1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

The work of breathing required during maximal exercise compromises blood flow to limb locomotor muscles and reduces exercise performance. We asked if force output of the inspiratory muscles affected exercise-induced peripheral fatigue of locomotor muscles. Eight male cyclists exercised at ≥ 90% peak O2 uptake to exhaustion (CTRL). On a separate occasion, subjects exercised for the same duration and power output as CTRL (13.2 ± 0.9 min, 292 W), but force output of the inspiratory muscles was reduced (−56%versus CTRL) using a proportional assist ventilator (PAV). Subjects also exercised to exhaustion (7.9 ± 0.6 min, 292 W) while force output of the inspiratory muscles was increased (+80%versus CTRL) via inspiratory resistive loads (IRLs), and again for the same duration and power output with breathing unimpeded (IRL-CTRL). Quadriceps twitch force (Qtw), in response to supramaximal paired magnetic stimuli of the femoral nerve (1–100 Hz), was assessed pre- and at 2.5 through to 70 min postexercise. Immediately after CTRL exercise, Qtw was reduced −28 ± 5% below pre-exercise baseline and this reduction was attenuated following PAV exercise (−20 ± 5%; P < 0.05). Conversely, increasing the force output of the inspiratory muscles (IRL) exacerbated exercise-induced quadriceps muscle fatigue (Qtw=−12 ± 8% IRL-CTRL versus−20 ± 7% IRL; P < 0.05). Repeat studies between days showed that the effects of exercise per se, and of superimposed inspiratory muscle loading on quadriceps fatigue were highly reproducible. In conclusion, peripheral fatigue of locomotor muscles resulting from high-intensity sustained exercise is, in part, due to the accompanying high levels of respiratory muscle work.

During maximal exercise, the respiratory muscles require ∼10% of the total oxygen consumption inline image in moderately fit subjects and up to 15% in highly fit subjects (Aaron et al. 1992). Furthermore, a substantial portion of the cardiac output (up to 14–16%) is directed to the respiratory muscles to support their metabolic requirements (Manohar, 1986; Harms et al. 1998b). This high demand for respiratory muscle blood flow compromises blood flow to working limb locomotor muscles because of sympathetically mediated vasoconstriction (Harms et al. 1997). Thus, when the work of breathing is reduced by over 50% using a mechanical ventilator there is a commensurate increase in limb blood flow and vascular conductance (Harms et al. 1997). Conversely, when the work of breathing is increased by a comparable amount using inspiratory resistors, limb blood flow and vascular conductance fall (Harms et al. 1997).

Mechanical unloading of the respiratory muscles during heavy sustained exercise inline image prevented diaphragm fatigue (Babcock et al. 2002) and resulted in a 14% improvement in time to exhaustion in trained male cyclists with a reduction in inline image and the rate of rise in perceptions of respiratory and limb discomfort (Harms et al. 2000). Collectively, these findings indicate that the work of breathing normally encountered during sustained heavy exercise has a significant influence on exercise capacity. However, the reasons for this influence are complex because both the intensity of effort perceptions and local O2 transport to working limb muscle are affected. Limb muscle force output is responsive to even small changes in limb muscle blood flow under conditions of high intensities of muscle contraction (Barclay, 1986; Rabischong & Guiraud, 1993; Hogan et al. 1998; Cole & Brown, 2000). Thus, the purpose of the present study was to determine the effect of inspiratory muscle work on exercise-induced quadriceps muscle fatigue which, for the purpose of the present study, was defined as a loss of ability to produce force or power with a muscle which is reversible by rest (Bigland-Ritchie & Woods, 1984; NHLBI Workshop, 1990). Muscle fatigue has been divided into two components, peripheral fatigue and central fatigue (Bigland-Ritchie et al. 1978; Gandevia, 2001). In the present study, we were primarily concerned with changes in peripheral muscle fatigue; that is, the loss of force due to processes occurring at or distal to the neuromuscular junction. Thus, we hypothesised that partial unloading of the inspiratory muscles would attenuate exercise-induced peripheral fatigue of the quadriceps muscle and, conversely, that inspiratory muscle loading would exacerbate peripheral fatigue of this muscle.

Some of the results of this study have been reported previously in the form of an abstract (Romer et al. 2005).


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix


Eight endurance-trained male cyclists, with resting pulmonary function within normal limits, volunteered to participate in the study. Informed consent was obtained in writing from each subject and the Institutional Review Board of the University of Wisconsin-Madison approved all procedures. The study conformed to the standards set by the Declaration of Helsinki. Descriptive characteristics of the subjects (mean ±s.e.m.) were: age 25.1 ± 1.4 years, body mass 71.6 ± 3.3 kg, peak O2 uptake inline image.


At a preliminary visit to the laboratory subjects were thoroughly familiarized with the procedures used to assess quadriceps muscle function and performed a maximal incremental exercise test (33 W every 3 min starting from 98 W) on an electromagnetically braked cycle ergometer (Elema, Sweden) for the determination of peak power output inline image. On a separate occasion, subjects performed a 5 min warm-up at inline image followed by a step-wise increase in power output inline image that was sustained until they were unable to maintain a pedal cadence above 60 r.p.m. despite strong verbal encouragement (CTRL). At a separate visit, subjects repeated the constant-load exercise, at the same intensity and for the same duration as in CTRL, but the force output of the inspiratory muscles was reduced using a proportional assist ventilator (PAV) (see below). Subjects also exercised to exhaustion on two separate occasions while the force output of the inspiratory muscles was increased via inspiratory resistive loads (IRL) (see below). Two additional exercise tests were performed whereby subjects exercised at the same power output and for the same duration as in IRL, but with breathing unimpeded (IRL-CTRL). The purpose of the two IRL and the two IRL-CTRL trials was to determine the between-day reliability of the effects of increasing respiratory muscle work on limb fatigue. The order of the trials was randomized and counterbalanced. The participants could not be blinded to the treatments, but they were unaware of the experimental hypotheses and naive to the purpose of the study. Contractile function and membrane excitability of the quadriceps muscle in response to supramaximal magnetic stimulation of the femoral nerve were assessed before and up to 70 min after each of the exercise trials (see below). Each exercise session was separated by at least 48 h and was completed at the same time of day. Subjects refrained from caffeine for 12 h and stressful exercise for 48 h before each exercise test. Ambient temperature and relative humidity were not different between conditions.

Exercise responses

Ventilation and pulmonary gas exchange were measured breath-by-breath using apparatus and techniques that have been previously described (Harms et al. 1998a). Arterial O2 saturation was estimated using a pulse oximeter (SpO2, Nellcor N-595, Pleasanton, CA, USA) with adhesive forehead sensors (Nellcor Max-Fast). Heart rate (HR) was measured from the R–R interval of an electrocardiogram using a three-lead arrangement. Ratings of perceived exertion (dyspnoea and limb discomfort) were obtained every 2 min using Borg's modified CR10 scale (Borg, 1998). Capillary blood samples were collected from an earlobe for subsequent determination of total whole blood lactate concentration ([La]B) using an electrochemical analyser (YSI 1500 Sport, OH, USA). We have previously shown a close correspondence between measures of lactate concentration in arterial blood and those in capillary blood when obtained simultaneously during heavy exercise (Romer et al. 2006). During the CTRL trial, and during the PAV and one of the IRL trials, gastric (Pga) and oesophageal (Poe) pressures were measured via nasopharyngeal balloons using standard procedures (Baydur et al. 1982). Transdiaphragmatic pressure (Pdi) was obtained online by subtracting Poe from Pga. Poe and Pdi were integrated over the period of inspiratory flow and the results were multiplied by respiratory frequency (fR) and labelled the inspiratory muscle pressure–time product and the diaphragm pressure–time product, respectively (see Fig. 1).


