in response to intense, long-term physical training. For example, increases in V̇o_2max are accompanied by enhanced circulating blood volume and Hb mass, increased left ventricular compliance and volume, reduced vascular resistance, and increased muscle mitochondrial volume and capillary density and therefore O₂ extraction (118). Should one also expect the lung parenchyma, airways, respiratory muscles, pulmonary vasculature, and ventilatory control system to undergo the appropriate phenotypic plasticity to match the demands for greater O₂ transport imposed by the higher metabolic rates achieved in the endurance-trained individual? After all, these functions and structures appear to be overbuilt and/or regulated to a near-perfect extent in the untrained, healthy, young adult.

First, lung volumes and diffusion surface areas in mammalian species are capable of undergoing phenotypic plasticity in response to several types of chronic stressors, at least during maturation. These include long-term exposure to hypoxic environments (23, 80), partial pneumonectomy (80), and even severe starvation and refeeding (23, 123). Furthermore, in some mammals, such as the pronged-horn antelope with V̇o_2max three- to fourfold greater than that of sedentary species of similar body mass, the lung’s diffusion capacity rises in proportion to other enhanced cardiovascular and muscle metabolic capacities (193, 209). Genetic influences were also revealed via artificial selection for running endurance in rodents. Even though 7 generations of selective breeding yielded no pulmonary changes, 15 generations provided lung volumes and diffusion capacities that were increased out of proportion to body mass (99). However, this plasticity is not forthcoming in the healthy lung’s response to physical training, either during or following maturation (168, 177, 180). Even human endurance-trained athletes with superior V̇o_2max may or may not have marginally enhanced diffusion capabilities, lung volumes, and maximal flow-volume capacities (30, 33, 168, 194). The absence of training effects on lung morphology is especially surprising given the adaptive role of chronic lung stretch (for example, with pneumonectomy) and metabolism (for example, with nutritional changes) in altering lung volume and diffusion surface (164). In contrast, the aerobic capacity of respiratory muscles, is, like limb locomotor muscle, uniformly increased in response to intense training (187). Thus, given the limited plasticity in the lung with training and even via genetic endowment in the Olympian, the airways, diffusion surface, pulmonary vasculature, and respiratory muscles need a greater “safety margin” to accommodate the increased demand driven by the more malleable controllers of O₂ transport and utilization. We now discuss the accumulated evidence to date, which documents successes and failures and highlights those groups that are most susceptible to exercise-induced respiratory system limitations.

UNDERBUILT/MALADAPTED AIRWAYS IN HIGHLY TRAINED ENDURANCE ATHLETE

Capacity for Flow-Volume Less Than Demand

Figure 2 shows mean values in a group of highly trained runners in whom the requirements for ventilation during heavy to maximum exercise meet or exceed their maximum volitional flow-volume loop. The corresponding esophageal pressure-volume loops demonstrate that the tidal expiratory pressures at these high work rates often exceed maximum “effective” pressure, where further increases in expiratory muscle effort provide no additional flow because of positive, collapsing transmural airway pressures. The inspiratory muscles are required to produce forces within 5–10% of their maximum dynamic pressure available for velocity of shortening and force production at the flows and volumes achieved at maximum exercise. Peak intrathoracic pressures (ITPs) are in the ± 30–35 cmH₂O range and gastric pressures in the ± 15–25 cmH₂O range, as active expiration is initiated during even mild exercise with subsequent progressive recruitment of rib cage and abdominal expiratory muscles at rising exercise intensities.

The consequences of the “excessive” ventilatory demand and associated intrathoracic and abdominal pressures during heavy to maximum exercise in trained subjects include the following:

• • Hyperinflation, which allows increased flow rates but places the inspiratory muscles at a shorter and less optimal length for force production and requires tidal volume to be generated on a less linear, more inefficient part of the pressure-volume relationship. Dynamic hyperinflation does not always occur, even in the presence of complete expiratory flow limitation, although this circumstance also results in excessive expiratory intrathoracic and abdominal pressures.
• • Excessive inspiratory and expiratory work of breathing leading to respiratory muscle metabolite accumulation and fatigue eliciting associated reflex effects on limb vascular resistance (see respiratory muscle demand greater than capacity).
• • Constrained hyperventilatory response to heavy exercise, as demonstrated by the negligible or reduced ventilatory response to superimposed inspired CO₂ and/or hypoxia (vs. those responses obtained at lower exercise intensities) (29, 94, 126). Increases in the hyperventilatory response to exercise and/or superimposed chemical stimuli were observed when the maximum flow-volume envelope was enlarged and expiratory flow limitation reduced via low-density He:O₂ inhalation (39, 126, 133).
• • Limited numbers of studies have manipulated intrathoracic and abdominal pressures in exercising humans and canines and observed the following cardiovascular consequences. During inspiration, reducing the negativity of ITP reduced stroke volume (67), reflecting the importance of preload, and raising abdominal pressure increased resistance and impeded the return of blood flow in the femoral vein (130). During expiration, with small superimposed increments in ITP, stroke volume is reduced presumably because a decreased transmural pressure across the ventricles impedes diastolic filling (130, 190). Together, these findings point to a significant respiratory constraint of venous return and stroke volume in heavy exercise secondary to a combination of positive intra-abdominal pressures and especially positive intrathoracic pressures during expiration: effects that would be exacerbated via expiratory flow limitation.

Using the overlap metric of tidal to maximum flow-volume (62),¹ a significant prevalence of underbuilt maximum flow-volume envelopes in response to the demands of ventilation during heavy-maximum intensity exercise has been reported in endurance-trained athletes of all ages and especially in the master athlete and in females (see exceptional susceptibility to respiratory system limitations). In addition to excessive ventilatory demand driven by high work rates in the highly trained subject, exercise-induced flow limitation is also precipitated (even at relatively low levels of ventilation) by compromised airway structure, as observed in females versus males and in the lungs of elderly master athletes (see below). Alternatively, those athletes (e.g., swimmers) with especially large lung volumes and airways and very large maximum flow-volume envelopes have excessive safety margins in lung morphology and will avoid expiratory flow limitation and its sequelae even at high ventilatory demands (8). The large lung volumes seen in competitive swimmers may act to provide selective advantage, with studies indicating that lung volume appears superior even at a young age and that growth is not enhanced by subsequent training exposure (19).

Underbuilt Intrathoracic Airways

We refer here to airways extending from the main bronchi beginning with generation eight of the tracheal bronchial tree with 2 mm or less diameter, lacking cartilaginous support, and lined with smooth muscle and surfactant. In most healthy untrained subjects, these airways maximally bronchodilate with exercise (see above). Furthermore, moderate-intensity exercise training regimens commonly reduce airway inflammatory markers and airway hyperresponsiveness in mildly asthmatic individuals (128, 202). Training has also been shown to prevent cigarette smoking-induced chronic obstructive pulmonary disease in rodents by inhibiting local oxidative stress and proinflammatory cytokine release in the airways combined with release of anti-inflammatory cytokines from contracting skeletal muscle (196).

