INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Progressive inspiratory threshold loading has been used to study aspects of inspiratory muscle function in health (1) and disease (4). The method involves a progressive increase in the threshold pressure (Pth) that must be developed to initiate inspiratory flow until "task failure" occurs when the subject is unable to continue. The maximal load achieved (Pth_max) has been used to define the maximal endurance capacity of the inspiratory muscles. However we and others have noted a systematic increase in Pth_max with the first few exposures to progressive threshold loading in naïve subjects, following which highly reproducible measurements are obtained (2, 3, 5, 10). The reasons for this improvement in performance are unclear. It does not appear to be attributable to improved inspiratory muscle strength, nor does fatigue of inspiratory muscles appear to be an important element of task failure in experienced subjects (2).

We hypothesized that the breathing pattern may change systematically over these early exposures to progressive threshold loading in a way that would lessen the work of breathing or the sensation of respiratory effort (15). If this were the case, then the improvement in performance (Pth_max) may represent the result of neuropsychological adaptation to the task rather than respiratory muscle conditioning. The distinction is important particularly as the potential for improvement in respiratory muscle strength or endurance with training or treatment could be overstated if the existence of such a "learning effect" was unrecognized. Thus the aim of this study was to investigate the mechanisms responsible for the increase in Pth_max with repeated exposures to progressive threshold loading in naïve, healthy individuals.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject Characteristics

Eighteen healthy male subjects participated in the study. All procedures were approved by the Human Rights Committee of the University of Western Australia. Informed consent was obtained in writing from each subject.

The following respiratory function data were acquired for each subject: total lung capacity (TLC) (body plethysmograph; Collins Inc., Braintree, MA); forced vital capacity (FVC), residual volume (RV), and forced expiratory volume in one second (FEV₁) (digital pneumotachograph, model 400VR; Hewlett-Packard, Waltham, MA); and transfer factor for carbon monoxide diffusing capacity (DL_CO) (model 1182; P.K. Morgan Ltd, Gillingham, UK).

Progressive Threshold Loading

Each subject performed three progressive threshold loading tests until task failure (each > 24 h apart). Each subject was naïve to the loaded breathing protocol prior to the first test.

Before each test respiratory muscle strength was estimated from measurements of the maximal inspiratory pressure (PI_max) developed during maximal inspiratory efforts at RV against a closed airway. Subjects were closely monitored to ensure that buccal muscles were not used in generating the inspiratory pressures. Each maneuver was repeated until three reproducible efforts were obtained (within 5% of each other). PI_max was determined from the highest pressure recorded during this sequence.

Following measurement of PI_max each subject underwent progressive inspiratory threshold loading (1). Subjects breathed through a mouthpiece attached to a low resistance variable orifice pneumotachograph (P.K. Morgan Ltd) connected in series to a modified inspiratory threshold valve which required development of a negative Pth before inspiratory airflow was achieved (18, 19). The Pth required to be developed with each inspiration could be increased by addition of weights to the valve.

Breath-by-breath measurements of oxygen uptake (O₂), carbon dioxide output (CO₂), minute ventilation (E) and its component tidal volume (VT) and breathing frequency (f), inspiratory time (TI), expiratory time (TE), total breath time (Ttot), and end-tidal CO₂ tension (PET_CO2) were collected continuously using an automated exercise metabolic system (Morgan Benchmark Exercise Test; P.K. Morgan Ltd). Before each test the pneumotachograph was calibrated over a range of flows using a 3-L syringe. Fast responding infrared CO₂ and zirconia cell O₂ analyzers were also calibrated before each test using gases of known concentrations.

