Skeletal muscle fatigue has been defined as a reduction in the force- and/or velocity-generating capacity of a muscle that has been under load that is relieved by rest ^(17,18,34). Whether fatigue, as defined, occurs in the respiratory muscles as a result of heavy whole-body exercise has been debated for many decades^{(13,15,35,44)}. The debate has persisted primarily owing to a lack of a definitive test to assess respiratory muscle function. Recently, the technique of bilateral phrenic nerve stimulation(BPNS) has been applied to exercise, providing a nonsubjective, non-motivation-dependent test to determine diaphragmatic function^{(7,8,26,27,29)}. Although several other respiratory muscles are recruited with whole-body exercise (i.e., external intercostals, scalenes, and sternocleidomastoid muscles), the diaphragm is the primary inspiratory muscle and the most effective pressure generator for increasing alveolar ventilation and thus provides the best index of respiratory system muscle function. The following discussion reviews recent studies using the technique of BPNS to determine the degree to which the diaphragm fatigues with heavy whole-body exercise in healthy humans. In addition, the discussion focuses on factors that may contribute to the exercise-induced diaphragmatic fatigue and attempts to address the more difficult question of whether respiratory muscle fatigue influences human performance.

INDICES OF RESPIRATORY MUSCLE RECRUITMENT

Whether respiratory muscle fatigue occurs is dependent on the demands placed on these muscles with heavy exercise (i.e., forces produced, velocity of muscle shortening, and length changes)^(3,11,41) versus the capacities of the muscles to respond (i.e., fiber type, oxidative capacity, capillary density, and recruitment order) ^(37,38,45). Since it is difficult to measure respiratory muscle force generation directly, it is often inferred from measurements of mouth (Pm), pleural (Pe), or transdiaphragmatic(Pdi) pressures determined through noncompliant tubing placed at the mouth and esophageal and gastric balloons. Other ways of inferring the degree of respiratory muscle recruitment during exercise are related to changes in lung volume, which have been shown to reflect changes in respiratory muscle length^(12,40), and changes in airflow, which have been shown to correlate to changes in velocity of muscle shortening⁽⁴²⁾. Although indirect, these measures provide some index of the degree of respiratory muscle recruitment during exercise, particularly when expressed as a percent of the maximal values or capacities available or when expressed within a given time period. More recently, an index of respiratory muscle force production that is obtained by calculating the time integral of the pleural and transdiaphragmatic inspiratory pressure waveforms multiplied by the breathing frequency (i.e., the time integral for Pe and Pdi · fb) has been used^(26,27,29). The change in time integral of Pdi and Pe during exercise provides an index of diaphragmatic versus total respiratory system muscle recruitment.

THE TECHNIQUE OF BPNS

Bilateral phrenic nerve stimulation involves stimulation of the phrenic nerves (usually transcutaneously in the neck or through needle or wire electrodes) bilaterally behind the sternocleidomastoid muscles in the neck^(5,6,10,25). Typically, twitches with short pulses (100-150 μs) are used. However, a few studies have applied short-duration tetanic stimulation either bilaterally or unilaterally^(9,26). The reliability of the technique is critically dependent on several key elements, including supramaximal stimulation, quasi-isometric conditions, muscle length, and abdominal compliance. Studies using this technique must carefully control for each of these potential sources of error.

Application of this technique to studies examining resistive breathing or resting hyperpnea have demonstrated that the human diaphragm can be fatigued^(6,9,10); i.e., when subjects are asked to achieve a given respiratory effort over a prolonged period of time, the diaphragm can be driven to the point of fatigue. These resting, volitional-task studies, however, differ markedly from the state of whole-body exercise, when respiratory muscle recruitment and subsequently ventilation are driven by precise neurohumoral regulation of arterial blood gases balanced with the respiratory load ^(15,47).

Levine and Henson ⁽²⁷⁾ were the first to apply the technique of BPNS to exercise and demonstrated that the human diaphragm was not fatigable with short-term progressive exercise to exhaustion unless a resistive load was added. Importantly, this study demonstrated that a portion(≈50%) of normal healthy humans will drive themselves to the point of diaphragm fatigue during short-term heavy exercise without stopping“prior to the fatigue” due to an increased sensation of dyspnea,“impending respiratory muscle fatigue,” or the increased load.