In contrast, elite endurance-trained athletes have a high prevalence of airway narrowing during or immediately following high-intensity exercise (98, 111, 163, 185). Furthermore, high-intensity training in human cross-country skiers or in horses or sled dog athletes causes disruption and remodeling of the airway epithelial layer, often accompanied by hypersensitized airway reactivity (95). The high ventilatory demands requiring flow rates in excess of ten times resting levels in the endurance-trained athlete appear to induce injury and remodeling of the airway epithelium via two types of stress (7). First, dehydration stress occurs over the first 10–12 airway generations as a high rate of evaporative water loss occurs and the osmolality of the very thin airway surface layer increases. This leads to a sloughing/detachment of the epithelial layer (7, 64). Second, mechanical sheer stress, via repeated stretch and compression of airway epithelium, likely invokes release of cytokines and epidermal growth factors (151). In the elite endurance athlete who trains daily in a cold, dry environment, the repeated release of these molecules forms an integral part of the injury-repair process in the subepithelial environment, promoting a hypersensitization of the contractile properties of the airway smooth muscle (97, 98, 151). This process seems to be at least partially reversible, with studies indicating that airway inflammation and hyperresponsiveness can revert to normal following a period of detraining in elite swimmers. Indeed, this has led some authors to propose that the development of lower airway dysfunction in some groups of elite athletes may be considered an occupational lung disease (74, 162).

Although exercise-induced bronchoconstriction (EIB) is primarily manifested immediately following intense, sustained exercise, airway resistance is also usually compromised, leading to an increased work of breathing during exercise, along with a constrained alveolar ventilation, V̇a/Q̇ nonuniformity (138), and O₂ and CO₂ exchange (71, 72). Several weeks of inhaled corticosteroid treatment have been shown to improve breathing mechanics and gas exchange in these subjects along with an enhancement of exercise performance (72).

Underbuilt Upper Extrathoracic Airway

The upper or extrathoracic airway, defined anatomically as the airway section lying above or proximal to the thoracic inlet, has generally been overlooked in exercise performance science (163). This is remarkable, given the fact that this airway segment contributes a high proportion of the total resistive load on the respiratory system and indeed is often described as the “bottleneck” of the airway (31, 179). Moreover, it does not function as a simple airflow conduit and has evolved intricate structural and neuromuscular capability to ensure effective and often synchronous delivery of a number of physiological functions, including swallowing, vocalization, and cough (81).

During vigorous exercise and as ventilatory demand progressively increases, there is evidence of a transient dilatation of the laryngeal inlet, likely driven by heightened neural traffic to the primary laryngeal abductor muscles (83). In all subjects, however, the laryngeal inlet remains a distinct “choke” point within the airway conduit. Movement of air across the upper airway is governed by the same laws of aerodynamics that dictate the behavior of any fluid traversing a tubular structure. Specifically, both the Venturi effect (linear movement through a tube causes decrease in lateral pressure) and Bernoulli principle (pressure is least where the velocity is greatest) dictate that as airflow increases, so negative transmural pressure gradients encourage inward collapse of surrounding (i.e., laryngeal) structures. Any compromise in the dimensions of the laryngeal space, even slightly beyond normal parameters, can thus cause a substantial increase in airway resistance, and as the glottic aperture narrows, this encourages any vulnerable structures to approximate medially or collapse inwardly (52) (see Fig. 4).

Fig. 4.

Schematic diagram highlighting the interplay of factors relevant to exercise-induced laryngeal obstruction (EILO). Laryngeal images taken during exercise of increasing intensity (images acquired at rest and moderate- and peak-intensity exercise), showing EILO. Images on schematic approximated to a figure adapted from Olin et al. (146), highlighting the close relationship between exercise-intensity and onset of EILO. [Adapted from Olin et al. (146) with permission of the © ERS 2020.]

At rest or in states of mild to moderate hyperpnoea, the movement of laryngeal structures appears to be of very limited or no immediate relevance. However, in athletic individuals capable of increasing ventilation considerably, the potential for flow-driven pressure changes across the glottic inlet becomes increasingly relevant. Indeed, it has been postulated that the laryngeal structures exist in a state of “force balance,” such that if airflow is modeled to increase indefinitely, then there would come a point in all individuals whereby the forces promoting glottic opening would be overcome and some form of medial or inward laryngeal collapse would occur (15, 37).

Although seemingly hypothetical, exercise-associated upper airway closure is an increasingly well-recognized and highly prevalent issue in athletic individuals. This transient phenomenon, termed exercise-induced laryngeal obstruction (EILO) (although still called exercise-induced vocal cord dysfunction in some centers), is now recognized to affect between 5 and 7% of all adolescents (90), ~7% of randomly selected Northern Europeans under 25 yr old (27), and 15–20% of athletes with apparently unexplained exertional breathlessness (142). There appears to be a female preponderance in most studies, and this appears to reflect the demographic characteristics encountered in real-life clinical practice; however, some epidemiological studies reveal no gender differences (90).

Laryngeal closure during exercise is associated with the development of dyspnea and stridor (82), an increased resistive work of breathing, heightened respiratory neural drive (207), and arterial hypoxemia (69). These deleterious consequences for physiological performance are transient and abate rapidly on exercise cessation (146). In an extreme form, however, the presence of fixed laryngeal narrowing (e.g., with a glottic stenosis) results in significant hypercapnia during exercise and a progressive hypopnoea, driven by the consequent mandatory prolongation of the inspiratory phase of the respiratory cycle (3).

The precise etiology of EILO is currently unknown; however, it seems likely that a relative deficiency in the structural integrity of some laryngeal structures (e.g., the supraglottic structures) is relevant in many cases (144). Studies extrapolating and modeling increasing flow, from axial imaging allied with computational flow dynamics, indicate that glottic narrowing is predictable, based on the aforementioned Venturi principle, and closure has significant up- and downstream consequences for airflow (54) and most likely for the work of breathing. Overall, this would tend to support a viewpoint that the upper airway, and specifically the larynx, could in many cases be considered relatively underbuilt for the very high ventilatory demand state. It is conceivable that extreme adaptation in other physiological systems (e.g., in cardiovascular function) places the larynx in an unfavorable or maladaptive position under states of extreme hyperpnoea in some (susceptible) individuals. In support of this argument is the finding that EILO appears to be most highly prevalent in adolescent female athletes and, indeed, at least in some series appears to regress to a degree with further maturation (119). Indeed, although there are limitations inherent to making accurate flow measurements in this area, it is generally agreed that in the majority of young individuals, EILO arises as the consequence of an imbalance between forces acting to maintain abduction (i.e., structural and neuromuscular support) and the airflow-associated pressure changes acting to promote glottic adduction. Moreover, when athletes are reevaluated following laryngeal corrective surgery (for anatomical excessive tissue in the arytenoid area of the larynx) (50), there is an apparent improvement in symptom burden but also a corresponding reduction in markers of resistive loading and work of breathing (50, 207). Overall, therefore, it is conceivable that in individuals with either a structural or functional predisposition to having an underbuilt upper airway (e.g., as may be seen in some young female athletes), the forces promoting laryngeal abduction become challenged and often overwhelmed by exercise hyperpnoea and the associated physical pressure changes and airway turbulence that ensues. This acts to encourage the upper airway to dynamically adopt behavior that leads to inspiratory flow limitation and, akin to that seen in expiratory flow limitation, it may be possible to establish this by evaluating the influence of further pressure modulation on flow in this region. An analogous physiological scenario is that of obstructive sleep apnea, whereby for afflicted individuals there is typically structural upper airway abnormalities that then predispose an individual to airway collapse during a state of reduced tonic input to the airway dilator musculature; i.e., at the onset of sleep.

