DIABETICMedicine DOI: 10.1111/dme.12817

Systematic Review or Meta-analysis Effect of obstructive sleep apnoea on diabetic retinopathy and maculopathy: a systematic review and meta-analysis W. B. Leong1,2, F. Jadhakhan3, S. Taheri4,5, Y. F. Chen6, P. Adab7 and G. N. Thomas7,8 1 School of Clinical and Experimental Medicine and Birmingham and Black Country NIHR CLAHRC, University of Birmingham, Birmingham, UK, 2Specialist Weight Management Services, Heart of England NHS Foundation Trust, Birmingham, UK,, 3Primary Care Clinical Sciences, University of Birmingham, Birmingham, UK, 4 Department of Medicine, Weill Cornell Medical College in New York, NY, USA and Doha, Qatar, 5Department of Medicine, King’s College London, London, UK, 6 Division of Health Sciences, University of Warwick, Coventry, UK, 7Public Health, Epidemiology and Biostatistics, University of Birmingham, Birmingham, UK and 8 Institute of Public Health, Social and Preventive Medicine, Mannheim Medical Faculty, Heidelberg University, Mannheim, Germany

Accepted 26 May 2015

Abstract Aims To summarize the association between obstructive sleep apnoea and diabetic retinopathy and diabetic maculopathy, and to examine the effects of oxygen desaturation index, mean and minimum oxygen saturation and time spent with < 90% oxygen saturation on diabetic retinopathy and diabetic maculopathy. Methods A systematic search was performed for papers published from inception to January 2014 in MEDLINE, EMBASE and the Cochrane Database of Systematic Reviews using indexed terms and free text. Additional searches were carried out for grey literature. Two authors conducted the study selection and quality assessment. Data extraction was performed by the main author and checked by the other authors.

One cohort study and 15 cross-sectional studies were included for narrative synthesis and three for metaanalyses. There was no convincing evidence that obstructive sleep apnoea was associated with diabetic retinopathy, although some evidence suggested that obstructive sleep apnoea was associated with greater severity of diabetic retinopathy and advanced diabetic retinopathy in people with Type 2 diabetes. Only six studies examined the impact of obstructive sleep apnoea on diabetic maculopathy and our narrative review suggests there is an association in Type 2 diabetes. Oxygen desaturation index, mean oxygen saturation or time spent with < 90% oxygen saturation were not associated with diabetic retinopathy, and insufficient evidence was available to draw conclusions on their effects on diabetic maculopathy; however, there was evidence from both narrative synthesis and meta-analysis that minimum oxygen saturation had an impact on diabetic retinopathy (pooled odds ratio 0.91, 95% CI 0.87–0.95; I2 = 0%). Results

Conclusions There is a need for large cohort studies with long-term follow-up data to examine the long-term effects of obstructive sleep apnoea and other sleep variables on advanced retinal disease in diabetes.

Diabet. Med. 33, 158–168 (2016)

Introduction Obstructive sleep apnoea (OSA) is characterized by chronic intermittent nocturnal hypoxaemia, arousal and sleep fragmentation. It has been found to affect between 58 and 86% of people with diabetes [1,2] and is increasingly recognized as a contributor to insulin resistance. One potential mechanism is that OSA may alter glycaemic control through the activation of inflammatory and oxidative stress pathways, such as increased advanced glycation end products [3]. Moreover, advanced glycation end products have been Correspondence to: Shahrad Taheri. E-mail: [email protected]

158

implicated in the pathogenesis of vascular complications in diabetes [4]. It is unsurprising, therefore, that some studies have shown an association between OSA and the presence of diabetic retinopathy (DR) [5,6], leading to the hypothesis that, in people with diabetes, the presence of OSA may predispose to the development and progression of retinal disease [2]. If OSA does accelerate the progression of DR, it would strengthen the need to assess OSA among people with diabetes; however, the evidence supporting this hypothesis is inconsistent. One study showed that chronic intermittent hypoxaemia was associated with worse proliferative DR among Japanese individuals [5]. By contrast, our previous work found no significant association between DR and OSA

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

Systematic Review or Meta-analysis

What’s New: • Obstructive sleep apnoea (OSA), accompanied by hypoxaemia, is common amongst patients with type 2 diabetes mellitus. Recently, it has been proposed that OSA and hypoxemia are associated with retinal disease in diabetes. A systematic review was conducted to examine this association. • The systematic review highlighted several methodological pitfalls in previous studies, but identified that minimum nocturnal oxygen saturation was associated with diabetic retinopathy. Thus, there is a need to consider hypoxemia as an important contributor to retinal disease in diabetes.

[7]. These differences may be attributable to methodological variations, such as the definition of OSA used, or the population under study, or may be masking small effects that are not identified within single studies. We conducted a systematic review to summarize the association between OSA and the risk of diabetes retinal disease (retinopathy and maculopathy). The measures used to assess OSA include the apnoea-hypopnoea index, oxygen desaturation index, time spent below 90% oxygen saturation at night (percentage of time spent with < 90% oxygen saturation), mean oxygen saturation, and minimum oxygen saturation. By using the different measures to assess OSA, we aimed to further understand the pathophysiological pathways mediating the impact of OSA on DR and/or diabetic maculopathy.

Methods The systematic review was conducted according to the Preferred Reporting Items for Systematic Review and Metaanalysis (PRISMA) and the Meta-analysis of Observational Studies in Epidemiology (MOOSE) statements. The protocol is registered in the PROSPERO database (CRD42014008757).

