INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As in peripheral muscles, it is well known that adaptations to training occur in the respiratory muscles. In the past, many studies have been performed with a diversity of protocols. Studies were done in rats where a chronically imposed resistance was used by means of tracheal banding (1, 2) or a tracheal cannula (3). Biochemical, electromyographic, contractile, and morphological properties of the diaphragm were investigated. The adaptive abilities of the diaphragm and other respiratory muscles to intermittent inspiratory work loads, as used in patient treatment programs, however, have received little investigative attention.

Respiratory muscles can be regarded as mixed from the viewpoint of fiber types and have the capacity to respond to both endurance and strength stimuli (4, 5). The aim of training is to induce increases in maximal inspiratory pressure (PI_max) which would lower the value of the ratio inspiratory pressure generated per breath to PI_max (PI/PI_max) and the tension-time index, TTI, thereby increasing endurance and decreasing the probability of fatigue (6).

In a previous study in rats using an inspiratory resistive load with a frequency of 5 ×/wk, 30 min/d during 8 wk, we showed that with loading, in vitro diaphragmatic force during the force-frequency curve did not increase, TPT was shorter, and fatigue resistance remained unchanged. Loading resulted in a hypertrophy of mainly fast-twitch fibers of the diaphragm (7). We speculated that the absence of an increase in force might be due to overload caused by the high training frequency.

Therefore, the present study was designed to investigate whether low load training imposed by the Hans-Rudolph valve alone may induce training effects, and whether loading at the frequency of 3 ×/wk would increase the in vitro force-generating capacity of the diaphragm. The effects of inspiratory muscle training were evaluated on rat diaphragm contractile properties and diaphragm and scalene morphological characteristics and were compared with a control group that was breathing and moving freely.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Model

Fifteen male Wistar rats, 15 wk old and weighing 360 to 410 g, were randomly assigned to control (n = 5) and training groups (n = 10). After 9 d of conditioning (adaptation to the training setup) the training groups were submitted to an intermittent inspiratory resistance, during 30 min/d, 3 ×/wk, 8 wk while control animals were moving and breathing freely. Training rats were put in a cylinder and were breathing through a Hans-Rudolph device, consisting of an inspiratory and expiratory valve, which was connected to a mask. Five of them only had to overcome the small resistance (SR) of the inspiratory valve, while five other rats had a moderate additional load (MR) by means of a resistance (needle) placed in the inspiratory limb of the Hans- Rudolph device, leaving the expiratory limb free. During the training period the resistance was gradually increased by reducing the diameter of the needle. In the beginning of the training session the internal diameter was 0.8 mm, the final internal diameter was 0.4 mm. The rationale for a progressive load was that animals tended to become apneic if the load was too high from the beginning on. Animals were restrained by means of a fixator. The experimental setup has been previously described (7).

Inspiratory Resistance

To measure the flow resistance of the orifice, we used a rotameter (Kobold instrumentals NV-SA, Type KDG 20, 4 to 40 L/h) and a differential pressure transducer (Validyne MP45, range ± 250 cm H₂O) connected to an amplifier. Signals were recorded on a hot pen recorder (W&W Electronics, Switzerland). Air was sent through the orifice. The pressure was measured at different flow rates (4, 5, 10, 20, 30, and 40 L/h). Finally, the resistance of the needle was expressed as pressure difference per unit flow (cm H₂O/ml/s).

Data Recording during Training

Animals were observed and constantly motivated to keep their snout in the mask during training to reduce instrumental dead space ventilation. Snout pressure was continuously measured by means of a pressure transducer (Validyne MP45, range ± 250 cm H₂O). The signal was amplified and recorded on a hot pen recorder and on a computer using Labdat software (Labdat/Anadat; RHT-Infodat, Montreal, Canada). Signal analysis was done with Anadat.