Despite significant progress in our understanding and awareness of the role of the upper airway in exercise physiology, over the past decade, the field still remains at a somewhat embryonic stage. A considerable amount of further work is needed now to better decipher the pathophysiology underpinning the laryngeal structural deficiencies seen in so many young individuals with EILO, i.e., why it is that the upper airway in some athletic individuals is underbuilt. We also need to progress our understanding of the factors dictating airflow, turbulence, and pressure change in the upper airway (55) and start to routinely employ objective visualization techniques (114) to ensure physiologists and clinicians have an accepted and standardized approach in assessing this problem; only then will we start to better appreciate whether the extrathoracic airway is over or underbuilt for exercise.

UNDERBUILT PULMONARY VASCULATURE

Effects of High Cardiac Output Combined with Limited Vasodilation

The benefit of the low resistance, high compliance pulmonary vascular circuit is that it reduces right ventricular (RV) work. The work, and oxygen demand, of the right ventricle is highly dependent on afterload (pulmonary arterial pressures) and to a lesser extent on heart rate, inotropy, and preload. Thus, low pulmonary artery pressures at rest enable the RV to do very little work, and the heart has adapted to these requirements such that the RV has approximately one-quarter the mass and contractile reserve of the left ventricle (LV) (108). In other words, the RV is built to deal with low pressures and the LV is built for moderate to high pressures. As outlined above, this arrangement works beautifully under resting conditions and during mild to moderate activities with modest increases in cardiac output.

Pulmonary artery pressures increase in a near-linear manner with exercise. Multiple studies using echocardiographic estimates of pulmonary artery pressures and direct invasive measures have demonstrated that mean pulmonary artery pressures increase by ~1 mmHg for every liter of cardiac output (113). Thus, in a well-trained athlete who increases cardiac output from 5 L/min to more than 30 L/min, the mean pulmonary artery pressure would be expected to increase nearly threefold, as compared with the systemic mean blood pressure that typically increases 30–40% to maximal exercise intensity. As summarized in Fig. 5, the work demands of the RV during exercise increase disproportionately to those of the LV. This concept was interrogated by La Gerche et al. (106), who used a combination of cardiac imaging and direct invasive pressure measurements at rest and at peak exercise intensity in athletes and nonathletes. RV end-systolic wall stress increased in athletes more than in nonathletes, but in both cohorts the increase was far greater than for the LV (125% vs. 14%, P < 0.0001). At rest, the systolic wall stress in the RV was less than half that of the LV and was therefore matched to its lesser mass and contractile reserve. However, at peak exercise the contractile forces, work, and oxygen demands of the RV would seem to exhaust the reserve given its lesser myocyte mass. This disproportionate RV work during exercise has also been elegantly demonstrated in exercising dogs, in which the venous effluent from the RV and LV can be sampled separately. Although coronary blood flow to the two ventricles is similar, the coronary venous Po₂ falls considerably more in the RV, reflecting the greater increase in work (68, 199).

Fig. 5.

Comparison of the presystemic and systemic circulation demands during exercise as compared with rest. The low resistance, high compliance pulmonary circulation imposes little work on the right ventricle (RV) at rest, but during exercise there is limited capacity to reduce resistance further. With the distension of the vasculature during exercise, compliance reduces. Thus, the pulmonary and RV pressures increase 2- to 3-fold. In the systemic circulation, the vascular resistance falls markedly, leading to a marked attenuation of the increase in ventricular work. Figure is based on data and concepts from La Gerche et al. (106).

A logical extension of the physiological constraints of the pulmonary circulation during intense prolonged exercise is that one might expect to observe RV dysfunction or fatigue. That is exactly what has been observed. La Gerche et al. (104) were the first to observe marked dilation and dysfunction of the RV after intense endurance exercise, and this has been validated in numerous subsequent studies using both echocardiography and cardiac MRI (28, 102, 136, 148, 150, 198). The degree of dysfunction appears related to both the intensity and duration of exercise as might be expected from the physiology: the intensity mandating high RV load and the duration exhausting the metabolic reserves and possibly resulting in some secondary inflammatory damage and interstitial damage (47). In the longer term, the RV tends to partially adapt by increasing its myocardial mass (and contractile reserve). In athletes, the RV-to-LV mass has been observed to be greater than in nonathletes (17, 106).

Clinical Consequences of Right Heart Remodeling

There are important clinical consequences of the underbuilt pulmonary vasculature and overstressed RV in endurance athletes. The repeated physiological stressors and resultant acute myocardial damage can result in chronic adverse remodeling that predisposes some athletes to potentially life-threatening arrhythmias (73, 103, 201). It has been well established that endurance exercise is associated with an increase in RV arrhythmias in people with a genetic predisposition to certain cardiomyopathies, suggesting that there is a potential environmental-genetic interaction in which intense exercise can exacerbate abnormalities in athletes with a weakness in cardiac structure and function (87, 178, 208). Increasingly accepted is the concept that a similar process may develop in some endurance athletes who have no identifiable genetic predisposition (109, 181) in a condition coined “exercise-induced right ventricular cardiomyopathy” by Hein Heidbuchel in 2003 (73) (see Fig. 6). The concept of exercise-induced RV fatigue raises another intriguing hypothesis: could this limit cardiac performance during prolonged exercise? Could it explain the “cardiac drift” in which heart rate and perceived exertion tend to increase and output plateaus or decreases? This has seldom been interrogated, but in a small cohort of athletes, Claessen et al. (28) studied biventricular function during exercise at the completion of an endurance cycle race and observed a relative failure of RV augmentation. This deserves further attention and may provide novel insights into central fatigue during intense prolonged exercise.

Fig. 6.