Eligibility criteria

The full inclusion and exclusion criteria for the systematic review are summarized in Table S1. Briefly, the review had no language restriction and included papers that met the following criteria: 1) participants: adults with Type 1 or Type 2 diabetes; 2) exposure: diagnosis of OSA either in accordance with the American Academy of Sleep Medicine guidance [8] or using either the apnoea-hypopnoea index from polysomnography or the oxygen desaturation index from pulse oximetry; 3) comparator: people with no OSA; 4) outcomes: development of DR or diabetic maculopathy (DR was categorized into overall DR and advanced DR. Advanced DR was defined as either preproliferative or proliferative DR. Diabetic

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

DIABETICMedicine

maculopathy consisted of both macular oedema and ischaemia); and 5) study design: observational studies (crosssectional, cohort or case–control studies). We also included OSA prevalence studies if data on diabetes microvascular complications were available [6,9–13].

Search strategy

Indexed terms, such as Medical Subject Headings (MeSH) terms, and free text were used to capture relevant literature. The search terms used are described Table S2. No restriction was applied to the searches performed from inception until January 2014 of the following electronic databases: MEDLINE, EMBASE and Cochrane Library. OpenGrey and Zetoc were also searched for grey literature. Finally, we hand searched citations from included papers.

Study selection

Two authors independently screened the titles and abstracts then reviewed the full texts of the eligible studies. Disagreements were resolved through discussion and a third author arbitrated, if required. In cases of duplicate or ‘kin’ studies, we chose the most recent study with the most complete data available.

Data extraction

A standardized form was designed based on the STrengthening the Reporting of OBservational studies in Epidemiology statement (STROBE) and was piloted on five studies. Improvements were made before formal use. Data extraction was performed by one author and checked by other authors. When additional information was needed, we contacted the study authors.

Study quality

A quality assessment form was devised from the Newcastle Ottawa Scale for non-randomized studies in meta-analyses [14]. Study quality was assessed according to five components: selection bias; respiratory measurement; blinding of the assessor performing DR or sleep analyses; study methods; and statistical analysis (Table S3). Each component had a set of criteria and each criterion was rated ‘yes’, ‘no’ or ‘unclear’. An overall judgment of either ‘weak’, ‘moderate’ or ‘strong’ was made for each component. Subsequently, an overall judgment rating of ‘weak’, ‘moderate’ or ‘strong’ was determined as per recommendations by the Cochrane Collaboration. This was carried out independently by two reviewers.

Data synthesis and analysis

The findings were synthesized according to the outcome reported, including studies which reported on overall DR,

159

Effect of obstructive sleep apnoea on diabetic retinopathy and maculopathy  W. B. Leong et al.

DIABETICMedicine

Results The initial search identified 851 studies (814 from databases and 37 from grey literature). After removal of duplicates, there were 731 records and 678 were excluded after title and abstract screening. A total of 53 full-text articles were reviewed and 37 were excluded; the remaining 16 studies were included in the systematic review (Fig. 1).

Study characteristics

The characteristics of the included studies, study population, methods of assessing OSA and the outcome measures used are summarized in Table 1. Only one study [15] was longitudinal, and the remainder were cross-sectional. The majority of the studies were from Europe and overall, included a total of 2731 participants with diabetes (2636 with Type 2 diabetes). Only two studies recruited participants with diabetes from the community [10,16]. There was

Included

Eligibility

Screening

Identification

advanced DR (preproliferative or proliferative DR) and diabetic maculopathy. Statistical analysis was carried out using STATA 13 (StataCorp LP, College Station, TX, USA). To minimize the influence of potential confounders, we performed random effects meta-analyses using adjusted odds ratios (ORs), in which at least three studies provided data for similar measures of OSA and retinopathy outcomes. These included the effect of OSA on advanced DR and the effect of minimum oxygen saturation on DR. Because of the small number of studies, no meta-analysis was carried out for the effects of oxygen desaturation index, mean oxygen saturation and percentage of time spent with < 90% oxygen saturation. Funnel plots were used to assess publication bias and small study effects for metaanalysis which contained more than five studies. We also carried out j statistic analysis to measure the agreement between the two authors. The inter-rater agreement was high (к=0.95) for study selection and substantial (к=0.79) for quality assessment.

FIGURE 1 Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) flow chart on study selection. OSA, obstructive sleep apnoea; DR, diabetic retinopathy.

160

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

Systematic Review or Meta-analysis

DIABETICMedicine

Table 1 Characteristics and results of included studies Study ID (country)

Study design

Storgaard et al. 2014¶ (Denmark) [13]

Patients

Demographics

OSA assessment

Results

CS

180 Type 2 diabetes

Berlin questionnaire then level III sleep test (ApneaLink) +/- level IV sleep test (Embletta)*

Univariate analysis for DR - No difference in DR between OSA (26.4%) vs. non-OSA (19.4%) (P > 0.05) - No difference in apnoea-hypopnoea index between DR 17  20 vs. non-DR 17  25 events/h (P > 0.05)

Nishimura et al. 2013§ (Japan) [27]

CS

136 Type 2 diabetes

Setting = diabetes outpatient Mean age = 59 years‡ Females = 42%‡ Mean HbA1c = 7.4%‡ Mean diabetes duration = 8.2 years‡ Mean BMI = 32.2 kg/m2 ‡ Setting = unclear Mean age = NA Females = NA Mean HbA1c = NA Mean diabetes duration = NA Mean BMI = NA

NA

Furukawa et al. 2013 (Japan) [22]

CS

513 Type 2 diabetes

Univariate analysis for DR - No significant difference in oxygen desaturation index for DR 17.1  15.2 vs. non-DR 15.4  14.7 events/h (P = 0.54). - No significant difference in apnoeahypopnoea index for DR 17.8  15.1 vs. non-DR 15.2  13.6 events/h (P = 0.33) Multivariate analysis for DR - Association between minimum oxygen saturation and DR (OR 0.89, 95% CI 0.83–0.95). Multivariate analysis for DR - No association between OSA and DR (OR 1.00, 95% CI 0.60–1.68)

Banerjee et al. 2013 (UK) [7]

