Inspiratory Muscle Training Potentiates the Beneficial Effects of Proportional Assisted Ventilation on Exertional Dyspnea and Exercise Tolerance in COPD: A Proof-of-Concept Randomized and Controlled Trial

Abstract

During pulmonary rehabilitation, a subset of subjects with COPD requires adjunct therapy to achieve high-intensity training. Both noninvasive ventilation (NIV) and inspiratory muscle training (IMT) are available to assist these subjects. We aimed to prime the respiratory muscles before NIV with IMT, anticipating additive effects for maximal exercise tolerance (T_lim) and dyspnea/leg fatigue relief throughout the exercise as primary outcomes. Changes in the respiratory pattern were secondary outcomes. COPD subjects performed a total of four identical constant work rate tests on a cycle ergometer at 75% of maximum work rate, under control ventilation (SHAM, 4 cm H₂O) or proportional assisted ventilation (PAV, individually adjusted), before and after 10 sessions of high-intensity IMT (three times/week) during 30 days. Two-way RM ANOVA with appropriate corrections were performed. Final analysis in nine subjects showed improved T_lim (Δ = 111 s) and lower minute-ventilation (Δ = 4 L^.min⁻¹) at exhaustion, when comparing the IMT effects within the PAV modality (p = 0.001 and p = 0.036, respectively) and improved T_lim for PAV vs. SHAM (PAV main-effect, p = 0.001; IMT main-effect, p = 0.006; PAV vs. IMT interaction, p = 0.034). In addition, IMT + PAV association, compared to PAV alone, resulted in lower respiratory frequency (IMT main-effect, p = 0.009; time main-effect, p < 0.0001; IMT vs. time interaction, p = 0.242) and lower inspiratory time related to duty cycle (IMT main-effect, p = 0.018; time main-effect, p = 0.0001; IMT vs. time interaction, p = 0.004) throughout exercise. The addition of IMT prior to a PAV-supported aerobic bout potentiates exercise tolerance and dyspnea relief and induces favourable changes in ventilatory pattern in severe COPD during high-intensity training (Brazilian Registry of Clinical Trials, number RBR-6n3dzz).

Keywords:

• COPD
• noninvasive ventilation
• inspiratory muscle training
• cardiopulmonary exercising test
• endurance
• dyspnea

Introduction

Chronic obstructive pulmonary disease (COPD) is a devastating disease with characteristic reductions in physical activities of daily living – which impact increasing cardiovascular morbidity [1,2]. In addition, subjects are trapped in a vicious circle of inactivity, disarranged respiratory neuromuscular networks, and comorbidities, all closely linked to activity-related dyspnea [3]. Although dyspnea and exercise intolerance share a multifactorial basis, key physiological components are increasingly recognized, such as dynamic hyperinflation (DH) and intrinsic respiratory muscle weakness, the latter of which has been gaining renewed interest [3,4].

In this context, noninvasive ventilation (NIV) and inspiratory muscle training (IMT) as stand-alone interventions might reduce the neural process of dyspnea during exercise in COPD subjects. Both share overall common effects in reducing neural respiratory drive [5–7] and improving breathing pattern [8–10], producing a favourable respiratory muscle recruitment pattern and equalizing respiratory muscle energy demands to increase energy supplies to peripheral muscles [3,11,12]. Of note, it has been hypothesized that both modalities individually contribute to decreasing total inspiratory muscle work, after acute utilization of NIV [6] or 8 weeks of IMT [9] in COPD. Accordingly, further potential advantageous interactions could hypothetically arise from the coordinated combination of (i) reduction in inspiratory muscle activation with NIV support and (ii) structural muscle remodelling by inducing an increase in oxidative type I fibres after IMT, proved by external intercostal muscle electromyography [10], or muscle biopsy [13], respectively. Moreover, severe deoxygenation [14] and restricted blood flow [15] of intercostal inspiratory muscles during exercise have previously been described and, conversely, increased oxygenation of intercostal muscles after IMT in heart failure [16]. Hence, we can expect an additive effect of priming the respiratory muscles with IMT, optimizing the overall respiratory muscle condition to apply NIV during the training.