Acute right ventricular (RV) load during exercise leading to chronic remodeling and arrhythmias. The disproportionate load on the RV during exercise can predispose to chronic remodeling that may be adaptive or maladaptive. These different responses may be related to training, genetics, or other unknown factors. The maladaptive athlete’s heart is predisposed to arrhythmias that predominantly arise from the remodeled RV. ARVC, arrhythmogenic RV cardiomyopathy. [Adapted from La Gerche and Heidbuchel. (105) with permission from Wolters Kluwer Health.]

Thus, a picture emerges of a degree of central limitation during intense exercise with potentially severe complications in a minority of athletes. However, as opposed to earlier descriptions of exercise physiology focusing on the systemic circulation and LV, research over the past few decades has narrowed much of the constraint to the pulmonary circulation and its perfusing pump, the RV. There are, however, several ways in which the body attempts to attenuate the pulmonary vascular limitations during exercise. As discussed in the earlier sections, pulmonary blood flow is not uniform; rather, there are zones of relative hypoperfusion at rest: those of greater ventilation and others of greater perfusion. Flow down larger diameter capillaries (such as the subpleural capillaries) is prioritized during exercise, and this phenomenon has been demonstrated with the novel technique of contrast echocardiography in which bubbles of 15–25 µm are injected into the venous circulation and fail to pass through the lung microvasculature at rest (45, 116). During exercise, these bubbles are observed to pass through the pulmonary circulation, and in those subjects in whom this phenomenon is observed, it is associated with greater reductions in pulmonary vascular resistance, greater cardiac outputs, and enhanced exercise capacity (107, 173). This represents an intriguing way in which the body attenuates the constraint on flow and, to some extent, helps protect the RV against the physiological stresses imposed during strenuous exercise.

This also leads to the sensible hypothesis that pharmacological modification of the pulmonary vasculature could be used to attenuate the limitation imposed by the pulmonary vasculature and protect the RV from damage. There are several medications, such as phosphodiesterase-5 and endothelin inhibitors, that promote vasodilation of the pulmonary arteries in a relatively specific manner. This hypothesis has been addressed in a number of trials of healthy nonathletic and athletic individuals at sea-level, altitude, and normobaric hypoxia with very mixed results (22). A simple explanation for the minimal or absent efficacy of these therapies is that strenuous exercise induces near-maximal pulmonary vasodilation in healthy subjects such that pulmonary vasodilators have minimal additional effect.

In summary, the RV and pulmonary circulation represent something of an Achilles’ heel that constrains central output during intense exercise, particularly when the exercise is prolonged, and may result in RV damage and arrhythmias in a minority of athletes. Although there are some inbuilt mechanisms to attenuate the stress, it stands to reason that the presystemic circulation favors energy preservation and efficiency and that the RV has evolved to be better suited to resting and low/moderate activity than to intense endurance exercise.

UNDERBUILT ALVEOLAR TO ARTERIAL GAS EXCHANGE

In the endurance-trained athlete (male and female, children, young and older adults), arterial HbO₂ desaturation amounting to 4–12% below resting values occurs during heavy intensity incremental or constant load exercise, with a prevalence ranging between 30 and 70% among the various groups studied to date (n = ~20–100), using temperature-corrected measures of arterial blood gases or pulse oximetry (32, 36, 39, 42, 66, 67, 145, 159). In susceptible athletes, the HbO₂ desaturation almost always first appears during submaximal steady-state exercise, usually at intensities at or slightly in excess of the lactate threshold and typically reaching a nadir at maximal exercise (see Fig. 7B). It is attributable in about equal amounts to a 10–30 mmHg reduction in $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ below rest combined with a pH and temperature-induced decrease in HbO₂ affinity. In turn, the reduced $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ correlates most strongly with excessively widened A-aDO₂ (~30–50 mmHg) in most cases and secondarily to a constrained hyperventilatory response ($\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ ~36–43 mmHg) (36). Exercise-induced arterial hypoxemia (EIAH) and its determinants are highly reproducible within subjects upon repeat testing (189). Finally, a consistent observation across studies is the marked variability in EIAH among athletes of equivalent V̇o_2max (see Fig. 7).

Fig. 7.

Major features of exercise-induced arterial hypoxemia (EIAH). A: pulmonary gas exchange in a group of 17 female competitive runners during constant load treadmill running at 90% of maximal oxygen consumption (V̇o_2max; 14 km/h, 3% grade) to exhaustion (preceded by 3 min at each of 50 and 75% of V̇o_2max). Subjects were 27 ± 7 yr old, with V̇o_2max of 50 ± 4 (44–56) mL·kg⁻¹·min⁻¹. Subjects were divided into those with >15 vs. <10 mmHg exercise-induced reductions in arterial partial pressure of oxygen ($\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$), with no differences in V̇o_2max between EIAH and non-EIAH groups. Note the reductions in $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ and raised alveolar-arterial partial pressure differences (A-aDO₂s) within the first minute of exercise at 90% V̇o_2max, which remained constant through exhaustion. Further reductions in arterial oxygen saturation ($\text{S}{\text{a}}_{{\text{O}}_2}$) were secondary to a Bohr effect rightward shift in oxyhemoglobin HbO₂ dissociation as arterial pH fell to 7.26 and esophageal temperature rose 2.3°C. B: mean ± SD values for the EIAH and non-EIAH groups during 3 min of treadmill walking/running at each of 50, 75, and 90% V̇o_2max. Note significant EIAH was already present at 75% V̇o_2max. Nadirs in $\text{S}{\text{a}}_{{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ with A-aDO₂ widened to maximum values occurred at 90–100% V̇o_2max. *Significant difference between groups (P < 0.05) [Adapted from Wetter et al. (211)].