CS

93 Type 2 diabetes

Altaf et al. 2013§ (UK) [15]

Cohort

199 Type 2 diabetes

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

Setting = multicentre diabetes outpatients Mean age = 62 years Females: 43% Mean HbA1c = 7.3% Mean diabetes duration = 11.7 years Mean BMI = 25.2 kg/m2 Setting = specialist weight management clinic Mean age = 52 years Females = 59% Mean HbA1c = 8.1% Mean diabetes duration = 7.6 years Mean BMI = 47.3 kg/m2

Oximetry (PULSOX-3Si)*

Setting = diabetes outpatient Mean age = NA Females = 43% Mean HbA1c = NA Mean diabetes duration

Level III sleep test (Alice PDX)*~

Level III sleep test (Embletta)*~

Multivariate analysis for DR - No association between OSA and DR (OR 1.00, 95% CI 0.98–1.02) - No association between minimum oxygen saturation and DR (OR 0.93, 95% CI 0.67–1.02) - No association between mean oxygen saturation and DR (OR 0.93, 95% CI 0.86–1.01) - No association between percentage of time spent with < 90% oxygen saturation and DR (OR 1.03, 95% CI 1.00–0.06) - No association between OSA and advanced-DR (OR 1.01, 95% CI 0.97– 1.04; unpublished data) Multivariate analysis for Diabetic maculopathy - No association between OSA and diabetic maculopathy (OR 1.01, 95% CI 0.98–1.04) - No association between mean oxygen saturation and diabetic maculopathy (OR 0.80, 95% CI 0.59–1.07) - Association between minimum oxygen saturation and diabetic maculopathy (OR 0.79, 95% CI 0.65–0.95) - No association between percentage of time spent with < 90% oxygen saturation and diabetic maculopathy (OR 1.03, 95% CI 0.99–1.08) Multivariate analysis for advanced DR - Association between OSA and advanced DR (OR 3.90, 95% CI 1.02– 15.30) - Followed-up for 4.4 years, OSA independently associated with

161

Effect of obstructive sleep apnoea on diabetic retinopathy and maculopathy  W. B. Leong et al.

DIABETICMedicine

Table 1 (Continued) Study ID (country)

Study design

Patients

Demographics

OSA assessment

Results progression of advanced DR (OR 6.60, 95% CI 1.20–13.10) Multivariate analysis for diabetic maculopathy - Association between OSA and diabetic maculopathy (OR 4.5, 95% CI 1.8–11.4) - Follow-up 4.4 years, OSA not associated with progression of diabetic maculopathy (effect size not reported) Multivariate analysis - Association between total retinopathy scores and OSA (P = 0.008) - No association between OSA and advanced DR (OR 12.6, 95% CI 0.62– 255.76) - No association between OSA and diabetic maculopathy (effect size not reported) Multivariate analysis for DR - No association between OSA and DR (effect size not reported)

= NA Mean BMI = NA

Rudrappa et al. 2012 (UK) [18]

CS

31 Type 2 diabetes

Setting = diabetes outpatient Mean age = 55 years Females = 0% Mean HbA1c = 9.2% Mean diabetes duration = 11.0 years Mean BMI = 35.3 kg/m2

Inpatient polysomnography (Visilab)

Mehta et al. 2012§ (India) [17]

CS

80 Type 2 diabetes

Inpatient polysomnography

Mason et al. 2012 (UK) [12]

CS

80 Type 2 diabetes with macular oedema

Setting = outpatient retinal clinic Mean age = NA Females = NA Mean HbA1c = NA Mean diabetes duration = NA Mean BMI = NA Setting = macular oedema outpatient Mean age = 65 years Females = 50% Mean HbA1c = 7.8% Mean diabetes duration = 15.0 years Mean BMI = 30.2 kg/m2

Kosseifi et al. 2012†,¶ (US) [19]

CS

98 Type 2 diabetes

Level III sleep test (NovaSom QSG)

West et al. 2010 (UK) [16]

CS

118 Type 2 diabetes

Shiba et al. 2010 (Japan) [20]

CS

219 Type 2 diabetes

Setting = outpatient sleep clinic Mean age = 61 Females = NA Mean HbA1c = 6.5% Mean diabetes duration = NA Mean BM = 33.7 kg/m2 Setting = 1 diabetes outpatient and 5 primary care centres Mean age = 67 years‡ Females = 0% Mean HbA1c = 8.0%‡ Mean diabetes duration = 10.0 years‡ Mean BMI = 30.2 kg/ m2‡ Setting = hospital inpatient (eye disease) Mean age = 63 years‡ Females = 43% Mean HbA1c = 7.3% Mean diabetes duration = 11.5 years Mean BMI = 24.2 kg/m2

162

Level IV sleep testing (Apnealink)ǂ~

Univariate analysis for DR - No associated between OSA and DR (P = 0.32) Univariate analysis for diabetic maculopathy - No association between macular thickness and apnoea-hypopnoea index - No association between macular thickness and percentage of time spent with < 90% oxygen saturation - No dose–response association between severity of oxygen desaturation index and macular thickness Univariate analysis for DR - Apnoea-hypopnoea index in DR group (44.2) higher than non-DR group (P < 0.05)

Oximetry and in some level III sleep test (Embletta)~

Multivariate analysis - Retinopathy scores associated with OSA (r2 = 0.19, P < 0.0001) - OSA associated with diabetic maculopathy (r2 = 0.30, P < 0.0001)

Oximetry (PMP-200G)

Multivariate analysis for advanced DR - Association between advanced DR with neovascularisation and OSA (OR 1.09, 95% CI 1.01–1.16) - Association between minimum oxygen saturation and advanced DR (OR 0.88– 0.99) - No association between percentage of

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

Systematic Review or Meta-analysis

DIABETICMedicine

Table 1 (Continued) Study ID (country)

