Myopia and diabetic retinopathy: A systematic review and meta-analysis

DIAB-6496; No. of Pages 9 diabetes research and clinical practice xxx (2015) xxx–xxx

Contents available at ScienceDirect

Diabetes Research and Clinical Practice journ al h ome pa ge : www .elsevier.co m/lo cate/diabres

Review

Myopia and diabetic retinopathy: A systematic review and meta-analysis Xiang Wang a,b, Luosheng Tang c,d, Ling Gao c,d, Yujia Yang c,d, Dan Cao c,d, Yunping Li c,d,* a

Cancer Research Institute, Central South University, Changsha 410078, Hunan, China Department of Surgery, Second Xiangya Hospital, Central South University, Changsha 410011, Hunan, China c Department of Ophthalmology, Second Xiangya Hospital, Central South University, Changsha 410011, Hunan, China d Eye Research Institute, Second Xiangya Hospital, Central South University, Changsha, Hunan 410011, China b

article info

abstract

Article history:

Aims: Myopia may have protective effects against diabetic retinopathy (DR). However, the

Received 8 June 2015

data from epidemiologic studies are inconsistent. We aimed to examine the association

Received in revised form

between myopia and DR by conducting a meta-analysis.

5 October 2015

Methods: We identified studies by searching the PubMed and EMBASE databases. Study-

Accepted 11 October 2015

specific odds ratios (ORs) were pooled using a fixed or random effects model. Myopic eyes

Available online xxx

were defined as having a spherical equivalent (SE) <

0.5 diopters (D). Myopic SE, each

diopter decrease in SE toward myopia, and each millimeter increase in axial length (AL) were Keywords:

used as independent surrogate variables for myopia.

Myopia

Results: Data from 6 population-based and 3 clinic-based studies were included in the

Diabetic retinopathy

analyses. Myopic SE (compared with emmetropic eyes) and each millimeter increase in

Axial length

AL were associated with a decreased risk for DR (pooled odds ratio [OR], 0.80 and 0.79, respectively;

Odds ratio

95% confidence interval [CI], 0.67–0.95 and 0.73–0.86, respectively; P = 0.011 and 0.000, respective-

Association

ly). Each millimeter increase in AL was also associated with a decreased risk for vision-threatening diabetic retinopathy (VTDR) (pooled OR, 0.70; 95% CI, 0.60–0.82; P = 0.000). No significant association between each diopter decrease in SE and DR was observed. Conclusions: Our meta-analysis suggests that individuals with myopia exhibit a decreased risk of developing DR or VTDR. An increased AL plays a critical role in this protective effect. # 2015 Elsevier Ireland Ltd. All rights reserved.

Contents 1. 2.

Introduction . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Search strategy . . . . . . . . . . . . . 2.2. Inclusion and exclusion criteria

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

000 000 000 000

* Corresponding author at: Department of Ophthalmology, Second Xiangya Hospital, Central South University, Changsha 410011, Hunan, China. Tel.: +86 731 85295138; fax: +86 731 82567957. E-mail address: [email protected] (Y. Li). http://dx.doi.org/10.1016/j.diabres.2015.10.020 0168-8227/# 2015 Elsevier Ireland Ltd. All rights reserved.

Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9

2

diabetes research and clinical practice xxx (2015) xxx–xxx

3. 4.

1.

