Acute effects of combined exercise and oscillatory positive expiratory pressure therapy on sputum properties and lung diffusing capacity in cystic fibrosis: a randomized, controlled, crossover trial

We conducted a prospective, randomized, controlled crossover trial (Clinicaltrials.​gov, NCT02750722). Adults with CF were invited to our laboratory facility on three different occasions. At the first study visit, the patients provided a sputum sample, performed pulmonary function testing and cardiopulmonary exercise testing (CPET). At the second and third study visit, patients provided sputum samples, performed pulmonary function testing and, depending on randomization, performed either continuous cycling exercise at moderate intensity without Flutter® (Experiment A, control condition) or moderate intensity interval cycling exercise incorporating Flutter® therapy (Experiment B, experimental condition). The detailed exercise protocol including measurements and assessments during study visit 2 and 3 are shown in Fig. 1.

Individuals with CF were recruited from the Adult CF Center at the University Hospital Zurich, Switzerland, between June 2016 and January 2017. Patients aged 18 years and older with a confirmed diagnosis of CF able to provide sputum samples were included. Exclusion criteria were as follows: i) listing for lung transplantation or status post lung transplantation, ii) chronic pulmonary infection with Burkholderia cepacia complex, iii) unstable clinical condition (i.e., major hemoptysis or pneumothorax within the last 3 months, acute pulmonary exacerbation [13], intravenous antibiotic treatment during the last 4 weeks, change in pulmonary medication during the study period); iv) cardiac arrhythmias with exercise and v) requirement of additional oxygen with exercise. Ethical approval was obtained from the Cantonal Ethics Committee of Zurich (2015–00153), Switzerland. All patients provided written informed consent.

We used central randomization to randomly allocate patients to the different experimental conditions (Experiment A or B). A person not involved in the study generated a list of random numbers using the statistical software package STATA (Version 12, StataCorp. 2012, College Station, Texas, USA). The list consisted of 30 random numbers, where even numbers represented the experimental condition ‘A’ and odd numbers the condition ‘B’. The order of the experiments at study visit 2 and 3 was either ‘A-B’ or ‘B-A’. At the first study visit, after written informed consent was obtained, the principal investigator (TR) called the person creating the list of random numbers to obtain information on the order of the exercise experiments for visit 2 and 3, respectively. The patient was immediately informed about the testing order for the following study visits.

All tests were scheduled at the same time of the day (+ − 1 h deviation) to avoid any potential impact of diurnal variation on pulmonary function measurements [14]. The time period between study visit 1 and 3 was 8 ± 2 days, respectively. Patients were told to abstain from fatty meals (for 3 h), caffeine-containing substances (for 4 h) and to avoid vigorous physical exercise during the last 24-h prior to the study visits, respectively. Moreover, patients were told to abstain from their regular inhalation and airway clearance therapy on the day of the study visits. The following provides a short description of assessments methods.

The average of all sputum samples (n = 45) obtained at rest during each of the three visits showed spinnability (Fig. 2a) and viscoelastic behavior (Fig. 2b-d) characteristic for CF sputum [7, 30‐32]. Details can be found in the online supplements (Additional file 3: Table S4). At high frequencies (ω > 10 rad s^− 1) G’ and G” as well as η^* (complex viscosity) and the phase angle increased due to instrument inertia, which causes artifacts in the sample signal. Consequently, the analysis of sputum viscoelasticity (G’ and G”) was restricted to an angular frequency of 10 rad s^− 1. Instrument inertia is further debated in the Discussion section and online supplements.

Table 2 shows changes in sputum properties for each experimental condition. No time-course changes were found during either experiment A or B, except for sputum solids content during experimental condition A (P = 0.038). Individual raw data for G’ at 1 and 10 rad s^− 1 are shown in Additional file 4: Figure S1 and Additional file 5: Figure S2 in the online supplements. In the experimental condition A, one study participant had very high values for G’ at 1 and 10 rad s^− 1 immediately post-exercise. Of note, there was no technical problem during the rheological measurements, but the sputum sample was purulent and thick. A summary of all sputum rheological data excluding this particular participant is shown in Additional file 6: Table S3 in the online supplements.

No differences in sputum rheological properties (viscoelastic moduli G’ and G”, yield stress, solids content) were found between experiments (Table 3), except differences in sputum spinnability comparing pre- versus 45′ post-exercise values between experiments. During recovery (absolute change, pre- vs. 45 min post-exercise), we noticed differences in patient reported ease of sputum expectoration, favoring experimental condition A (P = 0.016).

Individual raw data for DLNO, DLCO, V_A and V_cap at different time points during experiments A and B are shown in Additional file 7: Figure S3 and Additional file 8: Figure S4 in the online supplements. Time course changes in pulmonary diffusing capacity were observed during both experiments, while changes in DLNO and V_A were only observed in experimental condition A (Table 2). No differences in pulmonary diffusing capacity were found between the two experimental conditions (Table 3), except a higher V_A comparing post-exercise and pre-exercise changes favoring experiment A (P = 0.032).