According to this theoretical basis, we designed this cross-over proof-of-concept study aiming to explore the potential for additive effects through a dual-approach, first applying ten sessions of high-intensity inspiratory training in chronically stressed and overloaded respiratory muscles and further unloading them with noninvasive proportional assisted ventilation (PAV), anticipating improved exercise tolerance and dyspnea mitigation throughout the exercise.

Methods

Participants and design of study

This study was evaluated and approved by the local University ethics committee and strictly followed the precepts of the latest Declaration of Helsinki, good medical practice, and CONSORT (Transparent Reporting of Trial) recommendations. This is a prospective, randomized and cross-over study, carried out from March to August 2016. In total, 54 subjects with COPD were screened consecutively for inclusion/exclusion criteria, after approval by the ReBEC (Brazilian Registry of Clinical Trials, number RBR-6n3dzz), and followed up at a specialized clinic. Subjects over 40 years of age, former smokers of both sexes and at least 4 weeks free of exacerbations, without comorbidities such as bronchial asthma, heart failure, bronchiectasis, and pulmonary interstitial disease, except for controlled systemic arterial hypertension (SAH), and with maximal inspiratory pressure (MIP) <100 cm H₂O (see supplementary material), were admitted to the study. Subjects who were unable to perform the proposed stress tests, nonattendance rate >20%, active participation in a rehabilitation program, and severe intercurrences (e.g. angina cordis) were excluded. The protocol and critical reasons for exclusion and loss of follow-up are outlined in the adapted CONSORT flow chart (Figure 1). After the signature of the consent term and inclusion of the subject, the study was carried out through three stages. In the first stage (Figure 1), which lasted approximately 2 weeks, the subjects underwent clinical evaluations, including weight/height measurements with calibrated instruments, NIV familiarization tests on a cycle ergometer, pulmonary function tests, including meticulous MIP assessments. Next, an incremental cardiopulmonary exercise test (CPET) was performed on a cycle ergometer up to the limit of tolerance, intending to attain the maximum power (watts, W). After randomization by a computer-generated sequence of NIV tests, the first stage was concluded on two separate days – within 1 week – exclusively designed for endurance exercise at 75% of the maximal power attained at CPET, under distinct NIV support. During the second stage (Figure 1), subjects were submitted to high intensity IMT (80% of the MIP) for 10 sessions, within a maximum of 30 days. In the final stage (Figure 1), at the end of the IMT, in a maximum of 1 week, repeated cycle ergometer endurance tests were performed, with the same sequence of randomization as the first stage for PAV against a control condition (SHAM) (Figure 1).

Figure 1. Flow-chart adapted from CONSORT 2010.

Clinical evaluation, lung function tests and familiarization

Clinical assessment, including a detailed imaging study (thorax Rx, and CT as available) and verification of comorbidities using interview and clinical tests, was performed by a pulmonologist. Thereafter, anthropometric measurements, familiarization tests, and pulmonary function assessment – always in the morning – were performed by other involved researchers. Clinical stability and optimization of the drugs for COPD were strictly considered. Regular use of medications was mandatory and strictly checked before the exercise tests. The MIP and MEP measurements were performed following the ATS/ERS recommendations [17], in a digital manuvacuometer device recently calibrated by the manufacturer (M300, GLOBALMED, Porto Alegre, Brazil, 2010). Thus, we considered the mean-value measure during 1 s at the peak taken by the Müller maneuver for MIP. Reference values for the Brazilian population were used [18] and other details have been previously published [19]. Complete spirometry tests with a bronchodilator (400 μg Salbutamol^®) and lung diffusion capacity for carbon monoxide (DLco) were obtained, both following standard recommendations [20,21] and using previously published reference values [22,23]. Familiarization with cycle ergometry and PAV were performed together; the subjects attended on a separate day, and after the appropriate device settings (presumed to also contribute importantly to mask interface desensitization), were invited to perform a cycle ergometry bout to the limit of tolerance with PAV. Short-time bouts with PAV preceded the bout to the limit of tolerance, aiming to instruct the subjects about the ideal pedaling rate, dyspnea and leg effort scores, and criteria for exercise termination (see Supplementary Material for more details).

