Rationale and Design of Hyperglycemic Profiles in Obstructive Sleep Apnea Trial: The HYPNOS Trial

Abstract

The Hyperglycemic Profiles in Obstructive Sleep Apnea (HYPNOS) randomized clinical trial was conducted in adults with type 2 diabetes and moderate-to-severe obstructive sleep apnea (OSA) to determine whether treatment with positive airway pressure (PAP) therapy is associated with improvements in glycemic measures. Participants were randomly assigned to PAP therapy with lifestyle counseling or lifestyle counseling alone. While observational and experimental evidence indicate that intermittent hypoxemia and recurrent arousals in OSA may alter glucose metabolism and worsen glycemic measures, the effect of treating OSA with PAP therapy on these measures in type 2 diabetes is uncertain. Adequately powered randomized clinical trials have yet to be performed to demonstrate whether PAP therapy for OSA in patients with type 2 diabetes can improve glycemic measures and glycemic control. The HYPNOS trial was designed to determine whether PAP therapy for OSA in patients with type 2 diabetes over 3 months leads to improvements in glycemic measures including glycemic variability (standard deviation) based on Dexcom G4 Platinum continuous glucose monitoring. Secondary objectives were to assess the effects of PAP therapy for OSA on measures of: (1) glycemic variability based on Abbott Freestyle Pro Libre continuous glucose monitoring; (2) point-of-care hemoglobin A1c (HbA1c); (3) degree of post-prandial hyperglycemia as determined by the 7-point self-monitoring of blood glucose; (4) clinic and ambulatory blood pressure; and (5) endothelial function. The HYPNOS trial was designed to address gaps in our understanding of the effects of PAP therapy on glucose metabolism in adults with type 2 diabetes and moderate-to-severe OSA.

Trial Registration:

ClinicalTrials.gov Identifier NCT02454153

INTRODUCTION

Numerous studies suggest that quantity^1,2 and quality^3,4 of sleep can alter glucose homeostasis. Observational and experimental research suggest that habitual short sleep duration^5,6 and sleep disorders such as obstructive sleep apnea (OSA)^7-11 can contribute to insulin resistance, glucose intolerance, and type 2 diabetes. Given the abundance of data linking OSA to abnormalities in glucose metabolism, treatment in OSA with positive airway pressure (PAP) therapy may have favorable effects on glycemic variability and control in type 2 diabetes.

Experimental studies indicate that intermittent hypoxemia and recurrent arousals, the two pathophysiological concomitants of OSA, may worsen glycemic variability and control. Sustained or intermittent hypoxia has been shown to induce hyperinsulinemia in rodent models.^12-16 Experimental studies in human subjects have demonstrated impairments in insulin sensitivity when exposed to sustained or intermittent hypoxia.^17-19 Sleep disruption due to recurrent arousals in OSA can adversely affect glucose metabolism.^20-22 Intermittent hypoxemia and recurrent arousals may increase the normal variability in daily glucose profiles through their effects on sympathetic nervous system activity particularly during the night. Other potential mechanisms (Figure 1) linking OSA to abnormal glucose metabolism include reactive oxygen species formation and increases in circulating inflammatory adipo-cytokines (i.e., interleukin-6, tumor necrosis factor-α, leptin, and resistin).

Figure 1.

Conceptual framework of mechanisms connecting obstructive sleep apnea and worsening glucose metabolism

