Training with inspiratory pressure support in patients with severe COPD

Study population

Criteria for inclusion in the present study were: 1) diagnosis of COPD according to the criteria set by the Global Initiative for Chronic Obstructive Lung Disease (GOLD) scientific committee 12; 2) forced expiratory volume in one second (FEV₁) <60% predicted; 3) a breathing reserve at maximal exercise <20% of maximal voluntary ventilation (defined as 37.5 * FEV₁) at peak exercise; 4) peak minute ventilation <50 L·min⁻¹; 5) resting arterial oxygen tension>8 kPa (60 mmHg); 6) pulse oximetry oxygen saturation at maximal exercise ≥85%; 7) aged between 40 and 75 yrs; and 8) written informed consent. The study was approved by the Ethical Committee of the VU University Medical Centre, Amsterdam.

Intervention

Patients were randomly allocated (sealed envelopes) to training with IPS₁₀ or 5 cmH₂O (IPS₅). The exercise programme consisted of exercising on a calibrated cycle ergometer for 8 weeks, three times per week. Each treatment session lasted 45 min, during which time patients were encouraged to exercise for as long as possible. For the first training session, training intensity was set at 65% of maximum work (W_max). When, during this or a subsequent training session, a patient was able to exercise >15 min, the training intensity of the next session was increased by 5% of W_max. During the study, other components of pulmonary rehabilitation, such as breathing retraining, education, relaxation or peripheral muscle resistance training, were not provided. Due to practical constraints, it was not possible to blind the three physiotherapists to the inspiratory pressure support (IPS) intensity patients were training with, but patients were kept naïve with respect to randomisation outcome. All measurements were performed by one independent investigator (A. van 't Hul), who was not involved in the training and who was kept blinded to randomisation outcome.

Noninvasive ventilatory support

Patients received IPS from a flow-triggered ventilator (Raphael; Hamilton AG, Thäzüns, Switzerland). The flow trigger was set at 3 L·min⁻¹, i.e. the lowest possible value, and the inspiratory oxygen fraction at 21%. This ventilator has a capacity to support minute ventilation up to 50 L·min⁻¹ with a response time of 145 ms and a maximal inspiratory flow rate of 180 L·min⁻¹. Patients were attached to the ventilator via corrugated tubing, and a mouthpiece and nose clip were applied. The dead space of the mouthpiece was <15 mL. Ventilator inspiratory and expiratory breathing circuits were separated, which prevented the possibility of rebreathing. On the basis of previous studies, IPS₁₀ was considered adequate pressure support to increase exercise tolerance in COPD patients 13–15. Exercise with IPS₅ was considered appropriate control intervention because IPS₅ during exercise appeared not to affect exercise endurance in patients with COPD 16.

Baseline assessments

Pulmonary function variables were measured (Vmax229; Sensormedics, Yorba Linda, CA, USA) according to the European Respiratory Society recommendations 17. The method of Black and Hyatt 18 was used for the measurement of maximal respiratory pressures. A symptom-limited incremental maximal cycle exercise test was conducted according to the European Respiratory Society recommendations 19. The test was performed on a calibrated, electrically braked cycle ergometer (WLP 904; Lode, Groningen, the Netherlands) at a pedalling rate of 60 revolutions·min⁻¹. During testing, work rate was increased by 10% of the estimated maximal workload until exhaustion. Maximal heart rate was determined using the R–R interval from a three-lead electrocardiogram (HP78351A; Hewlett Packard, Andover, MA, USA). Peak minute ventilation (V′_Epeak in L·min⁻¹ at body temperature and ambient pressure, and saturated with water vapour) and peak oxygen uptake (V′_O2peak in mL·min⁻¹ at standard temperature and pressure, and dry) were measured with a breath-by-breath automated exercise metabolic system (Vmax229). Arterial blood samples were drawn from an indwelling catheter in the radial artery at rest and at maximal exercise and analysed (Rapidlab_▒840; Bayer AG, Leverkusen, Germany). W_max and V′_O2peak were related to the reference of Jones and coworkers 20, 21; V′_Epeak was related to the predicted maximal voluntary ventilation 22. At baseline, before the start of the training period, the acute effects of IPS₁₀ and IPS₅ were evaluated during exercise on constant work rate cycle endurance in all patients. Data on the acute responses to IPS during exercise in COPD of 21 patients were included in the current paper 16.