BPNS AND ENDURANCE EXERCISE (DEMAND VS CAPACITY)

We recently studied a group of subjects (33 ± 3 yr) using BPNS with a range of fitness levels (˙VO₂ = 61 ± 4 ml·kg^-1·min^-1, range 40-80) prior to and immediately after exercise to exhaustion at 85 (N = 8) and 95%(N = 11) of ˙VO_2max⁽²⁶⁾. On average, subjects were able to exercise for 31 and 14 min, respectively, and ventilation averaged 126 and 149 l·min^-1, respectively, near the end of exercise (i.e., 70 and 83% of their estimated maximal voluntary ventilation [MVV]). During exercise, tidal volume increased by >2.5 times the resting values, while breathing frequency increased by >4 times resting values. Figure 1 shows the changes in tidal volume, flow rate, and inspiratory and expiratory pleural pressures relative to their respective capacities. As shown, several factors suggest a significant recruitment of the respiratory muscles. Peak inspiratory pressures over the last few minutes of exercise at 85 and 95% of ˙VO_2max averaged 58 and 67% of the available dynamic capacity for pleural pressure generation(which accounts for changes in lung volume and flow rate) and 29 and 34% of the static capacity for pleural pressure generation, respectively. For both exercise intensities, peak inspiratory Pdi averaged 60% of the available dynamic P_dimax, peak inspiratory flow rates were 8-10 times resting levels, and end inspiratory lung volume averaged >85% of total lung capacity, the latter increasing the elastic load presented to the inspiratory muscles. Finally, flow limitation was present over 28 and 42% of the tidal breath during exercise at 85 and 95% of ˙VO_2max, respectively, causing a small increase in end expiratory lung volume (EELV), which adds to the inspiratory elastic load. The lengthtension and force-velocity relationship of the diaphragm is such that as lung volume increases (muscle length shortens), the ability to produce pressure decreases. Similarly, as breathing frequency increases (velocity of muscle shortening), the ability to produce pressure also decreases. Heavy exercise requires increased pressure demands at lung volumes and flow rates where the ability to produce pressure is significantly reduced. Thus, even small increases in EELV reduce the dynamic capacity available to the inspiratory muscles, as well as add to the load presented to these muscles.

Figure 2 shows the change in the pressure-time integral for Pe and Pdi over the course of exercise at each intensity. As shown, on the onset of exercise there is a sharp rise in pressure output by the diaphragm and the total respiratory muscle system. However, during both exercise intensities, diaphragmatic pressure output begins to plateau relative to total respiratory muscle pressure output (∫Pdi/Pe). Although speculative, this may suggest that the diaphragm contributes less to total respiratory muscle pressure output as exercise is progressing. Therefore, by the end of exercise, it is estimated that the pressure output of the diaphragm is increased≥400% over resting values (for both exercise intensities), while total respiratory muscle pressure output is increased ≥500 and 700% (for exercise at 85 and 95% of ˙VO_2max, respectively).

CHANGES IN STIMULATED PRESSURES (BPNS)

Bilateral phrenic nerve stimulation performed prior to exercise at 85 and 95% of ˙VO_2max averaged 28 and 25 cm H₂O at functional residual capacity (FRC) using a single supramaximal twitch, 55 and 46 cm H₂O with 10 Hz stimulation, and 90 and 78 cm H₂O with 20 Hz stimulation, respectively. Repeat BPNS within 10 min post-exercise at 95% of˙VO_2max resulted in significant declines in Pdi at all frequencies of stimulation and using a single twitch at different lung volumes (residual volume to total lung capacity), averaging 20%. After exercise at 85%˙VO_2max, significant declines in Pdi were only noted at FRC with a single twitch and averaged 15%. Figure 3 shows the decline in stimulated pressures after exercise at 95% of ˙VO_2max at the different lung volumes and frequencies of stimulation and the time course of recovery. The fall in Pdi with BPNS after exercise at 95% of˙VO_2max ranged from 8 to 32% and, on average, did not return to near pre-exercise levels until >60 min post-exercise. Thus, heavy whole-body exercise to exhaustion in relatively fit normal subjects does result in significant diaphragmatic fatigue (i.e., a reduction in the force-generating capacity of the muscle that is relieved by rest). Since Levine and Henson ⁽²⁷⁾ did not observe diaphragmatic fatigue with short-term progressive exercise, these studies would suggest not only that diaphragm muscle recruitment and exercise intensity are important but also that exercise duration plays a role in the fatigue process (see Factors Contributing to Diaphragm Fatigue).

