Insomnia and sleep-disordered breathing in FKRP-related limb-girdle muscular dystrophy R9. The Norwegian LGMDR9 cohort study (2020)

Previously, we identified 153 individuals (135 adults, ≥ 16 years) with a genetically confirmed LGMDR9 in Norway [3]. They were all invited to participation in «The Norwegian LGMDR9 cohort study» at the National Neuromuscular Centre, Norway (NMK), University Hospital of North Norway (UNN). Participants provided clinical information by completing a project-specific questionnaire and by consenting to retrieve patient notes from the specialist centers. All adult participants in the cohort study were invited to respectively a sleep survey, a HRQoL survey, and a clinical observational study consisting of a 2-day visit at UNN for examinations according to a study protocol. Clinical participants without mask-based therapies for respiration disorders were additionally invited to a polysomnography (PSG) recording. The present study includes data from the two survey studies, the clinical study, and the PSG study (Fig. 1).

The surveys were administered by regular mail. The sleep survey (consisting of instruments 1–4 described below) was administered 4 weeks after the HRQoL survey (consisting of instruments 5 and 6 below). The Fatigue Severity Scale and the respiratory questionnaire were also completed during the hospital visit. The schedule for the clinical study included capillary blood gas, clinical neurological examination, Body Mass Index (BMI), echocardiography with semi-automated estimation of the biplane left ventricular ejection fraction (EF), the 32-item Motor Function Measure, and pulmonary function tests (Table 1). For eligible participants, PSG with capnometry was performed within the frame of the clinical study. PSG equipment was attached in the afternoon by a technician at the Department of Clinical Neurophysiology, and the recording was conducted ambulatory at the patient hotel. SDB included sleep-related hypoventilation, sleep-related hypoxemia, and sleep apnea (SA) as defined below. EF was reported as EF < 50% (impaired) or ≥ 50% (normal). Data from the clinical study were collected and managed using REDCap¹electronic data capture tools hosted at UNN [16, 17].

Spirometry was performed according to The American Thoracic Society and the European Respiratory Society technical guidelines 2019 [31]. In addition to the standard sitting position, spirometry was also performed in the supine position. This has been recommended for patients with NMD since diaphragmatic dysfunction tends to cause a relatively low performance in the supine position [32]. We calculated the relative drop in Forced Vital Capacity (FVC) from sitting to supine position (∆FVC), and FVC percent predicted in both positions (FVC% and FVC% supine), using reference equations for standard position from The Global Lung Function Initiative (GLI) 2021 [33]. Maximal inspiratory pressure (MIP) was performed in a sitting position and measured three times, but additional one or two times when the technique was inadequate. The best of repeated measurements was used and recorded as percent predicted (MIP%) using reference equations from GLI [34].

MFM32 is a clinician-reported quantitative scale of motor function in individuals with NMD [35]. The scale is adapted to all degrees of severity; for walking and non-walking patients. Each item is scored on a 4-point Likert scale from 0 (cannot initiate the task) to 3 (performs the task fully). Total score ranges from 0 to 96 points, based on three subdomain scores: D1—Standing and transfers (39), D2—Axial and proximal motor function (36), and D3—Distal motor function (21). The result is expressed as percentage of the maximum possible score. The assessors were MFM32-certified physiotherapists with special expertise in neuromuscular disease.

SOMNOscreen equipment and Domino version 2.7.0 software (Somnomedics, Randersacker, Germany) were used for PSG. PSG recording was performed in accordance with the guidelines of The American Academy of Sleep Medicine (AASM) version 2.4 (2017) [36]. Six electroencephalographic leads (F3/F4, C3/C4, O1/O2), right and left electrooculography, and submental electromyography were used for sleep scoring. Pressure flow oral-nasal cannula, inductive thoracic and abdominal belts, and oximetry (Nonin) were used for respiratory assessment. Body position was monitored by an accelerometer incorporated in the PSG device attached to the chest. PSG scoring was divided between two consultant clinical neurophysiologists at UNN (co-authors) and conducted according to AASM [36]. Total sleep time (TST), sleep efficiency, amount of REM sleep and deep (N3) sleep (minutes and %TST), time spent in supine position (%TST), and mean oxygen saturation (SpO₂) were reported and included in the analyses. Desaturation was defined as a ≥ 3% decrease in SpO₂. Apnea was scored when there was ≥ 10 s duration of ≥ 90% air flow reduction, and classified as obstructive, central, and mixed [36]. Hypopnea was scored when there was ≥ 10 s duration of ≥ 30% air flow reduction associated with a ≥ 3% decrease in SpO2 and/or an EEG arousal [36] and were not classified considering both the general risk of misclassification of hypopneas [37] and the additional risk in NMD where events related to respiratory muscle weakness may resemble or compound obstructive or central events [12, 14]. Oxygen desaturation index (ODI), apnea–hypopnea index (AHI), apnea index (AI), obstructive AI (oAI), central AI (cAI), and microarousal index (MAI) were calculated as events per hour of sleep. AHI was additionally calculated differentially in Non-REM (NREM) and REM sleep, and in supine and non-supine position, and the respective ratios were reported among candidates with SA. We defined SA as AHI ≥ 5 independent of symptoms and comorbidities. SA severity was defined as mild (AHI 5.0–14.9), moderate (AHI 15.0–29.9), or severe (AHI ≥ 30.0) [36]. Sleep-related hypoxemia was defined as SpO2 ≤ 88% for > 5 consecutive minutes in accordance with ICSD-3 (AASM 2014) [6], but we used the definition regardless of concurrent capnometry outcomes. Additionally, we assessed time spent with SpO2 < 90% (minutes and %TST) and the presence of Cheyne-Stokes respiration.

