Study design and participants
The present crossover study included 22 healthy subjects who voluntarily participated in the study (10 men and 12 women; age: 18–29 years; body mass index [BMI]: 18 to 25 kg/m2; and no history of smoking and respiratory, cardiac, or neuromuscular diseases). Individuals with forced vital capacity (FVC) and forced expiratory volume in the 1 second (FEV1) < 80% and FEV1/FVC ratio < 85% of the predicted ratio, highly active individuals (assessed by the The International Physical Activity Questionnaire short version-IPAQ)27 and those who had nasal congestion or deviated septum were excluded. The study followed the recommendations of the Declaration of Helsinki and was approved by the Research Ethics Committee of Hospital Universitário Onofre Lopes (HUOL-EBSERH/BRASIL) according to protocol number 3.084.956.
Pulmonary function
Pulmonary function was assessed using the KoKo DigiDoser spirometer (nSpire Health, Inc.; Longmont, CO, USA). FEV1, FVC, FEV1/FVC ratio, and forced expiratory flow between 25–75% of the FVC maneuver were assessed. The technical procedures followed the recommendations of the American Thoracic Society/European Respiratory Society28 and the predicted values were calculated using the reference values for the Brazilian population29. Additionally, the slow vital capacity maneuver and the voluntary maximum ventilation test for 15 seconds (MVV) were performed.
Respiratory muscle strength
Evaluation of respiratory muscle strength was performed using a digital manometer (NEPEBLabCare/UFMG, Belo Horizonte, MG, Brazil) with the participants in a seated position and feet on the floor. The maximum inspiratory pressure (MIP) was obtained from the residual volume and the maximum expiratory pressure (MEP) was obtained from the total lung capacity. To obtain the predicted values, regression equations for healthy Brazilian population were used30. Additionally, nasal expiratory pressure (SNEP) and SNIP were also measured. For the SNIP test, previously described equations were used to obtain the reference values31.
Sniff curve analysis
All sniffs were performed at functional residual capacity with the individuals seated on a chair without back support. The contraction and relaxation properties were evaluated from the sniff traces. For the contraction parameters, the contraction time (expressed in ms) was calculated as the time to reach the peak pressure11, the maximum rate of pressure development (MRPD) was calcultated as the negative peak of the first derivative of the pressure-time curve (MRPD normalized to the sniff peak pressure [MRPD/sniff peak], expressed in ms-1)32, and the time to peak shortening (TPS) (expressed in ms) was calcualed as the time to reach MRPD33.
For the relaxation parameters, the half relaxation time (½RT) (expressed in ms) was calculated as the half-time of the relaxation curve, the maximum relaxation rate (MRR) (expressed in ms-1) was defined as the positive peak of the first derivative of the pressure-time curve (normalized for the peak sniff pressure)15. The time constant (τ, tau) of the later monoexponential phase of pressure decay (over 50–70% of the pressure decay curve) was also calculated (y=exp-t/τ). The correlation coefficient of the regression line (pressure vs. time) should be greater than 0.98 for the value of τ to be acceptable. The criteria for selecting appropriate sniffs for further analysis were: (1) sniff performed from functional residual capacity, (2) peak pressure maintained for less than 50 ms, (3) duration of inspiratory effort < 500 ms, and (4) SNIP waveform with a smooth decay curve16. The SNIP curve was analyzed using MATLAB software (The MathWorks Inc., Natick, MA, USA).
Chest wall volumes and compartments
For the assessment of chest wall volumes the OEP system (BTS Bioengineering, Quincy, MA, USA) was used, which included eight photosensitive cameras that captured the movement variation of 89 retro-reflective markers placed over predefined regions of the subject’s chest and abdomen35. Before each measurement, the device was calibrated in static and dynamic ways using a frequency of 60 frames per second to recognize the markers.
The data considered for the OEP analysis included changes in the volume of the chest wall (ΔVCW), pulmonary ribcage (ΔVRCp), abdominal ribcage (ΔVRCa), abdomen (ΔVAB), and inspiratory time (Ti). Using these values, the shortening velocity index of the inspiratory ribcage (∆VRCp/Ti) and the shortening velocity index of the diaphragm (∆VAB/Ti) were calculated, while the global shortening velocity index of the inspiratory muscles (ΔVcw/Ti) was obtained through the sum of these two parameters. In addition, the products of the pressure generated in the SNIP maneuver according to the shortening velocities (ΔVCW/Ti, ∆VRCp/Ti, and ∆VAB/Ti) denoted the mechanical power of the global inspiratory muscles (Ẇinsp), inspiratory ribcage muscles (Ẇrcm), and diaphragm (Ẇdi), respectively34. All of these data were analyzed during the SNIP maneuver.