Figure 1. Representative within-breath pressure–time traces for one subject ensemble averaged over 1 min for CTRL, IRL and PAV at exercise isotime. Poe, oesophageal pressure; Pdi, transdiaphragmatic pressure. Pressure–time products for the inspiratory muscles (A) and for the diaphragm (B) were calculated by integrating the respective pressure over the period of inspiratory flow and then multiplying the integral by the respiratory frequency.

Download figure to PowerPoint

Inspiratory muscle unloading and loading

A feedback-controlled mechanical ventilator with a proportional assist ventilation (PAV) mode was used to reduce the work of the inspiratory muscles during exercise (Younes, 1992). Briefly, subjects breathed through a two-way low-resistance non-rebreathing valve (7200 series, Hans Rudolph) that was connected on the inspiratory side to the ventilator and on the expiratory side to a solenoid switch. During inspiration, the solenoid switch closes and the ventilator delivers positive pressure in proportion to flow and volume. During expiration, the solenoid switch opens and allows for unimpeded expiration with no pressure delivery. The amount of assist was set at the maximum level that each subject could tolerate (∼2.5 cmH2O l−1 s−1 for flow assist, ∼3.0 cmH2O l−1 for volume assist). Inspiratory work was increased by adding mesh screens to the inspiratory line with resistances of 3–7 cmH2O l−1 s−1. A pilot study indicated that mild hypoxaemia occurred during IRL in some subjects compared with unimpeded breathing. We have shown previously that even mild hypoxaemia can affect the magnitude of quadriceps muscle fatigue (Romeret al. 2005a); therefore, inspired air during the IRL trials was supplemented with mild hyperoxic gas mixtures such that Smath image did not differ from CTRL values. The PAV and inspiratory resistance settings were chosen to cause decreases and increases in inspiratory muscle work similar in magnitude to our previous studies, which had shown significant effects of changing respiratory muscle work on limb blood flow and exercise performance (Harms et al. 1997, 1998b, 2000). Subjects participated in practice sessions to familiarise themselves with the inspiratory unloading and loading.

Contractile function

Electromyography Quadriceps electromyograms (EMGs) were recorded from three pairs of skin surface electrodes (Kendall H59P, Mansfield, MA, USA) positioned over the vastus lateralis, rectus femoris and vastus medialis. When the femoral nerve was stimulated, quadriceps compound muscle action potentials (M-waves) were captured on paper using a multichannel chart recorder. Membrane excitability was inferred from the peak-to-peak amplitude of the M-waves. There was no effect of the recording site on the relative changes in M-wave amplitudes; therefore, data for all three muscles were pooled. The position of the EMG electrodes was marked with indelible ink to ensure that they were placed in the same location at subsequent visits.

Magnetic stimulation Subjects lay supine on a table with the hip angle set at ∼3.14 rads (180 deg), the right knee joint angle at 1.57 rads (90 deg) of flexion and the arms folded across the chest. A non-compliant strap was attached around the subject's right leg just superior to the malleoli of the ankle joint. The strap was connected to a load cell (Interface, Model SM 1000, Scottsdale, AZ, USA) that was calibrated after each test with weights of known amounts. Care was taken to ensure that the knee angle did not change, the ankle strap and load cell were parallel to the floor, and the ankle strap position remained constant throughout the experiment. Two magnetic stimulators (Magstim 200, Jali Medical Inc., Newton, MA, USA), connected to a transformer (TwinCap Module, Jali Medical Inc.) and a double 40 mm coil, were used to stimulate the femoral nerve (Polkey et al. 1996; Kufel et al. 2002). We used single and paired stimuli to help discriminate between low and high frequency fatigue (Yan et al. 1993; Polkey et al. 1997). For the paired stimuli the two stimulators were synchronised by a separate module (BiStim Module, Jali Medical Inc.).

The area of stimulation associated with the largest Qtw and M-wave amplitudes was located by positioning the coil head high in the femoral triangle just lateral to the femoral artery (Polkey et al. 1996). This position was marked with indelible ink to ensure that the coil was placed at the same location for the remainder of the study. To determine whether nerve stimulation was supramaximal, three single twitches were obtained every 30 s at 50, 60, 70, 80, 85, 90, 95 and 100% of maximal power output for one of the stimulators at the start of every experiment. A near plateau in baseline Qtw and M-wave amplitudes with increasing stimulus intensities was observed in every subject, indicating maximal activation of the femoral nerve. Specifically, the mean changes in Qtw across all conditions between 85 and 90, 90 and 95, and 95 and 100% of maximum stimulator power output were 2.0 ± 0.5, 1.3 ± 0.3 and 0.6 ± 0.2%, respectively. The mean changes in M-wave amplitude (mean for 3 muscles) between 85 and 90, 90 and 95, and 95 and 100% of maximum stimulator power output were 2.2 ± 0.7, 1.4 ± 0.5, and 0.7 ± 0.2%, respectively. Twitch force at 100% of the stimulator's maximal power output measured at the beginning of the progressive increase in power output was not different from that obtained at the end, indicating that the protocol did not elicit twitch potentiation.

Assessment of fatigue Peripheral fatigue and central fatigue were assessed based on measurements of contractile function and voluntary activation, respectively. Subjects rested for at least 10 min, after which stimulus power was set at 100% of maximum and paired stimuli were given at interstimulus intervals of 10, 20 and 100 ms, corresponding to stimulation frequencies of 100, 50 and 10 Hz, respectively. Paired stimuli were separated by 30 s and were repeated four times each. Next, eight single stimuli were given, each separated by 30 s. The potentiated quadriceps twitch (Qtw,pot) is more sensitive for detecting fatigue than is the non-potentiated twitch (Qtw,np), particularly when the degree of fatigue is small (Kufel et al. 2002). Accordingly, we measured quadriceps twitch force 5 s after a 5 s maximal voluntary contraction (MVC) of the quadriceps and repeated this procedure six times such that six Qtw,pot were obtained. Like others (Polkey et al. 1996; Kufel et al. 2002), we found that the degree of potentiation was slightly smaller after the first and, to a lesser extent, after the second MVC; therefore, we discarded the first two measurements. Activation of the quadriceps during the MVC manoeuvre was assessed using a superimposed twitch technique (Strojnik & Komi, 1998). Briefly, the force produced during a superimposed single twitch on the MVC was compared to the force produced by the potentiated single twitch delivered 5 s afterwards (Strojnik & Komi, 1998). A correction was applied to the superimposed twitch because it did not always occur at maximal volitional force (Strojnik & Komi, 1998). A decline in voluntary activation after exercise is indicative of a reduction in central drive to the muscle (i.e. central fatigue). The entire assessment procedure took 15 min to complete and was performed before exercise (∼30 min) and at 2.5, 35 and 70 min after exercise. For all conditions, the order of stimulation frequencies was the same pre- versus postexercise.

Reliability At separate visits to the laboratory, subjects were removed from the testing apparatus after baseline measurements of muscle function had been obtained and rested in a chair for 30 min without contracting the quadriceps, after which they were reattached to the testing apparatus and measurements of quadriceps muscle function were repeated. There was no systematic bias in the baseline measurements either within or between days. Within-day coefficients of variation were 3.8% for peak quadriceps twitch force (Qtw,peak) and 4.7% for the force during the second twitch (Qtw,T2; mean for all frequencies), 9.7% for MVC force, 4.7% for voluntary activation and 4.1% for M-wave amplitude (mean for 3 muscles averaged over all frequencies of stimulation). Between-day coefficients of variation were 5.5% for Qtw,peak, 6.5% for Qtw,T2, 10.5% for MVC and 7.2% for voluntary activation.