Outcome measures

Walking distance was measured using the incremental shuttle walking test, which was carried out twice at baseline and also at post-training. Results of the second tests were used in the analysis 23. Symptom-limited exercise endurance was determined with a constant work rate cycle endurance test at 75% of the previously determined W_max. End-exercise physiological responses from the baseline test were compared with post-training responses at the isotime (the very moment that the baseline test was ended). The outcome of this endurance test has been shown to be highly reproducible in patients with COPD 24. Health status was assessed using the St. George's Respiratory Questionnaire (SGRQ). The SGRQ is a self-report questionnaire and scores health status in three areas: symptoms (dyspnoea, wheezing and coughing), activity (the severity to which activities of daily living are impaired by dyspnoea) and impacts (the influence of respiratory symptoms on social participation) 25. An improvement of more than 4% in total score (expressed as a percentage of the maximum score) is considered to be clinically relevant 26.

At baseline, all measurements were performed within 2 weeks prior to the start of the training period. Post-training measurements took place within 2 weeks after the final training session. All exercise tests were performed on separate days to avoid a sequential influence of test order. The assessor actually carrying out all measurements in the current study was not informed about randomisation outcome before all post-training assessments were completed in all patients.

Statistical analysis

Between-group differences were evaluated by means of independent-samples t-test or Mann-Whitney U-test. Within-group differences were evaluated by means of a paired t-test or Wilcoxon signed ranks test. An intention-to-treat analysis was carried out assuming that patients who dropped out had the same change as the average improvement in the IPS₅ group. Pearson's correlation coefficients were calculated to evaluate associations between variables. Values are expressed by their means and sd. For all tests a two-sided significance level of 0.05 was used.

Study population

Baseline physical characteristics, pulmonary function and maximal exercise performance of the patients completing the protocol are provided in tables 1⇓ and 2⇓. At baseline, no statistically significant between-group differences were present in age, stature, pulmonary function, exercise performance or health status. In the IPS₁₀ group, six patients had a Medical Research Council (MRC) score of 3, and eight patients had a score of 4. In the IPS₅ group, five patients had an MRC score of 3 and 10 had a score of 4. All patients received (inhalation) drug therapy; this was not changed during the study. None of the patients was familiar with NIVS before participation in this study and no patient was on home mechanical ventilation.

Acute responses to IPS₅ and IPS₁₀

In the 29 patients completing the protocol, exercise endurance time with IPS₁₀ was statistically significantly higher compared with exercise without IPS (7.6±4.8 versus 4.6±2.6 min; p<0.001), as well as compared with exercise with IPS₅ (7.6±4.8 versus 4.7±2.5 min; p<0.001). Endurance time did not differ significantly between tests without IPS and IPS₅ (4.6±2.6 versus 4.7±2.5 min; p = 0.10).

Effects on exercise training intensity

In figure 1⇓, average exercise intensity during training is depicted for both groups during the 8-week training period. Training intensities expressed as a percentage of the baseline W_max differed statistically significantly between both groups in each week. During the first week, training intensity was increased, according to the protocol, in five patients in the group training with IPS₁₀ (three patients at the first session and another two at the second session) because they were able to exercise for longer than 15 min on the initial load (65% of W_max). In contrast, in the IPS₅ group, only one patient was able to do so. During the first week of training, this already resulted in a significant between-group difference in average training work rate. In both groups, patients exercised effectively for ∼25 min during each session. A moderate but significant correlation coefficient was found between the increase in work rate during the training period in the IPS₁₀ group and the acute increase in cycle endurance resulting from the application of IPS₁₀ measured at baseline (r = 0.45; p = 0.02).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