ROLE OF FITNESS IN DIAPHRAGM FATIGUE

Does ˙VO_2max, as an index of overall fitness, play a role in diaphragm fatigue during exercise, and is there any indication that the respiratory muscles adapt to long-term whole-body exercise training? Previous investigations have attempted to answer these questions by comparing tests of ventilatory muscle endurance between untrained and trained subjects or by performing volitional maneuvers post-exercise in these two populations^{(4,14,19,32,44)}. Using the variability of fitness levels among our subjects, BPNS provides additional insight into these interesting questions. Although the sample size is small, when subjects are separated into highly and average fit subjects (i.e.,˙VO₂ = 73 ml·kg^-1·min^-1, N = 6, vs ˙VO₂ = 49 ml·kg^-1·min^-1,N = 5), no difference was found in the degree of diaphragmatic fatigue post-exercise at 95% of ˙VO_2max. However, as shown inFigure 4, large differences were noted in peak Pdi, peak Pdi as a percent of the dynamic P_dimax, the time integral for Pdi·min^-1, and total diaphragmatic pressure output over the exercise duration. Thus, although a high level of fitness does not protect the diaphragm muscle from fatigue during heavy exercise (95% of˙VO_2max), it likely leads to a training effect on the diaphragm since, for the same degree of diaphragmatic fatigue, a substantially greater amount of diaphragmatic “work” was performed in the fitter subjects. Although not tested, it can be assumed that at matched levels of diaphragmatic power output, the more highly fit subjects would incur less diaphragmatic fatigue relative to the less fit subjects. An alternative explanation for the similar degree of fatigue between the highly fit and average fit subjects is that the diaphragm reaches a given level of fatigue and is subsequently de-recruited or not further recruited in an attempt to spare this primary inspiratory muscle. We do note a plateau in peak Pdi and in our index of diaphragmatic pressure output relative to the total ventilatory output early in both exercise sessions (see Fig. 2,∫Pdi/Pe). However, within the subjects tested, the drop in BPNS Pdi was quite variable, ranging from 8 to 32%, and in subsequent studies we noted decreases as high as 50% post-exercise ⁽⁸⁾. Thus, it seems unlikely that an upper limit of diaphragm fatigue exists and likely does not explain the similar degrees of fatigue in the highly fit versus average fit subjects.

The influence of fitness on diaphragm fatigue is further emphasized in a study by Mador et al. ⁽²⁹⁾, who demonstrated diaphragmatic fatigue using BPNS in unfit subjects after only 8 min of exercise at a work intensity that elicited, on average, only 80% of˙VO_2max. In this particular study, minute ventilation reached only 50% of the 12-s MVV, although ˙VO₂ increased to >100% of the pre-test ˙VO_2max by the end of the exercise bout⁽²⁹⁾. It is interesting, however, that despite the relatively low levels of minute ventilation in these less fit subjects, the peak ∫Pdi was similar to those achieved in our highly fit subjects, suggesting a similar degree of diaphragmatic recruitment. However, due to the greatly reduced exercise duration, the cumulative “work history” of the diaphragm was substantially less.

FACTORS CONTRIBUTING TO THE DIAPHRAGMATIC FATIGUE

Several factors demonstrated a good correlation with the amount of diaphragm fatigue observed post-heavy endurance exercise. These included the percent of ˙VO_2max subjects reached near the end of the endurance exercise runs (r² = -0.67), the peak diaphragmatic pressures produced as a percent of the dynamic Pdi_max (r² = -0.52), and the relative increase in the pressure-time product of the Pdi waveform expressed per minute (r² = -0.67 and -0.85 for 85 and 95% ˙VO_2max, respectively). These correlations would imply that diaphragmatic pressure generation (i.e., work or recruitment) plays a critical role in the fatigue process and that factors related to the exercise intensity (e.g., competition for blood flow) likely also play a key role.