Arterial O₂ saturation (Sa_O2) (ear probe, pulse oximeter, Biox 3700; Ohmeda, Boulder, CO) was measured continuously. Pth was continuously monitored at the mouthpiece (differential pressure transducer, model PM131; Statham Instruments, Oxnard, CA). Rib cage and abdominal motion were continuously monitored by respiratory inductance pneumography (Respitrace, Ardsley, NY) using two motion transducers, one at the level of the nipples and the other at the level of the umbilicus. These signals were calibrated by an isovolume maneuver and electronically summed to provide a measure of total chest wall displacement. Change in lung volume was determined from chest wall displacement by referencing the summed Respitrace signal at end-expiratory lung volume (EELV) to measurements of FRC and RV obtained before and after loaded breathing. EELV was expressed as a percentage of expiratory reserve volume (FRC = 0%, RV = 100%). We have previously validated this method of measuring EELV against measurements obtained using body plethysmography (2). Motion signals were collected at 100 Hz on a 486DX personal computer using the LabVIEW data acquisition package (National Instruments, Austin, TX).

The loaded breathing protocol required subjects to sit quietly and breathe for 5 min through the unloaded threshold valve while baseline metabolic and chest wall motion measurements were collected. Inspiratory load was then increased by 100 g (approximately 20 cm H₂O) each 2 min until the subject was no longer able to sustain the breathing task (task failure), despite strong encouragement from the investigator. No instructions were given to the subject regarding what breathing pattern to adopt, nor were subjects given feedback on their previous performance while doing the next test. The mean duration of the progressive threshold loading tests was 12 ± 4 min.

Throughout each test, perception of effort was estimated during the final 10 s of each minute using the 15-grade scale for rating of perceived exertion (RPE) (20). The scale was held momentarily in front of the subject who was asked to point to the number that best corresponded to their effort sensation. On each occasion the investigator emphasized that the subject specifically describe "sensation of effort" and not "sensation of breathlessness." Immediately after task failure PI_max was remeasured.

Data Analysis

Breath-by-breath measurements of all variables were transferred to a spreadsheet package where the mean value of each variable was calculated for an integral number of breaths over the final 20-s epoch of each minute. Breaths that were marred by artifact (i.e., swallowing, coughing, or sighing) were excluded from analysis.

During each progressive threshold loading test the following measurements were derived: (1) maximal Pth (Pth_max), defined as the mean Pth generated over the final 20 s of the highest load tolerated for at least 60 s; (2) respiratory muscle O₂ consumption (O_2resp), calculated from the difference between the O₂ at any given load and the resting O₂, expressed as kg · m/min; (3) rate of external inspiratory mechanical work (_ext), calculated from the product of Pth · VT · f (21), expressed as kg · m/min; (4) respiratory muscle efficiency, defined as _ext/O_2resp (3); and (5) sensory magnitude (SENS), calculated from the formula Pth^1.3 · TI^0.52 · f ^0.19 (17).

Measurements of O_2resp and _ext were normalized for variations in ventilation and EELV, respectively, to aid comparison between differing conditions. Normalization of O_2resp was performed by dividing it by the E at which it was measured and expressing this ratio as ml O₂/L ventilation. _ext was normalized to allow for the effect of EELV on inspiratory muscle mechanics, so that the derived index might better reflect inspiratory metabolic work. It was assumed that pressure-generating capacity (that is PI_max) increased linearly by 18% between resting FRC and RV (2) as a result of improved mechanical advantage. Thus normalized Pth at each load where EELV was less than resting FRC was reduced in proportion to the calculated increase in PI_max at that EELV.

Statistics

Because PI_max was unchanged throughout the study (see RESULTS), in all tests inspiratory load was expressed as a percentage of each subject's mean PI_max (%PI_max) which was derived from the average of all PI_max measurements. As the inspiratory load was increased by approximately 20 cm H₂O, as described previously, the number of data points obtained during a threshold loading test varied from subject to subject according to the individual's PI_max. Therefore, to facilitate comparisons between subjects data obtained during a test were averaged into bins corresponding to increments in PI_max of 15 to 20%.