Analysis of twitch data The eight non-potentiated single Qtw responses were averaged over 1 s and then digitally subtracted, using a computer, from the averaged paired response at each stimulation frequency (Yan et al. 1993; Polkey et al. 1997). The amplitude of the resultant response (T2) was measured from baseline to peak. Accurate temporal alignment of the force signals was achieved using the marker signal produced by the discharge of the first magnetic stimulator. The within-twitch responses to each single supramaximal 1 Hz stimulus (non-potentiated and potentiated) and to each of the 100 Hz paired stimuli were analysed for peak force, contraction time, maximal rate of force development, one-half relaxation time, and maximal relaxation rate (Lepers et al. 2002).

Statistical analyses

Repeated measures ANOVA was used to test for within-group effects across time. Following significant main effects, planned pairwise comparisons were made using the Bonferroni method. Results are expressed as the mean ± standard error of the mean (s.e.m.). Statistical significance was set at P < 0.05. Statistical analyses were performed using the 11.5 release version of SPSS for Windows (SPSS Inc., Chicago, IL, USA).


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

Inspiratory muscle force output and oxygen uptake

Group mean values are shown for throughout exercise in Fig. 2 and for the final minute of constant-load exercise in Table 1. Subjects cycled at the same power output (292 ± 13 W) and for the same duration (13.2 ± 0.9 min) under CTRL and inspiratory muscle unloading (PAV) conditions. Despite identical power outputs, the time to volitional exhaustion was less during inspiratory muscle loading (IRL) than CTRL (7.9 ± 0.6 versus 13.2 ± 0.9 min, respectively; P < 0.01), which represented a 39 ± 6% (range 22–49%) decrease in exercise time. Throughout the entire PAV exercise period the pressure–time product for both the diaphragm and the inspiratory muscles was substantially reduced relative to control (P < 0.01) (−37 ± 7% and −56 ± 6%, respectively). Conversely, throughout exercise with IRL, pressure–time products for both the diaphragm and the inspiratory muscles were increased above control (51 ± 10% and 80 ± 9%, respectively). inline image during CTRL exercise rose throughout the exercise period and terminated at inline image during the final minute (Table 1). Reducing the pressure–time product of the inspiratory muscles via PAV resulted in a 8.2 ± 2.1% decrease in whole-body inline image when comparisons were made versus CTRL at the same duration (i.e. final minute) of exercise. Increasing the pressure–time product of the inspiratory muscles via IRL resulted in a 5.7 ± 2.6% increase in inline image above control.


Figure 2. Group mean change for diaphragm pressure–time product (iPdi×fR) and inspiratory muscle pressure–time product (Poe×fR) at baseline (Pre), during the fifth minute of the warm-up (WU) at 40% peak power output inline image and every minute thereafter at inline image for CTRL versus PAV (A and B) and Inspiratory Resistive Load (IRL) versus IRL-CTRL (C and D). Both measures of respiratory force output were reduced during PAV versus CTRL and increased during IRL versus IRL-CTRL. Values are mean ±s.e.m.n= 8 per group. **P < 0.01, significantly different at the same time.

Download figure to PowerPoint

Table 1.  Response to final minute of constant-load exercise for control (CTRL), inspiratory muscle unloading (PAV), inspiratory resistive loading (IRL) and inspiratory resistive loading-control (IRL-CTRL)
  1. Mean ±s.e.m.n= 8 per group. *P < 0.05; **P < 0.01, significantly different from CTRL. †P < 0.05; ††P < 0.01, significantly different from IRL. The IRL and IRL-CTRL values are the mean of two trials performed on different days. Poe×fR, pressure–time product for the inspiratory muscles; Pdi×fR, pressure–time product for the diaphragm; SpO2, arterial O2 saturation via pulse oximetry; HR, heart rate; TiTtot, inspiratory duty cycle; fR, respiratory frequency; VT, tidal volume; inline image, minute ventilation; inline image uptake; inline image production; [La]B, blood lactate concentration; RPE, ratings of perceived exertion.

Power output (W)292 ± 13    292 ± 13   
Exercise time (min)13.2 ± 0.9    7.9 ± 0.6**
Poe×fR ((cmH2O s) × min)648 ± 48  285 ± 37**1131 ± 58  636 ± 44††
Pdi×fR ((cmH2O s) × min)844 ± 53  534 ± 55**1252 ± 57  845 ± 48††
SpO2 (%)93.1 ± 1.094.7 ± 0.793.0 ± 1.393.7 ± 0.7 
HR (beats min−1)180 ± 4 178 ± 3 175 ± 4 176 ± 3  
Ti/Ttot 0.47 ± 0.01 0.43 ± 0.02 0.56 ± 0.01  0.48 ± 0.01††
fR (breaths min−1)52.6 ± 3.6 43.3 ± 3.3*44.3 ± 2.644.1 ± 3.1 
VT (l) 2.88 ± 0.15   3.77 ± 0.23** 2.92 ± 0.17  3.25 ± 0.15††
inline image148.4 ± 5.9 159.4 ± 8.8 126.3 ± 4.2 140.8 ± 7.0† 
inline image 4.00 ± 0.14   3.66 ± 0.13** 3.90 ± 0.11 3.69 ± 0.13†
inline image96 ± 2  88 ± 1**94 ± 288 ± 2†
inline image37.2 ± 1.3 43.8 ± 2.7*33.6 ± 1.036.2 ± 1.5 
inline image37.5 ± 1.140.7 ± 1.9 30.4 ± 0.632.8 ± 1.2†
[La]B (mm)10.0 ± 1.1 9.2 ± 0.6 9.0 ± 0.79.2 ± 0.7
RPE (dyspnoea)10.0 ± 0.3  8.4 ± 0.6* 9.8 ± 0.3  7.9 ± 0.5††
RPE (limb)10.2 ± 0.1  9.0 ± 0.4* 8.8 ± 0.4 7.9 ± 0.5†

Contractile function

CTRL versus inspiratory muscle unloading (PAV): 292 ± 13 W, 13.2 ± 0.9 min Quadriceps EMG M-wave amplitude did not differ pre- versus postexercise for either CTRL or PAV. Immediately after CTRL exercise, Qtw,peak and Qtw,T2 were reduced at all stimulation frequencies compared with pre-exercise baseline values (mean decrease for all stimulation frequencies =−28%; P < 0.01) and remained reduced up to 70 min postexercise (Fig. 3A and B). Significant reductions were also noted for non-potentiated and potentiated single twitches (−30 ± 3 and −38 ± 4%, respectively), MVC (−28 ± 9%) and voluntary activation (−8 ± 4%).


Figure 3. Group mean peak quadriceps twitch force (Qtw,peak) and the force of the second twitch (Qtw,T2) in response to supramaximal stimuli at 1 Hz (single twitch), 10 Hz (100 ms interstimulus duration), 50 Hz (20 ms) and 100 Hz (10 ms) pre- and up to 70 min postexercise for CTRL (A and B) and IRL-CTRL (C and D). Exercise was 292 ± 13 W for 13.2 ± 0.9 min in CTRL and 292 ± 13 W for 7.9 ± 0.6 min in IRL-CTRL. In CTRL, the Qtw,peak response was significantly reduced at all frequencies and times after exercise; immediately postexercise the Qtw,T2 response was significantly reduced at all stimulation frequencies; at 35 min postexercise the Qtw,T2 response at 10 Hz was less than at 50 and 100 Hz. In IRL-CTRL, Qtw,T2 (mean for all frequencies) was reduced immediately after versus before exercise. n= 8 per group. Values are mean ±s.e.m.

