Effect of continuous positive airway pressure treatment of obstructive sleep apnea-hypopnea in multiple sclerosis: A randomized, double-blind, placebo-controlled trial (SAMS-PAP study)

The study was a randomized, double-blinded, placebo-controlled trial conducted at the McGill University Health Centre (primary center) and the Centre Hospitalier de l’Université de Montréal from January, 2013 through October, 2018. The study was approved by our institutional research ethics boards (ID no. NEU-11-043 and 2013-3229, CE 12.099-BSP), and all patients provided written informed consent. This trial was registered at ClinicalTrials.gov (NCT01746342).

We recruited patients from our MS Clinics with an established diagnosis of MS¹⁶ and Expanded Disability Status Scale (EDSS)¹⁷ score of 0 to 7, who were relapse-free for at least 30 days and stable on disease modifying treatment for 3 months prior to screening. Additional inclusion criteria were poor sleep quality by PSQI⁶ score > 5, substantial fatigue by FSS¹⁵ score ⩾ 4, no more than minimal cognitive impairment with a Montreal Cognitive Assessment (MoCA)¹⁸ score of ⩾ 26, and a diagnosis of OSAH by PSG with a total apnea-hypopnea index (AHI) ⩾ 15 events/hour of sleep. Exclusion criteria were pregnancy, psychiatric conditions which preclude informed consent or study requirements, or other significant neurological or active medical conditions, clinically significant laboratory abnormalities on screening blood tests, current OSAH treatment, or other symptomatic sleep disorder. Patients with severe OSAH (AHI > 30) with either a 4% oxygen desaturation index ⩾ 15 events/hour, work in a safety–critical position or severe sleepiness Epworth Sleepiness Scale Score (ESS)¹⁹ ⩾ 15 were excluded for ethical reasons.

Randomization (DACIMA software) was via a randomized block scheme with two stratification factors: study center and baseline fatigue level (FSS).

Participants, evaluating physicians, study coordinators, and statisticians were blinded to treatment assignment. However, the sleep physicians, sleep technologists, and respiratory therapists installing CPAP were not blinded. The latter were specially trained not to break the blind.

The study protocol is outlined in Figure 1. At screening, patients underwent a complete medical and standardized sleep history, including a question on morning fatigue (Are you fatigued on awakening in the morning?), Restless Legs Syndrome diagnostic questionnaire,²⁰ physical exam by a study Neurologist or Physiatrist (P.D., Y.L., D.A.T.) with determination of EDSS, MoCA, PSQI, FSS, and ESS, and blood sampling. Eligible patients then underwent complete overnight in-laboratory PSG.²¹ The subsequent randomization visit included evaluation by Sleep specialists (J.K., S.K., and V.J.) and additional questionnaires including Fatigue Scale for Motor and Cognitive Functions (FSMC),²² pain visual analog scale (VAS) scores, Multiple Sclerosis Quality of Life-54 (MSQOL-54),²³ Center for Epidemiological Studies-Depression Scale (CES-D),²⁴ and Tower of London-Drexel University (TOLDX) assessment (evaluates executive function).²⁵ After randomization, CPAP teaching, and installation, participants underwent manual CPAP titration PSG. Patients were instructed to use their CPAP nightly for 6 months. Study assessments were completed at 3 and 6 months and included an interval history, neurological exam with EDSS, relapse assessment, study questionnaires and assessments (MoCA, TOLDX, morning fatigue question only at 6 months), CPAP satisfaction and side effect questionnaire, CPAP compliance, and Sleep specialist interview. PSG on study CPAP machine was obtained prior to the final visit.

Active and sham CPAP units (Philips Respironics System One) were identical in appearance. Active CPAP was set based on PSG titration with subsequent adjustment as needed to optimize download AHI. Objective compliance was determined regularly from the CPAP memory and recorded at 3 and 6 months. Patients with suboptimal compliance received intermittent phone calls; those with persistent difficulties were referred to a blinded clinical sleep nurse. A blinding questionnaire was completed at 6 months.