Surface electromyography
The electromyographic activity of the respiratory muscles was obtained simultaneously with the assessment of chest wall volumes using a trigger. An electromyograph (TeleMyo DTS Desk Receiver®; Noraxon USA Inc., Scottsdale, AZ, USA) was used to acquire the signals with 16-bit resolution. The signal was captured at a sampling frequency set at 1500 Hz, with a low-pass filter of 500 Hz, gain of 1000×, and a common-mode rejection index greater than 100 dB. The data were stored in the software MR version 3.8 (Noraxon USA Inc., Scottsdale, AZ, USA). Ag/AgCl bipolar surface electrodes were attached to the skin in the direction of the muscle fibers after the skin was prepared with abrasion, followed by trichotomy and cleaning with 70% alcohol according to the Surface Electromyography for the Non-Invasive Assessment of Muscles recommendations (SENIAM)35. The procedure was performed on the following muscles: 1) sternocleidomastoid (SCM), at the lower third of the distance between the mastoid process and the sternoclavicular joint36; 2) scalene muscles (ESC), at 5 cm from the sternoclavicular joint and 2 cm above this point37; and 3) parasternal (PS) muscle, at the second intercostal space and 3 cm from the sternum38.
SEMG processing and analysis
SEMG signals were processed using a 20-400 Hz Butterworth bandpass filter and analyzed in the time and frequency domains to calculate root mean square (RMS) and MF, respectively. During the SNIP maneuvers, each portion of the SEMG signal corresponding to a SNIP maneuver was subjected to RMS and MF analysis. During RET, only the MF was taken into account, which was calculated through continuous wavelet transform technique using Daubechies4 mother in 5-second windows. For analysis, MF and RMS were normalized for each patient by expressing them relative to values obtained at the beginning of the fatigue protocol (i.e., mean of the initial 10 s) and plotted as a function of the total time of RET. All sEMG analyses were performed off-line using MATLAB software (The MathWorks Inc., Natick, MA, USA).
Tissue oxygenation
Assessment of tissue oxygenation was performed using the NIRS device (Portamon; Artinis Medical Systems BV, Elst, Netherlands). The technique is based on the application of light with near-infrared wavelength, considering the principles of absorption and dispersion based on the spatially resolved spectroscopy method25. The Portamon is a non-invasive, portable, wireless tool that contains a receiver and three light-emitting diodes spaced at 30, 35, and 40 mm, which capture the absolute concentrations of O2Hb, HHb, and tHb. Thus, tissue oxygenation and local blood volume were estimated from these variables. The device uses wavelengths of 760 and 850 nm and a bluetooth connection that allows online monitoring while the individual performs various activities. Moreover, it does not suffer from interferences in the presence of equipment such as the EMG machine. The equipment was fixed on the subjects’ skin over the left SCM using adhesive tapes after the skin was cleaned with 70% alcohol. It was fixed in a position similar to that described by Shadgan et al39.
Respiratory endurance test
Two types of respiratory muscle training devices (SpiroTiger, Idiag®, Fehraltorf, Switzerland and POWERbreathe, HaB International Ltd, Southam, UK) were used to perform the NH and the IPTL protocols, respectively. For the NH test, the parameters were based on a previous study40 with the size of the rebreathing bag established at 50% of the individual’s vital capacity, minute volume (VE) adjusted to 70% of the 15-second MVV, and the respiratory rate (RR) defined according to the manufacturer’s recommendations using the formula: Respiratory rate = AMV/(Bag size × 1.2) (1/min), where AMV is target ventilation per minute. The MVV level was set at 70%, as previous studies have reported task failure with this type of ventilation41. Subjects were asked to breathe holding the pre-determined VE and task failure was defined when the subjects reached volitional exhaustion or when they were unable to maintain RR and VE after three warnings from the evaluator. For the IPTL test, the medium-resistance POWERbreathe® Classic (POWERbreathe; HaB International Ltd., Southam, UK) was used and the test was performed with a load of 80% of the MIP, based on the recommendations of a recent systematic review6. Task failure was defined when the subjects reached volitional exhaustion or were unable to overcome the load after 3 warnings from the evaluator, thus ending the test.