Download figure to PowerPoint

The effect of partially unloading the inspiratory muscles (via PAV) on contractile function is summarised in Table 2 and Figs 4A and 5. With PAV, the group mean decrease in Qtw,T2 (mean for all frequencies) immediately postexercise was attenuated by almost one-third (−28 versus−20% for CTRL versus PAV, respectively, P < 0.01; Fig. 5A) and this effect of PAV occurred in all eight subjects. The difference between CTRL and PAV exercise conditions in the percentage reduction in Qtw,T2 from baseline was statistically significant at 1, 50 and 100 Hz (Fig. 4). PAV also attenuated the pre- to postexercise reduction in Qtw,peak (mean for all frequencies) (−28 versus−23% for CTRL and PAV, respectively; P < 0.01), Qtw,peak (1 Hz,np) (−30 versus−20%; P < 0.05) and Qtw,peak (1 Hz,pot) (−38 versus−33%, P < 0.05). The effect of PAV on attenuating the reduction in contractile function persisted up to 70 min postexercise for Qtw,peak (1 Hz,np and the mean for all frequencies) and for Qtw,T2 (mean for all frequencies). There was no effect of PAV on either MVC or voluntary activation at the end of exercise.

Table 2.  Group mean changes in contractile function measurements from baseline at 2.5, 35 and 70 min postexercise for control (CTRL) and inspiratory muscle unloading (PAV)
Baseline2.5 min35 min70 minBaseline2.5 min35 min70 min
  1. Mean ±s.e.m.n= 8 per group. *P < 0.05; **P < 0.01, significantly different from baseline value. †P < 0.05; ††P < 0.01, significantly different from CTRL at corresponding time. Values in parentheses represent percentage changes from baseline values. Qtw,peak, peak quadriceps twitch force; np, non-potentiated; pot, potentiated; Qtw,T2, peak quadriceps twitch force for the T2 response; MVC, highest force averaged over 1 s during maximal voluntary contraction; Activation, level of voluntary drive during MVC measured using twitch interpolation.

 1 Hz,np (N)152 ± 13105 ± 7** 124 ± 9** 133 ± 10**145 ± 14108 ± 10**125 ± 12*136 ± 12 
 (%Δ) (−30 ± 3)**(−18 ± 3)**(−12 ± 2)** (−25 ± 3)**†(−13 ± 3)**†(−6 ± 3)†
 1 Hz,pot (N)204 ± 17124 ± 9**  162 ± 13** 175 ± 13**199 ± 18131 ± 9** 166 ± 13**181 ± 15*
 (%Δ) (−38 ± 4)**(−20 ± 2)**(−13 ± 2)** (−33 ± 3)**†(−16 ± 3)** (−8 ± 3)*
 Mean of 1, 10, 50, 100 Hz (N)221 ± 16 155 ± 11**  183 ± 13** 193 ± 14**211 ± 17162 ± 15** 186 ± 17**201 ± 17 
 (%Δ) (−28 ± 4)**(−16 ± 3)**(−12 ± 2)** (−23 ± 4)**††(−12 ± 3)**(−5 ± 3)†
 Mean of 1, 10, 50 100 Hz (N)137 ± 9  97 ± 8** 112 ± 6**120 ± 9**130 ± 10104 ± 11*117 ± 11*128 ± 11 
 (%Δ) (−28 ± 5)**(−17 ± 4)**(−12 ± 3)** (−20 ± 5)**††(−10 ± 3)* (−2 ± 3)†
 10 Hz/100 Hz 1.08 ± 0.06 1.00 ± 0.070.92 ± 0.05** 0.99 ± 0.051.07 ± 0.060.98 ± 0.080.91 ± 0.05**1.00 ± 0.04
 (%Δ) (−7 ± 5) (−14 ± 3)**(−7 ± 3)  (−9 ± 5) (−15 ± 3)**(−5 ± 3)  
 MVC (N)567 ± 53 410 ± 68*500 ± 43523 ± 44560 ± 65417 ± 55*484 ± 50*495 ± 44 
 (%Δ) (−28 ± 9)* (−9 ± 7)(−6 ± 7)  (−22 ± 8)* (−12 ± 3)**(−8 ± 6) 
 Activation (%)92 ± 1 85 ± 4* 85 ± 3*91 ± 290 ± 285 ± 3 91 ± 3 93 ± 2 
 (%Δ) (−8 ± 4)* (−8 ± 3)*(−2 ± 2)  (−5 ± 4) (1 ± 2)(3 ± 2)

Figure 4. Group mean change for the second quadriceps twitch amplitude (Qtw,T2), expressed as percentage change from baseline values, for CTRL versus PAV (A) and IRL versus IRL-CTRL (B) over a range of stimulation frequencies. Data were collected 2.5 min postexercise. n= 8 per group. Values are mean ±s.e.m.*P < 0.05; **P < 0.01, significantly different at the same frequency of stimulation. Note: the IRL and IRL-CTRL values are the mean of two trials performed on different days.

Download figure to PowerPoint


Figure 5. Group mean change across all stimulus frequencies for the second quadriceps twitch amplitude (Qtw,T2), expressed as percentage change from baseline values, for CTRL versus PAV (A) and IRL versus IRL-CTRL (B). Trial 1 and Trial 2 were performed on different days at identical power outputs and exercise durations for each of the IRL and IRL-CTRL conditions. Data were collected 2.5 min postexercise. n= 8 per group. Values are mean ±s.e.m.

Download figure to PowerPoint

Inspiratory resistance loading (IRL) versus IRL-CTRL: 292 ± 13 W, 7.9 ± 0.6 min The absolute values for quadriceps force–frequency curves before and following the control exercise condition for IRL are shown in Fig. 3C and D. Similar to the effects of the control exercise period for PAV (see Fig. 3A and B), Qtw was reduced at all frequencies of stimulation immediately following exercise and gradually recovered over 70 min. The control exercise trials for PAV and for IRL were at the same work rates, but the exercise time for IRL-CTRL was shorter than for PAV (7.9 versus 13.2 min), thus the postexercise decreases in evoked quadriceps forces were significantly less.

There was no difference in M-wave amplitude pre- versus postexercise for either of the IRL-CTRL or IRL conditions. Immediately after exercise in IRL-CTRL, Qtw,T2 (average for all stimulation frequencies) was 12 ± 8% less than baseline (P < 0.05) (Fig. 3C and D). Decreases were also noted immediately postexercise for non-potentiated and potentiated single twitches (−18 ± 4 and −31 ± 5%, respectively). In IRL, the group mean decrease in Qtw,T2 (average for all frequencies) was almost twice as much compared with IRL-CTRL (−20 ± 6 versus−12 ± 8%, respectively; P < 0.05). The percentage decrease in Qtw,T2 was greater in 7 of the 8 subjects in IRL versus IRL-CTRL. The largest differences in Qtw,T2 were at the lower stimulation frequencies (Fig. 4B). The effect of IRL on twitch measurements persisted up to 70 min postexercise (Table 3).

Table 3.  Group mean changes in contractile function measurements from baseline at 2.5, 35 and 70 min postexercise for inspiratory resistive load (IRL) and control (IRL-CTRL)
Baseline2.5 min35 min70 minBaseline2.5 min35 min70 min
  1. Mean ±s.e.m.n= 8 per group. *P < 0.05; **P < 0.01, significantly different from baseline value. †P < 0.05; ††P < 0.01, significantly different from IRL at corresponding time. Values in parentheses represent percentage changes from baseline values. Definitions as in Table 2. The IRL and IRL-CTRL values are the mean of two trials performed on different days.