SenTec Digital Monitoring System (SenTec AG, Therwil, Switzerland) was used for nocturnal transcutaneous PCO2 (PtcCO2) monitoring. The sensor was placed on the forehead. Current AASM criteria for sleep-related hypoventilation were applied: an increase in PtcCO2 to a value > 7.33 kPa (55 mmHg) for ≥ 10 min and/or ≥ 1.33 kPa (10 mmHg) increase in PtcCO2 during sleep in comparison to awake supine values to a value exceeding 6.67 kPa (50 mmHg) for ≥ 10 min [38]. Mean and maximal PtcCO2, and maximal rise in PtcCO2 from initial values were used as variables.

Data were analyzed using IBM SPSS Statistics for Windows (Version 28.0. Armonk, NY: IBM Corp.). Distribution of continuous data are described using median and interquartile range (IQR). Categorical variables are presented with frequencies. Simple group comparisons from continuous variables were performed with independent t-test with bootstrapping (5000 resamples). Group comparison from categorical variables were assessed with Pearson chi-square or Fischer exact test with mid p-correction, as appropriate, and significant findings presented with odds ratio (OR) with confidence interval (CI). Correlations were examined with Spearman rank correlation (r_s) or Pearson correlation (r) according to the distributions and inspected with scatter plot and curve estimation. In the assessment of predictors of AHI, multiple linear regression (with backward elimination method) was used due to several relevant relationships with potential confounding effects: age, sex, and BMI as well-established risk factors for SA [39] and among LGMDR9-related variables (pulmonary function indices, EF < 50%, macroglossia, and dysarthria or dysphagia), EF < 50% was found relevant based on p value in t test (p < 0.10). Assumption of normal distribution of the residuals was assessed with P–P plot, skewness, and kurtosis, homoscedasticity by scatter plotting predicted versus residual values, and influential cases in terms of Cook’s distance and DeltaFit. The alpha level was set to p < 0.05. Considering the exploratory nature of the study, correction for multiple testing was not used in this study [40].

Figure 1 provides an overview of the recruitment and inclusion of the participants. In total, 77/90 (86%) adults participated in the sleep survey, and the response rate of the four instruments ranged 72–75/90 (80–83%). The response rate of the HRQoL survey was 78/90 (87%). The inclusion rate for the clinical study was 49/95 (52%). Comparable levels of age, sex distribution, wheelchair dependency, and established mask-based therapies indicated that the samples participating in the two surveys and the clinical study were representative of the cohort. In the clinical study, 35/49 (14/22 females and 21/27 males) did not receive mask-based therapy and were thus eligible for PSG. However, six participants (five females and one male) either declined the invitation to PSG or were not invited due to practical inconvenience, and three recordings (of one female and two males) were unsuccessful, hence excluded. The 26 successful PSG registrations represented 8 of 14 (57%) eligible females and 18 of 21 (86%) eligible males.

Background data and outcomes of the sleep survey are presented in Table 2. There was a comparable number of female and male participants. The subgroup with mask-based therapies comprised eight respondents with continuous positive airway pressure support (CPAP) (i.e., 10% of all respondents) and 16 with NIV (i.e., 21% of all respondents) as bi-level positive airway pressure or ventilator, of which five also used NIV at daytime. The same subgroup had a female preponderance (63% vs 38%), was older (median age 56 vs 37 years), and had a higher prevalence of wheelchair dependency (67% vs 17%).

