The effect of tongue elevation muscle training in patients with obstructive sleep apnea: A randomised controlled trial

Abstract Background Oropharyngeal myofunctional therapy is a multi‐component therapy effective to reduce the severity of obstructive sleep apnoea (OSA). However, existing protocols are difficult to replicate in the clinical setting. There is a need to isolate the specific effectiveness of each component of the therapy. Objective To assess the effects of a 6 weeks tongue elevation training programme in patients with OSA. Methods We conducted a multicentre randomised controlled trial. Eligible participants were adults diagnosed with moderate OSA who presented low adherence to continuous positive airway pressure therapy (mean use <4 h per night). The intervention group completed a 6 weeks tongue elevation training protocol that consisted in anterior tongue elevation strength and endurance tasks with the Iowa Oral Performance Instrument. The control group completed a 6 weeks sham training protocol that involved expiratory muscle training at very low intensity. Polygraphy data, tongue force and endurance, and OSA symptoms were evaluated pre‐ and post‐intervention. The primary outcome was apneoa‐hypopnea index (AHI). Results Twenty‐seven patients (55 ± 11 years) were recruited. According to modified intention‐to‐treat analysis (n = 25), changes in AHI and c did not significantly differ between groups. Daytime sleepiness (Epworth Sleepiness Scale) and tongue endurance significantly improved in the intervention group compared to the control group (p = .015 and .022, respectively). In the intervention group, 75% of participants had a decrease in daytime sleepiness that exceeded the minimal clinically important difference. Conclusion Six weeks of tongue elevation muscle training had no effect on OSA severity.


| BACKG ROU N D
Obstructive sleep apnoea (OSA) is a common sleep disorder with an estimated prevalence of 19% in the general population of Lausanne (Switzerland). 1 Untreated, OSA is an important cause of morbidity and mortality. 2 Continuous positive airway pressure (CPAP) is the first-line treatment option offered to these patients. 3 However, between 17% and 85% of patients with OSA do not comply with CPAP therapy and remain untreated. 4 Alternative treatment options have therefore been investigated, including myofunctional therapy (MFT) of the upper airways. 5 The rationale for MFT stems from evidence showing that upper airway dilator muscles are dysfunctional and play a role in the pathophysiology of OSA. 6 Indeed, this disorder is characterised by recurrent upper airway collapse that lead to partial or complete airway occlusion. Upper airway dilator muscles, including genioglossus which forms the bulk of the tongue and is the largest upper airway dilator, are pivotal in the maintenance of upper airway patency during sleep. 7 However, poor genioglossus muscle activation during sleep is a physiologic feature contributing to OSA which, in susceptible individuals, may lead to the narrowing or collapse of upper airways. 8 Moreover, histological changes in the genioglossus and reduced cerebrovascular reactivity in the motor areas that control the upper airway musculature are found in patients with OSA. 9,10 These neuromuscular deficits are associated with sensorimotor defects of the tongue 8,11 and result in a greater propensity to upper airway muscle fatigue. 9 The MFT is a multi-component intervention that typically comprises a combination of exercises covering various oropharyngeal structures, such as tongue, palate, pharynx, or epiglottis, 12 and aims to act positively on the impaired sensorimotor deficits in the upper airway muscles encountered in patients with OSA. 9,11,[13][14][15] The efficacy of MFT in reducing the severity of OSA, daytime sleepiness and quality of life has been demonstrated in several reviews and meta-analyses. 5,[16][17][18][19] Two previous randomised controlled trials conducted in patients with moderate OSA showed that, compared to sham therapy, MFT including soft palate, tongue, facial muscle exercises, as well as stomatognathic function exercises (i.e. suction, breathing, speech, swallowing, chewing) resulted in a reduction of apnoea-hypopnea index (AHI) by 50% of its initial value. 12,20 The improvement in AHI seems related to upper airway remodelling, thereby reducing the collapsibility of upper airways during sleep.
However, since MFT protocols involve a variety of exercises, 20,21 the specific effectiveness of each exercise is unknown. Besides, the frequency and duration of existing protocols are challenging to replicate in the clinical setting. 20 In addition, the time-consuming aspect of this therapy might alter treatment adherence on the long run, which is considered as the main drawback of MFT. 22 Identifying the effective parts of MFT and deciphering the underlying mechanisms of this multi-component therapy may lead to a better understanding of how the therapy works and facilitate its implementation in the clinical setting. Assessing the efficacy of few, simple and easy exercises is thus needed. Because the genioglossus is the main upper airway dilator muscle, 23 focusing on the tongue may represent an interesting anatomical target. Another argument that endorse the strategy to focus on the tongue is the presence of lower lingual tone in children and adults with sleep-disordered breathing compared to healthy controls. 24-26 Interestingly, recent studies have shown that the improvement of AHI with MFT correlates significantly with the improvement of tongue strength. 14,27,28 Therefore, we aimed to assess the specific effectiveness of a tongue elevation muscle protocol in reducing OSA severity. We hypothesised that a 6 weeks tongue strength and endurance training programme would reduce AHI in patients with OSA.