During the training period of 30 min, each breath was registered. Successive samples of 30 s were obtained over the 30-min training period for each animal. For each sample the following measurements were made on the middle and the last day of training because the resistance applied was similar at these two dates: sum of all negative pressures generated and average per breath (P_neg); sum of all positive pressures generated and average per breath (P_pos); sum of all positive and negative pressures (pressure swing) generated and average per breath (P_snout); maximal inspiratory pressure swing (P_max), being the maximum negative pressure; and breathing rate (BR). Afterwards, the sum and average of each of the abovementioned variables was made for a given training session (30 min) and a given animal. Finally, randomized samples were taken to determine inspiratory time (TI) and duration of respiratory cycle (Ttot). These data were used to determine the TTI of the diaphragm defined as TI/Ttot × P_neg/PI_max with PI_max being the predicted maximal pressure generated by the inspiratory muscles of the rat. PI_max was assumed to be 80 cm H₂O.

Removal and Dissection of Muscle Bundles

After 8 wk, animals were anesthetized with sodium pentobarbital (Nembutal 60 mg/kg, intraperitoneally). They were tracheotomized and a cannula was inserted. They were mechanically ventilated with an O₂-enriched gas mixture (Harvard pump respiratory). The diaphragm was quickly removed through a laparotomy, and immediately immersed in a cooled and curarized, oxygenated Krebs solution containing (in mmol/L): NaCl 137, KCl 4, CaCl₂ 2, MgCl₂ 1, KH₂PO₄ 1, NaHCO₃ 12, glucose 6.5. Two small rectangular bundles (width < 2 mm) from the middle part of the lateral costal region of each hemidiaphragm were obtained by careful dissection parallel to the long axis of the fibers. Both ends of each bundle were tied with silk sutures to serve as anchoring points. The bundles were suspended in a tissue bath containing Krebs solution and continuously aerated with 95% O₂ and 5% CO₂. Temperature was maintained at 37° C using a thermostatically controlled water pump. One end of the bundle was tied to a rigid support, while the other was fastened to an isometric force transducer (Maywood Ltd, Hampshire, UK) connected to a micrometer. Signals were amplified and recorded on computer via analog to digital conversion (DT-2801A) using Labdat (Labdat/Anadat, RHT-Infodat). Both bundles were placed in between two large platinum electrodes. Stimulations were delivered through a Harvard 50-5016 stimulator (Edenbridge, Kent, UK), connected to a power amplifier made from Power One Model HS24-4.8, developed by the Computer Technology Resource Center, University of Virginia (R. J. Evans, 1983). Optimal muscle length (L_o) for peak twitch force was established for each bundle. All subsequent measurements were made a L_o. A 15-min thermo-equilibration period was allotted before starting stimulations.

In Vitro Contractile Properties of the Diaphragm

The following procedures were performed in sequential order: twitch characteristics, maximal tetanic tension, force-frequency curve, and fatigue run.

Twitch characteristics. Twitch tension (P_t), time-to-peak tension (TPT), and half-relaxation time (1/2RT) were obtained as the average values of two successive twitch stimulations (1 Hz). Time-to-peak tension was defined as the time from the start of the contraction to the point where peak tension was achieved. Half-relaxation time was the time from peak tension to the point where tension dropped by half.

Maximal tetanic force and twitch-tetanus ratio. Bundles were stimulated twice at 160 Hz, during 250 ms, with a 2-min interval. Each pulse had a duration of 0.2 ms. Tetanic force (P_o) was recorded as the maximal tension elicited at 160 Hz. Twitch-to-tetanus ratio (P_t/P_o) was calculated for each muscle bundle.

Force-frequency curve. Bundles were stimulated at the following frequencies: 25, 160, 50, 160, 80, 160, 120, and 160 Hz. Each stimulus was separated by a 2-min interval. Force-frequency data were expressed as a percentage of the force at 160 Hz, calculated from the interposed 160-Hz stimulation.

Fatigue run. During 5 min, each bundle was stimulated with 25 Hz at a train duration of 330 ms, every 3 s. Developed force was measured every 30 s.