In total, 32% had insomnia and 10% EDS. Overlap of insomnia and EDS occurred in two respondents (Fig. 2). The prevalence of fatigue or frequent daytime tiredness ranged 37–40%. Frequent nightly awakenings and/or non-restorative sleep were common (36–49%), whereas 10% reported frequent morning headache and 3% frequent nocturnal dyspnea.

Insomnia was equally prevalent in the subgroup with nocturnal masks as in those without. Only one patient with mask-based therapy had EDS, i.e., severe residual sleepiness. Compared to the subgroup without masks, those who used masks showed a tendency towards a higher prevalence of nightly awakenings (55% vs 38%) and wake after sleep onset (> 30 min) (36% vs 12%) but a relatively lower prevalence of sleep onset latency (18% vs 31%). Prevalence of early awakening and dissatisfaction with sleep in the two groups was comparable (22%).

The prevalence of insomnia was 36% in females and 27% in males (p = 0.40) and unrelated to wheelchair dependency (p = 0.81). Level of insomnia symptoms was not correlated with age (r_s = 0.07) or physical or social aspects of HRQoL but to poorer outcomes on mental aspects of HRQoL, especially vitality, pain, and fatigue (Table 3). Correspondingly, the prevalence of insomnia was increased in fatigued compared to non-fatigued patients; 54% vs 21%; (OR 4.41, CI 1.52–12.79, p = 0.005) (Fig. 2). Concurrently, the prevalence of EDS was relatively low (10%) and not significantly increased in fatigued patients (p = 0.30) (Fig. 2).

Background data and outcomes of the PSG study and additional assessments of the 26 PSG candidates during their hospital visit are shown in Table 4. Females and males had comparable age and BMI, but females showed relatively poorer outcomes on pulmonary function tests and a higher proportion of males had an impaired EF. Two subjects were wheelchair dependent. The PSG candidates with indications of SDB, except mild SA, are presented individually in Table 5. One of the PSG candidates, a male with comorbid chronic obstructive pulmonary disease and an EF < 50% had tried mask-based therapy previously but not tolerated it (Table 5, Subject 1). PSG showed severe SA, and he was the only PSG candidate with Cheyne-Stokes respiration and a SpO2 < 90% exceeding 5% TST. He also met the criteria for insomnia in the survey, which may have contributed to the lack of tolerance to the mask. Nineteen PSG candidates had a successful capnometry. Mean PtcCO2 levels were median 5.69 kPa in females and 5.99 kPa in males and the highest mean value recorded was 6.62 kPa. Maximum PtcCO2 levels were median 6.13 kPa in females and 6.58 kPa in males and the highest level recorded was 7.25 kPa (Table 4). Only one met the applied criteria for sleep-related hypoventilation (Table 5, Subject 9). No correlation between PtcCO2 levels and PFT for any of the variables included in the study (see method section) was found. None met the applied criterion for sleep-related hypoxemia. SA was detected in 16/26 PSG candidates (4/8 females and 12/18 males) and was moderate or severe in 8/26 (Table 4). Among the 16 subjects with SA, median oAI was 6.8/h and cAI 0.2/h and the oAI comprised > 50% of AHI in one half, whereas the hypopneas predominated in the other half. A cAI ≥ 5 was only recorded in one subject. This subject had an implanted cardiac resynchronization therapy defibrillator and an EF < 50% but also a predominant, severe OSA (Table 5, Subject 2).

All four PSG candidates with a BMI > 30 kg/m² and all eight with EF < 50% had SA. AHI was relatively increased in those with an EF < 50% compared to those with a normal EF (p = 0.096). AHI was uncorrelated with the PFT metrics (r_s = 0.18 to − 0.06, p ≥ 0.40) and unrelated to macroglossia (n = 4) and dysarthria and/or dysphagia (n = 3). Multiple regression showed that AHI correlated with advancing age and an EF < 50% but not with BMI or sex (Table 6). We assessed the relationships between AHI and sleeping position and REM sleep, respectively. Respiratory events tend to increase in the supine sleeping position and patients with LGMDR9 may be more prone to the supine position due to mobility difficulties. REM sleep tend to aggravate OSA and events related to respiratory muscle weakness due to physiological REM atonia, as mentioned, whereas central apneas usually occur in NREM sleep [41]. Among the subjects with SA, median AHI ratio supine/non-supine was 3.2 (range 1.3–16.4) and median AHI ratio REM/NREM was 2.7 (range 0–6.2).