Given the strong biological basis for the deleterious effects of OSA on glucose metabolism, a handful of studies have evaluated whether PAP treatment for OSA in type 2 diabetes is associated with improvements in glucose homeostasis and associated glycemic measures.^23-37 Most initial studies that have evaluated PAP treatment for OSA in patients with type 2 diabetes have been observational or uncontrolled pre-post assessments after the initiation of PAP therapy.^28-36 These studies have shown improvements in insulin sensitivity,^30,31 postprandial hyperglycemia,²⁹ glycemic variability,²⁸ and HbA1c.^21-23 The results from these uncontrolled studies are insufficient to provide guidance as to whether treatment of OSA, a common comorbidity in type 2 diabetes, has favorable metabolic implications. Several randomized clinical trials have examined the effect of PAP therapy on glycemic measures in adults with type 2 diabetes and previously untreated OSA.^23-27,37,38 One trial reported beneficial effects of PAP therapy on glycemic control among adults with type 2 diabetes,²³ with the rest finding no effect.^24-27,37,38 However, these trials had important limitations including inadequate power and suboptimal adherence to PAP therapy. Recently, secondary analyses in the Sleep Apnea Cardiovascular End Points (SAVE) Trial found no effect of PAP therapy on HbA1c;³⁷ yet, adherence to PAP therapy was again suboptimal.

The primary objective of the HYPNOS trial was to determine whether PAP therapy was associated with improvements in glycemic variability (standard deviation) based on Dexcom continuous glucose monitoring (CGM) in adults with type 2 diabetes with untreated moderate-to-severe OSA. Secondary objectives included assessing whether treatment of OSA with PAP therapy would lead to improvements in measures of: (1) glycemic variability based on the Abbott CGM system; (2) glycemic control (i.e., HbA1c); (3) degree of post-prandial hyperglycemia for each meal as determined by 7-point self-monitoring of blood glucose (SMBG); (3) clinic and ambulatory blood pressure; and (4) endothelial function.

METHODS

Overview of Study Design

We conducted a single center, randomized controlled trial of PAP therapy with lifestyle counseling vs. lifestyle counseling alone in adult patients with type 2 diabetes. Inclusion and exclusion criteria are detailed in Table 1. Briefly, inclusion criteria included a diagnosis of type 2 diabetes and age 21-75 years with previously untreated OSA. Exclusion criteria included insulin therapy use, a HbA1c < 6.5%, and apnea hypopnea index (AHI) less than 15 events per hour. Other exclusion criteria included restless legs syndrome, insomnia, shift work, circadian phase advance or delay as suggested by self-report of the timing and/or duration of sleep patterns, poor sleep hygiene (erratic or inconsistent sleep habits), or sleep duration of < 6 hours. After the initial screening phase, eligible participants completed a run-in phase during which participants were given an opportunity over 5-7 nights to demonstrate PAP therapy adherence, defined as PAP therapy use for at least three consecutive nights with at least four hours per night. Eligible participants who were able to tolerate PAP therapy and meet minimum adherence requirements, were randomly assigned to one of the two treatment groups. The institutional review board at the Johns Hopkins School of Medicine approved the study protocols. All participants provided written informed consent prior to enrollment in the screening phase of the study.

Table 1.

Inclusion and exclusion criteria for the Hyperglycemic Profiles in Obstructive Sleep Apnea (HYPNOS) trial

Inclusion Criteria

Age 21-75 years Type 2 diabetes mellitus HbA1c ≥ 6.5% Obstructive sleep apnea (apnea-hypopnea index [AHI] ≥ 15 events/hr)

Exclusion Criteria

Inability to consent or commit to the required visits Use of insulin for diabetes Weight change of > 10% in last six months Use of oral steroids in the last six months Pulmonary disease (e.g., asthma, COPD) requiring ongoing oral or parenteral therapy (e.g. steroids, mepoluzimab, omaluzimab), or hospitalization within last 3 months for pulmonary disease Renal or hepatic insufficiency Recent MI or stroke (< 3 months) Poor sleep hygiene (erratic or inconsistent sleep habits), sleep duration of < 6 hours, or other known sleep disorders (restless legs syndrome, sleep-related hypoventilation, insomnia, shift work, circadian phase advance or delay) Obesity-hypoventilation syndrome Body mass index ≥ 50 kg/m2 Occupation as a commercial driver or operator of heavy machinery Active substance use Untreated thyroid disease Pregnancy Any history of seizures or other neurologic disease Central sleep apnea Participants not suitable for the study based on the clinical judgment Use of investigational drug within the past 30 days Participating in another study

Recruitment and Screening Procedures

Adults with type 2 diabetes from the local Baltimore-Washington area were recruited between December 2014 and December 2019. Those responding to study advertisements completed a comprehensive telephone interview conducted by trained research study staff using a structured assessment to identify those who would be eligible. If a participant was eligible based on the telephone assessment, additional screening assessments included a point-of-care HbA1c measurement and home sleep apnea test, followed by a PAP run-in period.