Two-way analysis of variance (ANOVA) with repeated measures was used to evaluate differences in PI_max before each of the loaded breathing tests. For each variable, between-test differences for the three tests were analyzed by two-way ANOVA or two-way analysis of covariance (ANCOVA) (regression analysis) using the best fit data from single or squared functions. ANOVA with repeated measures was used to compare the changes in each variable with progressive inspiratory loading, measurements of each variable at task failure with successive tests, and resting measurements obtained prior to each test. For the purpose of clarity, each data point in graphs 2 and 3 is shown as mean ± SEM, otherwise all data are reported as mean ± SD; p < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject Characteristics

Anthropometric measurements and lung function data are shown in Table 1. In all subjects resting respiratory function was within normal limits. Table 2 shows the measurements of breathing pattern and gas exchange obtained at rest prior to each of the three successive progressive threshold loading tests: there were no significant differences between the different testing days, and measurements were highly reproducible with the coefficient of variation (= SD/mean) of all variables = 8.1 ± 3.8%, range 0.5 to 13.5%.

Respiratory Muscle Strength (PI_max)

There was no significant difference between PI_max measured before and immediately after any of the three progressive threshold loading tests; as such, the mean of these two measurements was used to represent the daily PI_max. This daily PI_max was unchanged over the 3 testing days (Figure 1, open bars).

View larger version (38K): [in this window] [in a new window]   Figure 1.   Respiratory muscle strength (PImax, open bars) and maximal threshold pressure (Pthmax, hatched bars) with successive tests; n = 18 per test; error bars, ± SD. *Significantly different (p < 0.05) from Pthmax achieved in Test 1.

Progressive Threshold Loading

In contrast to measurements of PI_max, which were similar on all testing days, Pth_max increased in all subjects with successive tests (69 ± 17, 77 ± 16, 86 ± 11% of PI_max) (p < 0.05) (Figure 1, hatched bars).

Systematic changes in breathing pattern were observed both with increasing inspiratory load and from test 1 to test 3 (Figures 2 and 3). Both E and VT increased markedly up to 30% PI_max before slowly declining with further increases in inspiratory load. At lower loads f tended to decrease, but then increased with increasing load, particularly above 30% PI_max. TI increased with initial inspiratory load during the first and second tests. Both TI and TE fell with increasing load above 30% PI_max, accounting for the increase in f. The relative changes in TI and TE were such that TI/Ttot was stable with increasing workload during the first test, but declined substantially with moderate loads on the second and, particularly, the third tests. With the increase in VT and decrease in TI, VT/TI increased with increasing workload up to 30% PI_max, then slowly decreased in parallel with VT with the application of higher loads.

View larger version (27K): [in this window] [in a new window]   Figure 2.   Effect of progressively increasing inspiratory threshold load (expressed as a percentage of maximal inspiratory pressure [%PImax]) on E, VT, TI, TE, f , and EELV. Mean data were from three separate tests  24 h apart; n = 18 per test; error bars, ± SEM.

View larger version (30K): [in this window] [in a new window]   Figure 3.   Effect of progressively increasing inspiratory threshold load (expressed as %PImax) on ext (normalized for changes in EELV ), O2resp (normalized for change in E), change in arterial O2 saturation (SaO2), change in end-tidal CO2 (PETCO2), RPE, and SENS. Mean data were from three separate tests  24 h apart, n = 18 per test; error bars, ± SEM.

With successive tests E increased at equivalent workloads as the result of increasing VT (p < 0.05). Breathing frequency tended to decrease, hence the time taken for each breath (Ttot) increased from test to test. Within-breath timing also changed, with a decrease in TI and increase in TE, hence a decrease in TI/Ttot with successive tests (p < 0.05). With the increase in VT and decrease in TI, VT/TI increased with successive tests (p < 0.05).

EELV progressively decreased with increasing load. Although the decrease in EELV tended to be less at equivalent workloads with successive tests (p < 0.05), it was not significantly different at Pth_max with each test.

O_2resp and _ext both increased with increasing workload, and, in keeping with the increase in E, were greater at equivalent workload with succeeding tests (p < 0.05). These changes were comparable so that respiratory efficiency was not significantly different from test 1 to test 3 (2.2 ± 1.4, 2.3 ± 2.1, and 2.5 ± 2.7 %, respectively).

Up to an inspiratory load of 30% PI_max there was a decrease in PET_CO2 and marginal increase in Sa_O2. Beyond this workload PET_CO2 progressively increased and Sa_O2 decreased. At equivalent workloads Sa_O2 was higher and PET_CO2 lower with succeeding tests (p < 0.05).