 1 Hz,np (N)149 ± 11111 ± 7**123 ± 8**133 ± 8* 144 ± 11116 ± 7*  125 ± 9*  132 ± 9* 
 (%Δ) (−25 ± 4)** (−17 ± 3)**(−10 ± 3)** (−18 ± 4)**††(−12 ± 3)**††(−7 ± 3)* 
 1 Hz,pot (N)207 ± 12126 ± 9** 161 ± 12**  179 ± 12**197 ± 16132 ± 10**163 ± 11* 180 ± 11 
 (%Δ) (−39 ± 4)** (−22 ± 3)**(−14 ± 3)** (−31 ± 5)**† (−16 ± 4)**  (−7 ± 3) 
 Mean of 1, 10, 50, 100 Hz (N)218 ± 15167 ± 12* 182 ± 12** 192 ± 11*210 ± 15172 ± 14* 183 ± 14* 193 ± 13*
 (%Δ) (−22 ± 5)** (−16 ± 3)**(−11 ± 3)** (−16 ± 6)*   (−13 ± 4)**   (−8 ± 2)*†
 Mean of 1, 10, 50, 100 Hz (N)135 ± 9 107 ± 9* 112 ± 8**120 ± 7* 131 ± 8 113 ± 10  118 ± 8*  121 ± 8* 
 (%Δ) (−20 ± 6)* (−17 ± 3)**(−11 ± 3)** (−12 ± 8)†   (−10 ± 3)**†† (−7 ± 3)*†
 10 Hz/100 Hz 1.05 ± 0.050.97 ± 0.060.95 ± 0.05*1.03 ± 0.041.05 ± 0.051.06 ± 0.07 0.97 ± 0.06*1.04 ± 0.05
 (%Δ) (−8 ± 4) (−9 ± 4)*(−1 ± 3)  (−0 ± 4)  (−8 ± 2)* (−1 ± 1) 
 MVC (N)589 ± 42407 ± 66* 478 ± 50*521 ± 44579 ± 50439 ± 74  499 ± 48  548 ± 37 
 (%Δ) (−30 ± 10)*(−18 ± 8)  (−11 ± 5)   (−22 ± 11)  (−12 ± 7)   (−4 ± 4) 
 Activation (%)88 ± 373 ± 7*84 ± 4 86 ± 385 ± 375 ± 8  83 ± 5  86 ± 4 
 (%Δ) (−19 ± 6)* (−5 ± 3) (−2 ± 1)  (−12 ± 7)*   (−3 ± 4)  (−1 ± 3) 

The IRL-CTRL and IRL exercise trials were repeated in all subjects on two separate test days. The effects of IRL-CTRL and IRL on quadriceps muscle fatigue were nearly identical across the different test days (Fig. 5B).

Within-twitch measurements There was no effect of PAV or IRL on any of the within-twitch parameters (data not shown).

Ventilation, lactate and effort perceptions

Near end-exercise, minute ventlilation inline image was slightly higher (albeit non-significant) in PAV versus CTRL and lower in IRL versus IRL-CTRL (P < 0.05) due primarily to increases and decreases in tidal volume (VT), respectively (Table 1). Due to concomitant changes in inline image and rate of CO2 production inline image there were small but significant changes in inline image (increase with PAV) and inline image (decrease with IRL) compared with control values. The inspiratory duty cycle (Ti/Ttot) was unchanged with PAV but was significantly increased during IRL due to a prolongation of inspiratory time. Neither absolute values nor rates of rise of blood lactate concentration were different between CTRL and PAV or IRL and IRL-CTRL, although the rate of rise was higher in IRL versus PAV (P < 0.01) (Fig. 6). At end-exercise CTRL, limb discomfort was the primary symptom in three subjects, dyspnoea was the primary symptom in another three subjects, whereas limb discomfort and dyspnoea were equally prominent in two of the subjects. Throughout most of exercise, the perceptual ratings of limb discomfort and dyspnoea were lower in PAV and higher in IRL compared with respective control values (Fig. 7). The only difference in the rate of rise of ratings of perceived exertion (RPE) was for dyspnoea during IRL versus IRL-CTRL (1.160 ± 0.085 versus 0.844 ± 0.060 units min−1; P < 0.05).


Figure 6. Blood lactate concentration ([La]B) measured at baseline (Pre), after 5 min of warm-up (WU) at inline image, every 2 min of exhausting exercise at inline image, and after ∼10 min of recovery (Rec) for PAV (A) and IRL (B). n= 8 per group. Values are mean ±s.e.m. Note: the IRL and IRL-CTRL values are the mean of two trials performed on different days.

Download figure to PowerPoint


Figure 7. Ratings of perceived exertion (RPE) for limb discomfort (top panels) and dyspnoea (bottom panels) are shown for PAV versus CTRL (A and B) and IRL versus IRL-CTRL (C and D). n= 8 per group. Values are mean ±s.e.m.*P < 0.05; **P < 0.01, significantly different from corresponding time value. Note: the IRL and IRL-CTRL values are the mean of two trials performed on different days.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

This study determined the effect of inspiratory muscle work upon exercise-induced quadriceps muscle fatigue in healthy, physically trained humans. When cycle exercise was continued to the limit of tolerance with breathing unimpeded (CTRL), quadriceps twitch force was significantly reduced below baseline, at all frequencies of single and paired supramaximal magnetic stimulation of the femoral nerve. During repeat exercise at identical power output and duration, reducing the normally occurring force output of the inspiratory muscles by 56% via mechanical ventilation (PAV) attenuated the magnitude of exercise-induced quadriceps muscle fatigue by almost one-third, and reduced the perceptions of dyspnoea and limb discomfort. Increasing the force output of the inspiratory muscles by 80% above control, via inspiratory resistive loading (IRL), exacerbated the quadriceps fatigue and increased effort perceptions compared with identical exercise with breathing unimpeded (IRL-CTRL). These findings indicate a significant effect of the work of breathing during sustained, high-intensity exercise upon peripheral fatigue.

Technical considerations

The interpretation of our finding of a 20–28% difference in peripheral fatigue between control and loading/unloading conditions is critically dependent on our ability to detect relatively small systematic changes in our measures of neuromuscular function within and between days. This degree of sensitivity was achieved through our use of a non-volitional technique of assessing contractile function, i.e. magnetic femoral nerve stimulation (Polkey et al. 1996; Babcock et al. 1998, 2002; Mador et al. 2000, 2001; Kufel et al. 2002). Compared to electrical stimulation techniques, magnetic stimulation is better tolerated by subjects and more amenable to reproducible stimulation of the femoral nerve because of a more diffuse spread of current and less reliance on the pressure applied over the nerve. Thus, coefficients of variation for within-subject reproducibility were ≤ 7% for all measurements of evoked force. These reliability coefficients are similar to those reported previously (Polkey et al. 1996; Kufel et al. 2002; Romer et al. 2006). It was especially important that we could reproduce, both technically and biologically, the effects of heavy intensity exercise and of a superimposed inspiratory resistive load on peripheral fatigue. Thus, we found no systematic change in the magnitude of peripheral fatigue when either control or IRL trials were repeated on different days. Equally important in being able to interpret the changes in fatigue was that we used a study design whereby work rate, exercise duration and arterial O2 saturation were matched for each of the unloading/loading comparisons while the force output of the inspiratory muscles was the only parameter manipulated. Another consideration is whether the exercise-induced increases in muscle temperature, which are known to influence muscle fatigue (De Ruiter et al. 1999; De Ruiter & De Haan, 2000), affected our ability to interpret the apparent differences in fatigue between conditions. However, such changes in muscle temperature would not be expected to affect our fatigue comparisons because neither environmental temperature nor exercise duration was different for each of the unloading/loading comparisons (i.e. CTRL versus PAV and IRL-CTRL versus IRL).