At the screening visit, a point-of-care HbA1c was measured using the DC Vantage Analyzer using 1 μl of blood from a finger stick in a non-fasting state. For the home sleep apnea test, a home self-applied monitor (ApneaLink® Resmed Sydney) was provided to record for one night of sleep. Using the Apnealink® monitor, airflow was recorded with a nasal cannula connected to a pressure transducer. Pulse oximetry was used to assess oxyhemoglobin saturation and respiratory effort was measured with a pneumatic sensor attached to an effort belt. A minimum recording threshold of four hours duration was required on the limited channel study. After completion of the overnight recording, each study was manually scored and reviewed by a physician (RNA) board certified in sleep medicine. Oxyhemoglobin desaturations of at least 3% were identified and an oxygen desaturation index (ODI) for ≥ 3% was calculated as the number of oxygen desaturations events per hour of recording time. Apneas and hypopneas were defined using the following criteria. Apneas were identified if there was a 90% or greater reduction in airflow for at least 10 seconds. Hypopneas were identified if there was a ≥ 30% reduction in airflow for at least 10 seconds which was associated with an oxyhemoglobin desaturation of ≥ 3%. The apnea-hypopnea index (AHI), the disease defining metric for OSA, was computed as the number of apneas and hypopneas per hour of recording time. Participants with HbA1c < 6.5% or AHI < 15.0 events/hour were excluded. Those with a HbA1c ≥ 6.5% and an AHI ≥ 15.0 events/hour were subsequently required to demonstrate adequate use of PAP therapy during a run-in period. During this period, the PAP device needed to be used for 4 hours/night or more on at least 3 consecutive nights out of 5-7 nights for continued eligibility in the trial.

Randomization and Blinding

After confirming eligibility, eligible participants were randomized 1:1 using a simple randomization scheme to PAP therapy with lifestyle counseling or lifestyle counseling alone. Participants were not masked to treatment allocation. There were several considerations in designing a study without a sham-PAP device in the control arm. While sham-PAP itself is a relatively safe intervention, it is not without significant participant burden. Moreover, there are data to suggest that the use of sham-PAP is associated with impaired sleep quality and worse sleep efficiency,³⁹ which may influence glucose homeostasis. Given these potential issues, sub-therapeutic or sham-PAP was not used as the control intervention.

Study Interventions

The intervention included PAP therapy, titrated using a computerized auto-titration protocol embedded in the Philips Respironics System One RemStart Auto-PAP machine. Study participants had an opportunity to try various PAP mask interfaces to optimize fit. Verbal and written directions on the use of these materials were provided, with instructions reinforced after the initial period of PAP titration was completed, and then at scheduled telephone follow-up calls. After each night of use, the recorded data on usage were automatically transmitted from the participant’s PAP machine to a central database using a wireless modem communication. The research team reviewed the transmitted data and judged its acceptability based on usage for at least 4 hours/night for at least 3 consecutive nights during the run-in period which was 5-7 nights. After randomization, PAP data were visually reviewed by one of the investigators within 7-14 days to identify the optimal pressure, defined as the 95^th percentile of pressure distribution that eliminated airflow limited events.

Lifestyle counseling included education on nutrition, exercise, and adherence to medical therapy for improving glycemic control based on the American Heart Association⁴⁰ and American Diabetes Association Guidelines.⁴¹ Study staff also provided counseling on sleep hygiene using recommendations made by the American Academic of Sleep Medicine to suggest ways to improve the regularity and duration of sleep. Specifically, participants were counseled on targeting 7-8 hours of sleep at night and maintaining a regular sleep schedule.