Training intensities expressed as percentage of maximal power output (W_max) achieved during the incremental maximal cycle test at baseline. Values and error bars represent the mean and sd. In each week of training, patients trained with noninvasive ventilatory support in the form of inspiratory pressure support of 10 cmH₂O (▪) were able to train, on average, at significantly higher intensities (p<0.05) than patients trained with inspiratory pressure support of 5 cmH₂O (▴). All patients started training at 65% of W_max, but, during the first week, five patients in the group training with inspiratory pressure support of 10 cmH₂O were able to increase their training intensity. In contrast, in the group trained with sham inspiratory pressure support of 5 cmH₂O, only one patient was able to do so. This resulted in a significant between-group difference in average training work rate during the first week of training.

Effects of training on exercise performance

The between-group difference in improvement in walking distance was statistically significant expressed either as a distance (31±21 versus 14±31 m; p<0.05) or as a percentage of the baseline test (16±17 versus 3±13%; p<0.05). Also, in the intention-to-treat analysis, the between-group difference remained statistically significant (p<0.05). A post hoc analysis revealed a statistical power of 66% to detect this difference in walking distance. Walking distance improved in 12 out of the 14 patients from the IPS₁₀ group (on average from 253±132 to 284±133 m; p<0.01) and in eight of the 15 patients from the IPS₅ group (on average from 241±107 to 255±128; p = 0.10).

The improvement in constant work rate cycle endurance time was larger in the IPS₁₀ group compared with the IPS₅ group (7.4±5.4 versus 3.9±6.0 min; p = 0.07), but only reached the level of significance when expressed as percentage improvement of baseline (164±124 versus 88±128%; p<0.05). In addition, in an intention-to-treat analysis, this between-group difference in improvement remained significant (p<0.05). In the IPS₁₀ group, 13 of the 14 patients improved (on average from 4.5±2.5 to 9.9±6.1 min; p<0.001), whereas, in the IPS₅ group, 12 of the 15 patients increased their cycle endurance (on average from 4.5±2.8 to 8.4±6.3 min; p<0.05). Figures 2⇓ and 3⇓ show the improvements in incremental shuttle walking distance and cycle endurance, respectively, in both groups.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2—

Change in incremental shuttle walking distance with training. *: p<0.05.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3—

Change in constant-load cycle endurance with training. *: p<0.05.

Table 3⇓ shows the results of the physiological responses during the constant work-rate cycle endurance test. Data of end-exercise responses during the baseline test and at isotime of the post-training test are provided. A statistically significant between-group difference was found for the reduction in minute ventilation (−4.0±3.2 versus −0.7±2.9 L·min⁻¹; p = 0.01). Also, in the intention-to-treat analysis, this between-group difference in reduction in minute ventilation was statistically significant (p<0.05). In contrast to the IPS₅ group, in the IPS₁₀ group, significantly lower values were found for minute ventilation, respiratory rate, oxygen uptake and carbon dioxide output after training (p<0.05).

Effects of training on health status

Table 4⇓ shows the results of the SGRQ. No significant between-group differences were observed at baseline. After training, a significantly lower score in the activity component was only observed in the IPS₁₀ group. No significant between-group difference was observed after training in any component of the questionnaire.

Patient selection

Patients were selected for the current study on the basis of the existence of a significant ventilatory impaired exercise capacity without marked pulmonary oxygen transport deficit. This was based on the authors' assumption that the application of IPS₁₀ might especially benefit this subgroup of patients. Randomisation resulted in well-balanced groups with respect to possible prognostic factors for outcome of exercise training, such as disease severity and maximal exercise parameters. In addition, the included patients had a marked inspiratory muscle weakness. Since weakness makes these muscles susceptible to fatigue this might have contributed to the positive outcome of training with IPS₁₀ in the present study 27.