| Design and participants
This is a multicentre randomised controlled study. Participants were either allocated to receive a tongue muscle training protocol (therapy group) or a sham protocol (control group) for 6 weeks. Adult patients previously diagnosed with moderate OSA (AHI via polysomnography between 15 and 30 events/h) who were registered in the database of the Sleep Investigation and Research Center of the Lausanne University Hospital and the Sleep Medicine Center of the Cliniques universitaires Saint-Luc (Brussels, Belgium) were eligible if they presented a low adherence to CPAP therapy (mean use <4 h per night). Change in OSA severity classification between the diagnostic polysomnography and the AHI measured at inclusion was allowed.
Exclusion criteria were craniofacial malformations, current use of hypnotic medications, history of stroke, a concurrent neuromuscular disease or a severe obstructive nasal disease. If participants were still on CPAP therapy, they were invited to discontinue this therapy at least 1 week before the start of the protocol. 29 30 and reference data have been published. 31 The two models of the IOPI were used for the purpose of this study: the IOPI Pro (Model 3.1) and the IOPI Trainer (Model 3.2). The IOPI Pro device was used for the measurements.
It consists of a portable device measuring the pressure that an individual can produce by squeezing a small air-filled bulb with the tongue against the hard palate with an upward movement of the anterior part of the tongue. The pressure obtained is displayed on the LCD screen of the instrument and is expressed in kilopascals (kPa).

| Training protocol
The training protocol consisted in anterior tongue elevation strength and endurance tasks. Participants were instructed to perform the training exercises once a day, 4 times a week for 6 weeks.  Table 1.
After the strength task, the participants were invited to complete a tongue isometric endurance elevation task. Participants were asked to maintain an isometric lingual pressure on the IOPI bulb equivalent to 50% of the baseline strength value until task failure, corresponding to a pressure drop >10% (visible by the LED lights of the IOPI device) for more than 2 s. During the first 3 weeks, participants had to perform 2 sets of this task interspersed with 2 min rest intervals. In the the last 3 weeks, the number of sets increased to 3 ( Table 1).
The first session was realised face-to-face with the investigator.
Oral and written instructions were given to patients to assist them in setting up the IOPI Trainer device independently. In addition, participants were contacted by phone once a week to ensure proper utilisation of the IOPI device.

| Sham protocol
The sham protocol consisted in exhaling 10 repetitions for 3 consecutive sets with 2 min rest between series in a positive expiratory pressure device (Threshold PEP) with a resistance set at the lowest output (4 cmH 2 O), 4 sessions per week for 6 weeks. The resistance increased slightly from 4 to 6 cmH 2 O over the 6 weeks to simulate a real training programme and favour adherence of participants randomised in the control group.

| Adherence
All patients had to fill a diary recording adherence to exercises (yes or no). Adequate adherence was defined by the completion of at least 75% of all exercise sessions. In the therapy group, all exercise sessions were recorded in the IOPI Trainer device, which was then additionally controlled.

| Tongue muscle assessment
The IOPI Pro device was used to assess tongue muscle elevation strength and endurance. Tongue strength is measured by obtaining maximal tongue elevation pressure. Anterior tongue elevation strength was assessed by asking the participants to squeeze as hard as possible and for 3 s the IOPI bulb positioned as mentioned above. Three trials interspersed by a 2 min resting period were allowed and the greatest value will be registered. Tongue endurance was assessed 5 min later by measuring the duration the participants were able to sustain the bulb squeeze at 50% of the maximal strength value recorded. Timing began when the participants achieved the target pressure and was discontinued when the recorded pressure dropped below 10% of the target for more than 2 s. The mean duration of two trials was reported. To ensure accurate measurement, calibration was checked and adjusted, if necessary, prior to obtaining measurements from each participant.

| Questionnaires
Subjective daytime sleepiness was measured by the Epworth Sleepiness Scale (ESS) which evaluate the propensity to sleep from 'no chance of falling asleep' (scored 0) to 'high chance' (scored 3) in eight different situations. Total score greater than 10 indicates excessive daytime sleepiness. 35 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). Total score greater than 5 indicates poor sleep quality. 36 Finally, fatigue was assessed through the Pichot questionnaire. 37 The questionnaire comprised 24 items including three homogenous sub-scales of eight items each, measuring, respectively, the depressive mood, the asthenia-fatigue and the anxiety dimensions. Only the asthenia-fatigue scale was used in the protocol which consisted of eight questions scored from '0' (not at all) to '4' (extremely).