Study design

Patients

Schober et al. 2011¶ (Germany) [10]

CS

58 Type 1 diabetes & 498 Type 2 diabetes

Borel et al. 2010¶ (France) [9]

CS

37 Type 1 diabetes

Unver et al. 2009 (USA) [24]

CS

44 diabetes

Laaban et al. 2009¶ (France) [11]

CS

303 Type 2 diabetes

Merritt et al. 2007§ (UK) [21]

CS

44 Type 2 diabetes

Demographics

OSA assessment

Setting = 14 primary care centres Mean age = 60 years Females = 48% Mean HbA1c = 7.6% Mean diabetes duration = 9.3 years Mean BMI = 31.9 kg/m2 Setting = diabetes outpatient Mean age = 43 years Females = 32% Mean HbA1c = 7.8% Mean diabetes duration = 23.0 years Mean BMI = 24.7 kg/m2 Setting = outpatient retinal clinic Mean age = 60 years‡ Females = NA Mean HbA1c = 7.5%‡ Mean diabetes duration = 16.3 years‡ Mean BMI = NA

Level IV sleep test (ApneaLink)ǂ~

Setting = hospital inpatient (diabetes) Mean age = 61 years Females = 49% Mean HbA1c = 9.2% Mean diabetes duration = 14.5 years Mean BMI = 32.0 kg/m2 Setting = diabetes outpatient Mean age = 61 years Females = 32% Mean HbA1c = 8.0%‡ Mean diabetes duration = 16.5 years‡ Mean BMI = 30.4 kg/m2‡

Level IV sleep test (CID 102)~

Oximetry

NA

Oximetry (Minolta 300i)

Results time spent with < 90% oxygen saturation and advanced DR (effect size not reported) - No difference in mean oxygen saturation between OSA (97.6%  1.2) and non-OSA (97.3%  1.5; P = 0.14) Univariate analysis for DR - No difference in DR between OSA (13.9%) vs. non-OSA (11.5%; P = 0.45)

Univariate analysis for DR - Significant difference in DR between pathological recording (73.3%) vs. borderline recording (11.1%; P < 0.05)

Univariate analysis for DR - Significant difference between DR and OSA (OR 143.18, 95% CI: 7.41 to 2767.79) - 10 eyes had pan-retinal photocoagulation in OSA group vs. 1 in non-OSA group Univariate analysis for Diabetic maculopathy - Association between OSA and diabetic maculopathy (OR 14.44, 95% CI; 2.34 to 147.60) - 7 OSA patients had laser treatment for macular oedema and most worsened after treatment with some needing ≥ 2 treatment. Nine non-OSA patients had laser treatment for macular oedema and majority resolved without further progression Univariate analysis for DR - Significant association between OSA and DR (OR 0.56, 95% CI 0.35–0.92)

Univariate analysis for advanced-DR - No significant difference in oxygen desaturation index between advancedDR (15.4  14.7) vs. nonadvanced-DR (17.1  15.2) (P = 0.54) - Significant difference in percentage of time spent with < 90% oxygen saturation between advanced DR (12.6%  19.7) and non-advanced DR (1.8%  2.6) (P = 0.03)

OSA, obstructive sleep apnoea; CS, cross-sectional study; NA, information not available; DR, diabetic retinopathy; OR, odds ratio. Bold text indicates significant results. *Minimum 4 h recording; †Veteran study with low female participants; ‡Average values of the 2 groups; § Conference abstracts; ¶Prevalence studies

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

163

DIABETICMedicine

Effect of obstructive sleep apnoea on diabetic retinopathy and maculopathy  W. B. Leong et al.

much heterogeneity between studies in the way that OSA was assessed and reported. This included in-hospital polysomnography in two studies [17,18] and portable home sleep studies [7,10–12,15,19] using a range of devices and pulse oximetry [9,20–22]. Furthermore, the criteria used to define OSA and related measures were very variable (Table S4). Only two of eight studies that provided information on the definitions of apnoea, hypopnoea or oxygen desaturation used the American Academy of Sleep Medicine criteria [7,11].

Definition of outcomes

The definition of DR was variable, and in six studies no clear definition was provided [9–11,23,24]. Definitions included use of diagnosed DR through the English National Screening Program for Diabetic Retinopathy [7,15,16,18], or from medical records [17,20]. Other measures undertaken were measurement of macula thickness using a dual laser spectraldomain optical coherence tomography [12], and florescence fundoscopy by ophthalmology specialists [22].

advanced DR status (15%) compared with participants without OSA (3%; P = 0.01). Furthermore, it was reported that OSA predicted the development of advanced DR (OR 6.6, 95% CI 1.2–35.1) after adjustment for potential confounders. Among the other three studies, only one found a significant association [24]. In a meta-analysis of three studies that reported adjusted ORs for advanced DR [7,15,25], pooled estimates did not show a significant association between advanced DR and OSA (pooled OR 1.05, 95% CI 0.95–1.16); however, a significant level of heterogeneity was present (I2 = 73.2%). In conclusion, from our narrative review, there was some evidence that OSA was associated with greater severity of DR as well as advanced DR among people with Type 2 diabetes. The meta-analysis performed for the cross-sectional studies, however, did not support the association between advanced DR and OSA. The significant level of heterogeneity as well as the small number of studies mean that the pooled results may not be reliable and need to be interpreted with caution [26].

Association between diabetic retinopathy and other Association between obstructive sleep apnoea and diabetic retinopathy in Type 2 diabetes

Eight studies (n = 1134) reported the effect of OSA on overall DR [7,10–13,17,19,24]; only two (n = 173) of which used multiple logistic regression analyses [7,17] to adjust for confounders (Table 1). These studies, and four of the six studies that reported non-adjusted ORs, reported no significant association between OSA and DR [10–13]. Only two small studies from the USA (n = 142) [19,24], in which no adjustments for potential confounders were made, reported significant associations; however, these unadjusted models were likely to be exposed to multiple confounders such as gender and BMI, therefore, overall, there was no convincing evidence that OSA was associated with DR in their Type 2 diabetes population.