Interpretation of the effects of training with IPS₁₀

As observed by Hawkins et al. 9, it was found that patients training with IPS₁₀ were able to achieve significantly higher training intensities throughout the 8-week training period. In the last week of training, intensities were 76±14 and 94±27% of the baseline's W_max, in the IPS₅ and the IPS₁₀ group, respectively. These intensities are considerably higher compared with the maximal tolerable exercise intensity during an 8-week training programme in a study by Maltais et al. 5, who found a mean training intensity of 60% during the final week. Recently, Emtner et al. 28 reported exercise intensities up to 96% of baseline's W_max in patients with a similar degree of airway obstruction, training on room air. These between-study differences in maximal tolerable exercise intensity are not readily explained, but may be due to a variation in study populations, differences in the amount of encouragement by the supervisors of the training, or the way W_max was determined 29.

In the group training with IPS₁₀, a two-fold larger increase in constant work rate cycle endurance was found. Unfortunately, no data are available with respect to the minimal change required to represent a clinically meaningful change. However, the increase in cycle endurance appeared to be higher than in other studies of COPD patients with a similar airway obstruction receiving training on a cycle ergometer without NIVS 4, 30, and was also considerably higher than the effect of a long-acting bronchodilator on exercise endurance 31. Similar to the results of the current study are the greater improvements in cycle endurance obtained with training with NIVS in the studies of Hawkins et al. 9 and Costes et al. 10. By contrast, Bianchi et al. 8 did not observe a significant differential effect in either training intensity or exercise performance between COPD patients training with and without proportional assist ventilation (PAV). Possible explanations for the contrasting findings between Bianchi's study and the current and other studies 7, 9–10 are patient characteristics, such as the relatively mild ventilatory impairment and the relatively preserved exercise performance (W_max = 85 watt). Furthermore, the intolerance for NIVS during exercise training in Bianchi's study might have resulted from the use of a nasal mask instead of a full-face mask or mouthpiece. During strenuous exercise, patients may prefer to breathe through their mouth rather than through their nose.

Dyspnoea scores were not rated during exercise in the current study. Despite the reported reliability of exertional dyspnoea measurement by some, it is the experience of the authors, as well as of others, that measures to rate symptoms (Borg scores, visual analogue scales) may vary to a great extent within individuals during similar exercise loads and duration on repeated testing. In addition, it was anticipated that ratings were impractical during exercise with NIVS because of difficulties with verbal communications, acknowledging that ratings of dyspnoea may interfere with patients' necessary concentration to exercise with NIVS. The reduction in minute ventilation at identical work rate and/or duration may improve exercise performance in COPD, because of the reduction in dynamic hyperinflation 32. This reduction might be induced by the lower respiratory rate found in the IPS₁₀ group. Interestingly, the effects of training with IPS₁₀ on breathing pattern, that is, a reduction in respiratory rate, show great similarities with the acute effects of IPS₁₀ 13, 16. This suggests that training with IPS₁₀ may induce a breathing retraining effect. Alternatively, training with IPS₁₀ might have resulted in a superior improvement in lower limb muscle function compared with training with IPS₅. Unfortunately, muscle function tests were not included in the assessment protocol of the current study.

Noninvasive ventilatory support type and settings

Different modalities of NIVS have been used successfully to improve exercise tolerance acutely in patients with COPD, such as continuous positive airway pressure, IPS and PAV, either in combination with positive end expiratory pressure or not. In addition, marked differences in pressure intensities have been used. In a systematic review on the acute effects of NIVS in patients with COPD, no significant heterogeneity was observed between the effect sizes of these modalities 6. This suggests equal efficacy of continuous positive airway pressure, IPS and PAV, in improving exercise training tolerance. In the studies reporting positive effects of training with NIVS in COPD patients, two studies used IPS 7, 10, and in one study PAV was used 9. In the present study, the effects of training with two fixed levels of IPS, that is 5 and 10 cmH₂O, were contrasted. It was decided the IPS intensities would not be individualised for methodological reasons, i.e. standardisation of the intervention. It was also anticipated that it would not be feasible to titrate individual inspiratory pressure or positive end expiratory pressure settings in the absence of the possibility to measure actual respiratory mechanics. IPS₅ was anticipated to represent an adequate control intervention because it was found not to affect exercise endurance significantly in patients with COPD, despite a small improvement in breathing efficiency 16. Also, in the patients participating in the present study, no difference was observed between exercise endurance with IPS₅ compared with exercise without IPS. Future research may be directed at the effect of individualised inspiratory pressure settings and/or the addition of positive end expiratory pressure on the efficacy of NIVS during exercise training.