RPE and SENS increased progressively with increasing workload. Subjects perceived any given load to require less effort with successive tests (p < 0.05). The change in the pattern of breathing with successive tests was also associated with a tendency to a lower SENS at any given workload. Both RPE and SENS reached similar maximal values at Pth_max with each test.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Progressive threshold loading is a commonly used test of inspiratory muscle function in both health (1) and disease (4- 14). A reason for its popularity is the assertion in Martyn and coworkers' original study (1) that the method assesses muscle endurance without the need for a learning or familiarization period. While other studies have supported this view (11, 22- 24), our experience (2) and that of others (3, 5, 10) is that a learning period, consisting of between one and three tests, is required before reproducible measurements are obtained. The reason for these different findings is uncertain, but may be related to differences in subject groups, the presence of an undocumented familiarization period, the degree of encouragement offered by the investigators, how maximal load is defined, or possibly to differences in the performance characteristics of the threshold valve itself (19). The finding in the present study that Pth_max increased over the first three exposures to the test in previously naïve subjects strongly suggests that a learning period is needed before reproducible results can be obtained. Failure to recognize this need could lead to incorrectly attributing improvement in performance after training or treatment to improvement in respiratory muscle strength or efficiency. This observation may be relevant to other tasks involving progressive loading of respiratory muscles, besides threshold loading.

A previous study of ours (2) showed that beyond a learning period performance is highly reproducible. Consistent with this earlier work, the finding that Pth_max was 86 ± 11% of PI_max on the third test indicates that, after a learning period, it is primarily the strength of the respiratory muscles which limits performance during progressive inspiratory threshold loading. The relatively low Pth_max achieved during the first and second tests (69 ± 17% and 77 ± 16% of PI_max, respectively) indicates a substantial respiratory muscle strength reserve at the point of task failure during these early tests. The unchanged PI_max over the three loaded breathing tests also suggests that factors other than respiratory muscle strength were responsible for performance limitation in the initial tests.

While varying in degree from test to test, systematic changes in breathing pattern were observed in all tests in response to progressively increasing inspiratory threshold load (Figures 2 and 3). VT and E initially increased and then decreased, f progressively increased, and EELV systematically decreased. Hypoxemia, and to a lesser degree hypercapnia, were evident at task failure. We have previously reported such changes, which appear structured to optimize endurance of inspiratory muscles and their capacity to develop force (2). As in the present study, breathing pattern was unconstrained to allow study of the compensatory mechanisms that preserve ventilation with increasing threshold load within a given test, and the changes that occur from test to test.

In general, with successive exposures to the task VT increased and T_I and f tended to decrease at comparative loads. This pattern of breathing allowed TE to increase, potentially increasing endurance capacity by increasing the time available to the inspiratory muscles for recovery before the onset of the next breath (25). However, several observations suggest that it is not the endurance capacity of respiratory muscles themselves that is the primary factor limiting Pth_max during the initial tests. First, because the Pth_max achieved is related to strength only after the first few exposures to progressive threshold loading, the substantial reserve in strength at task failure during the initial tests implies that there is also substantial endurance reserve.

Second, the change in breathing pattern with successive tests increased E with each load, and with it _ext suggesting that the muscles were working harder at each load. While a decrease in TI reduces the pressure-time index (Pth · TI) for each breath, the effect of the concurrent increases in VT and f is to increase _ext and with it O_2resp. It should be noted that pressure-time index is an unsatisfactory estimate of energy consumption in the presence of threshold loads presumably because it fails to account for the component of energy use that relates to velocity of muscle shortening. This is because, unlike with resistive loads, pressure (Pth) and velocity of shortening (flow) are largely independent of each other.

Third, there was no change in respiratory muscle efficiency with the change in breathing pattern between tests. This, together with the finding that PI_max did not change suggests that conditioning of muscles was not a major factor in increasing Pth_max with successive tests.