Participants in both groups received a phone call from a research team member twice a month after the first month visit. The goal of these calls was to maintain communication with participants, help address issues related to PAP adherence, and enhance continued participation in the trial. Adherence to all routine medications was emphasized.

Data Collection Schedule

Participants had routine scheduled visits including the randomization visit at baseline followed by visits at 2-weeks, 1-month, 2-months, and 3-months (Figure 2). The baseline and 3-month assessments included the following: (a) self-administered questionnaires; (b) actigraphy; (c) blood draw and urine collection; (d) application of continuous glucose monitoring devices (DexCom G4 Platinum and Abbott Freestyle Pro Libre) and instruction in self-monitoring of blood glucose; (e) anthropometry; (f) 24-hour and clinic blood pressure measurement; and (g) peripheral arterial tonometry (Endo-PAT). The 2-week, 1-month, and 2-month visits included measurement of blood pressure, body weight, and a blood draw.

Figure 2. Screening and intervention study design.

Abbreviations: PAP, positive airway pressure; CGM, continuous glucose monitoring; SMBG, self-monitoring of blood glucose

Questionnaires

Baseline characterization included an assessment of demographic factors (race, marital status, education, work status, smoking history, alcohol history, and caffeine consumption), Epworth Sleepiness Scale,⁴² and prevalent medical conditions. Participants met with a physician prior to randomization to compile a medication list, which was then transcribed and coded. Standardized questionnaires were administered at the baseline and 3-month visits. Quality of life was ascertained using the Medical Outcomes Short-Form 36 (SF-36),⁴³ Functional Outcomes of Sleep Questionnaire (FOSQ),⁴⁴ and the Sleep Apnea Quality of Life Index (SAQLI).⁴⁵ To determine a participant’s habitual activity level, the Minnesota Leisure Time Physical Activity Questionnaire (PAQ) was used.⁴⁶

Actigraphy

Prior to randomization and then at the 3-month visit, participants were asked to collect actigraphy data. At the time the participant dropped off the PAP machine, he/she was shown how to use the actigraph (Actiwatch, Phillips Respironics). The participant was instructed to wear the actigraph on the non-dominant wrist for at least seven 24-hour periods (minimum of 5 nights, including at least one weekend/non-workday period). Actigraphy recordings were downloaded and all epochs were reviewed by at least one of the investigators. All raw data was marked for bed time (“rest”), sleep onset, sleep, and wake. Multiple parameters of sleep/wake, e.g. total sleep time, wake time, sleep efficiency, sleep latency, were calculated.

Blood Draw and Urine Collection

Venous blood was collected in vacutainer tubes, and serum/plasma was extracted. Approximately 50 ml of venous blood was drawn and equally distributed across 6-7 vacutainer tubes with approximately 7 ml in each tube. Blood draws occurred at each of the following visits: baseline, 2-weeks, 1-month, 2-months, and 3-months. A spot urine sample was collected at baseline and the 3-month follow-up visit. All specimens were stored in a −80°C freezer.

Continuous Glucose Monitoring (CGM) and Self-Monitoring of Blood Glucose (SMBG)

Participants wore two continuous glucose monitoring (CGM) devices simultaneously for up to 14 days (DexCom G4 Platinum and Abbott Freestyle Pro Libre), which were applied by a trained research technician prior to randomization. Local sites were disinfected prior to application. The G4 sensor was placed on the right lateral abdominal wall. Interstitial glucose was measured every 5 minutes for up to 14 days on the Dexcom device. Calibrations for the G4 sensor were performed using capillary blood glucose values measured at least twice per day (per the manufacturer’s instructions) with a FreeStyle InsuLinx glucometer (Abbott Diabetes Care, Inc, Alameda, CA). The Abbott Freestyle Pro Libre CGM sensor was inserted into subcutaneous tissue on the back of upper arm on the same side as the Dexcom sensor. Interstitial glucose was measured every 15 minutes for up to 14 days on the Abbott device. At the 3-month visit, a trained research technician applied both CGM devices a second time. Sensors that detached or malfunctioned were replaced as needed.