Obstructive sleep apnoea and retinopathy scores/advanced diabetic retinopathy in Type 2 diabetes

Five studies examined the association between OSA and retinopathy scores as well as advanced DR [7,15,16,18,24]. Two studies (n = 149, only male participants) reported a significant association between OSA and total retinopathy scores using adjusted models [16,18]. Four studies (n = 367) [7,15,18,24], including one cohort study [15] examined the relationship between OSA and advanced DR. The study, which was longitudinal with a follow-up of 4.4  1.0 years and had adjusted for potential confounding factors, found significant associations, which were clinically important. The cohort study also found that participants with OSA were more likely to progress from no or background DR status to

164

respiratory sleep measurements in Type 2 diabetes

Four studies used oximetry to assess the effect of oxygen desaturation index on DR in Type 2 diabetes populations (n = 949) [21,22,25,27]. Only one study reported a significant association in people with Type 2 diabetes [25]. The others did not show any associations [21,22,27]. In summary, the evidence on the effect of oxygen desaturation index on DR was inconclusive in Type 2 diabetes populations. Apart from oxygen desaturation index and apnoea-hypopnoea index, sleep studies also recorded percentage of time spent with < 90% oxygen saturation and mean and minimum oxygen saturations. Two studies (n = 312) assessed the effect of mean oxygen saturation and neither found significant associations with DR [5,7]. Three studies (n = 356) [5,7,21] examined the effect of percentage of time spent with < 90% oxygen saturation on DR, with only one study reporting a significant association [21]. There was insufficient evidence, therefore, that mean oxygen saturation or percentage of time spent with < 90% oxygen saturation was associated with DR in populations with Type 2 diabetes. Three studies (n = 448) [7,20,27] examined the effect of minimum oxygen saturation, with two reported significant associations using adjusted models in populations with Type 2 diabetes [20,27]. A meta-analysis was carried with a pooled OR of 0.91 (95% CI 0.87–0.95, I2 = 0%; Fig. 2). In summary, there was evidence from both the narrative synthesis and the meta-analysis that minimum oxygen saturation had an impact on DR in a population with Type 2 diabetes. Because of the limited available studies reporting on the effects of minimum oxygen on DR, publication bias may be present.

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

Systematic Review or Meta-analysis

DIABETICMedicine

Study

%

ID

OR (95% CI)

Weight

Banerjee 2013

0.93 (0.67, 1.02)

4.27

Shiba 2011

0.93 (0.88, 0.99)

54.37

Nishimura 2013

0.89 (0.83, 0.95)

41.36

Overall (I-squared = 0.0%, p = 0.620)

0.91 (0.87, 0.95)

100.00

NOTE: Weights are from random effects analysis .67

1

1.49

FIGURE 2 Forest plot of the pooled estimate of the effects of minimum oxygen saturation on diabetic retinopathy using a random effects model.

Diabetic maculopathy and respiratory sleep measurements

Six studies (including one cohort) explored the impact of OSA on diabetic maculopathy (n = 565) in people with Type 2 diabetes [7,12,15,16,18,24], with three reporting significant associations [15,24,28]. The cohort study reported a significant association at baseline, but no excess risk of diabetic maculopathy was observed after 4 years of followup among those with OSA compared with those without [15]. No association was found in two studies that examined the effect of percentage of time spent with < 90% oxygen saturation (n = 173) [7,12] nor in one study (n = 80) that examined the effect of oxygen desaturation index on diabetic maculopathy [12]. Likewise, only one study (n = 93) reported on the effect of both mean and minimum oxygen saturations on diabetic maculopathy [7] in a population with Type 2 diabetes. Although mean oxygen saturation was not found to have an impact; minimum oxygen saturation was a predictor of diabetic maculopathy (adjusted OR 0.79, 95% CI 0.65–0.95) [7]. In summary, there was some evidence that OSA was associated with diabetic maculopathy but insufficient evidence to draw conclusions on the effect of mean oxygen saturation, percentage of time spent with < 90% oxygen saturation, minimum oxygen saturation and oxygen desaturation index on diabetic maculopathy in people with Type 2 diabetes.

Associations between obstructive sleep apnoea and other sleep variables on diabetic retinopathy and diabetic maculopathy in Type 1 diabetes

Only one study (n = 37) examined the effect of oxygen desaturation index on DR and reported a significant association [9]; however, this was a small study and did not adjust for any confounders so no conclusion can be drawn. No study has examined the association between OSA and DR or diabetic maculopathy in Type 1 diabetes. Similarly, no study has investigated the effects of minimum and mean oxygen saturation, percentage of time spent with < 90% oxygen

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

saturation and DR or diabetic maculopathy in Type 1 diabetes.

Quality assessment

Table 2 summarizes the quality assessment of the studies included in the present review. The majority of studies were rated as either moderate or strong for study methods (n = 11) and the overall quality scores were weak to moderate (n = 14). Only two studies were rated as strong quality. Selection bias was an issue in the majority of the studies (n = 10 rated as weak). Methods of recording sleep data were not presented in several studies (n = 7 were weak), in particular, the criterion used for the diagnosis of OSA was poorly described. Only two studies blinded the sleep record assessor, while seven studies blinded the DR assessment.