| Randomisation
Patients were randomised in a 1:1 ratio arranged into block sizes of 6. Allocation concealment was performed using sequentially numbered sealed opaque envelopes prepared by an independent researcher not involved in the trial. The investigators (WP, NC, JT) opened the envelope after the collection of baseline measurements.

| Effects of the tongue training protocol
No changes in weight, neck and abdominal circumferences during the study period were observed in each group. In addition, there was no significant changes in AHI and other polygraphy-derived parameters F I G U R E 1 Study flow diagram between groups at the end of the 6-week trial period. In the control group, only tongue force significantly improved. In the therapy group, tongue force and endurance as well as subjective sleepiness, quality of sleep and fatigue significantly improved (Table 3, Figure 2).
Comparing the changes between groups after 6 weeks of intervention, only the ESS score and tongue endurance differed significantly, in favour of the intervention group (Table 3, Figure 2). Per-protocol analysis yielded to the same results.

| DISCUSS ION
This study showed that a 6 weeks tongue elevation strength and endurance training programme did not influence OSA severity.
However, improvements in tongue endurance and daytime sleepiness were observed.
The effects of the tongue task training protocol proposed is this study is not comparable to the acknowledged benefits of MFT on OSA severity. A recent review reported that AHI is reduced by approximately 50% in adults and 62% in children receiving MFT exercises. 16 This therapy was also shown to reduce episodes of desaturation. 16 However, the main drawback of MFT is adherence 22 which has been reported to be around 80%, but it can be as low as 50%. 18 week. 28 It is worth noting that adherence to MFT programmes are related to improvements in AHI, OSA symptoms and tongue muscle strength. 28,42 Since we obtained good patient adherence with our training programme, the lack of change on AHI may be due to the fact that only the tongue was targeted in this study, whereas previous MFT studies proposed a combination of exercises covering various oropharyngeal structures. Therefore, we assume that focusing on tongue elevation tasks may not represent a sufficiently well-rounded programme to elicit a detectable effect on polygraphy measurements. Interestingly, although we did not find a change in AHI or oxygen desaturation index, a large improvement in the ESS score was found in the tongue training group. In addition, 9 out of 12 participants (75%) had an ESS score decrease that exceeded three points, the minimum clinically important difference. 43 Although this result should be interpreted with caution given that it is a secondary endpoint, it still argues in favour Parameters Control (n = 13) Therapy (n = 12)

p-value Median [IQR] Median [IQR]
Age (  committees. However, this duration was also based on a scientific rationale. As highlighted above, short tongue-training protocols of 1 week have been found to reduce AHI and to induce neuroplastic changes in the corticomotor area of the tongue musculature. 48,49 In accordance with these findings, we observed an increase in the Second, selection bias could not be avoided. Inclusion of patients with moderate OSA was attempted based on a recent meta-analysis that evaluated the impact of MFT. 16 The vast majority of patients had moderate OSA, and the group data showed an average of 50% reduction in AHI. Based on these findings, we attempted to recruit this category of patients based on their initial diagnostic polysomnography to assess the effectiveness of an isolated tongue muscle elevation task protocol. However, some included patients had mild or severe OSA. The discrepancies between expected and measured baseline AHI could be due to different measurement methods (polysomnography at diagnosis vs polygraphy at start of the present study) 52 or the natural course of the disease due to time elapsed between diagnosis and study inclusion. Nonetheless, recent studies have found that MFT also brings benefits in patients with mild or severe OSA. 14,53,54 Therefore, albeit MFT protocols should not be confounded with our tongue task training protocol, we hypothesise that this selection bias had a negligible impact on our results. Third, we did not assess the phenotype traits of our cohort. Presumably, the best candidates for MFT are those who present low pharyngeal muscle responsiveness. 8 Future studies aiming to identify individuals who will potentially benefit the most from this therapy are needed. 55 Finally, we did not include anatomical variables in our exclusion criteria such as Mallampatti score, dental malocclusion and tongue tie. How these variables interact with the effects of tongue training in patients with OSA is unknown.

| CON CLUS ION
Our data suggest that 6 weeks of isolated tongue muscle elevation task has no effect on OSA severity.

ACK N OWLED G EM ENTS
The authors thank the company Nomics from Belgium for their kind support by providing the ambulatory polygraph free of charge during the time of the study. They also thank all the patients for their participation in this study. Finally, the authors thank the HES-SO University of Applied Sciences and Arts Western Switzerland, the European Respiratory Society, the Belgian Respiratory Society and the Foundation Saint-Luc for having supported this study. Open Access Funding provided by Haute Ecole Specialisee de la Suisse Occidentale.

FU N D I N G I N FO R M ATI O N
This was a researcher-initiated study, supported with fundings from the HES-SO University of Applied Sciences and Arts Western Switzerland, the European Respiratory Society, the Belgian Respiratory Society and the Foundation Saint-Luc.

CO N FLI C T O F I NTE R E S T
All authors declared no conflict of interest.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/joor.13369.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.