2.3. Data extraction and quality assessment . 2.4. DR assessment and definition . . . . . . . . . 2.5. Statistical analysis . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

Introduction

Diabetic retinopathy (DR), a common and specific microvascular complication of diabetes, remains the leading cause of preventable blindness in working-aged people [1]. Previous studies have demonstrated that myopia protects against DR [2,3]. However, population-based studies have revealed inconsistent and conflicting results. For example, the Singapore Malay Eye Study reported that eyes with more severe myopia were less likely to exhibit DR [2]. In contrast, the Beijing Eye Study reported no association between myopia and DR [4]. Recently, an observational review from Man et al. indicated that axial elongation, not myopia, may be primarily responsible for the protective relationship between myopia and DR [5]. A better understanding of the relationship between myopia and DR may provide insight into the pathophysiology of DR. However, a meta-analysis of the results of all available studies that have evaluated the association of myopia with DR has not been performed to date. The aim of this systematic review and meta-analysis is to examine the association between myopia and DR based on data from available population-based and clinic-based studies. Where data were available, we evaluated this relationship using the myopic spherical equivalent (SE), each diopter decrease in SE toward myopia, and each millimeter increase in AL as independent surrogate variables for myopia.

2.

Methods

2.1.

Search strategy

First, we performed a systematic search of PubMed and EMBASE to identify all relevant population-based and clinicbased studies published up to March 2015 using the following search items: (‘‘myopia’’ [MeSH Terms] OR ‘‘myopia’’ [All Fields]) OR (myopic [All Fields]) OR (‘‘refractive errors’’ [MeSH Terms]) OR (‘‘refractive’’ [All Fields] AND ‘‘errors’’ [All Fields]) OR (‘‘refractive errors’’ [All Fields]) OR (‘‘refractive’’ [All Fields] AND ‘‘error’’ [All Fields]) OR (‘‘refractive error’’ [All Fields]) OR (‘‘axial length’’ AND (DR [All Fields] OR proliferative DR (PDR) [All Fields]) OR (‘‘diabetes’’ [MeSH Terms] and ‘‘retinopathy’’ [MeSH Terms] OR ‘‘diabetic macular edema’’ [All Fields]). English-language articles were retrieved, and duplicate citations were excluded after a review of the titles and abstracts. The full texts of the remaining articles were reviewed to ensure that the studies met the inclusion and exclusion criteria. In addition, the reference lists of all of the identified studies were examined. Two authors (WX and YYJ) independently conducted the search; any disagreements were

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

000 000 000 000 000 000 000

resolved by adjudication with two additional reviewers (GL and TLS).

2.2.

Inclusion and exclusion criteria

Studies were included if they met the following criteria: (i) explored the associations among myopia, AL and DR; (ii) used DR as an outcome measure, which was assessed based on fundus photographs according to standardized protocols, such as the Early Treatment Diabetic Retinopathy Study (ETDRS) or the Airlie House classification system; and (iii) reported a measure of the association either as an odds ratio (OR) or a hazard ratio (HR) with a 95% confidence interval (CI) or allowed for the calculation of such metrics from the raw data presented in the article. We excluded (i) studies published in non-English languages and (ii) studies without a clear threshold definition of myopia or lacking fundus photography results according to standardized protocols. When multiple publications from the same study population were available, we identified any duplicate analyses and included only the most recent publication.

2.3.

Data extraction and quality assessment

Using a standardized data extraction sheet, the following information (if available) was extracted from the studies and recorded: (i) last name of the first author, (ii) year of publication, (iii) study name, (iv) study design, (v) race/ ethnicity of the study population, (vi) number of subjects included in the analysis, (vii) age range of the study participants, (viii) case definition of DR and myopia, (ix) effect estimate(s), and (x) the confounding factors for which adjustment was performed. We assessed the study quality using the tool described by Sanderson and colleagues [6]. The variables examined included the methods for selecting the study participants, the methods for measuring exposure (myopia) and outcome (DR), design-specific sources of bias (excluding confounding variables), the methods that were used to control for confounding variables, the statistical methods (excluding the control of confounding variables), and potential conflicts of interest.

2.4.

DR assessment and definition

In all of the studies, DR was graded using fundus photographs according to the modified ETDRS grading scale or the Airlie House classification system. DR severity was categorized as non-proliferative DR (NPDR; levels 20–53) or proliferative DR (PDR) (level 60). Diabetic macular edema (DME) was categorized as absent or present. The primary outcomes for this study were based on the severity in the worse eye or in the

Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9 diabetes research and clinical practice xxx (2015) xxx–xxx

single eye that was photographed. Any DR was defined as the presence of NPDR, PDR, DME, or any combination thereof; vision-threatening diabetic retinopathy (VTDR) was defined as the presence of severe NPDR, PDR, and/or DME.