Discussion Our review found that there was some evidence from crosssectional studies and a cohort study, that OSA was associated with greater severity of DR as well as advanced DR amongst people with Type 2 diabetes. Data on diabetic maculopathy were more limited and inconclusive. Whilst overall, larger cross-sectional studies suggested an association with OSA, this was not confirmed in a longitudinal study [15]. This may be attributable to the relatively short duration of follow-up giving insufficient time for development of complications. We also reported evidence from both our narrative synthesis and our meta-analysis that the level of hypoxaemia was associated with DR in Type 2 diabetes populations. Additionally, one study found that minimum oxygen saturation was associated with diabetic maculopathy [7] in a population with Type 2 diabetes. Because of the limited availability of studies that report on the effects of minimum oxygen saturation on DR, publication bias of positive results cannot be excluded. More studies are needed to examine the effect of minimum oxygen saturation, especially in people with Type

165

DIABETICMedicine

Effect of obstructive sleep apnoea on diabetic retinopathy and maculopathy  W. B. Leong et al.

Table 2 Quality assessment of included studies Components Study

Selection bias

Respiratory measurement

Blinding

Study methods

Analysis

Overall

Storgaard et al. 2014 [13] Nishimura et al. 2013 [27] Furukawa et al. 2013 [22] Banerjee et al. 2013 [7] Altaf et al. 2013 [15] Rudrappa et al. 2012 [18] Mehta et al. 2012 [17] Mason et al. 2012 [12] Kosseifi et al. 2012 [19] West et al. 2010 [16] Shiba et al. 2010 [20] Schober et al. 2011 [10] Borel et al. 2010 [9] Unver et al. 2009 [24] Laaban et al. 2009 [11] Merritt et al. 2007 [21]

Weak Moderate Weak Strong Weak Weak Weak Weak Weak Moderate Moderate Weak Weak Moderate Moderate Weak

Moderate Weak Moderate Strong Strong Weak Weak Moderate Weak Moderate Moderate Moderate Weak Weak Strong Weak

Weak Weak Moderate Strong Weak Moderate Weak Moderate Weak Strong Moderate Weak Weak Weak Moderate Moderate

Strong Moderate Weak Moderate Weak Strong Moderate Strong Moderate Strong Weak Strong Moderate Weak Moderate Weak

Weak Moderate Strong Strong Moderate Moderate Moderate Weak Weak Strong Strong Weak Weak Weak Weak Weak

Weak Moderate Weak Strong Moderate Moderate Weak Moderate Weak Strong Moderate Weak Weak Weak Moderate Weak

2 diabetes and diabetic maculopathy. It will be of particular interest to know if low oxygen saturation in those with mild OSA has any impact on DR as well as diabetic maculopathy in the long term, given that this group of individuals is not likely to be offered continuous positive airway pressure treatment according to current guidelines and clinical practice. Currently, there is a dearth of evidence on the effects that OSA and other sleep variables have on DR in populations with Type 1 diabetes. There was a high degree of heterogeneity between studies in the devices used to record sleep measurements. The ‘gold standard’ for the diagnosis of OSA is a laboratory polysomnography with an attending technician. Given the increasing demand for sleep studies and the scarcity of facilities and technicians, portable devices are a useful cost-effective alternative. In the present review, the majority of the devices had three channels except for three studies, which used a device with only two channels. Studies also differed greatly in their definitions of apnoea/hypopnoea and oxygen desaturation as well as in the diagnostic criteria used for OSA. The information regarding the minimum recording period was also generally unavailable. As the majority of the studies were cross-sectional, it is difficult to infer causality; however, the study by Altaf et al. [15] was prospective and reported significant worsening of DR after 4 years. It is likely that OSA resulted in the development and greater progression of DR. The impact of OSA on DR may be mediated in several ways, including sympathetic activation, hypertension and through exacerbation of glycaemic control via insulin counter-regulatory hormone release, and increasing activation of deleterious molecular pathways in target organs. Downstream pathways include excess production of superoxide leading to the activation of the polyol, advanced glycation end products,

166

protein kinase C and hexosamine pathways [4]. These pathways subsequently activate oxidative stress and inflammatory reactions resulting in endothelial dysfunction. The visual process is a highly demanding process, therefore, the retina is very vulnerable to hypoxia [29,30]. Nocturnal hypoxaemia accompanying OSA may also cause a direct insult to the retina. When the retina is exposed to hypoxaemia, a cascade of inflammatory and oxidative stress reactions occur. Exposure of rats to intermittent hypoxia at different levels of oxygenation showed that those with the greatest level of hypoxia (60%) had the highest glucose, insulin and inflammatory marker levels [31]. Oxygen therapy in rats exposed to hypoxia has likewise been shown to reduce inflammation [32]. Similarly, in skin biopsies of patients with OSA, those with oxygen saturations < 75% had greater levels of inflammatory gene expression [33]. It is also plausible that diabetic microvascular complications promote the development of OSA. People with DR usually also have other complications, including diabetic neuropathy. Diabetic neuropathy can affect pharyngeal muscle control and increase upper airway collapsibility. A case series of individuals with Charcot-Marie-Tooth disease found that the neurological condition correlated highly with OSA (r = 0.69, P = 0.029) [34]. Similarly, people with diabetes with autonomic neuropathy were more likely to have OSA compared with those without the condition [28]; therefore, the association between OSA and DR is probably bi-directional. Our systematic review has some limitations. The reliance on the results from cross-sectional studies is a concern. Cross-sectional studies can be subject to selection bias. This was certainly reflected in our quality assessment in both selection bias and blinding of sleep analysis assessor; however, four studies were conference abstracts and two were