In the present study, we found no differences in sputum viscoelasticity between the two experimental conditions, thus rejecting our initial hypothesis. Two previous studies reported altered sputum viscoelasticity after an acute bout of treadmill exercise or Flutter® and treadmill exercise compared to resting breathing in adults with CF [6, 11]. Treadmill but not cycling exercise reduced sputum viscoelasticity and trunk oscillations during treadmill running/walking have been postulated as a possible underlying mechanism [11]. One could argue that the chosen exercise mode in our study was not sufficient to improve viscoelastic properties of sputum in our patients and/or that the magnitude of effect from Flutter® was not large enough to produce greater changes in sputum viscoelastic properties compared to cycling alone. During exercise, our patients achieved approximately 47–49% of their maximal minute ventilation (90–97% of their maximal tidal volumes and 50–55% of their maximal respiratory frequencies, see Additional file 2: Table S2) suggestive of a high ventilatory demand. Higher exercise intensities (e.g., 90% of maximal heart rate) may induce significant airway narrowing [12] unlikely to be maintained by many patients when exercising for longer periods, in particular, in patients with advanced lung disease. With respect to Flutter®, our patients performed 6–10 breathing maneuvers during each of four cycles. Higher volumes of Flutter® therapy (6 cycles à 15 breaths) combined with huffing and coughing (forced expiration technique, FET) improved sputum viscoelasticity compared to resting breathing in adults with CF [6]. It is important to note that we deliberately did not follow existing CF physiotherapy recommendations [27] suggesting to individually determine the angle of the Flutter® along with the performance of forced expiratory maneuvers. In this study, we were mainly interested in mechanistic effects of airway oscillations on sputum viscoelasticity [7] on top of high ventilation during exercise. We did not implement forced expiratory maneuvers due to potential interference with our cycling protocol and pulmonary gas exchange measurements. The authors hypothesize that a more intense Flutter® therapy, together with individual adaptation of the Flutter® angle, but not the lack of forced expiratory maneuvers, could explain the absence of additional effect in our study. Given the fact that positive expiratory pressure and oscillations are thought to mechanically impact on biophysical properties of mucus (i.e., reduction of viscoelasticity), huffing and coughing following Flutter® maneuvers assist in mobilization and transport of secretions from peripheral to central airways, but should not substantially change viscoelastic properties of sputum in addition. This hypothesis is supported by an in vitro experiment demonstrating that Flutter® oscillations alone augment sputum elasticity after 15 and 30 min, respectively [7].

It is important to mention that in comparison to our experimental study design, the study by Dwyer et al. [6] compared Flutter® therapy with resting breathing (no intervention), which likely increases the chance to observe effects between experimental conditions. Nevertheless, given the high variability of sputum properties [33] and differing intra- and interindividual responses to airway clearance therapy, a “no intervention” visit would have probably provided further insights into within-patient treatment responses.

Interestingly, and in line with findings by Dwyer et al. [11], ease of sputum expectoration was higher during recovery from continuous cycling versus cycling exercise with Flutter®. This suggests that patients perceive the benefit of airway clearance therapy not immediately but during recovery from exercise. However, the clinical meaningfulness of these findings cannot be interpreted due to the lack of a minimal important difference for the visual analogue scale.

We observed occurrence of inertia at high frequencies, causing experimental artifacts (see Additional file 9 for further discussion). King and Macklem suggested first that sputum rheology at high frequency simulates cough studying dog tracheal mucus [34]. The concept was then further elaborated suggesting that low frequency (1 rad s^− 1) deformations are relevant when simulating ciliary transport [35] whereas high frequency (100 rad s^− 1) deformations are more characteristic for cough clearance [36]. The 1 rad s^− 1 / 100 rad s^− 1 concept was initially elaborated using active microrheology but was later applied in shear rheology testing [6, 11, 37, 38]. We found a strong increase in moduli (G’, G”) and phase angles at frequencies > 10 rad s^− 1 (Fig. 2c and d, Additional file 10: Figure S5a) and observed that sample torques deviated from the electrical torques (Additional file 10: Figure S5b), both strong indicators for inertia. The deviation of sample torque from electrical torque due to inertia is schematically represented in torque vector drawings in Additional file 10: Figure S5c. Also, we calculated the theoretical inertia limit of the rheometer, which agreed well with the observed onset of inertia. Further proof is obtained comparing our results to other findings that show a continuous trend of the dynamic moduli up to 50 Hz (≈ 314 rad s ^− 1) using passive microrheology [32], which is not affected by inertia at frequencies below 1 MHz [39]. Data generated in previous studies using shear rheology at 100 rad s^− 1 also showed signs of inertia [6, 11, 37, 38], revealed by a substantially increased loss modulus G” or complex modulus (G*) and high tan δ values compared to those measured at 1 rad s^− 1. Future rheology measurements with sputum should be carefully checked for inertia, as in the worst-case scenario, false positive or false negative outcomes occur. We propose restricting frequencies to 10 rad s^− 1 to simulate the high frequency behavior of sputum in shear rheology.

In the present study, DLNO and DLCO increased during continuous cycling exercise, whereas interval cycling exercise interspersed with use of the Flutter® increased only DLCO but not DLNO, most likely due to the absence of increase in V_A. It is well known that DLNO, DLCO, and V_cap increase linearly during exercise with respect to cardiac output [40]. In our study, continuous cycling increased post-exercise DLNO by about 5%, but the effect was not sustained during recovery. We extend previous findings reporting a remarkable increase in DLNO (39 ± 8% from rest using rebreathing measurements) during moderate intensity cycling exercise in CF adults, with the authors questioning the duration of beneficial effects [12]. Our data demonstrate that increase in DLNO is (only) an acute, exercise-induced effect, most likely the result of increased V_A and subsequently greater surface area for diffusion rather than improved mucociliary clearance mechanisms such as improved sputum hydration and/or viscoelastic properties.

This study has limitations. First, we did not measure expiratory flow during exercise [6, 11] to evaluate whether our experiments were sufficient to create an expiratory airflow bias, a mechanism potentially improving mucociliary clearance. Furthermore, the sputum was sheared at 50 Pa in simple shear experiments prior to spinnability measurements. The shear homogenized the samples, theoretically disrupting sputum microstructure, which could have caused the comparably low spinnability values. Finally, our study patient cohort was small, limiting the generalizability of our findings to the overall CF population.