This was a 6 week randomized controlled trial, including on the one hand a control group with cardiac rehabilitation only, and on the other hand an interventional group admitted for CR and IMT. This study was conducted in the cardiac rehabilitation department at Saint-Etienne university hospital (France). Eligible patients were randomly assigned in a 1:1 allocation.

During the initial evaluation, patients had a 24-h Holter ECG monitoring to assess cardiac arrhythmia and to evaluate their very low frequency power spectral density of heart rate increment (VLFI) which is a predictor of sleep disorder (Roche et al., 2002; Sforza et al., 2007). Patients with a high risk of OSA, attested by VLFI>4%, had a respiratory polygraphy recording to assess OSA severity more accurately. Adult patients initiating a cardiac rehabilitation, under the age of 80, with an AHI ≥15 and ≤ 30 events h⁻¹ were eligible to participate.

Exclusion criteria were recent thoracic surgery by sternotomy (<12 weeks), obstructive ventilatory disorder with an FEV-1/FVC ratio <70%, already treated for OSA.

The study was performed from May 2015 to July 2021. Approval and ethical clearance were obtained from an institutional review board (CPP Sud Est I 1408189-2015-A00030-49) which was in accordance with the principles embodied in the Declaration of Helsinki. Prior to study initiation, the objectives of the study were clearly explained to subjects in order to obtain written informed consent. The study was a priori registered at ClinicalTrial.gov (NCT02494648).

The CR at the Saint-Etienne University Hospital was a 7-week interdisciplinary program that combines interventions performed by cardiologists, nurses, rehabilitation physicians, physiotherapists, dieticians, and psychologists. At the time of enrollment, the patient’s functional status was assessed by a cardiopulmonary exercise test (CPET). Exercise training consists of 20 sessions, three times a week (Monday, Wednesday, and Friday) for 1.5 h. Each session starts with a 10-min warmup period followed by 25 min of cycling or treadmill (power output at ventilatory threshold defined during the initial CPET) and a 5-min cool down period. Resistance muscle training of upper and lower limbs is also offered. The workout was divided into 80% aerobic and 20% resistance exercises. The first week of CR, continuous endurance exercises were preferred, and from the second week, interval exercises were added. An expert physiotherapist and an adapted physical activity monitor supervised sessions with continuous heart rate and pulse oximetry monitoring. The workload progression was adjusted weekly according to the patient’s tolerance (Borg Perceived Effort Scale). Patients attended weekly hour-long group sessions with healthcare professionals aimed at reinforcing their health education. Particular emphasis is provided on understanding the pathophysiology of coronary artery disease, the role of cardiovascular risk factors, and its management, mainly through physical activity, anxiety control, smoking cessation, nutritional balance, and adherence to guidelines for recommended drugs.

The interventional group followed the CR program described above, as well as a 6-week IMT. To strengthen the inspiratory muscles, we used a resistive loading device (POWERbreathe^® Plus medium resistance, Southam, United Kingdom) with a flow-independent one-way valve to ensure consistent resistance and an adjustable specific pressure setting (from −23 to −186 cmH₂0). IMT consisted of performing 6 days a week, two sessions of 30 inspirations against a load of 70% MIP, during the entire duration of the CR period, i.e., nearly 40 sessions in 6 weeks. On days when the patient was present at the CR center, the IMT sessions were carried out under the supervision of the physiotherapist (PL). The third week, the MIP was reevaluated in the same conditions as during inclusion in order to adjust the training load. Patients were asked to keep a training notebook with indication of the difficulty in performing the session on a 0–5 scale.

Patients randomized in the control group participated only in the CR and performed the same evaluations as the IMT group at baseline and at follow-up.

All outcome measures described were performed at baseline and at 6-week follow-up.

Respiratory polygraphic data. AHI was the main outcome measured from respiratory polygraphy. Ambulatory polygraphy was performed at home using either a Nox-T3 (Nox Medical, Reykjavik, Iceland) or a VistaO₂flux (Novacor, Rueil Malmaison, France). For both devices, the signals acquired were the following: nasal flow with pressure transducer and oxygen saturation with a digital pulse oximeter (Nonin, United States). For Nox-T3, respiratory signals were recorded with chest and abdominal respiratory inductance plethysmography belts. For VistaO₂flux, respiratory signals were evaluated by thoracic impedance sensors embedded in a multimodal ECG Holter recorder (Roche et al., 1999; Poupard et al., 2008; Chouchou et al., 2013). For each patient, the second polygraphy at the end of the study was carried out with the same device as baseline in order to be more consistent. In accordance with the 2012 American Academy of Sleep Medicine guidelines (Berry et al., 2012), an apnea was defined as a cessation in airflow (≥90% baseline) for ≥10 s, and a hypopnea was defined as a reduction in airflow (≥30% baseline) for ≥10 s which resulted in a ≥ 4% oxygen desaturation. Sleep apnea severity was rated in accordance with the individual’s AHI, defined as the total number of apneas and hypopneas per hour of sleep and oxygen desaturation per hour of sleep; where an AHI of 5–15 events.h⁻¹ was defined as mild, 16–30 moderate, and >30 designated as severe OSA. Secondary polygraphic outcomes included oxygen desaturation index (ODI), central apnea index, the mean and the lowest oxyhemoglobin saturation (SpO₂), and the percentage of time with a SpO₂ under 90% (SpO2 < 90%).