Complete overnight PSG was performed using standard methods as described previously.^8,9,26,27 Data were acquired initially using Stellate systems (Natus, Inc., Harmonie 7.0 software), then from 2015 onward using Nihon-Kohden PSG 1100 (Polysmith v. software). PSG studies were scored manually by a certified technologist with sleep physician review using American Academy of Sleep Medicine (AASM) 2012 criteria,²⁸ with the exception of respiratory events which were scored using Chicago criteria.²¹ Summary PSG outcome measures were generated as previously described.^8,9 Manual CPAP titration studies were conducted using a standard approach²⁶ utilizing the CPAP flow signal as the primary respiratory signal for effective CPAP and the mask pressure and respiratory inductance signals for sham CPAP.

The primary outcome measure was the difference in FSS scores between baseline and 6 months of treatment. Based on our preliminary observational data for change in FSS following OSAH treatment in MS,⁹ considering a reported clinically important difference for FSS of 0.6 (95% CI, 0.3 to 0.9),²⁹ we calculated that a sample size of 58 patients would enable us to detect a between-group difference in FSS of 0.9, with 80% power and alpha error of 0.05. Projecting a drop-out rate of 10%, we aimed to include a total of 65 patients.

The primary analysis compared mean changes in the FSS (primary outcome) between treatment and control groups between baseline and 6 months. Secondary analyses included a comparison of mean changes between treatment and control groups for the secondary outcomes (FSMC, ESS, PSQI, pain VAS scores, EDSS, MSQOL-54 (physical and mental health composite summary; PCS and MCS), CES-D, MoCA, TOLDX, PSG variables, and objective CPAP compliance) at 6 months, and a comparison of mean changes between groups of all outcomes (when available) at 3 months. Exploratory analyses included comparison of changes in remaining outcome measures (morning fatigue question from the standardized sleep questionnaire, CPAP satisfaction questionnaire, and MS relapses) between treatment groups. Our primary analysis included all patients who completed the study protocol. This approach was adapted post hoc due to early drop-out and extensive missing data from both arms for these drop-outs (Figure 2). (Intention to treat analysis was performed secondarily and is presented in Supplemental Materials.) Missing values for outcome measures were assigned using multiple imputation through chained equations.^30,31 Continuous variables were evaluated with linear regression analysis and the binary outcome with the Wald test. The primary analysis was unadjusted, but a secondary analysis was adjusted for age and sex, but not other variables due to the relatively small number of patients. Another secondary analysis (“Per protocol analysis,” Figure 2) included only patients compliant with CPAP (using CPAP at least 4 hours per night for at least 70% of nights). Finally, in an exploratory analysis, mean disease alleviation (MDA) was calculated using the equation: [(diagnostic AHI – AHI on treatment)/Diagnostic AHI] × [CPAP Use (h/night)/Total Sleep Time].^32,33 MDA was calculated by two methods: MDA_p using AHI on treatment obtained from end of study PSG and MDA_c calculated using AHI on treatment obtained from 6-month CPAP download. MDA_c was only calculated for the treatment group since for the sham CPAP group, there was no AHI obtainable from download. The association of MDA with primary and secondary outcomes was evaluated with an adjusted linear regression model.

All tests were two-sided, and a p-value less than 0.05 was considered to be statistically significant. We did not adjust for multiple analyses since we identified a single primary outcome a priori. The data were analyzed with RStudio statistical software.³⁴

Of the 466 patients approached for the study, 103 completed screening (Figure 2). The main reasons for declining participation were lack of time or interest (n = 232). In total, 49 patients fulfilled study criteria and were enrolled. Of the patients excluded, 21 had AHI < 15, 12 exceeded the AHI and other study safety criteria and were treated emergently. 27 patients were randomly assigned to fixed CPAP treatment and 22 to sham CPAP. However, 10 patients in the active treatment group and five controls either never used CPAP or used the machine one to two times and then dropped out of the study. In total, 17 patients in each group completed the study protocol and were included in the primary analysis. Recruitment was halted prior to achieving study target population due to the depletion of operating funds.

Clinical characteristics at baseline and during the study period of patients who completed the study were similar in both groups (Table 1). Most patients were on Fingolimod, Natalizumab, or other newer MS immunomodulating medications. Patients had few other comorbidities in addition to MS (Supplementary Table S1). There were no differences in baseline characteristics of patients who completed the study and all patients who were randomized (Supplemental Table S2). However, patients who dropped out had significantly lower EDSS, lower FSS, and less severe OSAH than study completers (Supplemental Table S3).