An assumption of the paired stimuli technique is that the force response to the second stimulus is representative of the muscle's response to many stimuli at that frequency, as would occur with tetanic contractions. Tetanic and paired twitch techniques provide comparable measurements of diaphragm fatigue (Yan et al. 1993; Polkey et al. 1997; Babcock et al. 1998). In a single subject we extended this finding to show that the effect of IRL on quadriceps muscle fatigue in the immediate postexercise period was the same whether using magnetic or electrical tetanic stimulation of the femoral nerve (data not shown). Another assumption of the magnetic stimulation technique is that the motor nerve input to the quadriceps, via the femoral nerve, was both the same and supramaximal before and after exercise for each of the inspiratory muscle loading/unloading comparisons. Based on the plateau in quadriceps twitch force and M-wave amplitudes with increasing stimulus intensities that we observed in every subject, it is likely that stimulation occurred at a truly supramaximal intensity. An additional consideration is that repeated voluntary contractions increase the activation threshold of motor axons due to activity-dependent hyperpolarization (Vagg et al. 1998). This effect would be expected to reduce the population of axons excited by the same stimulus intensity after versus before exercise, regardless of whether M-wave amplitudes remain unchanged (Vagg et al. 1998). Thus, motor output to the muscle during magnetic stimulation following exercise was probably reduced, although comparisons between control and loading/unloading should not be influenced because exercise was performed at the same intensity and for the same duration in both conditions.

Characteristics and causes of muscle fatigue

The magnitude of peripheral fatigue was attenuated by almost one-third when the inspiratory muscles were unloaded (−28% for CTRL versus−20% for PAV) and exacerbated to an even greater extent when the inspiratory muscles were loaded (−12% for CTRL-IRL versus−20% for IRL). The changes in contractile function were not due to reductions in action potential transmission because M-wave amplitudes were not different pre- versus postexercise. That muscle function began to recover rapidly after exercise and that the differences in fatigue between unloading/loading and control conditions were most pronounced immediately after exercise suggest a role for metabolic processes.

Respiratory muscle metaboreflex We propose, as originally hypothesised (see Introduction), that the increases and decreases in peripheral fatigue with respiratory muscle loading and unloading, respectively, reflect corresponding changes in limb blood flow inline image and limb O2 transport. There are several findings that support this interpretation. First, our previous demonstration of a significant inverse relationship between the work of breathing and limb vascular conductance and blood flow during near maximum exercise would apply to the present findings, because the exercise intensities and the changes in the work of breathing achieved with loading and unloading were very comparable between studies (Harms et al. 1997). We note that the inspiratory muscle unloading increased inline image but also reduced the total cardiac output, whereas inspiratory muscle loading reduced inline image without an effect on cardiac output – probably because cardiac output was already at maximum achievable levels. Second, even though the effects on inline image were only in the range of 5 to 7% with unloading and −9 to −10% with loading (Harms et al. 1997), it has been repeatedly demonstrated that even small changes in blood flow to the contracting limb have major effects on muscle force output (Barclay, 1986; Rabischong & Guiraud, 1993; Hogan et al. 1998; Cole & Brown, 2000). These effects of changing inline image have been attributed to changes in O2 transport and/or to changes in the washout of local metabolites (Barclay, 1986; Frisbee et al. 1999). Third, we have recently shown that preventing even relatively small reductions in arterial O2 content (98 versus 91%Smath image) and therefore in O2 transport during heavy-intensity endurance exercise also prevented about one-half the amount of exercise-induced peripheral fatigue, when comparisons were made at identical work rates and durations (Romer et al. 2006).

The mechanisms underlying the effects of respiratory muscle work on limb vascular conductance are believed to involve metaboreflex feedback from the fatiguing diaphragm and/or expiratory muscles, leading to increased sympathetically mediated vasoconstriction (Dempsey et al. 2002). This effect on muscle sympathetic nerve activity and limb vascular conductance has been demonstrated through the use of voluntary increases in inspiratory and expiratory muscle work to the point of task failure in the resting human (St Croix et al. 2000; Sheel et al. 2001; Derchak et al. 2002). Similar effects have been shown more specifically by the reduced vascular conductance induced by lactic acid infusions into the phrenic artery in the resting and exercising canine – an effect that was prevented via pharmacological sympathetic blockade (Rodman et al. 2003). Accordingly, since respiratory muscle unloading prevents the exercise-induced diaphragmatic fatigue that normally occurs during heavy exercise (Babcock et al. 2002), we may also expect suppression of the sympathetic vasoconstrictor outflow, and vice versa with greater than normal loading.

Is the effect of changing respiratory muscle work on peripheral fatigue related to changes in the relative intensity of the exercise? Unloading caused an 8% mean reduction in total inline image and loading caused a 5% increase in inline image. Therefore, in terms of total body inline image, the relative intensities of the exercise were changed in a direction that would influence peripheral fatigue. However, this conventional interpretation is complicated by the fact that whilst increasing and decreasing the work of the respiratory muscles is probably the major influence on total body inline image, limb blood flow also changes in the opposite direction, so that limb inline image is actually slightly higher with unloading and reduced with loading (Harms et al. 1997, 1998b). Thus, at the level of the working locomotor muscle the conventional explanation of changes in ‘relative intensity’ may not be relevant. These effects contrast with a positive association of changes in relative exercise intensity at the level of the locomotor muscle with peripheral fatigue obtained in the face of changes in arterial O2 saturation and content (Romer et al. 2006).

Is the magnitude of peripheral fatigue that is associated with the changes in respiratory muscle work biologically significant? Cycling exercise at ≥ 90% of inline image to exhaustion caused a mean 28% (range 15–36%) reduction in maximum force output of the quadriceps in response to supramaximal stimulation and relieving 56% of the inspiratory muscle work prevented about one-third of this limb fatigue. This effect on peripheral fatigue is only a little more than one-half the effect we have recently found by preventing the 7–9% arterial O2 desaturation that can accompany heavy sustained exercise (Romer et al. 2006). Thus, this unloading effect, whilst consistent and statistically significant, appears to be relatively small. However, it is important to note that we reduced only slightly more than one-half the normal work of breathing so this effect probably represents an underestimation of what might be attributed to the total work of breathing of the inspiratory and expiratory muscles that occurs during heavy sustained exercise. An additional influence also not considered in our current study was the effect of relieving high expiratory pressures, which are often encountered in the face of expiratory flow limitation and hyperinflation during heavy exercise (Johnson et al. 1992) and which have a mechanical effect on reducing left ventricular stroke volume and cardiac output (Stark-Leyva et al. 2004).

The effect of artificially increasing inspiratory flow resistance on peripheral fatigue was much greater than unloading the inspiratory muscles. For example, 8 min of high intensity exercise caused an 18% reduction in limb force output, and superimposing an 80% increase in inspiratory muscle work caused a further 40% reduction in peripheral fatigue following exercise of equal duration and workload. Certainly part of this difference may be because the changes in the force output of the inspiratory muscles were not equal between unloading and loading (−50 versus+80%). However, another reason may be that increases in limb blood flow during unloading are typically less than the reductions in flow with loading (Harms et al. 1997), which is in part attributable to a reduction in total available cardiac output during unloading (Harms et al. 1998b).