Incremental CPET

The incremental CPET followed the routine of our laboratory and has been previously published [24]. Briefly, after strict recommendations for abstinence from stimulants and depressants, using only the habitual medication, the individuals were stimulated to pedal to 50 cycles min⁻¹, after 2-min of rest and 2-min of warm-up, toward maximum work rate tolerance, that is, when they could not pedal for more than 10 s over 40 cycles min⁻¹ under the strong stimulus of the examiner, and presented subjective signs of intense exhaustion. The initial power of “0” W during warm-up was increased by 5–10 W min⁻¹ when the forced expiratory volume in the first second (FEV₁) was less or greater than 1 L, respectively. Breath-by-breath oxygen consumption (V´O₂), exhaled carbon-dioxide (V´CO₂), minute-ventilation (V´_E), and respiratory-rate (f_R) and tidal volume (V_T) components were measured through an Vmax™ 229 Encore metabolic chart (SensorMedics, Yorba Linda, CA, USA, 2011), calibrated at two moments with high precision gases (GAMA-GASES, São Paulo, Brazil) before each test. Heart rate (HR) and rhythm were monitored using an ECG system (Cardiosoft^®, General Electric, Milwaukee, USA, 2012), integrated into the metabolic system and programed to control an electromagnetically braked cycle ergometer, Vsprint-200p model (Carefusion, Yorba Linda, CA, USA, 2011). Continuous peripheral digital oximetry monitoring (SpO₂) was performed using a DIXTAL DX2010™ system (Dixtal, Manaus, Brazil, 2010).

Constant work-rate cycle ergometry (CWC)

In the first and third stages of this study, each subject underwent two repeated endurance tests, with the same target power (75% of the maximum load in the incremental CPET), with the same saddle height, same professionals, identical monitoring system, NIV peripherals and supplies, and in very-close conditions of temperature and humidity in the laboratory environment (21 ± 0.9 °C). Subjects underwent tests up to maximum steady-state time-to-exhaustion (T_lim) and hemodynamic data (systemic blood pressure and heart rate) were collected every 2-min. Perception of leg fatigue and dyspnea (Borg 0-10) were scored each minute during the tests and at T_lim. The same criteria for incremental CPET for interrupting the tests were used for CWC.

Noninvasive ventilation

The NIV system was administered as described below. Using a V60™ ventilator (Philips Respironics, Carlsbad, CA, USA, 2015), the tubular connections were adapted to a naso-oral mask (small, medium, or large size, Hans Rudolph, Kansas, USA, 2012), tightly-fitted to the head by headstraps. The SHAM condition was adjusted to 4 cm H₂O, considered the minimum-value to overcome the resistance of the connections, according to the manufacturer (Philips Respironics^®). Initial parameters for PAV were individually predetermined based on the respiratory system resistance (R) and elastance (E), according to the runaway technique (see Supplementary Material). Individuals were familiarized with the complete system on a separate day.

Inspiratory muscle training

After the first stage, the subjects were submitted to 10 sessions of high intensity inspiratory training during 30 days, with a frequency of three times a week, always in the morning, composed of six training series, interspersed with a 1-min rest between each one. In the first session each series lasted 1 min and from the second session each series lasted 2-min, interspersed with a 1-min rest. After five training sessions, MIP was reassessed and new training ranges at 80% of the updated MIP were established up to training conclusion. The training was based on previous recommendations [9], using custom-built linear-load Threshold^® devices, combined in series, to provide a load range of 9–82 cm H₂O (see Limitations of the Study).

Data analysis and statistics

After downloading the CPET data to an Excel^® worksheet, the average V'O₂ and power (W) of the final 15 s of the exercise was considered to be representative of the subject’s peak. For the CWC tests, the minute-ventilation (V´_E), respiratory rate (f_R), tidal volume (V_T), positive inspiratory pressure support (PIP) and duty cycle (T_i/T_tot, %) parameters were accessed from the V60^® ventilator and analysed as an average of 15 s intervals and compared isotime in the four replicated tests, before and after IMT. Data are presented as mean ± SD. For comparisons of all the selected exercise data obtained at T_lim, we performed a two-way RM ANOVA, with their respective post-hoc analysis by the Holm–Šídák procedure, taking into account the standard sphericity. When this criterion was violated through the Mauchly test, we used the Greenhouse–Geisser correction. For isotime comparisons, we used the highest time with common measures, at a time-point which all subjects attained before the peak (3rd min). Sample size details are described in the Supplementary Material and significant differences were adjusted for p-value ≤0.05 or lower values, appropriately adjusted for multiple comparisons during the Sidak–bonferroni correction. PRISM 6.0 software was used for graphical and statistical analysis (GraphPad Software^®, La Jolla, California, USA).

