Sleep apnea in diabetes

J Sleep Res. (2015) 24, 425–431

Association between obstructive sleep apnea severity and glucose control in patients with untreated versus treated diabetes PASCALINE PRIOU1,2, MARC LE VAILLANT3, NICOLE MESLIER1,2, SYLVAINE CHOLLET4, THIERRY PIGEANNE5, PHILIPPE MASSON6, ACYA BIZIEUX-THAMINY7, MARIE-PIERRE HUMEAU8, FRANC ßO I S G O U P I L 9 ,  ER  IC GAGNADOUX1,2 THE IRSR P I E R R E - H E N R I D U C L U Z E A U 2 , 1 0 a n d F R ED SLEEP COHORT GROUP 1 partement de Pneumologie, Universite  d’Angers, CHU, Angers, France; 2INSERM U1063, Angers, France; 3CERMES, CNRS UMR8211De ^pital Laennec, Nantes, France; 5Unite  de Pneumologie, INSERM U988-EHESS, Villejuif, France; 4Service d’Explorations Fonctionnelles, Ho 6 ^le sante  des Olonnes, Olonne sur Mer, France; Service de Pneumologie, Centre Hospitalier, Cholet, France; 7Service de Pneumologie, Po Centre Hospitalier, La Roche sur Yon, France; 8Pneumologie, Nouvelles Cliniques Nantaises, Nantes, France; 9Service de Pneumologie, Centre Hospitalier, Le Mans, France and 10Departement d’Endocrinologie, Diabetologie, Nutrition, CHU, Angers, France;

Keywords diabetes, glycated haemoglobin, intermittent hypoxia, obstructive sleep apnea Correspondence de  ric Gagnadoux, MD, PhD, De partement de Fre Pneumologie, CHU, 4 rue Larrey, 49033 Angers Cedex 9, France. Tel.: 33 241353695; fax: 33 241354974; e-mail: [email protected] Accepted in revised form 20 December 2014; received 22 October 2014 DOI: 10.1111/jsr.12278

SUMMARY

The purpose of this study was to determine whether the association between obstructive sleep apnea severity and glucose control differs between patients with newly diagnosed and untreated type 2 diabetes, and patients with known and treated type 2 diabetes. This multicentre cross-sectional study included 762 patients investigated by sleep recording for suspected obstructive sleep apnea, 497 of whom were previously diagnosed and treated for type 2 diabetes (treated diabetic patients), while 265 had no medical history of diabetes but had fasting blood glucose ≥126 mg dL 1 and/or glycated haemoglobin (HbA1c) ≥6.5% consistent with newly diagnosed type 2 diabetes (untreated diabetic patients). Multivariate regression analyses were performed to evaluate the independent association between HbA1c and obstructive sleep apnea severity in treated and untreated patients with diabetes. In untreated diabetic patients, HbA1c was positively associated with apnea– hypopnea index (P = 0.0007) and 3% oxygen desaturation index (P = 0.0016) after adjustment for age, gender, body mass index, alcohol habits, metabolic dyslipidaemia, hypertension, statin use and study site. The adjusted mean value of HbA1c increased from 6.68% in the lowest quartile of the apnea–hypopnea index (61; P = 0.033 for linear trend). In treated patients with diabetes, HbA1c was associated with non-sleep variables, including age, metabolic dyslipidaemia and insulin use, but not with obstructive sleep apnea severity. Obstructive sleep apnea may adversely affect glucose control in patients with newly diagnosed and untreated type 2 diabetes, but may have a limited impact in patients with overt type 2 diabetes receiving anti-diabetic medications.

INTRODUCTION Obstructive sleep apnea (OSA) and type 2 diabetes are common co-morbid conditions (Foster et al., 2009). There is growing evidence in support of complex interactions between ª 2015 European Sleep Research Society

OSA and impaired glucose metabolism that may contribute to the increased cardiovascular morbidity in OSA (McNicholas and Bonsigore, 2007). Studies in animal and human models mimicking sleep-disordered breathing have identified several potential intermediate mechanisms linking OSA and impaired

425

426

P. Priou et al. (i) data on fasting blood glucose, HbA1c values and antidiabetic medications were available at inclusion in the database; (ii) baseline evaluation indicated the presence of type 2 diabetes as defined by fasting blood glucose ≥126 mg dL 1 and/or HbA1c ≥6.5% and/or the use of anti-diabetic medications regardless of fasting blood glucose and HbA1c values.

glucose metabolism, including intermittent hypoxia (Iiyori et al., 2007; Oltmanns et al., 2004) and reduced slow-wave sleep (Tasali et al., 2008). A vast body of literature has been published establishing a robust independent cross-sectional association between the presence and severity of OSA and pre-diabetic states such as insulin resistance and glucose intolerance in both population- and clinic-based cohorts of non-diabetic subjects (for review, see Reutrakul and Van Cauter, 2014). A recent metaanalysis of randomized controlled trials demonstrated that continuous positive airway pressure (CPAP) therapy improves insulin resistance in patients with OSA without diabetes (Iftikhar et al., 2013). Whether untreated OSA adversely affects glycaemic control in patients with type 2 diabetes remains uncertain, as previous studies evaluating the association between OSA and glycated haemoglobin (HbA1c) in patients with diabetes yielded conflicting results (Aronsohn et al., 2010; Einhorn et al., 2007; Grimaldi et al., 2014; Kent et al., 2014; Lam et al., 2010). A large clinic-based study showed that 30% of patients with newly diagnosed OSA have co-morbid type 2 diabetes with up to 40% of previously undiagnosed diabetes (Meslier et al., 2003). Previous studies evaluating the association between OSA and glucose control in patients with diabetes pooled subjects with and without oral hypoglycaemic medications or insulin (Aronsohn et al., 2010; Einhorn et al., 2007; Grimaldi et al., 2014; Kent et al., 2014), although anti-diabetic drugs markedly decrease HbA1c levels. This multisite cross-sectional study was designed to determine whether the association between OSA severity and glucose control differs between patients with known and treated type 2 diabetes and patients with newly diagnosed and untreated type 2 diabetes.