• • A few studies have examined potential mechanisms underlying the excessive A-aDO₂ by comparing subjects with EIAH versus those without, as summarized below. Application of the multiple inert gas elimination technique (MIGET) to athletes with severe EIAH (169) revealed approximately equal contributions to the widened A-aDO₂ from V̇a/Q̇ maldistribution, a constrained hyperventilatory response, and a combination of alveolar capillary diffusion disequilibrium perhaps combined with some degree of small intra/extrapulmonary shunts. The diffusion limitation may be explained by excessive exercise-induced decreases in the compound variable D/βQ, (D is diffusion capacity, β is the shape of the HbO₂ dissociation curve in the pulmonary capillary, and Q is blood flow) (153). In the EIAH athlete, a normal or even reduced maximum diffusion capacity combined with a high β (i.e., steeper slope of the HbO₂ dissociation curve) secondary to increased O₂ extraction and reduced mixed venous partial pressure of oxygen (${P\stackrel{\circledR }{\text{v}}}_{{\text{O}}_2}$) and a high Q̇ are common. Although exercise-induced intrapulmonary shunts do occur, it remains undecided whether they contribute significantly to EIAH (115).
• • Durand et al. (43) compared moderate EIAH versus non-EIAH trained-subjects, reporting higher Ppa (> 42 mmHg) and higher pulmonary vascular resistance, with equivalent diffusing capacity for carbon monoxide (DLco) during maximum exercise. These findings point to contributions to EIAH from V̇a/Q̇ maldistribution secondary to excessive Ppa precipitating interstitial edema, V̇a/Q̇ nonuniformity, and/or intrapulmonary shunt (191, 210). The presence of red cells in bronchoalveolar lavage fluid following intense exercise in trained cyclists was suggestive of a mechanical stress-induced disruption of the blood-gas barrier (77).
• • The onset of EIAH during submaximal exercise speaks against the concept of excessive demand (vs. capacity) as the sole determinant of EIAH in the athlete. Wetter et al. (211) tested the possibility that exercise-induced inflammation of the small, peripheral airways associated with chronic, intense physical training (see Underbuilt Intrathoracic Airways) might account for a widened A-aDO₂, perhaps via ventilation distribution nonuniformity. Although significant numbers of athletes did show evidence for increased airway reactivity and/or increased small airway resistance, these characteristics did not clearly distinguish those with EIAH. High-resolution imaging of ventilation distribution needs to be applied during exercise to address this question.
• • The constrained hyperventilatory response to maximum exercise will contribute to EIAH via a reduced overall V̇a/Q̇ in the face of a nonuniform V̇a/Q̇ distribution and falling concentration of mixed venous oxygen (${C\stackrel{\circledR }{\text{v}}}_{{\text{O}}_2}$). The ventilatory constraint is attributable to a significant extent in most athletes to mechanical limitations on both the expiratory (airways) and inspiratory (muscles) phases of respiration. However, there may also be an important constraint of neurally mediated “central drive” to breathe in play, as mechanical limits to V̇e are often approached but not reached in many of these EIAH athletes with a reduced hyperventilatory response (66, 94). This proposed reduction in central respiratory motor output likely reflects a broad interindividual continuum of responsiveness to both the multiple feedback and feedforward excitatory drives to breathe during heavy intensity exercise, in addition to the inhibitory effects of feedback from the lung and chest wall (53).

The consequences of EIAH have been quantified using increments in fraction of inspired oxygen (${\text{FI}}_{{\text{O}}_{\text{2}}}$) just sufficient to maintain HbO₂ saturation at resting levels. These studies have revealed a threshold of ~3% decrement in arterial oxygen saturation ($\text{S}{\text{a}}_{{\text{O}}_2}$) to begin to detect a significant negative effect on V̇o_2max, with a decrement of ~2% in V̇o_2max for each ensuing 1% fall in $\text{S}{\text{a}}_{{\text{O}}_2}$ (65, 143, 160). Thus, the maximum decrement in V̇o_2max attributable to EIAH is ~15% in humans (at 88% $\text{S}{\text{a}}_{{\text{O}}_2}$). Similar studies conducted during cycling time trials showed that preventing the EIAH reduced the level of locomotor muscle fatigue and improved endurance performance (175)².

RESPIRATORY MUSCLE DEMAND GREATER THAN CAPACITY

At high-intensity exercise, multiple primary and accessory inspiratory and expiratory muscles are recruited, requiring V̇o₂s and blood flows amounting to 15–20% of the maximum V̇o₂ and CO in trained subjects (2, 67, 121, 182). The use of electrical or magnetic supramaximal stimulation of the motor nerves (1–20 Hz) using transdiaphragmatic pressure (P_di) or gastric pressure (P_g) as a measure of force output showed that both the diaphragm and expiratory muscles undergo significant fatigue during high-intensity sustained exercise so long as the exercise intensity is in excess of 80% of peak work rate. Respiratory muscle fatigue begins around the midpoint and persists with further exercise to exhaustion and persists for 1–2 h in recovery (91, 206). Highly trained subjects required a higher absolute work rate than less-trained subjects did to experience diaphragm fatigue (11). Additional evidence that O₂ transport to the respiratory muscles did not meet metabolic requirements was obtained using near-infrared spectroscopy monitoring, which revealed deoxygenation of accessory respiratory muscles of the trunk during heavy intensity cycling exercise (112, 200).

Two lines of evidence in the exercising human have revealed the major factors contributing to exercise-induced metabolite accumulation and fatigue of the diaphragm (see Fig. 8, A and B). First, when a mechanical ventilator was used to reduce the inspiratory work of breathing by 40–60%, exercise-induced diaphragm fatigue did not occur (10). Second, when the level of diaphragmatic work incurred during high-intensity exercise was volitionally reproduced with the subject at rest, the resulting diaphragm fatigue was either nonexistent or greatly reduced and recovery was almost immediate; furthermore, a substantially higher level of voluntary diaphragmatic work was required to elicit diaphragm fatigue comparable to that incurred during the treadmill exercise (see Fig. 8B) (12). These findings reveal that both the absolute level of the work of breathing and the coincident locomotor muscle exercise, per se, contributed significantly to exercise-induced diaphragm fatigue: the former through imposing a high sustained demand for O₂ transport, and the latter likely via the effects of limb-exercise-induced sympathoexcitation on curtailing increases in respiratory muscle vascular conductance and blood flow (182). This proposed effect of heavy intensity limb exercise on curtailing diaphragmatic blood flow is analogous to the finding that substantial increases in leg vascular conductance during moderate to heavy intensity cycling were curtailed via increased sympathetically mediated vasoconstriction achieved by superimposing arm exercise (170, 192). Thus, although in vitro findings in rodent muscle vasculature support an attenuated vasoconstrictor responsiveness in the diaphragm versus limb vasculature (1), during high-intensity exercise the respiratory muscle vasculature appears not to be completely resistant to the influence of enhanced sympathetic vasoconstrictor activity emanating from the working limbs (157, 182).

Fig. 8.

Major contributions to exercise-induced diaphragm fatigue from work of the diaphragm (A) and locomotor muscle work (B). A: a proportional assist ventilator (PAV) was used to reduce the peak inspiratory esophageal and transdiaphragmatic pressures by 40–70% during exercise to exhaustion at >80% maximal oxygen consumption (V̇o_2max). This reduced work of breathing prevented diaphragm fatigue, presumably a reflection of reduced metabolite accumulation and therefore less stimulation of respiratory muscle group III-IV afferents. The consequences of this respiratory muscle unloading included: a) decreased blood flow to accessory respiratory muscles; b) reductions in both median nerve muscle sympathetic nerve activity (MSNA) in the resting arm and norepinephrine spillover across the working limb; c) increased vascular conductance and blood flow to the working limbs; and d) less limb fatigue and prolonged exercise performance [see summary in Sheel et al. (182)]. B: diaphragm fatigue measured as the transdiaphragmatic pressure (P_di) response to 1–20 Hz phrenic nerve stimulation before and after treadmill exercise to exhaustion at 90% V̇o_2max vs. before and after voluntary mimic at rest of the diaphragmatic work achieved during exercise. The greater diaphragm fatigue and the prolongation of recovery following the treadmill exercise reflects the effects of coincidental locomotor muscle exercise, per se (see text) [Adapted from Babcock et al. (12)]. fb, breathing frequency.