After these measurements, each muscle bundle was removed from the bath and its length, width, and thickness were measured at L_o. They were blotted dry and weighed. Cross-sectional area (CSA) was calculated by dividing weight by specific density (1.056) and muscle length. All forces were expressed per unit CSA. The remaining diaphragm tissue and other muscles, either respiratory or peripheral, were dissected, trimmed, blotted and weighed. The following muscles were studied: right medical scalene muscle, intercostal muscles (including sternum and chondral parts of the ribs), the musculus gastrocnemius, musculus soleus, musculus plantaris, and musculus extensor digitorum longus from the right hindlimb.

Muscle Area and Fiber Type Analysis

Muscle strips obtained from the costal region of the diaphragm, the middle part of the scalene and gastrocnemius were put on a cork holder, with the muscle fibers perpendicularly to the surface of the cork. They were quickly frozen in isopentane cooled by liquid N₂. Serial cross-sections parallel to the cork were cut at 10 µm thickness with a cryostat kept at 20° C. For the histological analysis two sections of each muscle were taken for routine hematoxylin-eosin staining. Moreover, on the basis of their staining reactions for myofibrillar ATPase, pH 4.5 and 4.3, muscle fibers were classified as either type I, type IIa, or type IIx/b fibers (8). Analysis of type IIx and IIb by using light paraformaldehyde fixation at pH 10.4 (9) revealed that type IIb fibers represented only 20% of the pooled type IIx/b population. Data of type IIb fibers were pooled with those of type IIx and reported as type IIx/b. Training affected type IIb and IIx fibers in a similar way. Fiber type proportions and CSAs were determined by use of a Leitz Laborlux S microscope (Wetzlar, Germany) at ×20 magnification, connected to a computerized image system (Quantimet 500; Leica, Cambridge Ltd, UK). Fiber diameters were automatically determined by measuring the greatest diameter perpendicular to the long axis. CSAs were determined from the number of pixels within the outlined borders. Between 100 and 250 fibers of each muscle were used to calculate mean diameter, CSA, and fiber proportion of all fiber types. The total interstitial area (also determined from the number of pixels within the outlined borders) was measured relative to fiber area. In addition, for all the muscles studied, CSAs were corrected for the shortening occurring from optimal length according to a formula obtained from previous studies performed in our laboratory. For the diaphragm, the correction factor obtained in our study on 32 rats was similar to that of Prakash and coworkers (10). Our data showed that L_o in the diaphragm was on the average 1.53 longer than excised length. The correction factor was 1.07 (n = 16) and 1.04 (n = 40) for the scalene and gastrocnemius, respectively.

Statistical Analysis

Values in the text, tables, and figures are mean ± SD. Results of two bundles from one animal were averaged and analyzed subsequently. Data from control and training animals were compared by one-way analysis of variance. Differences between means were assessed using a subsequent Gabriel/Dunnet multiple range test (11). Multivariate analysis of variance with repeated measurement was used to assess differences as a function of time or stimulation frequency. Statistical significance was set at p < 0.05. Statistical analysis was performed using the SAS statistical package (SAS Institute, Cary, NC). Statistical power was obtained by the formula of Snedecor and Cochran (12).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

One rat of the small-resistance group died during the training period. The cause of its death was not determined.

Validation of Training Model

Based on a previous study (7), it was known that in our training setup, animals did not generate hypercapnia. Indeed, end-tidal CO₂ and breathing rate monitored throughout the whole training session by means of a microcapnometer (Model 0151-003L; Columbus Instruments, Columbus, OH) showed that these values remained stable throughout the whole training session (7).

Inspiratory Resistance

From the first to last day of training, different needles with a progressively smaller internal diameter were used. On the first day, animals had to overcome a resistance of 0.7 cm H₂O/ml/s at a flow rate of 5 ml/s. On the last day, this was 6.4 cm H₂O/ ml/s at the same flow rate.

Respiratory Characteristics

Results of both groups in the middle (Day 12) and at the end of training are shown in Table 1. The inspiratory time and respiratory cycle were similar in both groups, whereas TTI was higher with a greater training load regardless of the training day (p < 0.01). BR was significantly slower in MR compared with SR but reaching statistical significance only on Day 12 (p < 0.01). Finally, the SR group breathed significantly slower on the last day of training compared with Day 12 (p < 0.05).