ROLE OF DIAPHRAGMATIC PRESSURE PRODUCTION IN FATIGUE

To determine the role that diaphragmatic pressure generation played in the fatigue process independent of the whole-body “exercise effect,” Babcock et al. ⁽⁸⁾ had subjects mimic the pressures produced during exercise (95% ˙VO_2max) at rest for an identical time period as produced during exercise (in subjects who demonstrated exercise-induced diaphragmatic fatigue). Interestingly, these subjects demonstrated only a small borderline significant decline in stimulated pressures post-mimic. Even when subjects produced diaphragmatic pressures at rest that were 2 times those produced during heavy exercise, the extent of diaphragm fatigue did not reach that observed post-heavy exercise.Figure 5 shows the relationship between diaphragm pressures produced during exercise relative to those produced with mimicking and the degree of diaphragmatic fatigue.

The results of this study would imply that factors related to exercise, in part independent of diaphragmatic pressure production, play a major role in the amount of diaphragmatic fatigue produced during exercise or that diaphragm fatigue is dependent on an interaction of several factors. Likely contributors to diaphragm fatigue that are independent of the work history are metabolites produced in locomotor muscles that may remain present in the arterial blood or competition for blood flow between locomotor muscles and the respiratory muscles. The latter might increase the metabolites produced by the diaphragm and/or locomotor muscles, such as lactic acid, hydrogen ions, or oxygen free radicals, or may exert an independent effect on diaphragm function^(39,46). Changes in the muscle environment independent of diaphragmatic pressure production may contribute to the degree of diaphragm fatigue. It has been shown in animal studies that acidosis alone will depress diaphragm function independent of work history⁽²⁰⁾. However, when Babcock et al.⁽⁸⁾ subsequently stimulated a nonexercised muscle (i.e., the first dorsal interosseous muscle of the hand) post-heavy whole-body exercise, no changes in stimulated forces were observed, suggesting that circulating metabolites alone may not contribute to the fatigue, at least not in an inactive muscle. Previous work by Fregosi and Dempsey^(21,22) had implied a progressive increase in lactate accumulation of the diaphragm with heavy exercise in rats; thus, it may be possible that the diaphragm becomes progressively acidic with exercise, contributing to the fatigue ⁽⁸⁾.

The alternative explanation for the apparent work-independent diaphragmatic fatigue may be that at high levels of metabolic demand cardiac output is limited and respiratory muscle blood flow requirements begin to compete with the locomotor muscles in a manner similar to adding arm muscle to leg muscle work. It is unlikely that cardiac output is sufficient to supply all the active vascular beds with blood flow, and previous studies have demonstrated a fall in muscle force output that parallels a reduction in blood flow^(43,46). Thus, it is possible that blood flow is not available for adequate oxygen delivery to the diaphragm or for removal of metabolic by-products during high-intensity whole-body exercise. Aaron et al.⁽²⁾ have shown that the ˙VO₂ of the respiratory muscles may approach 10-15% of total body ˙VO₂ with heavy exercise, which implies that a substantial portion of the total cardiac output is required by the respiratory muscles.

IMPLICATIONS FOR HUMAN PERFORMANCE

What are the implications of a 15-20% reduction in stimulated diaphragmatic pressure observed 10-60 min post-exercise? There are two potential ways respiratory muscle fatigue may limit human performance, through an inadequate ventilatory response to exercise (i.e., alveolar hypoventilation) or by an increased “sensation” of dyspnea. Alveolar hypoventilation may occur as a result of the respiratory muscles not being able to generate the needed pressures or when an altered breathing pattern, such as the tachypneic pattern sometimes associated with respiratory muscle fatigue, occurs. An increased sensation of dyspnea could occur as a result of an elevated pressure demand relative to the available pressure-generating capacity, circulating metabolites or metabolites produced within the diaphragm stimulating sensitive receptors, an altered breathing pattern (i.e., increased lung volume causing hyperinflation and increasing the elastic load), or altered respiratory muscle recruitment or motor unit recruitment within a given respiratory muscle. Thus, respiratory (diaphragm) fatigue could occur and influence performance with or without an effect on alveolar ventilation. Similarly, an altered alveolar ventilation or an increased sensation of dyspnea could occur independent of respiratory muscle fatigue.