The cardiovascular consequences of the normally occurring exercise-induced inspiratory and expiratory muscle work and metabolite accumulation were revealed through the use of mechanical ventilation (see summary in Fig. 8A). These included reduced sympathetic vasoconstrictor activity and increased blood flow to locomotor muscles, leading to reduced limb fatigue and increased exercise performance. These sympathetic and cardiovascular consequences of respiratory muscle unloading occurred during high-intensity exercise in which diaphragm fatigue was prevented but also during very brief duration high-intensity exercise and even during submaximal exercise, during which diaphragm fatigue was unlikely to have occurred (40, 67). Based on pioneering investigations of phrenic nerve afferents in rodent and canine preparations (76, 85, 88, 174), we propose that these cardiovascular responses in exercising humans resulting from the prevention of metabolite accumulation in the respiratory muscles (via a reduced work of breathing) were secondary to attenuation of the group III-IV metaboreceptor-induced increase in sympathetic vasoconstrictor activity. The reductions in limb fatigue and improvement in endurance exercise performance with respiratory muscle unloading are presumed to reflect the beneficial consequences of preventing reflex effects of metabolite accumulation in the respiratory muscles. This interpretation is consistent with the exacerbation of limb muscle fatigue and reductions in performance elicited by fatiguing the diaphragm (via voluntary hyperpnea at rest) before cycling exercise (213).

EXCEPTIONAL SUSCEPTIBILITY TO RESPIRATORY SYSTEM LIMITATIONS

There is growing evidence that many highly fit humans, the healthy elderly, females, athletes exposed to high altitudes, and select athletically gifted canines, felines, and equines experience an especially high prevalence of respiratory system limitations to exercise performance.

Thoroughbred Horse: Underbuilt and Inadequately Designed

Owing to highly selective breeding over 300 yr and originating in a small gene pool, this species provides the ultimate example of an underbuilt lung structure providing the primary source of exercise limitation (156). On the “demand” side, a massive maximum O₂ flux capacity (V̇o_2max 140–220 mL/kg) is supported by the sheer mass of active skeletal muscle (55% body mass), mitochondrial volume, and capillary density together with a very large compliant heart (1.5–2.0% body mass) capable of producing up to a 2-L stroke volume and >400 L/min cardiac output. This huge O₂ transport system is complemented by an O₂ extraction across the active muscle that exceeds 90% at V̇o_2max and an exercise-induced splenic contraction that doubles circulating red cell mass and O₂ carrying capacity. Despite the prodigious structural capacity of the thoroughbred equine lung, including a 2,400-m exchange surface area, it remains grossly underbuilt and inadequately designed, as demonstrated by the following responses to heavy intensity exercise: 1) Ppa exceeds 120 mmHg secondary to the high cardiac output and downstream effects of a raised left atrial pressure, with right ventricular wall thickness equal to or exceeding that of the left (108); 2) expansion of the pulmonary capillary blood volume is insufficient to maintain red cell transit time >0.3 s; 3) a high dead space ventilation occurs because of the horse’s extreme tracheal length and high breathing frequency coupled to locomotion, yet the resultant need for greater total minute ventilation is opposed by a high resistive work of breathing because of obligate nasal ventilation coupled with a partial collapse of the unsupported nasal airway on inspiration (156). The result is a marked alveolar hypoventilation (_{$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$} ~50–60 mmHg), i.e., “respiratory failure,” in the face of marked swings in intrapleural pressure; 4) a combination of a short red blood cell transit time plus severe hypoventilation induces severe EIAH in all thoroughbreds studied ($\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$ ~60–70 mmHg), and a right-shifted HbO₂ dissociation curve via the Bohr effect secondary to a marked respiratory plus metabolic acidosis exacerbates HbO₂ arterial desaturation (75–85% $\text{S}{\text{a}}_{{\text{O}}_2}$); and 5) the combined high intraluminal pulmonary capillary pressures plus marked intrapleural pressure swings rupture the alveolar capillary interface with “hemorrhage” of red cells into the alveolar space and airways (14, 122, 156).

When EIAH is prevented [via modest increases in ${\text{FI}}_{{\text{O}}_{\text{2}}}$ (203)] or the high resistive work of breathing is alleviated via nasal stents (156), exercise performance is substantially improved in all thoroughbred equines.

Aging: Does the Age-Dependent Decline in Respiratory System Capacity Parallel or Exceed That in Demand?

Beginning as early as the second decade, lung structure begins its inexorable decline, including loss of lung elastic recoil, alveolar duct enlargement, and rib cage calcification. Functionally, these changes are manifested in reductions in the maximum effective expiratory pressure and the maximum flow-volume envelope, leading to increased airway closing volume, hyperinflation, and reduced inspiratory and vital capacities as well as a less-compliant chest wall and pulmonary vasculature and a markedly reduced lung diffusion capacity (92). Accordingly, in the sixth through eighth decades, even at rest, both trained and untrained subjects show airway closure at higher lung volumes, hyperinflation, and a moderate widening of the alveolar to arterial Po₂ difference with 5–15 mmHg reductions in $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$.

During moderate- through maximum-intensity exercise, an increased V̇e/V̇co₂ with elevated breathing frequency and reduced Vt combined with mild hyperinflation results in increased work of breathing and dyspneic sensations, even in the untrained elderly subject with V̇o_2max of 25–30 mL/kg and maximum V̇e in the 60–90 L/min range (89, 92, 120, 132) (see Fig. 9). The augmented V̇e/V̇co₂ demonstrates a remarkable plasticity within the aging respiratory control system, which ensures homeostasis of V̇a/V̇co₂ and _{$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$} during exercise near resting levels. On the other hand, this compensation comes at the substantial expense of an increased work of breathing associated with exertional dyspnea, with some evidence at maximum exercise of ventilatory constraint, as shown by ventilatory increases with He:O2 plus reduced ventilatory responses to added ${\text{FI}}_{{\text{CO}}_{\text{2}}}$ (9, 26). This enhanced perception of breathlessness on exertion likely contributes to a curtailment of an active lifestyle with healthy aging. However, it is doubtful if the respiratory system becomes a major limitation to maximum exercise performance in the untrained elderly because the 35–40% reductions in forced expiratory volume in 1 s (FEV₁) and DLco at age 70 versus 25 yr are accompanied by 40% reductions in V̇o_2max, i.e., maximal demand. This reduced V̇o_2max is primarily driven by age-dependent reductions in ventricular compliance and muscle mitochondrial volume and therefore in maximum cardiac output and the maximum a-V̇o₂ difference (75, 134, 135, 165).

Fig. 9.