Animals who trained with a moderate inspiratory resistance generated overall higher pressures than animals that had to overcome only the small resistance of the valve. On the twelfth (middle) day of training, SR and MR groups generated a P_snout of 0.6 ± 0.06 (P_neg: 0.3 ± 0.03) and 4.8 ± 1.6 (P_neg: 3.0 ± 1.0) cm H₂O, respectively (p < 0.001). On the last day of training this was 0.5 ± 0.07 (P_neg: 0.2 ± 0.05) and 4.0 ± 1.5 (P_neg: 2.5 ± 1.1) cm H₂O, respectively (p < 0.01). In both groups, there was no difference in P_snout between the two days of training (Figure 1).

View larger version (12K): [in this window] [in a new window]   Figure 1.   The snout pressures (Psnout), detailed as positive and negative pressures, generated in the small resistance group (solid bars) and moderate resistance group (open bars) on the middle and last day of training. Within 1 d between groups: **p < 0.01.

The average inspiratory peak pressure, P_max on the twelfth day of training was 1.4 ± 0.1 and 14.4 ± 5.0 cm H₂O in SR and MR, respectively (p < 0.001). On the last day of training this was 1.1 ± 0.1 and 12.2 ± 3.9 cm H₂O, respectively (p < 0.0001), with peak pressures occasionally reaching values up to 10 and 30 cm H₂O, respectively. Only in the SR group was there a difference in P_max between both days of training (p < 0.01).

Body and Muscle Mass

Intermittent inspiratory resistive loading did not affect body weight nor diaphragm, scalenus medius, or gastrocnemius weight as shown in Table 2. Weight of intercostal muscles, soleus, plantaris, and extensor digitorum longus was also unaffected by training.

Diaphragm Geometric Properties

Diaphragm bundle length (C: 19.50 ± 1.45, SR: 19.34 ± 2.29, MR: 19.15 ± 1.90 mm), and thickness (C: 0.55 ± 0.13, SR: 0.59 ± 0.20, MR: 0.55 ± 0.09 mm) were also unaffected by training. Diaphragm width was similar in the three groups (C: 1.70 ± 0.30, SR: 1.56 ± 0.19, MR: 1.73 ± 0.22 mm).

Diaphragmatic Contractile Properties

Diaphragm twitch characteristics, tetanic tension and twitch-tetanus ratio are shown in Table 3. Little or no changes were seen between the three groups.

The force generated with increasing stimulus frequency was similar in all groups (Figure 2).

View larger version (10K): [in this window] [in a new window]   Figure 2.   In vitro diaphragmatic force generation as a percentage of maximal tetanic tension (Po) versus stimulation frequency in control (open circles), SR (closed circles), and MR (closed triangles).

Force decline during the fatigue run was similar in the three groups, either expressed in absolute terms or as a percentage of the initial force (C: 47 ± 6, SR: 51 ± 13, MR: 45 ± 5% of the initial force, NS). The decrease in maximal tetanic force during the force-frequency protocol was similar in the three groups (C: 15 ± 11, SR: 20 ± 21, MR: 16 ± 12%, NS).

Histology

Histological examination of routine hematoxylin-eosin stained slides of diaphragm, scalenus medius, and gastrocnemius showed no differences between the three groups.

Morphology

Diaphragm muscle fiber dimensions were affected differently according to the training stimulus (SR or MR). For type I fibers a significant hypertrophy was seen in the MR group compared with the SR (+25%) and control group (+44%) (p < 0.05). For type IIa fibers both trained groups showed a hypertrophy compared with the control group (MR: +34%, SR: +30%, p < 0.05). Only in the MR group type IIx/b fibers showed a hypertrophy compared with the control group (+32%, p < 0.05) (Figure 3). The same picture was seen in the diameter of the fibers. No differences were seen in fiber type proportions between the three groups. The total area taken up by interstitial tissue relative to diaphragm fiber area was significantly less in MR (8 ± 2%) and in SR (10 ± 4%) compared with that in C (17 ± 3%) (p < 0.001). For the musculus scalenus medius and gastrocnemius no differences in fiber dimensions between the groups were observed. The pooled value for the type IIx/b fiber cross-sectional area for the scalenus medius was 3,428 ± 476 µm². Pooled values for gastrocnemius internus were 2,452 ± 631, 2,094 ± 614, and 2,512 ± 568 µm² for type I, IIa, and IIx/b fibers, respectively.