In the previously cited studies documenting diaphragm fatigue, ventilation was generally appropriate for the given metabolic demand (i.e., Pet CO₂ values were ≤40 mm Hg, oxygen saturation near resting values). Thus, it is unlikely that an alveolar hypoventilation contributed to limiting exercise. Similarly, studies that have fatigued the respiratory muscles prior to performance of whole-body exercise have not observed an inadequate ventilatory response but typically no effect, decreased performance times, or altered breathing patterns ^{(16,28,31,33)}. On the other hand, as previously noted, we do have evidence suggesting that respiratory muscle recruitment may be altered. During exercise at both 85 and 95% of ˙VO_2max, diaphragm pressure production plateaued (seeFig. 2) after ≈6-10 min of exercise, while total minute ventilation and respiratory muscle pressure output continued to rise. This suggests that the diaphragm may have contributed less to total respiratory system force generation over time. Whether this is the result of the observed diaphragm fatigue or simply a normal muscle recruitment strategy by the respiratory system remains to be determined ⁽²⁴⁾.

UNLOADING THE RESPIRATORY MUSCLES

To date, the best technique to determine whether respiratory muscle fatigue influences human performance is to unload the respiratory muscles and observe whether performance improves. If no effect is observed, it would be concluded that the degree of diaphragm fatigue observed was not sufficient to alter performance. On the other hand, if an effect was observed, it may suggest that respiratory muscle fatigue had an influence on performance; however, it would be difficult to differentiate the respiratory “load” or the“sensation of dyspnea” from an effect owing purely to fatigue. To date, studies have used either reduced viscosity gases (helium) or pressure-assist ventilators to unload the respiratory muscles. Unloading the respiratory muscles could theoretically have three potential effects: 1) decreasing the work of breathing (i.e., reducing the reliance on fatigable motor units and reducing the metabolites produced), 2) decreasing blood flow demands and competition with the locomotor muscles, and 3) decreasing the sensation of dyspnea. The result of unloading the respiratory muscles would be an increase in endurance time and a decrease in whole-body ˙VO₂ secondary to the drop in respiratory muscle ˙VO₂. However, with very-high-intensity exercise, it is possible that with unloading there may be an increase in blood flow to the locomotor muscles, resulting in no or minimal change in whole-body ˙VO₂. If the respiratory muscles begin to extract a larger proportion of the total cardiac output as ventilatory demands increase with heavy exercise, one would expect a nonlinear rise in measured leg blood flow. However, a recent study by Poole et al.⁽³⁶⁾ did not observe a divergence from linearity in leg blood flow determined through the thermodilution technique as exercise intensity increased. However, measurements were not made sequentially at different workloads near maximal exercise, and there was a trend for the leg blood flow to have a reduced slope relative to whole-body ˙VO₂. Clearly, additional studies that load and unload the respiratory muscles need to be performed to determine the influence on locomotor muscle blood flow.