Two major contributions to increased work of breathing during exercise in healthy elderly trained subjects. A: airway closure and narrowing on expiration occurs at higher lung volumes with aging, causing nonuniform ventilation distribution and increased falling dead space to tidal volume ratio (Vd/Vt) at rest (0.39 old vs. 0.30 young) and during exercise (0.29 vs. 0.15). Thus, at any given exercise carbon dioxide production (V̇co₂), total minute ventilation (V̇e) is elevated in the elderly, which produces similar alveolar ventilation (V̇a) and partial pressure of carbon dioxide (_{$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$}) as in the young but requires an increased work of breathing. B: maximum expiratory flow at 50% total lung capacity (MEF₅₀) is reduced by 25–35% at age 70 vs. 30 yr and the maximum effective pressure for productive flow is reduced by 50–70%. Combined with the higher V̇e/V̇co₂, this limited capacity for effective expiratory flow means that most trained 70-yr-old subjects achieve maximum effective expiratory pressures (and exceed 90% of maximum inspiratory dynamic pressures) during exercise at an oxygen consumption (V̇o₂) that is 40% less and at a level of V̇e that is 50–75 L/min below that required in younger trained adults (compare with Fig. 2). At maximum exercise, many highly trained elderly showed little or no ventilatory increase with added chemoreceptor stimuli, and approximately one-third of the group showed alveolar-arterial partial pressure differences (A-aDO₂) >30 mmHg and arterial oxygen saturation in the 87–93% range [Figure based on findings published in Miller and Dempsey (129)].

On the other hand, might an imbalance of O₂ transport demand versus respiratory system capacity occur in highly trained elderly subjects whose trainable cardiovascular system and muscle metabolic capacities allow a V̇o_2max in the 40–60 mL·kg⁻¹·min⁻¹ range, requiring ventilations in the range of 100–130 L/min and cardiac outputs of 15+ L/min (49, 79, 92, 197). In many highly trained 70-yr-old subjects, their maximum capacities for flow-volume and diffusion are still capable of accommodating the levels of maximum ventilations and O₂/CO₂ exchange achieved by untrained 30-yr-old subjects. This accommodation, even after 40 yr of aging, attests to the substantial reserve available in the healthy young lung. However, as in their younger trained counterparts, significant numbers of master athletes will experience severe expiratory flow limitation during heavy intensity-maximum exercise, with pleural pressures exceeding maximum effective pressure and a high work of breathing with substantial dyspnea. Less frequently, these elderly athletes experience limited hyperventilation, widened A-aDO₂ in excess of 30–35 mmHg, $\text{P}{\text{a}}_{\text{O}}{}_{{}_2}$10–25 mmHg, and $\text{S}{\text{a}}_{{\text{O}}_2}$ 4–10% below resting values (9, 63, 92, 129, 161) (see Fig. 9).

A limited number of observations based on invasive catheterization revealed a marked pulmonary hypertension during exercise in healthy elderly subjects (61, 166). Even at cardiac outputs of 10–15 L/min, the pulmonary artery (as well as pulmonary wedge) pressures consistently rose to levels that were twice those in healthy younger subjects, i.e., Ppa ~25–40 mmHg. In young adults, these levels of Ppa are achieved at cardiac outputs in the range of 20–30 L/min (139). These remarkable elevations in pulmonary vascular resistance and pressures in the elderly were attributed to age-related myocardial stiffness, causing reduced diastolic compliance of the ventricular wall and contributing to high left ventricular filling pressures.

Some, but not all, cross-sectional comparisons between trained and untrained elderly subjects show superior lung function in the trained subjects (9, 63, 92, 93, 124). Furthermore, a longitudinal 7-yr study showed that habitual physical training did not provide a protective effect against age-induced declines in lung function at rest and exercise in the sixth and seventh decades (124). Thus, available evidence, although limited, suggests that the lungs of older athletes, like their younger counterparts, are untrainable, whereas the cardiovascular, hematological, and muscular systems remain highly trainable. Thus, significant numbers of the trained elderly experience respiratory system limitations to heavy intensity exercise. The major difference in the master athlete is that these limitations occur at much lower metabolic demands than in their younger counterparts.

Underbuilt Conducting Airways in Females

Based on measures of lung mechanics, Jere Mead hypothesized 40 yr ago that the airway calibers of adult women were of smaller diameter than those of men even when compared at equal lung volumes: a condition that he termed “airway dysanapsis” (127). More recently, high-resolution computed tomography has demonstrated that airway cross-sectional areas are comparable between the sexes throughout maturation through pubescence, but postpubescent females show a 20–30% reduction in the diameter of the trachea and most mainstem bronchi (101, 171, 183). A smaller lung size in adult females accounts for much of these sex differences in airway size, but even when a limited number of comparisons were made at equivalent lung volumes the postpubescent and adult female still had narrowed trachea and bronchi (183). Resting diffusion capacity and lung volumes are also lower in girls and women versus boys and men, even when adjusted for age, height, and Hb concentration (141).

These sex differences in airway morphology are manifested during moderate through heavy intensity exercise:

• • Partial expiratory flow limitation with constraint of the hyperpneic response begins to occur at 40–60 L/min V̇e in healthy adult females of all ages and training status. Complete expiratory flow limitation at maximum exercise is observed in trained females at maximum V̇e much lower than in men (62, 125, 132).
• • At any given V̇e during exercise, the resistive work and oxygen cost of breathing, dyspneic sensations, and diaphragm fatigue are increased in adult females 20–70 yr of age (41, 57). When young men and women were compared at identical levels of diaphragmatic work (achieved at rest via pressure-threshold loading), diaphragmatic fatigue development was not different between the sexes (57).
• • In significant numbers of fit adult women, EIAH occurs secondary to both constrained hyperventilation and widened A-aDO₂ during heavy intensity exercise (18, 39, 41, 66, 67, 211) as it does in fit men. The difference between the sexes was that EIAH commonly occurs in women with much lower V̇o_2max (45–60 mL/kg). Pulmonary diffusion capacity and capillary blood volume are also lower in fit young women versus men during exercise at comparable cardiac outputs, and these differences were attributable entirely to lower lung volumes in females (18).
• • Preventing flow limitation via He:O₂ breathing prevents hyperinflation and provides an increased hyperventilatory response (125). Substantially reducing the work of breathing (via mechanical ventilation) redistributes blood flow and reduces limb fatigue in women as it does in men, again at lower absolute exercise intensities in women (38, 41, 42, 57).

We interpret these data to mean that adult women of all ages and over a wide range of fitness levels have hormonally determined trachea, bronchi, and lung sizes that are underbuilt for the flow rates demanded by heavy intensity exercise. Because the consequences of airway dysanapsis are likely to exist in the majority of women and be manifested even in moderately heavy exercise, we would predict that adult women across the fitness range would experience a greater susceptibility to respiratory limitations to exercise than do men.