Results

Baseline characteristics

The general characteristics, lung function tests, and selected incremental CPET results are described in Table 1. The majority were individuals with COPD grade III/IV obstruction (8/9 subjects) and with P_I, max < 60 cm H₂O at baseline (6/9 subjects). After exclusions and drop-outs, nine subjects completed the study (Figure 1).

Table 1. General characteristics, lung function and CPET.

Variables	Values
General features
Age, years	62 ± 8
Sex, M/F	6/3
Weight, kg	65 ± 13
BMI, kg/m^2	24 ± 4
Smoking history, p/y	85 ± 60
mMRC, score	2.4 ± 1
[Hb], g/dl	15 ± 2
GOLD II/III/IV, n	1/6/2
Pulmonary function
FEV1, % pred	40 ± 15
FVC, % pred	75 ± 18
FEV1/FVC, %	43 ± 8
DLco, mL min^-1 mmHg^-1	9 ± 5
DLco, corrected % pred	46 ± 13
PI, max, % pred	46 ± 12
PE, max , % pred	56 ± 15
Cardiopulmonary E. testing
V´O2peak, % pred	71 ± 10
V´O2peak, mL/min/kg	15 ± 4
Wpeak, Watts	53 ± 29
Wpeak, % pred	45 ± 20
O2-pulsepeak, mL/beat	7 ± 2
V´E, L/min	34.7 ± 9
V´E /MVV, %	86 ± 17
HRpeak, bpm	126 ± 22
SpO2rest, %	94 ± 2
SpO2peak, %	88 ± 7
Dyspnea (Borg), peak (min-max)	7 ± 1 (4–9)
Leg effort (Borg) peak (min-max)	7 ± 2 (4–10)

Abbreviations: BMI = body mass index; mMRC = modified Medical R. Council; FEV₁ = forced expiratory volume in 1 s; FVC = forced vital capacity; DLco = lung diffusing capacity CO; P_I, max = maximal inspiratory pressure; P_E, max = maximal expiratory pressure; V′O_{2 =} oxygen uptake; W = work load (Watts); V′_E = minute-ventilation; MVV = maximal voluntary ventilation; HR = heart rate; CPET = cardiopulmonary E. testing; SpO₂ = pulse oximetry.

PAV effects at T_lim

At T_lim, there was a PAV main-effect pointing to PAV significantly increasing time-to-exhaustion, improving SpO₂ (%), increasing V_T (L) and V'_E (L/min), changing respiratory pattern (decreasing T_i/T_tot and f_R/V_T while increasing V_T/T_i) and thus reaching increased sensory scores, compared with SHAM (Table 2, p < 0.05 for all).

Table 2. Comparisons for main exercise results, both within NIV-modality condition (IMT) and between NIV-modality condition (SHAM or PAV) at T_lim.