View larger version (16K): [in this window] [in a new window]   Figure 3.   CSA (cross-sectional area) of the three fiber types of the diaphragm (type I, IIa, IIx/b) in control (open bars), small resistance (hatched bars), and moderate resistance groups (solid bars). A statistically significant hypertrophy was obtained in MR compared with SR and control for type I fibers, in MR and SR groups compared with control for type IIa fibers, and in the MR compared with the control group for type IIx/b fibers. *p < 0.05.

Power Analysis

The number of animals was chosen on the basis of a power analysis. According to this analysis, the statistical power of our data was 85 and 84% for P_o and P_t, respectively, for a difference between the control and training groups of about 10 and 14%, respectively, as observed. As a consequence, the type II error was 15 and 16% for P_o and P_t, respectively. Furthermore, the power analysis showed that 78 and 86 animals would be respectively needed for the increase in P_t and P_o obtained in the present study to reach statistical significance. For the diaphragmatic morphometric changes, power was 99% for type I CSA (increase of 44%), 100% for type IIa CSA (increase of 33%), and 94% for type IIx/b CSA (increase of 32%). The number of animals needed for these changes to reach statistical significance was relatively low (3, 1, and 5 animals for type I, IIa, and IIx/b, respectively) as was the type II error.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Present data show that there were no changes in body weight and muscle masses, and little or no changes in diaphragm contractile properties after 8 wk of inspiratory resistive loading at a frequency of 3 times a week. Nevertheless, morphological changes were found in the two training groups despite the fact that only low pressures were generated during loading. Cross-sectional area of all muscle fibers increased in the MR group, whereas in the SR group only type IIa fibers increased compared with control animals (Figure 3). Probably the training load and the respiratory characteristics as TI/Ttot and BR are determinant factors in the induced effects.

Bellemare and Grassino (13) have pointed out that the duty cycle is an important determinant of inspiratory muscle endurance. For the diaphragm, both the pressure and the duty cycle are incorporated in a tension-time index (TTI): TI/Ttot × PI/PI_max = TTI ~ 1/endurance.

As in a previous study (7), the magnitude of the load in the present study is mainly derived from a high TI/Ttot (0.59 versus 0.39 at rest) (14), and a high BR (> 145 breaths/min versus 109 at rest) (14), so that mainly endurance training is expected to be performed.

Different morphological adaptations in the diaphragm between the two training groups appeared to depend on the load, since TTI was higher in the MR group compared with the SR group in the overall training period. As TI/Ttot was similar in both groups, this discrepancy in TTI was the result of an increased P_snout in the MR group, as expected. At the end of the training period, P_snout ranged from 2.8 to 6.6 cm H₂O (versus 0.4 to 0.6 cm H₂O in the SR group), reaching an average P_max up to 12 cm H₂O (versus 1 cm H₂O in the SR group). This corresponds with a load on the respiratory muscles of on the average 5%, with peaks up to 15% of the assumed maximal inspiratory pressure (i.e., 80 cm H₂O). However, the pressure was measured at the airway opening and hence, it does not include the pressure necessary to overcome airway resistance and lung elastance. Compared with the study of Bisschop and colleagues (7), the load in the present study was lower. This was the result of both a lower frequency of training (3 ×/wk versus 5 ×/wk) and a lower resistance reached at the end of the training period (6.4 versus 18.4 cm H₂O/ml/s). Because the MR group had to overcome a higher resistance, the BR was slower along the whole training period and especially in the middle of the training period compared with SR (p < 0.01). At the end, however, differences disappeared because the SR group probably adapted to the training setup and consequently became less stressed and so breathed more slowly. Indeed, the MR group showed the slowest breathing rate with the longest inspiratory time (NS) which may explain a higher load put on the diaphragm (Table ), and also an overall hypertrophy of diaphragm fiber types.