Under what conditions might exercise-induced respiratory muscle work contribute to peripheral fatigue? First, it is likely that the exercise intensity must be high and sustained, as in the present study inline image. Our previous studies showed that the respiratory muscle work had to be increased to near fatiguing levels for muscle sympathetic nerve activity to be increased (St Croix et al. 2000; Derchak et al. 2002) or for blood flow to be decreased in a resting (Sheel et al. 2001) or an exercising limb (Harms et al. 1997; Wetter et al. 1999; Rodman et al. 2003). Furthermore, exercise-induced diaphragm fatigue in healthy subjects of varying fitness levels was only observed when exercise intensity exceeded 80–85% of inline image and the exercise was sustained to exhaustion (Johnson et al. 1993; Babcock et al. 1998). However, in patients with chronic obstructive pulmonary disease (COPD) the limb muscles are more fatigable compared with healthy control subjects (Mador et al. 2003) and the work of breathing during exercise is substantial (O'Donnell et al. 2001). There is also evidence that a significant fraction of COPD patients show a plateau in leg blood flow and leg inline image early during incremental exercise and that these patients have significantly higher levels of ventilation and dyspnoea, suggesting that blood flow may be redirected to the respiratory muscles (Simon et al. 2001). Additionally, patients with chronic heart failure who show an elevated ventilatory demand in combination with a blunted cardiac output response to exercise may also be candidates for major effects of respiratory muscle work on limb blood flow during even moderate intensity exercise. In health, exercise in the hypoxia of high altitudes would also be expected to exacerbate the effects of respiratory muscle work on exercise-induced limb muscle fatigue. This effect may be especially significant in chronic hypoxia in which exercise hyperventilation and respiratory muscle work are greatly magnified (Thoden et al. 1969) and yet exercise cardiac output is not elevated relative to sea level (Reeves et al. 1987).

Respiratory muscle work, limb fatigue and exercise performance

As shown in the present study and in previous studies, respiratory muscle loading and unloading affect high-intensity exercise performance (Demedts & Anthonisen, 1973; Harms et al. 2000). Since we found that the magnitude of peripheral fatigue was attenuated and exacerbated by inspiratory muscle unloading and loading, respectively, does this mean that changes in peripheral fatigue caused the previously mentioned changes in performance? In the present study, muscle fatigue was quantified as a decrease in isometric force output in response to supramaximal nerve stimulation during recovery from exercise. Thus, while we think it reasonable to link whole-body performance with this evidence of peripheral fatigue, we cannot be certain how the observed approximately one-third reduction (via PAV) or 40% increase (via IRL) in peripheral fatigue translate precisely into the subject's capability for sustaining a given (probably submaximal) power output during cycling.

It is also likely that changes in exercise performance with inspiratory muscle loading/unloading will be in part due to changes in motor output to the locomotor muscles during the exercise, i.e. central fatigue (Bigland-Ritchie et al. 1978; Gandevia, 2001). A significant 8% reduction in voluntary activation of the quadriceps was noted using twitch interpolation during MVC manoeuvres after exercise, which confirms earlier reports (Bentley et al. 2000; Lepers et al. 2001, 2002), and loading and unloading the respiratory muscles tended to exacerbate or reduce this effect to a small extent. However, since central fatigue is known to be task specific (Gandevia, 2001), our observations obtained during voluntary activation of an isometric contraction during recovery from exercise do not imply that changes in central fatigue contributed to the observed alterations in exercise performance with respiratory muscle loading and unloading. Voluntary activation was less reliable compared with quadriceps twitch force measurements such that small changes in central fatigue may not have been detectable using the superimposed twitch technique. More importantly, measurements of voluntary activation were performed after the twitch force measurements such that any effect of inspiratory muscle loading/unloading on central fatigue might have been attenuated during the postexercise recovery. This postulate is supported by previous research showing that voluntary activation recovers almost fully by 3 min after exercise (Bigland-Ritchie et al. 1986).

Perhaps more relevant were our findings that the perceptions of dyspnoea and limb discomfort were intensified during inspiratory muscle loading and were reduced during unloading. O'Donnell et al. (1999) in heart failure patients and Harms et al. (2000) in healthy subjects have also shown that reductions in inspiratory muscle work during exercise elicited reductions in the perceptual intensity of limb discomfort. These changes in effort perception might be expected to affect voluntary activation of the limbs and therefore performance time. In turn, since much of the increased effort perceptions with loading and their reduction with unloading probably originated from changing sensory inputs, it is reasonable to assume that at least some of the change in the intensity of perceived efforts was linked to peripheral fatigue of both the respiratory muscles and the locomotor muscles.