Variables	SHAM			PAV			p	p	p
	Pre IMT	Post IMT	Mean Diff	Pre IMT	Post IMT	Mean Diff	PAV	IMT	PAV vs. IMT
Tlim(s)	243 ± 61	283 ± 44	40	341 ± 118^††	452 ± 82^§§§§‡‡‡‡	111**	0.0010	0.006	0.0340
HR (bpm)	120 ± 19	117 ± 17	−3	125 ± 16	116 ± 19	−9	ns	ns	ns
SpO2 (%)	86 ± 8	85 ± 9	−1.1	89 ± 9^†δ	91 ± 6^§‡‡	1.8	0.0070	ns	ns
SBP (mmHg)	168 ± 45	173 ± 20	5	188 ± 33	187 ± 18	−1	ns	ns	ns
DBP (mmHg)	107 ± 15	98 ± 7	−9.2	104 ± 19	113 ± 16^‡	9	ns	ns	0.0470
Dyspnea (Borg score)	6.3 ± 1.7	5.7 ± 1	−0.6	6.7 ± 2	7.7 ± 1.9^‡	1	0.0100	ns	ns
Leg effort (Borg score)	6.3 ± 2.1	6.2 ± 1.6	−0.1	6.6 ± 2.2	7.8 ± 1.6^§	1.2	0.0300	ns	ns
VT (L)	1.08 ± 0.2	1.07 ± 0.3	−0.01	1.38 ± 0.2^††† δ δ	1.28 ± 0.3^§§‡‡	−0.1	0.0003	ns	ns
V´E (L)	30 ± 7	31 ± 7	−0.4	42 ± 10^†††† δ δ	38 ± 9^§§§‡‡‡	4.1*	0.0004	ns	0.0500
fR (rpm)	29 ± 6	29 ± 6	−0.1	31 ± 5	30 ± 5	1.4	ns	ns	ns
Ti/Ttot (%)	36 ± 4	34 ± 4	1.4	31 ± 3^†† δ	32 ± 3^‡	−1.1	0.0100	ns	ns
TE (s)	1.4 ± 0.4	1.4 ± 0.3	−0.02	1.4 ± 0.3	1.4 ± 0.3	0.003	ns	ns	ns
fR/ VT (rpm L^-1)	29 ± 12	29 ± 11	−0.3	23 ± 7^†† δ	25 ± 8^‡	−2.2	0.0490	ns	ns
VT/ Ti (L/s)	1.4 ± 0.2	1.5 ± 0.2	−0.05	2.3 ± 0.5 ^δ δ δ δ	2.0 ± 0.5^§§§	0.3*	<0.0001	ns	0.0300

Abbreviations: NIV = noninvasive ventilation; IMT = inspiratory muscle training; SHAM = control ventilation; PAV = proportional assisted ventilation; T_lim = maximum exercise tolerance time; HR = heart rate; SpO₂ = digital oximetry; SBP = systolic blood pressure; DBP = diastolic blood pressure; V_T = tidal volume; V´_E = minute-ventilation; f_R = respiratory frequency; T_i/T_tot = duty respiratory cycle. *p < 0.05 PAV post IMT vs. PAV Pre IMT; **p < 0.01 PAV post IMT vs. PAV Pre IMT; ^†p < 0.05. ^††p < 0.01. ^†††p < 0.001 and ^††††p < 0.0001 for PAV pre IMT vs. SHAM Pre IMT; ^δp < 0.05. ^{δ δ}p < 0.01 for PAV pre IMT vs. SHAM post IMT; ^§p < 0.05. ^§§p < 0.01. ^§§§p < 0.001 and ^§§§§p < 0.0001 for PAV post IMT vs. SHAM pre IMT; ^‡p < 0.05. ^‡‡p < 0.01. ^‡‡‡p < 0.001 and ^‡‡‡‡p < 0.0001 for PAV post IMT vs. SHAM post IMT.

The bold values are statistically significant at p < 0.05.

PAV and IMT effects at T_lim

We found a within PAV increase for T_lim (p = 0.001) and reduction in V´_E (p = 0.036, Table 2) in addition to a significant main-effect for T_lim owing to IMT (Table 2, p = 0.006) and favourable interaction between NIV vs. IMT toward an increase in T_lim (Table 2, p = 0.034). Additional differences between PAV vs. SHAM for the IMT condition are depicted in Table 2. IMT resulted in a significant increase in MIP at two-weeks (ΔMIP = 8.3 ± 4.7 cm H₂O, p < 0.01, Figure 2) and four-weeks (ΔMIP = 17.0 cm H₂O, p < 0.0001, Figure 2).

Figure 2. Change in MIP from baseline to 5 (intermediary) and 10 IMT sessions. **p < 0.01 intermediary vs. pre-IMT; ***p < 0.001 post vs. pre IMT; ^†p < 0.05 post IMT vs. intermediary.