Each participant was instructed to record a 7-point SMBG profile on 3 consecutive days at baseline and at 3 months. The 7-point SMBG profile included glucose measurements at the following time points: (a) morning fasting, (b) 2 hours post breakfast, (c) before lunch; (d) 2 hours after lunch; (e) before dinner; (f) 2 hours after dinner; and (g) bedtime. We calculated post-prandial – pre-prandial change in glucose values for each meal. A commercially available glucometer (Abbott) was provided by the study along with the glucose measuring strips. The values from the 7-point SMBG were recorded by the study participant on a paper diary. Participants were asked to document their dietary intake in a food diary for 3 days during the acquisition of the 7-point SMBG and CGM data.

Anthropometry

Standard methods were employed for measurements of weight, height and for neck, waist, and hip circumferences. Measurement of neck circumference was performed by using an inelastic tape that is applied around the neck just below the laryngeal prominence. Waist circumference was measured at the narrowest part of the torso, at the end of normal expiration. For obese participants, the smallest horizontal circumference between the ribs and the iliac crest was measured. Hip circumference was measured around the buttocks in a horizontal plane at the level of the maximal extension of the buttocks. Anthropometry occurred at the baseline and 3-month visits. Total body fat was also obtained using dual-energy x-ray absorptiometry (DEXA) scans at the baseline and 3-month visits. Fat mass and percent body fat were determined using a Hologic QDR-4500A DEXA scanner.

24-Hour and Clinic Blood Pressure

Subjects underwent ambulatory 24-hour blood pressure monitoring (Welch Allyn ABPM 6100) prior to randomization. Ambulatory 24-hour blood pressure data was collected again at the 3-month visit. Participants were instructed in the use of a 24-hour ambulatory blood pressure monitor. The device consists of a blood pressure cuff that is connected by rubber tubing to a small pressure monitoring device that weighs only 9 oz and fits easily in a shirt pocket. The device was programmed to measure blood pressure at hourly intervals for the subsequent 24 hours.

Clinic blood pressure measurements were performed at all visits. Participants remained on antihypertensives and all of their usual medications for all study visits. Resting arm blood pressure was obtained in triplicate using a conventional mercury sphygmomanometer. Cuff size was determined by measuring the circumference of the upper arm, measured at the midpoint, and identifying the appropriate bladder size from a standard chart.

Endothelial Function Using Peripheral Arterial Tonometry (PAT)

Endothelial function was assessed using Peripheral Arterial Tonometry (PAT) via an Endo-PAT device (Itamar Medical) at baseline and the 3-month visit to derive a reactive hyperemia index score. Participants were positioned supine in a quiet environment with dimmed lighting. One-time use only) probes were placed on the index finger of each hand. A blood pressure cuff was placed on the non-dominant arm. Following a brief period (5 minutes) of baseline recordings, the cuff was inflated to 60 mmHg above their systolic blood pressure determined earlier (to at least 200 mmHg but not more than 300 mmHg) to occlude blood flow in the arm. The period of occlusion was timed with a stopwatch for 5 minutes. At exactly 5 minutes, the pressure was rapidly released, and readings taken for a minimum of 5 minutes. The timing of the clinic BP and EndoPAT were standardized in terms of the order of tests performed at the study visits.

PAP Adherence

Study participants randomized to the PAP arm had their PAP usage monitored objectively using a remote monitoring system. Within 7-14 of randomization, participants in the PAP arm had their PAP data reviewed by a study investigator to determine the 95^th percentile of pressure distribution that would be used to fix PAP pressure. PAP usage on each participant was monitored weekly. Those participants with suboptimal adherence, high degree of air leak, and poor treatment efficacy (AHI ≥ 5 events/hour), were contacted to technical troubleshoot the problem. Issues related to mask discomfort and poor fit were addressed by having the participant come to the clinic for an alternative mask.