2.5.

Statistical analysis

We performed the meta-analysis using STATA version 13.0 (StataCorp LP, College Station, TX, USA). The fully adjusted study-specific ORs were combined to estimate the pooled OR and 95% CI. Statistical heterogeneity among studies was evaluated using the I2 statistic [7]. I2 values of 0–24%, 25–49%, 50–74%, and greater than 75% denote no, low, moderate, and high heterogeneity, respectively [8]. Where moderate or high heterogeneity was observed, a random effects meta-analysis was performed; otherwise, a fixed effects model was used. Heterogeneity caused by the study design was addressed by performing a meta-analysis separately for all studies combined and for population-based studies alone. For the Wisconsin Epidemiologic Study of Diabetic Retinopathy [3], which reported the results for younger-onset and older-onset cohorts separately, we combined the 2 ORs and subsequently included the pooled OR in the meta-analysis. For the study from the Diabetes Management Project in Australia [9], which reported the associations between AL and mild, moderate, and severe DR and DME, we combined the 2 ORs for severe DR and DME and subsequently included this pooled OR as the OR for VTDR in the meta-analysis. For myopia, myopic SE, each diopter decrease in SE toward myopia and each millimeter increase in AL were analyzed as independent variables in the model. For studies that exclusively reported stratified ORs or HRs, we pooled the ORs or HRs to obtain an overall estimate for myopia. If no refractive error thresholds were reported in the publication, we contacted the principal investigator to obtain the complete dataset and calculated the OR of the association of myopia with DR. SE was defined as the sphere plus half the negative cylinder. Myopic SE was defined as SE < 0.5 D compared with emmetropic eyes. We evaluated publication bias using the Egger regression asymmetry test [10] and Begg’s test [11]. A 2-sided P-value less than 0.05 was regarded as significant for all analyses.

3.

Results

Our literature search identified 2631 articles. After a review of the titles and abstracts, 17 articles were found to be potentially relevant for inclusion criteria and thus retrieved for full-text review. After a thorough review, 9 studies were excluded from the systematic review as they did not strictly meet the inclusion criteria. One additional article was identified from the reference lists. A total of 9 studies were included in the systematic review, including 6 population-based studies (4 cross-sectional studies and 2 cohort studies) and 3 clinicbased cross-sectional studies (Fig. 1). The extracted data are summarized in Table 1. The studies were published between 1985 and 2015. Among the 7 cross-sectional studies, 5 were conducted in Asia, 1 was conducted in Australia and 1 was conducted in the United States. Among the 2 cohort studies, 1