For the primary outcome of change in FSS at 6 months, the study showed no difference between active and sham CPAP groups (p = 0.64), with a mean reduction of FSS between baseline and 6 months of 0.32 in the treatment group and 0.2 in the control group (Table 2). There were no significant differences between treatment and control groups at 6 months with respect to the secondary outcomes of fatigue as assessed by the FSMC, somnolence (ESS), sleep quality (PSQI), pain (pain VAS), EDSS, PCS score of the MSQOL-54, depressed mood (CES-D), MoCA, and TOLDX (Table 2). For the secondary outcomes at 6 months, there was a slight improvement in mental quality of life in the control group which had a mean improvement of 5.82 in the MCS score of the MSQL-54 compared to −0.75 in the treatment group (p < 0.01, 95% CI of difference 2.85, 9.83). At 3 months, for secondary outcomes, we found a statistically and clinically³⁵ significant reduction in mean sleepiness measured by ESS of 3.4 in the treatment group compared to 0.8 in the control group (p = 0.03, 95% CI of difference 0.34, 5.20). There was a clinically significant²⁹ trend for improvement in the FSS with treatment with a reduction of mean 0.9 compared to 0.3 in controls (p = 0.09, 95% CI of difference −0.23, 1.20). At 3 months, there were no other significant differences between groups for outcome measures (Table 2). The adjusted analyses were similar to unadjusted analyses.

When all patients including drop-outs were included in an intention-to-treat (ITT) analysis, the results for the primary and secondary outcomes were similar to the results in study completers (Supplemental Table S4).

PSG data are presented in Table 3. Baseline PSG characteristics between groups were similar in both groups (Table 3). With treatment, there was a significant reduction in the mean total arousal index, AHI, obstructive hypopnea index, and 4% oxygen desaturation index, and a trend to increased total sleep time in the active CPAP group only. Sham CPAP did not yield significant changes in PSG parameters. AHI reduction was significantly greater with active CPAP compared to sham.

The active treatment group had better overall compliance than the control group (Table 4). Percent days with the usage for more than 4 hours were a mean of 70% in the active CPAP group and 35% in the sham group. In total, 11 patients in the treatment group and five controls achieved the minimum requirement of per protocol CPAP usage of at least 4 hours/night for 70% of nights.

In an exploratory outcome, on a standardized sleep questionnaire, a significant proportion of patients in the treatment group reported resolution of morning fatigue compared to the control group at 6 months (p = 0.03) (Table 2). There was no difference between treatment and control groups for the exploratory outcome of MS relapses.

On the exploratory outcome of the CPAP satisfaction questionnaire, at 6 months, patients in the fixed CPAP group were significantly more likely to report that they were better off since they started using CPAP than controls (p = 0.04). In addition, patients in the fixed CPAP group tended to report more frequently that they wished to use CPAP in the long run than controls (p = 0.06).

The chi-square p-value comparing the proportions of patients and investigators guessing that patients were on active treatment provided evidence for unblinding for patients but not investigators (p = 0.0028 and p = 0.16) (Supplemental Table S5).

Adverse effects were related to the physical effects of the machine and mask. Patients in the sham CPAP group had more difficulty waking up during the night due to CPAP and those using active CPAP had more difficulty with air leaking around the mask, but otherwise side effects were similar in the two groups (Table 5).

For the subset of patients meeting threshold CPAP compliance (Figure 2), there were no significant beneficial effects of active CPAP compared with sham on primary and secondary outcome measures (Supplemental Table S6).

MDA_p (mean ± SD) was 52.4% ± 35.5% in the active group and 0.73% ± 42.5% in the control group (Table 4). Results of MDA were similar when TST from the PSQI was used to calculate MDA. However, MDA_c for patients in the treatment group was higher at 85.5% ± 38.8%. There was no association of MDA_P and FSS at 3 and 6 months in CPAP-treated patients, and no association between the change in FSS and MDA_P at 3 and 6 months (Supplemental Table S7). However, at 3 and 6 months, reduced PSQI score was significantly associated with improved MDA_p (p = 0.01 and p < 0.01, respectively).