The AHI was unrelated to the items of the respiratory questionnaire. However, the two individuals with severe SA stood out with problems on three of five items (Table 5, Subject 1 and 2). No correlations between FSS and nocturnal measurements were detected including TST, amount N3 sleep (minutes and %TST), amount REM sleep (minutes and %TST), AHI, MAI, ODI, mean SpO2, and mean PtcCO2. Nevertheless, relationships between FSS and pulmonary function impairment were observed and most clearly between FSS and MIP% (Table 7, Fig. 3). In the subgroup with SDB, a correlation also between FSS and ∆FVC was found. In the whole group without mask-based therapies, FSS correlated both with MIP% and FVC % supine but not with MFM32 (D3: r = − 0.25, D2: r = − 0.20, D1: r = − 0.10, total score: r = − 0.15) and capillary PCO2, bicarbonate, or base excess. Comparatively, in the group with established mask-based therapies, no correlation between FSS and PFT metrics was detected and there was even a tendency of an inverse relationship between fatigue severity and ∆FVC (Table 7). This group consisted of four subjects with CPAP and ten subjects with NIV (one also partly daytime and another 24 hours) and had relatively higher levels of age, BMI, and wheelchair dependency as well as poorer outcomes on PFT compared to the group without mask-based therapy (Table 7).

Insomnia was more prevalent relative to two previous Norwegian population studies (32% vs 20%) [23, 42]. Co-existing EDS and insomnia was rare, which resonates with previous knowledge that sleepiness is usually not present in insomnia [7]. Potential contributors to an increased prevalence of insomnia in LGMDR9 are immobility, myalgia, mental distress, and SA [8, 43‐45]. This study showed that levels of insomnia were associated with mental aspects of HRQoL including pain. Previously, we found that both mental, physical, and social aspects of SF-36, but not pain, were significantly poorer in our LGMDR9 cohort compared to a reference population (Jensen SM et al., submitted paper). Insomnia was also related to the mental aspects SF-36 vitality and INQoL fatigue and the increased prevalence of insomnia in fatigued compared to non-fatigued patients substantiates this relationship. This finding aligns with the aforementioned study in slowly progressive muscular dystrophies [10] and with a study of a large clinical sample with insomnia [46]. In the latter study, a bidirectional relationship was proposed and depression was identified as a significant mediator. Worth noting is that CBT-I is also recommended for comorbid insomnia, particularly as it may potentially improve accompanying conditions such as pain or depression [7].

Patients who received mask-based therapy also had a high prevalence of insomnia, but the nocturnal pattern differed such that there were less difficulties with sleep initiation and more excessive wake time throughout the night, compared to patients without a mask. However, early morning awakening was comparable in both subgroups. Insomnia in these patients may reflect inappropriate mask-based therapy or the need for additional intervention for insomnia. As aforementioned, insomnia in patients with mask-based therapy is a particular concern since it may affect adherence and, resultantly, treatment outcomes. The potential need for interdisciplinary treatment in comorbid SDB and insomnia may be a limiting factor.

The diagnosis of SDB relies on objective measurements of respiratory functions during sleep. Symptoms of untreated or sub-optimally treated SDB may reflect in the respiratory questionnaires, gauging of morning headache and nocturnal dyspnea, and the ESS assessing EDS. Although EDS has heterogeneous causes, it is commonly associated with SDB [47]. The survey showed that the prevalence of these potential symptoms of SDB was low. EDS was even relatively low compared to a previous Norwegian general population study (9.6% vs 17.7%) [48].

AHI was elevated in a high proportion of our PSG candidates. This was not unexpected considering the high prevalence in general populations [39]. Furthermore, applying the specified criteria for sleep-related hypoventilation and hypoxemia, only one mild case of sleep-related hypoventilation was detected. Nevertheless, as previous studies have demonstrated, this highly depends on the diagnostic criteria [49‐51]. In Duchenne muscular dystrophy, disease-specific treatment criteria for hypoventilation, more liberal than AASM criteria, have been established, yet comparative trials remain [9, 51]. However, a low rate of sleep-related hypoventilation in our study can be explained by the sample. Compared to those with established mask-based therapy, the PSG sample represented a relatively younger age group and earlier stages of disease by a low rate of wheelchair dependency and more preserved pulmonary function. Additionally, the group with mask-based therapy had a relatively higher rate of obesity (median BMI > 30 kg/m²), which is an independent risk factor for hypoventilation (obesity-hypoventilation syndrome).