Baseline evaluation included anthropometric data, smoking habits, alcohol consumption, medical history and medication use. Fasting blood samples were taken for measurement of blood glucose, HbA1c and serum lipid levels. As previously described (Priou et al., 2012), HbA1c was assayed in accredited laboratories and was used as a clinical indicator of glucose control. Blood pressure was measured in the evening, ~2 h before the start of the sleep recording using a periodically calibrated mercury sphygmomanometer. Hypertension was defined by systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg, and/or the use of anti-hypertensive treatment. Waist circumference (WC; cm) was measured using a non-elastic measuring tape. Metabolic dyslipidaemia (MD) was defined by the combination of elevated triglyceride levels (≥150 mg dL 1) and low highdensity lipoprotein cholesterol levels (≤50 mg dL 1 for women and ≤40 mg dL 1 for men; Rana et al., 2010; Trzepizur et al., 2013). Excessive daytime sleepiness was assessed by the Epworth Sleepiness Scale (ESS; Johns, 1991). Depression was defined by a QD2A score ≥7 (Gagnadoux et al., 2014) and/or the use of anti-depressant medication.

MATERIALS AND METHODS

Sleep studies

Design and study population This multisite study was conducted on the Institut de  Respiratoire des Pays de la Loire Recherche en Sante (IRSR) sleep cohort. Since 15 May 2007, consecutive patients ≥18 years old investigated by overnight polysomnography (PSG) or overnight respiratory recording for suspected OSA in seven centres in the west of France have been eligible for inclusion in the IRSR sleep cohort. Approval was obtained from the University of Angers ethics committee  Consultatif sur le Traitement de l’Information and the ‘Comite re de Recherche dans le domaine de la Sante ’ en matie (C.C.T.I.R.S.; 07.207bis). The database is anonymous and complies with the restrictive requirements of the ‘Commission ’ (C.N.I.L.), the French Nationale Informatique et Liberte information technology, and personal data protection authority. All patients included in the IRSR sleep cohort have given their written informed consent. Between 15 May 2007 and 4 April 2014, patients from the IRSR sleep cohort were eligible for the present study when:

Surveys, measurements and questionnaires

Obstructive sleep apnea was diagnosed by overnight PSG or overnight respiratory recording. As previously described (Gagnadoux et al., 2014), PSG was performed with continuous recording of the following channels: electroencephalogram, electrooculogram, chin electromyogram, arterial oxygen saturation, nasal-oral airflow (pressure cannula and tracheal sounds), electrocardiogram, chest and abdominal wall motion, and body position. Overnight respiratory recordings were performed with continuous recording of oxygen saturation, nasal-oral airflow, chest and abdominal wall motion, and body position. Respiratory events were scored manually according to recommended criteria (American Academy of Sleep Medicine, 1999). Apnea was defined as cessation of airflow for ≥10 s. Hypopnea was defined as a ≥ 50% reduction of airflow or a < 50% reduction of airflow accompanied by a ≥ 3% decrease in SaO2 (and/or arousal in patients undergoing overnight PSG). OSA was defined by an apnea–hypopnea index (AHI) ≥5. OSA severity was evaluated by the AHI and the 3% oxygen desaturation index (ODI).

ª 2015 European Sleep Research Society

OSA and glucose control in diabetes Statistical analysis All statistical analyses were performed with SAS software (SAS/STAT Package 2002–2003 by SAS Institute, Cary, NC, USA). Continuous variables were described as mean [standard deviation (SD)] for variables with a normal distribution, and median [interquartile range (IQR)] for variables with a non-normal distribution. Characteristics of the study population were determined according to anti-diabetic treatment status. Subjects with previously diagnosed and treated type 2 diabetes were recorded as treated diabetic patients. Subjects with fasting blood glucose ≥126 mg∙dL 1 or HbA1c ≥6.5%, but no medical history of type 2 diabetes and no previous anti-diabetic medication were recorded as untreated diabetic patients. Untreated and treated diabetic patients were compared using two-sample t-test for continuous variables with a normal distribution, Mann–Whitney test for continuous variables with a non-normal distribution and Chi-square test for categorical variables. The primary dependent variable of interest was HbA1c. Primary independent variables included AHI and ODI. Multivariate regression analyses were performed separately for untreated and treated diabetic patients to characterize the independent associations between HbA1c and AHI (model 1), and between HbA1c and ODI (model 2). To adjust for potential confounders, the following covariates that are likely to interfere with glucose control were entered in the multivariate analysis: age, gender, body mass index (BMI), alcohol abuse, MD, hypertension, statin use and insulin use. Adjustment for study site was also performed using a generalized estimating equation model (Liang and Zeger, 1993). In order to estimate the effect size of increasing OSA severity on glucose control, adjusted means of HbA1c were calculated according to quartiles of AHI and ODI in treated and untreated diabetic patients. We verified that residuals of the regression analysis were normally distributed using a Kolmogorov–Smirnoff test (P > 0.15). A goodness-of-fit test (Fisher’s exact test) was also conducted in order to verify the adequacy of the models (P < 0.05 for all models; McCullagh and Nelder, 1989). Results were expressed as percentages and mean (SD), except as indicated. A two-tailed P-value

Association between obstructive sleep apnea severity and glucose control in patients with untreated versus treated diabetes.

The purpose of this study was to determine whether the association between obstructive sleep apnea severity and glucose control differs between patien...
188KB Sizes 0 Downloads 8 Views