Respiratory function assessment. Lung function was assessed by standard spirometry and included forced expiratory volume in 1 s (FEV-1), forced vital capacity (FVC), and peak expiratory flow in accordance with the guidelines of the European Respiratory Society (Laveneziana et al., 2019). Maximal inspiratory pressure (MIP) was obtained with a specific device (POWERbreathe KH2, POWERbreathe International Ltd., Southam, United Kingdom). The maneuvers were carried out with the patient seated and knees flexed to 90° and were performed using the residual volume to realize a forced inspiration. Maximal inspiratory pressure was determined as the highest average pressure over 1 s reached during exercise. We considered three valid maneuvers (coefficient of variation below 10%) and the highest value was used for IMT.

Surveys. We analyzed questionnaires evaluating subjective sleep quality and quality of life. Subjective daytime sleepiness was measured by the Epworth Sleepiness Scale (ESS) which evaluates the propensity to sleep from “no” (scored 0) to “intense” (scored 3) in eight different situations. A total score > 10 indicates excessive daytime sleepiness (Johns, 1991). Quality of sleep was evaluated with the Pittsburgh Sleep Quality Index (PSQI), which is a questionnaire that evaluates seven sleep components on a scale from 0 (no difficulty) to 3 (severe difficulty). The results are expressed as a global score (ranging from 0 to 21). A total score > 5 indicates poor sleep quality (Buysse et al., 1989). The 12-item Short Form Survey (SF-12) was used to assess self-reported health-related quality of life, which evaluates physical health with a range of 23.99 (poor) to 56.58 (good) and mental health with a range of 19.06 (poor) to 60.76 (good; Ware et al., 1996).

Descriptive and inferential analyses were performed using GraphPad Prism 6 (GraphPad, San Diego, United States). Characteristics of the sample and treatment effect between groups were assessed using Student’s t-test for independent samples. Pre-versus post-CR differences were examined using paired t-tests, and chi-square test corrected by Fisher’s exact test was used for the categorical variables. The results are shown as mean ± standard deviation, number and percentage, and as differences of means and confidence interval (95%). Statistical significance was set at a two-tailed p < 0.05.

The sample size calculation was based on the Puhan et al. (2006) study results. For an expected difference of 7 points of AHI between the two groups, with a statistical power of 90%, we calculated a sample size of 44 patients. To consider dropouts before the end of the rehabilitation program, we decided to include 48 patients.

During the inclusion period, 636 patients were admitted to our cardiac rehabilitation center and 191 patients (40%) were assessed for suspected OSA. After respiratory polygraphy, 44.5% (n = 85) had AHI > 30 events.h⁻¹ and 29.8% (n = 57) had AHI < 15events.h⁻¹, and 48 patients with an AHI between 15 and 30 events.h⁻¹ were included in our study. The flowchart of the study is depicted in Figure 1. Baseline clinical characteristics of the study population are summarized in Table 1. Analyses of characteristics showed no significant differences at baseline.

The post-intervention change in BMI was not significantly different between groups (Table 2). In the IMT group, neck circumference decreased by 0.5 ± 0.9 cm (p = 0.02), and the difference between groups was also significant after intervention (p = 0.01). MIP significantly increased with the IMT intervention compared to the control (p = 0.003; Table 2).

At the end of the cardiac rehabilitation, both groups showed a significant improvement in maximal oxygen uptake from 20.7 ± 4.9 before CR to 24.7 ± 5.6 (ml⁻¹min⁻¹kg⁻¹; p < 0.0001) after CR for the IMT group and from 20.6 ± 4.9 before CR to 23.7 ± 5.3 (ml⁻¹min⁻¹kg⁻¹; p = 0.0003) after CR for the control group. Maximum power output during CPET after CR increased by 23.6 ± 23.7 W (p < 0.0001) for the IMT group and by 27.5 ± 22.9 W (p < 0.0001) for the control group Table 2.

At baseline, polygraphy data between groups were similar (Table 3). Post-intervention, the mean reduction in AHI score was only significant for the IMT group, which showed a decrease of 6.5 ± 9.5 events.h⁻¹ (95% CI: 2.015 to 10.08, p = 0.005). By contrast, the control group which only carried out the CR showed a nonsignificant decrease of 0.7 ± 13.4 events.h⁻¹ (Table 3; Figure 2). The difference in AHI of 5.8 ± 3.2 points between the two groups was close to significance (p = 0.09). ODI was significantly decreased in the IMT group by 6.4 ± 8.02 events.h⁻¹ (95% CI: −9.910 to −2.977, p = 0.0009); for the CTL group, no significant change in ODI was observed (p = 0.14). If we only consider the obstructive part of sleep apnea (AHI minus central apnea index), we observed a significant decrease of 6.7 ± 8.5 events.h⁻¹ (p = 0.004).

ESS only decreased significantly for the IMT group (−2.2 ± 3.2 points, p = 0.0176), and PSQI was close to significant with a mean score below 5 (3.9 ± 1.6, p = 0.053). For the mental health score of SF-12, the IMT group showed a significant increase of 5.58 ± 7.4 points (p = 0.008), while it did not change in the CTL group (p = 0.7). As a result, we found for this mental health score a significant difference between groups of 6.5 ± 2.6 points (p = 0.01). We found no other significant difference in the questionnaires, either intragroup or intergroup (Table 3).