Limitations of the study

Obviously, the present study has limitations. First, the main methodological shortcoming of the present study is the lack of ability to blind the physiotherapists who provided the training to the treatment allocation. The resulting expectation bias may have resulted in differences in the extent to which patients were encouraged by the therapists during the exercise training. The amount of encouragement is likely to be a factor influencing a patient's motivation to train harder. However, to minimise this influence, physiotherapists were not informed about the acute effects of IPS₁₀ at baseline in individual patients. Moreover, the exercise protocol was standardised and strictly adhered to. Secondly, differential effects of training with NIVS directly after the training episode were evaluated. There was no follow-up. No data were available in the literature on differential effects of training with NIVS at the time when the authors were preparing their study. In addition, after the training period, patients were offered continuous care, including additional training interventions and educational sessions. This was offered on the basis of individual need and thus varied between individuals. Thirdly, a significantly lower value of the SGRQ activity component was observed only in the patients trained with IPS₁₀; no between-group difference was found. Apparently, the extra gain in exercise performance following training with IPS₁₀ produced no additional gain in perceived health status. This finding confirms that changes in exercise performance and quality of life resulting from rehabilitation are independent 1. Alternatively, the present study may be underpowered to detect statistically significant between-group differences in change due to training. Finally, the improvement in walking distance appeared to be low in this study. It must be noted, however, that patients were offered aerobic (cycle) exercise training as a monotherapy. Specificity of training might be another possible explanation, i.e. training on a cycle is likely to increase cycle endurance rather than walking distance. Also, other important components of pulmonary rehabilitation potentially contributing to improvements in health status, such as peripheral and respiratory muscle strength training, education and psychosocial support, were not provided. Therefore, if exercise training with NIVS is considered, the authors recommend combining it with other rehabilitation components, based on the individual needs, using the potential synergistic effects 33, 34.

Clinical implications

Practical drawbacks may accompany the use of NIVS as an adjunct during training. First, NIVS is a labour-intensive type of intervention, because of the required one-to-one patient–therapist ratio, and will add to the cost of rehabilitation programmes. Secondly, discomfort might be associated with the use of a mouthpiece or mask during NIVS. Although, in the present study, no patients dropped out on these grounds, the authors are aware that, in another study, a substantial number of patients was lost for this reason 8. During the recruiting phase of the present study, the authors only came across a few patients who, after a trial session, resigned from further participation in the study because of discomfort related to the application of the NIVS. Again, this may be due to the patients that were selected for this study, i.e. patients with a clear ventilatory impaired exercise capacity.

Despite the selection of patients on the basis of a ventilatory impaired exercise capacity, there was still a large variability found in the extent to which patients were able to benefit from training with IPS₁₀. In the current study, an association was observed between the increase in training intensity during the training period and the acute increase in training tolerance at baseline as a result of the application of IPS₁₀. This indicates that patients who showed markedly acute improved exercise tolerance on IPS₁₀ also tended to benefit most from training with IPS₁₀. Regrettably, the predictive validity of the acute response was low. Consequently, the authors suggest that a single trial of exercise performance on IPS₁₀ can be used to identify patients who might benefit from this intervention. Subsequently, training with IPS₁₀ could be offered to patients unable to tolerate a high training intensity and in whom the application of IPS₁₀ results in a significant increase in exercise tolerance.

In summary, the current study demonstrates the additional effects of training with inspiratory pressure support of 10 cmH₂O on exercise training intensity, constant work rate cycle endurance and shuttle walking distance in patients with chronic obstructive pulmonary disease and a ventilatory limited exercise capacity. Inspiratory pressure support of 10 cmH₂O may be considered as an adjunct during exercise training aimed at high-intensity training, yielding greater improvements in exercise performance.