Respiratory System Underbuilt/Maladapted for Hypoxic Environments

Over five decades ago, the Mexico City Olympics revealed the devastation that even apparently “mild” elevations [2,200 m; barometric pressure ~580 mmHg; inspired partial pressure of oxygen ($\text{P}{\mathrm{I}}_{\text{O}2}$) ~122 mmHg; arterial HbO₂ saturation ~93% at rest] rendered the endurance performances of elite sea-level athletes. Subsequent experiments have confirmed the decrements in V̇o_2max and endurance performance experienced by those athletes upon acute and especially short-term exposures to even moderate altitudes, beginning in the 600––1,300-m range (58, 60, 188). These performance decrements have been strongly linked with the following maladaptations: 1) the magnitude of exercise-induced arterial O₂ desaturation in hypoxic environments, secondary to diffusion limitation [whether these athletes did or did not experience significant EIAH in normoxia (24, 35, 58, 60, 110)]; 2) an excessive work of breathing and its effects on locomotor muscle fatigue and exertional dyspnea secondary to hyper-chemosensitivity induced by the synergistic effects on the drive to breathe of hypoxemia combined with exercise (5, 195) [in sharp contrast to these maladaptive responses to exercise in the sojourner, trained or untrained residents of high altitude have acquired a markedly reduced chemosensitivity plus an enhanced pulmonary diffusion capacity that allows them to protect their arterial Po₂ during exercise via a narrowed A-aDO₂ without excessive hyperventilation and dyspnea (23, 34, 117)]; and 3) enhanced pulmonary vascular resistance and Ppa secondary to hypoxic pulmonary vasoconstriction combined with the athlete’s high pulmonary blood flows (44, 46, 51, 140). In cases with enhanced heterogeneity of hypoxic-induced pulmonary vasoconstriction, capillary pressures during high-intensity exercise would be sufficiently high in some overperfused regions of the lung to provide excessive mechanical stress, loss of alveolar-capillary barrier integrity, and pulmonary edema, along with a worsening of arterial hypoxemia (44).

In summary, exercise performance in the highly trained athlete during sojourn to what are normally viewed as even “mild” and “moderate” high altitudes is now limited primarily by a pulmonary diffusion capacity, respiratory muscle work and aerobic capacity, neurochemical respiratory control system, and pulmonary vasculature that are underbuilt or excessively sensitive for the combined stresses of hypoxia and high-intensity exercise. In view of these multifaceted negative consequences to endurance performance at high altitude in the athletic sojourner, it appears paradoxical that such large numbers of elite athletes utilize training regimens requiring hypoxic exposure with the intent of enhancing their Hb mass and performance at sea level (131). The efficacy of this practice has recently been questioned (172, 184).

SUMMARY

Relative Importance of the Respiratory System to Exercise Limitation

In almost all healthy subjects, the evidence is clear that it is the cardiovascular system in general and maximum stroke volume specifically, together with circulating blood volume and Hb mass, that are the major gatekeepers regulating O₂ transport during exercise. The respiratory system “inadequacies” or constraints we have discussed previously (with some notable exceptions) do not occur in all healthy subjects, and when they do, they account for relatively minor limitations to performance. For example: 1) EIAH occurs to a variable extent in the highly trained subject and at most accounts for ~15% of the V̇o_2max and 15–25% of exercise-induced limb fatigue; 2) EIB occurs with high prevalence in the highly trained endurance athletes, but its effects on exercise performance are not well quantified and the underlying mechanisms for exercise limitation remain elusive; 3) EILO elicits severe dyspneic sensations, constrains ventilation and gas exchange, and presents a major limit to exercise performance; however, EILO prevalence among the highly trained subject awaits more precise characterization in larger populations of athletes across the world; 4) we cannot yet predict with certainty the effects of a limited pulmonary vascular recruitment on stroke volume or exercise performance, i.e., is maximum stroke volume really determined by the “weakest ventricle”?; and 5) positive expiratory intrathoracic pressures appear to exert a modulatory but likely relatively small constraint on maximum stroke volume.

Notable exceptions to these inconsistent influences of pulmonary system limitations include the following:

• • the dominant role of the lung’s failed gas exchange as a determinant of exercise performance in the equine thoroughbred and in the human athlete in hypoxic environments;
• • the consistent effect across healthy trained and untrained subjects of respiratory muscle work on locomotor muscle blood flow, fatigue, and endurance exercise performance carried out at high intensities;
• • aging and sex differences in lung morphology and flow-volume capacities offer another consistent yet unquantified respiratory system limitation, primarily via their effects on respiratory muscle work and its sequelae, especially in the highly trained subject.

FUTURE WORK

The uncovering of limits to the healthy respiratory system response to exercise has opened new avenues of exploration to further understand the pathophysiology of exercise performance limitation in chronic cardiovascular and respiratory diseases (6, 25, 56, 130, 137, 147, 186). Related unresolved problems that deserve in-depth scrutiny include the following:

• • longitudinal investigation of exercise training effects on the aging lung, conducted with state-of-the-art measures of lung physiology, endocrinology, and lung imaging and analogous to recently conducted training studies on aging with cardiovascular outcome measures (21, 78, 79). It is now well established that exercise training increases antioxidant defenses and anti-inflammatory mediator release from contracting muscle, capable of inhibiting inflammation and oxidative stress (53, 83). It is important that investigation of these exercise-induced mechanisms be extended to include the human lung during healthy aging and in chronic pulmonary diseases.
• • sex influences on lung morphology and pulmonary vasculature distensibility.
• • neural control of extrathoracic upper airway caliber during exercise and its contribution to EILO.
• • long-term pathophysiology of pulmonary vascular remodeling in the highly trained and especially in the master athlete.
• • potential negative impacts of long-term high-intensity training on the lung parenchyma: does injury to the alveolar capillary surface impact EIAH and explain its occurrence even during submaximal exercise?
• • quantification of influences of the respiratory versus locomotor muscle pumps on venous return and stroke volume in a setting of heavy intensity exercise.
• • comparison of respiratory muscle vs. locomotor muscle vascular responsiveness to sympathetic innervation and their effects on blood flow distribution during exercise. Also, does specific inspiratory muscle training influence blood flow distribution during exercise? If so, how?

Progress toward addressing these questions would benefit greatly from advances in methods for imaging the small airways and pulmonary vasculature and for quantifying diaphragmatic blood flow and heart volume dimensions in the human during heavy intensity exercise.

GRANTS

This work was supported in part by a grant from the National Heart, Lung, and Blood Institute (R21 HL137874 to J. Dempsey). A. L. Gerche is supported by a Future Leader Fellowship from the National Heart Foundation of Australia (102206).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.A.D. conceived and designed research; J.A.D., A.L.G., and J.H.H. prepared figures; J.A.D., A.L.G., and J.H.H. drafted manuscript; J.A.D., A.L.G., and J.H.H. edited and revised manuscript; J.A.D., A.L.G., and J.H.H. approved final version of manuscript.