PAV and IMT effects throughout exercise

For comparisons within SHAM and within PAV, IMT positively impacted only IMT + PAV, with a statistically significant increase in T_lim compared to PAV as a stand-alone intervention (Figure 3D, p = 0.001), with 6/9 individuals presenting an higher increase in T_lim than the recommended clinically-important value of 33%. In addition, the inflection of heightened curve tendency for dyspnea/leg fatigue throughout exercise was smoothed (down-shifted) during the course of PAV-supported exercise by the previous IMT, when comparing the dyspnea score for PAV post IMT (IMT main-effect, p = 0.235, time main-effect p < 0.0001 and IMT vs. time interaction p = 0.0002, Figure 3E) and leg effort (IMT main-effect, p = 0.237, time main-effect p < 0.0001 and IMT vs. time interaction p = 0.0003, Figure 3F) compared with PAV pre IMT values. Moreover, PAV post IMT led to significantly higher leg effort complaints compared to PAV pre IMT at T_lim (Figure 3F, p = 0.012)). In addition, PAV post IMT demonstrated a significantly favourable change for respiratory pattern when compared with PAV pre IMT, with a significant reduction in f_R (IMT main-effect, p = 0.009, time main-effect p < 0.0001 and IMT vs. time interaction p = 0.242, Figure 4H) and T_i/T_tot (IMT main-effect, p = 0.018, time main-effect p = 0.0001 and IMT vs. time interaction p = 0.004, Figure 3I) throughout exercise. Despite this respiratory pattern change, V´_E (IMT main-effect, p = 0.080, time main-effect p = 0.0001 and IMT vs. time interaction p = 0.402, Figure 4F) and V_T (IMT main-effect, p = 0.910, time main-effect p = 0.0001 and IMT vs. time interaction p = 0.556, Figure 3G) remained unchanged in this analysis. The levels of PAV support were comparable throughout the exercise pre- and post IMT (Supplementary material Figure S2, IMT main-effect p = 0.784, time main-effect p < 0.0001 and IMT vs. time interaction p = 0.220).

Figure 3. Panel representing individual T_lim change after IMT for SHAM (A) and PAV (D). Dyspnea scores (B and E) and leg effort sensation (C and F) are shown throughout exercise, respectively for SHAM and PAV. Abbreviations: IMT = inspiratory muscle training; SHAM = control ventilation; PAV = proportional assisted ventilation; small grey closed circles in (A) and (D) represent subjects with higher than 33% increase in T_lim. Closed circles and open circles in (B), (C), (E) and (F) represent IMT + PAV and PAV alone, respectively.

Figure 4. Respiratory pattern throughout exercise second IMT treatment (pre vs. post) for SHAM (above) and PAV (bellow). Closed circles and open circles represent IMT + PAV and PAV alone, respectively. *p < 0.05 pre vs. post IMT isotime; **p < 0.01 pre vs. post IMT isotime; ***p < 0.001 pre vs. post IMT isotime.

Discussion

In this study, we observed for the first time that IMT, performed early prior to the acute use of ventilatory assistance during high-intensity exercise, in individuals with severe COPD, leads to a reduction in dyspnea and increases exercise tolerance compared to NIV alone. Although we did not design this study specifically to unravel associated-mechanisms, short-time IMT before the use of PAV reduced the respiratory-frequency/inspiratory-time binomial throughout exercise, allowing a reduction in minute-ventilation compared to the control condition at exhaustion, supporting additional mitigation in neuroventilatory uncoupling and increasing peripheral subjective sensation at exhaustion.

Effects of isolated PAV

NIV has a limited role in pulmonary rehabilitation in COPD. However, in well-selected individuals (responders), it reduces dyspnea and increases tolerance to exercise [11,12,25]. It is currently considered that PSV and PAV are superior modes of NIV during exercise in COPD [26]. Overall, the results for PAV alone in this study (pre IMT) are broadly concordant with previous positive results for T_lim, V_T (L), V'_E (L/min), SpO₂ (%), respiratory pattern, and sensory changes, found by other authors [27–30]. Hemodynamic parameters (heart rate, cardiac output, and systemic arterial pressure) seem to be significantly modified mostly in an NIV-assisted aerobic training scenario, probably because these parameters depend on chronic adjustments in the autonomic nervous system, that occur mainly during global physical training [31].