3

was conducted in the United States, and 1 was conducted in China. The 7 cross-sectional studies comprised a total of 24,751 individuals aged >18 years, and the 2 cohort studies comprised a total of 4480 individuals. The response rate in each population-based cross-section study was greater than 70%. The sample sizes of the cohort studies ranged from 1878 to 2602 individuals, and all follow-up rates were greater than 60%. All of the studies graded DR according to standardized protocols such as the ETDRS or the Airlie House classification system. Refractive error was assessed by subjective refraction in most of the studies. The definition of myopia varied across studies. Myopia was defined as SE < 0.5 diopter (D) in most population-based studies except for the Korea National Health and Nutrition Examination Survey [12] and the Wisconsin Epidemiologic Study of Diabetic Retinopathy [3], which used SE < 1.0 D or < 2.0 D as the threshold, respectively. Table 2 summarizes the estimated ORs (95% CIs) obtained from each study for the association of myopia with DR and VTDR. The covariates for which adjustments were included in the individual multivariate analyses are also included. The association of myopic SE with DR was reported in 4 population-based cross-sectional studies, 1 cohort study and 2 clinic-based cross-sectional studies (Table 2). When combining the estimated effects based on these studies, myopic SE was significantly associated with a decreased risk for DR (pooled OR, 0.80; 95% CI, 0.67–0.95; P = 0.011; I2 = 36.7%) (Fig. 2). Excluding the clinic-based study, a borderline significant association was observed between myopic SE and DR in a meta-analysis of the 5 population-based studies (pooled OR, 0.83; 95% CI, 0.69–0.99; P = 0.043; I2 = 53.0%). The association of each diopter decrease in SE toward myopia with DR was reported in 2 population-based crosssectional studies and 1 cohort study (Table 2). When combining the estimated effects based on these studies, no significant associations were observed between each diopter decrease in SE toward myopia and DR (pooled OR, 1.01; 95% CI, 0.97–1.06; P = 0.597; I2 = 91.50%). Data on the association of each millimeter increase in AL and DR were available in 3 population-based cross-sectional studies, 1 cohort study and 2 clinic-based studies (Table 2). When combining the estimated effects based on these studies, each millimeter increase in AL was significantly associated with a decreased risk of DR (pooled OR, 0.79; 95% CI, 0.73–0.86; P = 0.000). Significant heterogeneity was observed among these 6 studies (I2 = 57.6%, P = 0.038). Sensitivity analysis showed that the Beijing Eye Study [13] influenced the pooled OR. After excluding the Beijing Eye Study, the pooled OR was 0.81 (95% CI, 0.74–0.89), with no evidence of heterogeneity (I2 = 17.4%; P = 0.304) (Fig. 3). Considering that 2 clinic-based studies were included in the above analysis, we also performed a meta-analysis using the 4 population-based studies alone (3 cross-sectional studies and 1 cohort study). The pooled OR was 0.79 (95% CI, 0.72–0.87; P = 0.000), and significant heterogeneity was observed (I2 = 70.5%; P = 0.017). Sensitivity analysis revealed that the Beijing Eye Study also influenced this pooled OR. After excluding the Beijing Eye Study, the pooled OR was 0.81 (95% CI, 0.74–0.90; P = 0.000), with low heterogeneity (I2 = 39.7%; P = 0.190). Data regarding the association of each millimeter increase in AL and VTDR were available for 5 cross-sectional studies

Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9

4

diabetes research and clinical practice xxx (2015) xxx–xxx

Fig. 1 – Flow diagram showing the selection process for the inclusion of studies in this meta-analysis. CI = confidence interval.

(3 population-based and 2 clinic-based studies) (Table 2). When combining the estimated effects based on these studies, each millimeter increase in AL was significantly associated with a decreased risk of VTDR (pooled OR, 0.70; 95% CI, 0.60–0.82; P = 0.000), with no evidence of heterogeneity (I2 = 0.0%; P = 0.826) (Fig. 4). For the 3 included population-based cross-sectional studies, the pooled OR was 0.72 (95% CI, 0.55–0.95; P = 0.02), with no evidence of heterogeneity (I2 = 0.0%; P = 0.591). In all of the analyses, no evidence of publication bias, as indicated by the non-significant results of the Egger tests (all P > 0.05) and Begg’s tests (all P > 0.05), was observed.

4.