Several previous studies have examined the influence of helium/oxygen breathing on human performance ^(1,48). Aaron et al.⁽¹⁾ tested the influence of helium breathing (79% He, 21% O₂) on a group of rowers (peak ˙VO₂ = 5.01 l·min^-1, ˙V_E = 152 l·min^-1) at 80 and 90-95% of ˙VO_2max to exhaustion. In this particular study (as shown in Table 1 and Fig. 6), exercise time and whole-body˙VO₂ did not differ significantly on or off helium at the lower work intensity; however, at the higher work intensity, exercise time increased by 40% while breathing helium. Ventilation at isotime periods (near the end of exercise, air breathing) were not different between helium and ambient air breathing, but peak ventilation was higher on helium. ˙VO₂ near the end of exercise was 10% lower breathing helium. Time to exhaustion on room air during repeat trials combining the 80 and 90-95% ˙VO_2max workloads were not significantly different (23.9 ± 4.8 min vs 24.4 ± 4.8 min, P > 0.3, coefficient of variation = ±1.1 min or±4.8%, correlation coefficient = 0.98). These results would suggest that respiratory muscle fatigue and/or the respiratory load played a significant role in limiting human performance at exercise intensities>90-95% of ˙VO_2max. Interestingly, in the study of Aaron et al.⁽¹⁾, maximal inspiratory pressures were not affected post-exercise at 90-95% of ˙VO_2max on room air or breathing helium, but maximal sustainable ventilations were reduced substantially to 38 and 41% of pre-exercise levels, respectively.

In contrast, recent studies by Gallagher and Younes⁽²³⁾ and Marciniuk et al. ⁽³⁰⁾ have used a different form of unloading, i.e., pressure assist. This device acts as a demand regulator, sensing changes in mouth pressure and responding by partially unloading the respiratory muscles. In these studies, performed at up to 85% of ˙VO_2max, no significant differences were observed in exercise time, whole-body ˙VO₂, ventilation, or perception of effort with the unloading, implying respiratory fatigue or that the load plays no role in overall performance.

Why the differences? Aaron et al. ⁽¹⁾, using helium, did not find a difference with exercise intensities <90% of˙VO_2max but only near maximal exercise intensities, implying a potential intensity-related difference. Helium has effects other than simply unloading the respiratory muscles. In “responders,” helium increases the flow-volume loop, likely increasing the breathing reserve and potential ventilatory capacity. In addition, helium may affect ventilation distribution, increasing alveolar ventilation or decreasing the alveolar to arterial oxygen difference (all of which improve arterial oxygenation, which is often found to decrease in trained endurance athletes), or it may simply alter the perception of load and not actually prevent respiratory muscle fatigue. Another concern has been that helium may act by unloading the external respiratory apparatus (tubing, pneumotachs, turbines, tissot, mixing chamber, mouthpiece, etc.), which often provides a small but significant resistance to breathing, rather than by having its major influence on the lungs and upper airways.

Similarly, the pressure-assist device may also alter the normal breathing mechanics. The pressure-assist device unloads by adding a positive pressure rather than a negative pressure on inspiration. Small increases in respiratory system resistance have been described with this type of unloading⁽²³⁾. There was also a trend toward a small increase in exercise time and a reduced ˙VO₂ in the unloading study performed at 80-85% of ˙VO_2max^(23,30). Thus, to date it appears that at workloads below 85% of ˙VO_2max, respiratory fatigue likely does not limit human performance; however, at the extremes of human performance, it remains undetermined. Clearly, when the degree of diaphragm fatigue is considered relative to available capacity, peak force production is encroaching on the capacity for pressure generation at the higher work intensities (see Fig. 7). Undoubtedly, heavy prolonged exercise in the heat or at altitude ⁽⁷⁾ that increases the competition for blood flow between vascular beds (i.e., skin blood flow) and increases the work of breathing further challenges the ability of the cardiovascular system to deliver blood flow to the heavily recruited respiratory muscle system.

Accordingly, over the last several years it has become clear that the diaphragm, like locomotor muscles, is susceptible to fatigue with heavy whole-body dynamic exercise. A combination of factors influence the degree of fatigue, including the exercise intensity, degree of diaphragm pressure production, exercise duration, and likely level of fitness. Whether the observed fatigue influences exercise performance appears doubtful except near the extreme levels of human performance.

To date, only changes in muscle force production have been tested using the technique of BPNS. It is becoming increasingly appreciated that other changes may occur in conjunction with the decreased ability to produce force, such as a change in the velocity of muscle shortening or in the ability to shorten under load. This may be particularly important during exercise that requires high velocities of muscle shortening (i.e., increased flow rates) in parallel with the demands for increased pressure production and shortening. Future studies will be necessary to determine the role these factors may play in the overall fatigue process and diaphragmatic function during whole-body dynamic exercise.