The work of breathing that normally occurs during high-intensity sustained exercise was shown to contribute significantly to peripheral fatigue and, conversely, loading of the respiratory muscles exacerbated peripheral fatigue. Based on previous findings we think it likely that these effects on limb fatigue were due to corresponding effects of respiratory muscle work on limb vascular resistance and blood flow. Further, we postulate that the significant effects of respiratory muscle work on exercise performance are in large part due to the direct effect of peripheral fatigue on limiting muscle force output in response to a given motor command and through its feedback effect on increasing sensory input to the central nervous system.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix
  • Aaron EA, Seow KC, Johnson BD & Dempsey JA (1992). Oxygen cost of exercise hyperpnea: implications for performance. J Appl Physiol 72, 18181825.
  • Babcock MA, Pegelow DF, Harms CA & Dempsey JA (2002). Effects of respiratory muscle unloading on exercise-induced diaphragm fatigue. J Appl Physiol 93, 201206.
  • Babcock MA, Pegelow DF, Taha BH & Dempsey JA (1998). High frequency diaphragmatic fatigue detected with paired stimuli in humans. Med Sci Sports Exerc 30, 506511.
  • Barclay JK (1986). A delivery-independent blood flow effect on skeletal muscle fatigue. J Appl Physiol 61, 10841090.
  • Baydur A, Behrakis PK, Zin WA, Jaeger M & Milic-Emili J (1982). A simple method for assessing the validity of the esophageal balloon technique. Am Rev Respir Dis 126, 788791.
  • Bentley DJ, Smith PA, Davie AJ & Zhou S (2000). Muscle activation of the knee extensors following high intensity endurance exercise in cyclists. Eur J Appl Physiol 81, 297302.
  • Bigland-Ritchie BR, Dawson NJ, Johansson RS & Lippold OC (1986). Reflex origin for the slowing of motoneurone firing rates in fatigue of human voluntary contractions. J Physiol 379, 451459.
  • Bigland-Ritchie B, Jones DA, Hosking GP & Edwards RH (1978). Central and peripheral fatigue in sustained maximum voluntary contractions of human quadriceps muscle. Clin Sci Mol Med 54, 609614.
  • Bigland-Ritchie B & Woods JJ (1984). Changes in muscle contractile properties and neural control during human muscular fatigue. Muscle Nerve 7, 691699.
  • Borg G (1998). Borg's Perceived Exertion and Pain Scales. Human Kinetics, Champaign , IL .
  • Cole MA & Brown MD (2000). Response of the human triceps surae muscle to electrical stimulation during varying levels of blood flow restriction. Eur J Appl Physiol 82, 3944.
  • Demedts M & Anthonisen NR (1973). Effects of increased external airway resistance during steady-state exercise. J Appl Physiol 35, 361366.
  • Dempsey JA, Sheel AW, St Croix CM & Morgan BJ (2002). Respiratory influences on sympathetic vasomotor outflow in humans. Respir Physiol Neurobiol 130, 320.
  • Derchak PA, Sheel AW, Morgan BJ & Dempsey JA (2002). Effects of expiratory muscle work on muscle sympathetic nerve activity. J Appl Physiol 92, 15391552.
  • De Ruiter CJ & De Haan A (2000). Temperature effect on the force/velocity relationship of the fresh and fatigued human adductor pollicis muscle. Pflugers Arch 440, 163170.
  • De Ruiter CJ, Jones DA, Sargeant AJ & De Haan A (1999). Temperature effect on the rates of isometric force development and relaxation in the fresh and fatigued human adductor pollicis muscle. Exp Physiol 84, 11371150.
  • Frisbee JC, Murrant CL, Wilson BA & Barclay JK (1999). Polycythemia decreases fatigue in tetanic contractions of canine skeletal muscle. Med Sci Sports Exerc 31, 12931298.
  • Gandevia SC (2001). Spinal and supraspinal factors in human muscle fatigue. Physiol Rev 81, 17251789.
  • Harms CA, Babcock MA, McClaran SR, Pegelow DF, Nickele GA, Nelson WB & Dempsey JA (1997). Respiratory muscle work compromises leg blood flow during maximal exercise. J Appl Physiol 82, 15731583.
  • Harms CA, McClaran SR, Nickele GA, Pegelow DF, Nelson WB & Dempsey JA (1998a). Exercise-induced arterial hypoxaemia in healthy young women. J Physiol 507, 619628.
  • Harms CA, Wetter TJ, McClaran SR, Pegelow DF, Nickele GA, Nelson WB, Hanson P & Dempsey JA (1998b). Effects of respiratory muscle work on cardiac output and its distribution during maximal exercise. J Appl Physiol 85, 609618.
  • Harms CA, Wetter TJ, St Croix CM, Pegelow DF & Dempsey JA (2000). Effects of respiratory muscle work on exercise performance. J Appl Physiol 89, 131138.
  • Hogan MC, Gladden LB, Grassi B, Stary CM & Samaja M (1998). Bioenergetics of contracting skeletal muscle after partial reduction of blood flow. J Appl Physiol 84, 18821888.
  • Johnson BD, Babcock MA, Suman OE & Dempsey JA (1993). Exercise-induced diaphragmatic fatigue in healthy humans. J Physiol 460, 385405.
  • Johnson BD, Saupe KW & Dempsey JA (1992). Mechanical constraints on exercise hyperpnea in endurance athletes. J Appl Physiol 73, 874886.
  • Kufel TJ, Pineda LA & Mador MJ (2002). Comparison of potentiated and unpotentiated twitches as an index of muscle fatigue. Muscle Nerve 25, 438444.
  • Lepers R, Maffiuletti NA, Rochette L, Brugniaux J & Millet GY (2002). Neuromuscular fatigue during a long-duration cycling exercise. J Appl Physiol 92, 14871493.
  • Lepers R, Millet GY & Maffiuletti NA (2001). Effect of cycling cadence on contractile and neural properties of knee extensors. Med Sci Sports Exerc 33, 18821888.
  • Mador MJ, Bozkanat E & Kufel TJ (2003). Quadriceps fatigue after cycle exercise in patients with COPD compared with healthy control subjects. Chest 123, 11041111.
  • Mador JM, Kufel TJ & Pineda LA (2000). Quadriceps and diaphragmatic function after exhaustive cycle exercise in the healthy elderly. Am J Respir Crit Care Med 162, 17601766.
  • Mador MJ, Kufel TJ, Pineda LA, Steinwald A, Aggarwal A, Upadhyay AM & Khan MA (2001). Effect of pulmonary rehabilitation on quadriceps fatiguability during exercise. Am J Respir Crit Care Med 163, 930935.
  • Manohar M (1986). Blood flow to the respiratory and limb muscles and to abdominal organs during maximal exertion in ponies. J Physiol 377, 2535.
  • NHLBI Workshop (1990). Respiratory muscle fatigue: Report of the respiratory muscle fatigue workshop group. Am Rev Respir Dis 142, 474486.
  • O'Donnell DE, D'Arsigny C, Raj S, Abdollah H & Webb KA (1999). Ventilatory assistance improves exercise endurance in stable congestive heart failure. Am J Respir Crit Care Med 160, 18041811.
  • O'Donnell DE, Revill SM & Webb KA (2001). Dynamic hyperinflation and exercise intolerance in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 164, 770777.
  • Polkey MI, Kyroussis D, Hamnegard CH, Hughes PD, Rafferty GF, Moxham J & Green M (1997). Paired phrenic nerve stimuli for the detection of diaphragm fatigue in humans. Eur Respir J 10, 18591864.
  • Polkey MI, Kyroussis D, Hamnegard CH, Mills GH, Green M & Moxham J (1996). Quadriceps strength and fatigue assessed by magnetic stimulation of the femoral nerve in man. Muscle Nerve 19, 549555.
  • Rabischong E & Guiraud D (1993). Determination of fatigue in the electrically stimulated quadriceps muscle and relative effect of ischaemia. J Biomed Eng 15, 443450.
  • Reeves JT, Groves BM, Sutton JR, Wagner PD, Cymerman A, Malconian MK, Rock PB, Young PM & Houston CS (1987). Operation Everest II: preservation of cardiac function at extreme altitude. J Appl Physiol 63, 531539.
  • Rodman JR, Henderson KS, Smith CA & Dempsey JA (2003). Cardiovascular effects of the respiratory muscle metaboreflexes in dogs: rest and exercise. J Appl Physiol 95, 11591169.
  • Romer LM, Haverkamp HC, Lovering AT, Pegelow DF & Dempsey JA (2006). Effect of exercise-induced arterial hypoxemia on quadriceps muscle fatigue in healthy humans. Am J Physiol Regul Integr Comp Physiol; 290, R365R375.
  • Romer LM, Pegelow DF, Haverkamp HC, Lovering AT & Dempsey JA (2005b). Effect of inspiratory muscle work on exercise-induced quadriceps muscle fatigue in healthy humans. J Physiol 565.P, C34.
  • St Croix CM, Morgan BJ, Wetter TJ & Dempsey JA (2000). Fatiguing inspiratory muscle work causes reflex sympathetic activation in humans. J Physiol 529, 493504.
  • Sheel AW, Derchak PA, Morgan BJ, Pegelow DF, Jacques AJ & Dempsey JA (2001). Fatiguing inspiratory muscle work causes reflex reduction in resting leg blood flow in humans. J Physiol 537, 277289.
  • Simon M, LeBlanc P, Jobin J, Desmeules M, Sullivan MJ & Maltais F (2001). Limitation of lower limb VO2 during cycling exercise in COPD patients. J Appl Physiol 90, 10131019.
  • Stark-Leyva KN, Beck KC & Johnson BD (2004). Influence of expiratory loading and hyperinflation on cardiac output during exercise. J Appl Physiol 96, 19201927.
  • Strojnik V & Komi PV (1998). Neuromuscular fatigue after maximal stretch-shortening cycle exercise. J Appl Physiol 84, 344350.
  • Thoden JS, Dempsey JA, Reddan WG, Birnbaum ML, Forster HV, Grover RF & Rankin J (1969). Ventilatory work during steady-state response to exercise. Fed Proc 28, 13161321.
  • Vagg R, Mogyoros I, Kiernan MC & Burke D (1998). Activity-dependent hyperpolarization of human motor axons produced by natural activity. J Physiol 507, 919925.
  • Wetter TJ, Harms CA, Nelson WB, Pegelow DF & Dempsey JA (1999). Influence of respiratory muscle work on VO2 and leg blood flow during submaximal exercise. J Appl Physiol 87, 643651.
  • Yan S, Gauthier AP, Similowski T, Faltus R, Macklem PT & Bellemare F (1993). Force-frequency relationships of in vivo human and in vitro rat diaphragm using paired stimuli. Eur Respir J 6, 211218.
  • Younes M (1992). Proportional assist ventilation, a new approach to ventilatory support. Theory. Am Rev Respir Dis 145, 114120.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix


Support for this project was provided by a National Heart, Lung, and Blood Institute (NHLBI) RO1 Grant (HL-15469-33). A. T. Lovering and H. C. Haverkamp were supported by a NHLBI Training Grant (T32 HL-07654-16). We thank Mr Benjamin Dempsey and Ms Sonia Gysland for valuable assistance with analysing the M-wave data, and Dr Magdy Younes for use of his PAV prototype.