In the present study, the changes in contractile properties observed after training did not reach statistical significance although, as shown in Table , P_t and P_o in trained animals were increased by about 14 and 10%, respectively, compared with control animals. The fact that these increases did not reach statistical significance might at first sight be attributed to the small sample size used in the study. However, power analysis revealed first that the probability of a type II error made with such small sample size was relatively low (16 and 15% for P_t and P_o, respectively), showing thereby that the number of animals was adequate. Second, it showed that 78 and 86 animals would have been respectively needed for these increases to reach statistical significance. Therefore, the absence of statistically significant changes in contractile properties after training did not appear to be due to inadequate sample size. It should also be noticed that the number of animals chosen in the present study was sufficient to observe significant changes in diaphragm morphometry, which were, however, not reflected in the contractile properties.

An overall hypertrophy of the diaphragm was seen in the MR group compared with C with no changes in muscle mass. Several explanations for this apparent discrepancy may be put forward. First, fibers might be fewer but larger in the trained diaphragm. However, muscle proportions did not change with training nor can training induce a decrease in fiber number. Second, fibers in the trained diaphragms might be shorter but larger in CSA, although geometric properties show no differences in diaphragm length between the groups in the present study. Third, increases in muscle fiber diameters might be due to decreases in interstitial tissue, yielding no net change in muscle mass. This latter speculation was confirmed because the area of interstitial tissue relative to fiber area was significantly less in SR (10 ± 4%) and MR (8 ± 2%) compared with C (17 ± 3%, p < 0.001).

In the study of Bisschop and colleagues (7) rats were trained in a similar way during 8 wk for 5 ×/wk and 30 min/d but a control group outside the cylinder was absent. In this study, the results showed a hypertrophy of fast-twitch diaphragmatic fibers in the inspiratory resistance group (comparable to our MR group), while forces did not increase compared with animals that had to overcome only the resistance of the Hans-Rudolph valves (comparable to our SR group). The latter observation was suggested to be due to overload. In the present training setup, where rats were trained at a lower load (because the resistance was smaller) and at a frequency of only 3 ×/wk, no increase in force was observed either. Nevertheless, an overall fiber hypertrophy in MR and a type IIa fiber hypertrophy in SR were still observed. However, a higher frequency (5 ×/wk) resulted in an additional type IIa and IIx/b fiber hypertrophy in MR compared with SR, which was absent in the present study. Thus, as in peripheral muscles, training intensity appears to determine diaphragmatic adaptations.

The intrinsic difference and/or differences in the load they had to bear between the diaphragm and scalenus medius was confirmed because muscle mass and morphometry of the latter muscle was unchanged with this type of loading. No changes were seen in the gastrocnemius, illustrating the specific response of the diaphragm to inspiratory resistive loading.

In other experimental models, chronic respiratory loads, as tracheal banding (1, 2) or a tracheal cannula (3) were directly imposed to the respiratory system. Keens and coworkers (1) showed that 5 to 6 wk of extratracheal banding in rats produced increases in oxidative enzyme capacity and proportion of type I muscle fibers in diaphragm. However, they did not assess the functional impact of these changes. Moreover, the characteristics of the load were not described in terms of muscular effort or airway pressure developed, but they were likely to be very high. Prezant and associates (2) reported a small increase in diaphragm mass to body weight ratio, a decrease in its contractility, and an increased endurance after long-term (24 to 28 wk) continuous respiratory resistive loading. This was associated with an increase in both the proportion and cross-sectional area of type I fibers. Further, increased maximal tension with unchanged endurance was observed after 5 wk of chronic increased resistive loading produced by implanted tracheal cannulas (3). These investigators, however, did not assess the histological effects on the diaphragm. Nevertheless differences in protocol may be responsible for the discrepant results obtained in these studies. First, in all of them, an inspiratory as well as expiratory load was chronically imposed. This was the most significant difference between these studies and our model. A frequency of 3 ×/wk, and a load imposed intermittently (30 min/d) could be used in a patient training model, whereas a chronically imposed load would be impossible to apply. Moreover, the magnitude of the resistive load might be quite different from one study to the next. Also, nutritional status of the animals and infections caused by surgery may contribute to divergent results.