Effects of isolated IMT

Although a small number of subjects did not demonstrate improvement in T_lim after IMT alone (SHAM effect), there was an overall main-effect for IMT on T_lim after considering both PAV/SHAM and IMT analysis. A previous meta-analysis showed no homogeneous improvement in the IMT group compared with the control group, and subsequent random effect models did not reach statistical significance for endurance exercise capacity after IMT [32], although a previous systematic review showed significant results for six-minute walk distance [33]. Of note, despite the absence of a control group for IMT, there was a substantial average increase in MIP after 10 sessions, equivalent to previously described.

Effects for combined IMT and PAV

In the within PAV modality comparison, IMT before the utilization of PAV resulted in important changes. According to our hypothesis, there was an expected additive effect of IMT on PAV, with a reduction in dyspnea and leg fatigue scores throughout the exercise. Several proposed favourable mechanisms are shared between NIV and IMT and may have provided optimization of PAV. For example, (i) reduction in neural drive and improvement in breathing pattern [5–10], (ii) better neural processing of dyspnea and favourable respiratory muscle recruitment pattern [3,25,34], and (iii) reduction in respiratory muscle energy demands and increased energy supplies to peripheral muscles [3,11,12].

Several studies with PAV have consistently shown variable f_R and V_T recruitment compared to SHAM, resulting in variable V´_E effects [27,28,35]. The addition of IMT to PAV led to a reduction in f_R and T_i/T_tot throughout exercise and lower V_T/T_i at T_lim, compared to PAV alone, following lower V´_E at T_lim when compared to the control condition; this suggests that, in addition to respiratory muscle support and reduction in ventilatory-drive, characteristically described for PAV, there was adoption of a respiratory pattern favourable to mitigation of neuroventilatory uncoupling, as a consequence of previous IMT.

Limitations of the study

Some limitations of this study have already been discussed. We add that despite the small number of individuals, the sample calculation for a design of repeated measures showed sufficient power for analysis. Undoubtedly, although the IMT time-course was shorter than generally recommended, 10 sessions were enough to significantly increase IMT and T_lim in this study. Another limitation refers to the use of PAV, which is a costly NIV technology, with an individualized time-consuming technique. However, our goal was to describe a concept of favourable overlapping effects using NIV and IMT and we believe that less costly and simpler technologies such as CPAP or BiPAP may have better results with a longer IMT time-course. We chose baseline MIP < 100 cmH₂O as an entry criterion to the study owing to the restricted operational range of the custom-built linear-load Threshold^® devices for muscle training (Supplementary material Figure S1), which do not support training pressure setting >82 cm H₂O. Larger baseline MIP (>100 cmH₂O) obviously demands training pressure settings >80 cm H₂O (80% of MIP). Accordingly, this limitation could be theoretically overcome beyond this proof-of-concept study.

Clinical implications

In this study, potentiation of NIV-supported exercise by a short time-course IMT were described. This concept opens a window for subjects with very-limited COPD exercise tolerance to achieve adequate levels of aerobic physical training. Future studies may also clarify better whether individuals who do not respond appropriately with increased tolerance to exercise from NIV, or even deteriorate, could respond after IMT.

Conclusions

The addition of IMT prior to a PAV-supported aerobic bout potentiates exercise tolerance and dyspnea relief and induces favourable changes in the ventilatory pattern in severe COPD during high-intensity training. Future studies with shorter or longer respiratory training schedules before acute NIV interventions and more detailed physiological invasive measurements are warranted to better understand this promising association.

Specific author contributions

Rodrigo Koch: Data Collection, manuscript preparation, analysis of data and review; Tiago Augusto: Literature search, manuscript preparation, and review; Alessandro Ramos: Literature search, data collection, manuscript preparation, and review; Paulo Müller: Study design, literature search, data collection, analysis of data, manuscript preparation, and review.

Ethics

The author and coauthors have contributed substantially to this original work and approved the final submission. This work is not being considered for publication, in whole or in part, in another journal, book or conference proceedings and the author and coauthors have no conflicts of interest. The author and coauthors reviewed the final stages of the manuscript.

Acknowledgments

The authors would like to thank Phillips Respironics, Inc. for supporting this work by providing the noninvasive V60 apparatus.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Data availability statement

The data are available upon request.