While the AHI in one half of those with SA was predominated by obstructive apneas, the other half had predominantly hypopneas, which could be obstructive, central, diaphragmatic, or compounded. AHI was related to advancing age, which is a general risk factor for OSA [39] but could also be related to progression of the disease. Concerning comorbid associations with SA, a Finnish nationwide registry-based case–control study found that SA (including OSA and CSA) is strongly associated with obesity, heart failure, and respiratory disease [52]. In concurrence, we found an association with heart failure. Previous studies have shown that heart failure is associated with both CSA and OSA and suggested that heart failure may promote OSA through CSA or upper airway edema [53], but the mechanisms have not been established. In the current study, central apneas were rare. Nonetheless, considering the unspecified nature of the hypopneas, this study cannot distinguish whether the correlation between AHI and an EF < 50% relates to a specific type of events. Further, the low BMI levels and relatively preserved pulmonary function among our PSG candidates, in general, may explain why BMI and pulmonary metrics did not turn out as significant correlates of the AHI. This may also indicate that SA is not underrecognized among patients with obesity or significant respiratory involvement since treated patients were not included in the PSG study. A report from a task force of the European Respiratory Society and the European Sleep Research Society advised to take the individual susceptibility into account in treatment decisions [54]. Comorbidities like heart failure, obesity, and respiratory muscle weakness may represent such vulnerabilities. Furthermore, SA treatment has shown potential to improve comorbidities such as obesity, metabolic disease, and heart failure [53, 55], which emphasizes the importance of considering morbidity profile in the monitoring of SDB.

AHI was related to the supine sleeping position and to REM sleep, which is well known in OSA [56] but could also be related to diaphragm weakness [14]. Since the sleeping position and the proportion of REM sleep during PSG registration may differ from habitual sleep without recording devices, these variables also need to be considered to avoid under or overdiagnosis. Additionally, the association with position means that positional therapy, preventing supine sleeping position, may be an option. Nevertheless, patients with muscle disease may need special consideration due to mobility or pain issues.

AHI was poorly correlated with sleep-related symptoms. However, research shows that SA is commonly sub-clinical, and both symptomatic and sub-clinical SA may be associated with cardiovascular disease [57]. This means that detection and intervention could be relevant independent of symptoms and consequently that regular sleep studies may be required in risk patients. Additionally, since symptoms and AHI are poor predictors of severity, there also exists a need for better biomarkers for treatment decisions [54].

Our study did not support the idea that fatigue in our LGMR9 cohort could be related to untreated SDB. However, we did find fatigue severity to be negatively correlated with inspiratory muscle strength (MIP). Since FSS was uncorrelated with metrics of blood gases and motor function, the mechanism of the FSS-MIP relationship seems more likely to be the work of breathing rather than an association with general motor function impairment or respiratory disturbances caused by the inspiratory weakness. The FSS-MIP relationship was absent in the group with established mask-based therapy. Potential explanations are the alleviation of breathing work due to ventilatory support and less physical exposure due to a higher rate of wheelchair dependency. Interestingly, a correlation between FSS and MIP has also been found in polio-myelitis [58] and several other studies have shown that MIP is a clinically meaningful outcome measure for NMD [59‐61].

This study provides insights into the prevalence of insomnia and unrecognized SDB in LGMDR9 and clinically relevant relationships with these sleep disorders and with fatigue. Important strengths of the surveys were sample sizes and response rates. The clinical study included a representative sample of the cohort and integrated assessments with standardized methods. PSG and PCO2 monitoring are considered gold standard methods for diagnosing SDB.

The study also has several limitations. Due to the exploratory design, we accepted a higher risk of conducting type I error (false positive), and instead of correcting for multiple testing, we urge for caution in assigning significance to p-values in the range of 0.01 to 0.05. Findings that are flagged as significant require replication. The prevalence of chronic insomnia disorder may be underestimated as the BIS does not cover all daytime impairments included in ICSD-3 (e.g., cognitive impairment, mood disturbance, and impaired motivation) or overestimated since a clinical interview is required to rule out exclusion criteria [7], such as insomnia due to poor sleep environment or insomnia mimics like circadian rhythm problems or restless legs syndrome. The PSG sample size was low, which increases the risk of type II errors (false negative conclusions), and females were relatively underrepresented. Identification of individuals with SDB relied on diagnostic criteria, which are mostly based on expert opinions [49, 60]. PSG may overestimate AHI by its tendency to increase the time in supine position [61] or underestimate it by reducing the proportion of REM sleep [62]. Sleep during PSG may have been impacted by an unfamiliar sleeping environment and setting and myalgia or tiredness after physical tests. However, TST, percentage of N3 sleep, and sleep efficiency indicated successful registrations. Technical issues limited the data completeness on capnometry and supine spirometry. Lastly, the study was conducted during the COVID-19 pandemic. Although the data were collected in relatively normal periods, this may have impacted the subjective measurements (HRQoL, sleep, and fatigue).