Fourth, even after accounting for changes in ventilation, there was an increase in O_2resp at task failure with successive tests (22.7 ± 12.6, 32.8 ± 12.6, and 34.8 ± 13.1 ml O₂/L ventilation) indicating that respiratory muscle oxidative capacity was not a limiting factor in the initial tests. Thus, it appears that factors apart from intrinsic muscle function are likely to have limited Pth_max during the initial tests and be responsible for its increase with subsequent tests.

An effect of the changes in breathing pattern with successive tests was to increase E at each inspiratory load so that PET_CO2 was lower and Sa_O2 higher at equivalent loads. It was notable that Sa_O2 and PET_CO2 at task failure were similar in each of the three tests, suggesting that these levels may have a role in determining the load at which task failure developed. The changes in breathing pattern had the effect of delaying the point at which these levels were achieved. While hypercapnia and hypoxia have been shown to decrease diaphragm contractility and endurance (26), the effects of these changes are unsatisfactory explanations for the task failure observed in the present study, as muscle function itself did not appear to be a limiting factor during the initial tests. Furthermore, we have previously shown that diaphragm contractility, as assessed by momentary bilateral phrenic nerve stimulation applied at end-expiration, is diminished only at the highest threshold loads (2), suggesting that contractile failure (fatigue) is unlikely to have been a limiting factor during the learning period. That study, using an identical protocol to that described in the present study, also demonstrated that preventing hypoxemia by provision of supplemental O₂ had no effect on Pth_max, indicating that hypoxemia has little effect on respiratory muscle function during progressive threshold loading.

It is also possible that hypercapnia or hypoxia may influence assessment of respiratory muscle endurance capacity through a sensory effect, with unpleasant associated sensations causing subjects to cease the loaded breathing task before respiratory muscle or ventilatory limitation is present. The finding that Sa_O2 and PET_CO2 were similar at task failure in each of the three tests for each subject suggests that these may be levels of chemical stimuli associated with a critical degree of sensory loading. As such, delay in achievement of these levels through changes in breathing pattern from test to test could be important in allowing the progressive increase in Pth_max. These similar levels at task failure from test to test suggest that increased endurance cannot be attributed to subjects learning to tolerate sensations associated with greater hypoxemia (30) or hypercarbia.

Sensation of respiratory effort is another source of sensory input that may be modified by changes in the breathing pattern adopted, independently of changes in blood gases. Killian and coworkers showed that the sensation of respiratory effort (SENS) is determined primarily by the peak pressure (Pm) developed during inspiration, but also significantly increases with TI, and to a lesser extent with f: SENS = Pm^1.3 · TI^0.52 · f ^0.19 (17). During progressive threshold loading the options for an individual to decrease SENS are limited to modifying TI and/or f, because the characteristics of a threshold-loaded breath dictate that Pm (i.e., Pth), which is the primary determinant of SENS, cannot be altered by changes in the breathing pattern. Both f and TI (and hence SENS) tended to decrease at a given load with successive tests, suggesting that the change in breathing pattern may have acted to reduce the sensation associated with increased inspiratory loads. Consistent with this notion is the finding that subjects perceived a given load to require less effort (RPE) with successive tests. The selection of a pattern of force development that minimizes distress is presumably a response conditioned by sensory experience which derives from receptors in the lungs, airways, and chest wall, from centrally generated signals of voluntary motor command, and from chemoreceptor activity (31). The present findings suggest that these responses can be modulated by learning.

Our findings indicate that sensory rather than muscle conditioning is responsible for the increase in Pth_max from test 1 to test 3. The observed changes in breathing pattern act to increase endurance capacity by decreasing the sensory load (a tendency for decreased TI and f, decreased PET_CO2, increased Sa_O2) and increasing recovery time for inspiratory muscles between contractions (increased TE). These changes are associated with a reduction in RPE and increases in E, O_2resp, and _ext, but all occur well within the limits of inspiratory strength capacity.

There has been considerable debate as to whether respiratory muscle training can actually strengthen muscles and whether stronger muscles can improve exercise tolerance in patients with respiratory disease (10, 12, 30, 32). It is possible that the conflicting results of studies addressing these questions are explained by failure to adequately distinguish improvements in respiratory muscle performance based on a learning effect achieved through sensory conditioning from a change in strength and endurance of the muscles themselves.