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

Systematic Review or Meta-analysis

short reports with limited information available and our quality assessment tool may not have comprehensively assessed the rigor of these studies. Although several studies adjusted for confounders, the issue of residual confounders remained. We also did not examine the possible association between sleep apnoea-related arousals and DR. The majority of the included studies were from European countries and there were a few from Asia with two small studies from North America. There were no large studies in North and South American, African or Asia-Pacific populations. The mechanistic effect of OSA on DR or diabetic maculopathy, however, should not differ in these populations, so our results should be generalizable. Although the present review explored the effects of OSA on DR, it is important to note that glycaemic control and hypertension are still the greatest risk factors for DR. This systematic review revealed key shortfalls in the reporting of the methodology used for sleep studies. We recommend that future studies should report the device used for sleep assessment, minimum recording time, definition for apnoea/ hypopnoea and criteria used for the diagnosis of OSA, as a minimum. In addition, several of the respiratory sleep measurement results were missing in majority of the studies. This could perhaps be avoided if core outcome reporting for sleep studies were introduced for observational studies. Another important consideration is blind scoring of sleep/ retinopathy data. Statistical analyses of such data should also report effect sizes after adjustment of major potential confounders such as glycaemic control and blood pressure. The results from our systematic review raise questions about the reliance of the apnoea-hypopnoea index as a sole criterion for the treatment of OSA. Given the high prevalence of OSA in people with Type 2 diabetes, the potential association between OSA and diabetes complications, and plausibility of shared pathophysiological mechanisms, there is a need for future large prospective studies with long-term follow-up data to examine the effects of OSA, including sleep apnoea-related arousals as well as other respiratory measurements on non-advanced as well as advanced DR in both Type 1 diabetes and Type 2 diabetes populations. This will help determine if the level of nocturnal hypoxaemia is a better predictor than the apnoea-hypopnoea index for the development or progression of diabetic microvascular disease in people with Type 2 diabetes. Likewise, it is also important to follow up OSA-free individuals with DR to determine if diabetes microvascular complications contribute to OSA development. Currently, it is unclear if continuous positive airway pressure can have a beneficial effect on DR; however, a recent proof-of-principle study reported potential improvement in visual acuity after continuous positive airway pressure treatment in those with diabetic maculopathy [35]. If this is proven to be effective in randomized control trials, then continuous positive airway pressure could potentially be incorporated as part of diabetic retinal disease management.

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

DIABETICMedicine

Funding sources

ST is funded by the Biomedical Research Program (BMRP) at Weill Cornell Medical College in Qatar, supported by Qatar Foundation.

Competing interests

None declared.

References 1 Resnick HE, Redline S, Shahar E, Gilpin A, Newman A, Walter R et al. Diabetes and sleep disturbances: findings from the Sleep Heart Health Study. Diabetes Care 2003; 26: 702–709. 2 Foster GD, Sanders MH, Millman R, Zammit G, Borradaile KE, Newman AB et al. Obstructive sleep apnea among obese patients with type 2 diabetes. Diabetes Care 2009; 32: 1017–1019. 3 Tan KC, Chow WS, Lam JC, Lam B, Bucala R, Betteridge J et al. Advanced glycation endproducts in nondiabetic patients with obstructive sleep apnea. Sleep 2006; 29: 329–333. 4 Pallayova M, Banerjee D, Taheri S. Novel insights into metabolic sequelae of obstructive sleep apnoea: A link between hypoxic stress and chronic diabetes complications. Diabetes Res and Clin Pract 2014; 104: 197–205. 5 Shiba T, Sato Y, Takahashi M. Relationship between diabetic retinopathy and sleep-disordered breathing. Am J Ophthalmol 2009; 147: 1017–1021. 6 Kosseifi S, Bailey B, Price R, Roy TM, Byrd RP, Jr., Peiris AN. The association between obstructive sleep apnea syndrome and microvascular complications in well-controlled diabetic patients. Military Med 2010; 175: 913–916. 7 Banerjee D, Leong WB, Arora T, Nolen M, Punamiya V, Grunstein R et al. The potential association between obstructive sleep apnea and diabetic retinopathy in severe obesity-the role of hypoxemia. PloS One 2013; 8: e79521. 8 Iber C, Ancoli-Israel S, Chesson A, Jr., Quan S. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. American Academy of Sleep Medicine: Westchester, IL; 2007. 9 Borel AL, Benhamou PY, Baguet JP, Halimi S, Levy P, Mallion JM et al. High prevalence of obstructive sleep apnoea syndrome in a Type 1 diabetic adult population: a pilot study. Diab Med 2010; 27: 1328–1329. 10 Schober AK, Neurath MF, Harsch IA. Prevalence of sleep apnoea in diabetic patients. Clin Respir J 2011; 5: 165–172. 11 Laaban JP, Daenen S, Leger D, Pascal S, Bayon V, Slama G et al. Prevalence and predictive factors of sleep apnoea syndrome in type 2 diabetic patients. Diabetes Metab 2009; 35: 372–377. 12 Mason RH, West SD, Kiire CA, Groves DC, Lipinski HJ, Jaycock A et al. High prevalence of sleep disordered breathing in patients with diabetic macular edema. Retina 2012; 32: 1791–1798. 13 Storgaard H, Mortensen B, Almdal T, Laub M, Tarnow L. At least one in three people with Type 2 diabetes mellitus referred to a diabetes centre has symptomatic obstructive sleep apnoea. Diabet Med 2014; 31: 1460–1467. 14 Wells G, Shea B, O’Connell D, Peterson J, Welch V, M Losos et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomized studies in meta-analyses. Ottawa Hospital Research Institute. Available at http://www.ohri.ca/programs Last accessed 1 January 2015. 15 Altaf Q, Dodson P, Ali A, Raymond NT, Wharton H, Fellows H et al. Obstructive sleep apnoea is associated with sight threatening

167

DIABETICMedicine

16

17

18

19

20

21

22

23

24

25

26

27

28

29

Effect of obstructive sleep apnoea on diabetic retinopathy and maculopathy  W. B. Leong et al.