Study protocol
All subjects were previously informed about the study methods and 10 SNIP maneuvers were performed before the beginning of the protocol for the purpose of learning effect. Data collection was performed on two different days separated by a period of 7 days and the order of the devices was randomized using a simple draw with an opaque envelope. On the first day, spirometry and manovacuometry were performed in addition to the initial assessment including filling the assessment form regarding anthropometric data and the level of physical activity. After this stage, a 20-minute rest period was provided before starting the first RET protocol with the selected device. On the second day of the data collection, subjects performed RET only with the second device. During the tests, heart rate (HR) and peripheral arterial saturation (SpO2) were monitored in addition to the application of the Borg effort scale before and immediately after RET.
The experimental protocol consisted of the following phases.
Pre-RET phase: subjects were asked to remain in a seated position in a chair without back support, while a single researcher positioned the SEMG electrodes and the retro-reflective OEP markers. Subsequently, the manometer plug was inserted in one of the nostrils, while the contralateral nostril remained unobstructed and the subjects were asked to perform 10 SNIP maneuvers with an interval of 30 seconds between consecutive maneuvers. Subjects were monitored simultaneously in this phase using OEP and SEMG. For each participant, the SNIP maneuver that generated the highest pressure peak was used to analyze the SEMG parameters, OEP parameters, and the parameters obtained from the sniff curve (pre-RET values).
RET: after the pre-RET phase, subjects remained seated at rest for 15 minutes while information regarding the protocol was provided. RET was performed using the device selected for that day. During RET, simultaneous signals from SEMG and NIRS were acquired. The duration of the test (Tlim) was recorded at the time of task failure according to the aforementioned criteria. Verbal encouragement was provided throughout the protocol.
Recovery phase: after the test, subjects were instructed to immediately remove the nozzle from the used device, place the plug of the manometer in the same nostril that was used previously, and perform 10 SNIP maneuvers with an interval of 30 seconds between consecutive maneuvers. Similar to the protocol in the pre-RET phase, the SEMG signal was captured simultaneously with the OEP. In this phase, all values obtained in the 10 maneuvers were considered for the analysis.
Statistical analysis
The normality of the data was verified using the Shapiro-Wilk test. In the descriptive analysis, mean and standard deviation were used for the data with normal distribution and median and interquartile range were used for the data with non-normal distribution. For parametric data, the comparisons between the moments and between devices were performed using the two-way repeated-measures analysis of variance (ANOVA). The comparison between the moments of each device (intragroup) was performed separately using the Friedman test in case of nonparametric data or one-way ANOVA in case of parametric data. To avoid type I errors due to multiplicity of post-fatigue moments, correction using two-stage false discovery rate test (using a threshold value of 5%) was applied in case of statistical significance instead of the post-hoc Bonferroni or Dunn test42. The comparison between the devices (intergroup) was performed using independent samples t-test for parametric data and Mann-Whitney U test for nonparametric data. To verify whether the inspiratory ribcage muscles were developing fatigue during the protocols, regression analysis was applied to the MF variable and regression curves adjusted to the maximum values in the sense of the minimum square were used as an index of fatigue development. For all regression analyses, the determination coefficients (r2), slopes, and time constants were calculated during the moments of task failure and recovery (TF and TRec, respectively). For the regression analysis during the recovery period, the starting point at time zero corresponded to the last point of task failure in each muscle. For linear regressions, TF and TRec were calculated as the inverse values of the regression line slope. For nonlinear regressions, the slopes were calculated as the derivatives of the exponential equation at the beginning of the task failure protocol.
Muscle fatigue was confirmed if the following two criteria were met: 1) negative slope in case of linear regressions43 and 2) decrease to levels below 60% of the values recorded at the beginning of the task failure44 in case of exponential regressions. The NIRS variables were monitored in real time and were subsequently analyzed using the Oxysoft software (Artinis Medical Systems BV, Elst, Netherlands). A moving Gaussian filter was applied and linear regression analysis was performed at intervals of 10% of the total RET duration.