Akabas and associates (15) and Bazzy and Haddad (16) showed that, in unanesthetized sheep, intermittent respiratory resistive loading improved diaphragm function (oxidative enzyme activity and in vivo diaphragm endurance). But they did not report the effect on diaphragm mass and contractile properties. The load imposed to the respiratory muscles was limited to avoid mechanical failure of the diaphragm (16). Indeed, diaphragm fatigue is associated with acute hypercapnia and failure of the respiratory pump (17).

Because training of the ventilatory muscles must follow the basic principles of training of any striated muscle with regard to the intensity and duration of the stimulus, and the specificity of training, it is surprising that low load (SR) and low training frequency in our model produced effects. Indeed, already a small resistance resulted in hypertrophy of type IIa fibers of the diaphragm; but, in addition, and corresponding to training principles, morphological changes were greater when the load (MR), or the frequency (5 ×/wk) were higher (7).

In humans, objective and quantitative data on the intensity of training are lacking (18). The load has to be high enough to induce training effects as mentioned before, but in published studies the training stimulus is extremely variable from one study to another. Thus, in the study of Larson and coworkers (18), patients trained with a load equivalent to 30% of PI_max had a higher inspiratory mouth pressure than those trained with a load equivalent to 15% of PI_max. However, Belman and Kuei (19) claimed that those two loads improved ventilatory strength and endurance to a comparable degree if appropriate statistical analysis was used in the study mentioned above. Dekhuijzen and coworkers, on the contrary, used a much higher load, equivalent to 70% of PI_max in patients (20), close to the 65% of PI_max in healthy volunteers used by Clanton and coworkers (21). Dekhuijzen and coworkers (20) showed that in patients with a ventilatory limitation of exercise capacity, inspiratory muscle training had an effect on PI_max and on exercise performance that was greater than that of exercise training alone. In models using whole-body endurance exercise (e.g., running, swimming) the intensity of the exercise performed by the diaphragm cannot be easily measured because exercise is not specific enough and not targeted directly to the diaphragm (8, 22). Nevertheless, ventilatory effects of exercise training in humans and animals were described, consisting of enhanced maximal sustainable ventilatory capacity, modest increases in maximal voluntary ventilation and PI_max or PE_max, and improvements in oxidative capacity of diaphragm fiber types (23).

According to Pardy and Rochester (6), one must ascertain that the inspiratory pressure per breath during training is between 50 and 70% of PI_max. In most of the reported studies, the duration of respiratory muscle training was between 4 and 8 wk (18, 28, 29). In a few studies, the training period lasted for as long as 16 to 20 wk (25, 28). Regardless of the type of training, the training intervention has generally been carried out for 15 min, two to three times a day, for 5 to 7 d per week. In the studies in which several evaluations were done throughout the training period, respiratory muscle function generally continued to improve with time (6, 30).

In the present study, we used an animal training protocol comparable to the protocol used in most of the clinical studies. However, we did not know precisely the relative load to put on the respiratory system because PI_max was not measured but assumed to be 80 cm H₂O.

In summary, 8 wk of intermittent inspiratory resistive loading at a rate of 3 times a week resulted in morphological type IIa hypertrophy of the diaphragm with, to the best of our knowledge, low loads. Higher loads resulted in an overall hypertrophy of all fiber types. The high number of repetitions (BR) and the high TI/Ttot presumably stressed the diaphragm. No decrease in diaphragm force was seen with a training frequency of 3 ×/wk.