retinopathy and predicts the development of preproliferative and proliferative retinopathy in patients with Type 2 diabetes: A longitudinal analysis. Diabet Med 2013; 30: E5. West SD, Groves DC, Lipinski HJ, Nicoll DJ, Mason RH, Scanlon PH et al. The prevalence of retinopathy in men with Type 2 diabetes and obstructive sleep apnoea. Diabet Med 2010; 27: 423–430. Mehta S, Chaudhry D, Singh SV, Atreja A, Sangwan V. Sleep disordered breathing (SDB) in patients of type 2 diabetes mellitus (DM) with and without retinopathy-a hospital based study. Thorax 2012; 67: A24. Rudrappa S, Warren G, Idris I. Obstructive sleep apnoea is associated with the development and progression of diabetic retinopathy, independent of conventional risk factors and novel biomarkers for diabetic retinopathy. Br J Ophthalmol 2012; 96: 1535. Kosseifi S, Bailey B, Price R, Roy TM, Byrd RP, Jr., Peiris AN. The association between obstructive sleep apnea syndrome and microvascular complications in well-controlled diabetic patients. Military Med 2010; 175: 913–916. Shiba T, Maeno T, Saishin Y, Hori Y, Takahashi M. Nocturnal intermittent serious hypoxia and reoxygenation in proliferative diabetic retinopathy cases. Am J Ophthalmol 2010; 149: 959–963. Merritt SP, Moxham J, Wong A, Steier J, Carroll P, Williams AJ. Sleep disordered breathing, a causative factor in the development of severe diabetic retinopathy? Thorax 2007; 62: A25. Furukawa S, Saito I, Yamamoto S, Miyake T, Ueda T, Niiya T et al. Nocturnal intermittent hypoxia as an associated risk factor for microalbuminuria in Japanese patients with type 2 diabetes mellitus. Eur J Endocrinol 2013; 169: 239–246. Koyama S, Takagi H, Otani A, Oh H, Nishimura K, Honda Y. Inhibitory mechanism of vascular endothelial growth factor (VEGF) by bucillamine. Br J Pharmacol 2002; 137: 901–909. Unver YB, Yavuz GSA, Stafford CA, Sinclair SH. A putative relation between obstructive sleep apnea and diabetic macular edema associated with optic nerve fiber layer infarcts. Open Sleep J 2009; 2: 11–19. Shiba T, Takahashi M, Hori Y, Saishin Y, Sato Y, Maeno T. Relationship between sleep-disordered breathing and iris and/or angle neovascularization in proliferative diabetic retinopathy cases. Am J Ophthalmol 2011; 151: 604–609. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. When does it make sense to perform a meta-analysis? In: Introduction to metaanalysis. Chichester: Wiley, 2009. Nishimura A, Kasai T, Sakamoto K, Ishiguro K, Nagasawa K, Okubo M et al. Relationship between sleep disordered breathing and diabetic retinopathy: Analysis in 136 diabetic cases. Diabetes 2013; 62: A159–A160. Bottini P, Redolfi S, Dottorini ML, Tantucci C. Autonomic neuropathy increases the risk of obstructive sleep apnea in obese diabetics. Respiration 2008; 75: 265–271. Wong A, Merritt S, Butt AN, Williams A, Swaminathan R. Effect of hypoxia on circulating levels of retina-specific messenger

168

30

31

32

33

34

35

RNA in type 2 diabetes mellitus. Ann N Y Acad Sci 2008; 1137: 243–252. Arden GB. The absence of diabetic retinopathy in patients with retinitis pigmentosa: implications for pathophysiology and possible treatment. Br J Ophthal 2001; 85: 366–370. He Q, Yang QC, Zhou Q, Zhu H, Niu WY, Feng J et al. Effects of varying degrees of intermittent hypoxia on proinflammatory cytokines and adipokines in rats and 3T3-L1 adipocytes. PloS One 2014; 9: e86326. Nacher M, Farre R, Montserrat JM, Torres M, Navajas D, Bulbena O et al. Biological consequences of oxygen desaturation and respiratory effort in an acute animal model of obstructive sleep apnea (OSA). Sleep Med 2009; 10: 892–897. Kaczmarek E, Bakker JP, Clarke DN, Csizmadia E, Kocher O, Veves A et al. Molecular biomarkers of vascular dysfunction in obstructive sleep apnea. PloS One 2013; 8: e70559. Dematteis M, Pepin JL, Jeanmart M, Deschaux C, Labarre-Vila A, Levy P. Charcot-Marie-Tooth disease and sleep apnoea syndrome: a family study. Lancet 2001; 357: 267–272. Mason RH, Kiire CA, Groves DC, Lipinski HJ, Jaycock A, Winter BC et al. Visual improvement following continuous positive airway pressure therapy in diabetic subjects with clinically significant macular oedema and obstructive sleep apnoea: proof of principle study. Respiration 2012; 84: 275–282.

Supporting Information Additional Supporting Information may be found in the online version of this article: Table S1 Eligibility criteria for systematic review. Table S2 Search terms for MEDLINE, EMBASE and Cochrane database. Table S3 Quality assessment form. Table S4 Criteria used for the diagnosis of obstructive sleep apnoea and diabetic retinopathy assessment as reported by the included studies. Table S5 Summary of the results categorised according to diabetic retinopathy outcomes. Figure S1 Forest plot on studies which reported unadjusted odds ratios and 95% confidence intervals on the effect of OSA on overall DR. Figure S2 Funnel plot on studies which reported unadjusted odds ratios. Figure S3 Forest plot of the effects of obstructive sleep apnoea on advanced-DR using results from studies, which reported adjusted odds ratios and 95% confidence intervals only.

ª 2015 The Authors. Diabetic Medicine ª 2015 Diabetes UK

Effect of obstructive sleep apnoea on diabetic retinopathy and maculopathy: a systematic review and meta-analysis.

To summarize the association between obstructive sleep apnoea and diabetic retinopathy and diabetic maculopathy, and to examine the effects of oxygen ...
245KB Sizes 0 Downloads 12 Views