Sample Size and Power

Sample size was calculated based on the magnitude of anticipated improvement in our primary outcome (e.g., standard deviation [SD] of Dexcom CGM glucose values). The primary assumption for determining sample size was an expected mean decrease in SD of CGM glucose values of 3.3 mg/dl (SD: 6.9 mg/dl) with PAP therapy with no expected change in the control group. Assuming a type 1 error and a power of 90%, a sample size of 92 participants would be needed in each arm (total N = 184). To account for modest dropout (~10%), the enrollment goal was 102 participants in PAP therapy arm and 102 participants in the control arm. A total sample size of 184 participants would also allow the study to detect a mean HbA1c difference of 0.36% between the two groups. Enrollment began in December 2014 and ended in December 2019 with a final sample size of 186 participants. In response to the COVID-19 pandemic, all clinical research activities ended in early March 2020.

Safety Monitoring

Adverse events related to either PAP use or any of the assessments were classified as mild, moderate or severe and determined to be: (a) unrelated to the study; (b) unlikely related to the study procedures; (c) possibly related to the study procedures; (d) probably related to study procedures; (e) definitely related to the study procedures. Any adverse event was defined and reported to the IRB and DSMB accordance with the rules regulating each severity class (expected vs. unexpected, serious vs. other; related vs. unrelated). The DSMB approved the protocol and consent procedures for the trial.

DISCUSSION

Emerging evidence suggests that obstructive sleep apnea (OSA) can alter glucose homeostasis and worsen glycemic control in patients with type 2 diabetes mellitus. While positive airway pressure therapy (PAP) is effective in treating OSA, its effects on glycemic variability, glycemic control, postprandial hyperglycemia, and endothelial function are not clear. Findings from this study will inform our understanding of PAP treatment among patients with type 2 diabetes. Although clinical guidelines from the American Diabetes Association mention OSA, little guidance is provided regarding its management.⁴⁷ This omission stems from the lack of data on whether PAP can improve glycemic measures. Thus, the most clinically relevant contribution of this trial will be the assessment of whether PAP therapy should be considered in the therapeutic armamentarium for type 2 diabetes.

Several limitations of this trial should be noted. First, the trial was not powered to detect small to modest effects nor evaluate subgroups. Second, oral glucose tolerance tests were not conducted due to participant burden. Third, the duration of the intervention may be not be long enough to find substantial changes in glycemic control. Fourth, CGM technology has rapidly evolved, and the technology used in the trial differs than what is currently on the market. Strengths of HYPNOS were the randomized clinical trial design which include the state-of-the-art intervention, rigorous data collection, and multi-faceted characterization of glycemia (e.g. using two CGM devices simultaneously, self-monitoring (fingerstick), HbA1c).

Findings from this trial have potential to impact the current standards of practice and provide data on whether OSA treatment can improve glycemic measures, and the feasibility of treating patients with type 2 diabetes not seeking help for sleep-related issues.

Table 2.

Data collection and visit schedule

		Month
		0	0.5	1	2	3
Interventions	Visits	X	X	X	X	X
	Adverse event assessment	X	X	X	X	X
Primary Outcome	Glycemic Variability on Continuous Glucose Monitor
	Dexcom G4 Platinum	X				X
Secondary Outcomes	Glycemic Variability on Continuous Glucose Monitor
	Abbott Freestyle Libre Pro	X				X
	Glycemic Control
	Point-of-care HbA1c	X				X
	Post-Prandial Hyperglycemia
	Self-Monitoring of Blood Glucose (SMBG)	X				X
	Blood Pressure and Endothelial Function
	24-hr blood pressure profile	X				X
	Clinical blood pressure	X	X	X	X	X
	Endothelial function (Endo-PAT)	X				X
Covariates	Usual Sleep Habits and Daily Activity Levels
	Actigraphy × 1 week	X				X
	Sleepiness and Quality of Life Assessments
	Subjective sleepiness (Epworth Sleepiness Scale)	X				X
	Quality of Life (SF-36, SAQLI, FOSQ)	X				X
	Body Composition and Nutrition
	Anthropometry (weight*, height, waist)	X				X
	DEXA scan (percent body fat)	X				X
	Food diary for calorie count and nutrient intake	X				X

^*Total body weight measured at all study visits