Discussion

In this systematic review and meta-analysis, we investigated the relationship between myopia and DR. We used myopic SE, each diopter decrease in SE toward myopia, and each millimeter increase in AL as independent surrogate variables of myopia. Using myopic SE, data from 6 cross-sectional studies and 1 cohort study revealed that myopia was

significantly associated with a decreased risk of DR (Fig. 2). When analyzed as a continuous variable for each diopter decrease in SE toward myopia, data from 2 population-based cross-sectional studies and 1 cohort study indicated that myopia was not associated with DR. However, when each millimeter increase in AL was analyzed as a continuous variable for myopia, data from 5 cross-sectional studies and 1 cohort study demonstrated that myopia was significantly associated with a decreased risk of DR (Fig. 3). Further analysis indicated that each millimeter increase in AL was significantly associated with a decreased risk of VTDR (Fig. 4). Only a few studies (3) were included in the analysis of the association between each diopter decrease in SE toward myopia and DR; moreover, the heterogeneity among studies was significant (I2 = 91.50%). For these reasons, we do not consider the results of this particular analysis to be reliable. Therefore, our results indicated that both myopic SE and each millimeter increase in AL were significantly associated with a decreased risk of DR. These further indicated that myopia is a protective factor against DR. The results are consistent with the hypothesis proposed by other investigators [14,15].

Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9

5

diabetes research and clinical practice xxx (2015) xxx–xxx

Table 1 – Characteristics of the included studies. Author (year)

Study design

Race/ethnicity

Sample size

Age (years)

Response or follow-up rate (%)a

Myopia threshold (SE, in diopters)

DR assessment

Population-based studies Lim et al. (2010) [2]

Cross-sectional study

Singapore

3280

40–80

78.7%

< 0.5

E

SiMES Pan et al. (2013) [13]

Cross-sectional study

Indians living in Singapore

3400

40–84

75.6%

< 0.5

A

SINDI Ganesan et al. (2012) [14]

Cross-sectional study

India

1058

40

85.8%

< 0.5

E

SN DREAMS 1 Jee et al. (2013) [12]

Cross-sectional study

South Korea

16,109

>40

77.8–82.8%

< 1.0

E

KNHNS Moss et al. (1994) [3]

Cohort study

USA

2990

82.3% and 77%

 2.0

A

WESDR Xu et al. (2014) [15]

Cohort study

China

2602

40

60.7% of the original cohort; 66.4% of the survivors

< 0.5

E

Cross-sectional study Cross-sectional study Cross-sectional study

Australia Taiwan USA

609 166 192

18 >18 <65

– – –

< 0.5 –  2.0

E E A

BES Clinic-based studies Man et al. (2012) [5] Yang et al. (2012) [16] Rand et al. (1985) [17]

Abbreviations: SE = spherical equivalent; DR = diabetic retinopathy; SiMES = Singapore Malay Eye Study; SINDI = Singapore Indian Eye Study; SN DREAMS 1 = Sankara Nethralaya-Diabetic Retinopathy Epidemiology and Molecular Genetic Study; KNHNS = Korea National Health and Nutrition Examination Survey; WESDR = Wisconsin Epidemiologic Study of Diabetic Retinopathy; BES = Beijing Eye Study; A = Airlie House classification system; E = Early Treatment Diabetic Retinopathy Scale (ETDRS). a The response rate is presented for the cross-sectional studies, and the follow-up rate is presented for the cohort studies.

Significant heterogeneity among studies was observed in the analysis of the association between each millimeter increase in AL and DR. The observed heterogeneity among the included studies could be due to the inclusion of the

Beijing Eye Study. When this study was removed from the analysis, the heterogeneity was reduced or eliminated. The Beijing Eye Study is the only cohort study among the included studies, and it found that the incidence of DR was significantly

Fig. 2 – Forest plot of the estimated effects of the association between myopia (myopic SE) and DR. CI = confidence interval; OR = odds ratio; SE = spherical equivalent; DR = diabetic retinopathy. Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9

6

diabetes research and clinical practice xxx (2015) xxx–xxx

Table 2 – ORs and 95% CIs of the association between myopia and diabetic retinopathy in the included studies. Author (year)

DR

Any myopia

VTDR each millimeter increase in AL

Variables adjusted in the analyses

Age, gender, education level, height, cataract, HbA1c level, hypertension, and total cholesterol level Age, gender, education level, body mass index, HbA1c, hypertension, and total cholesterol level in the generalized estimating equation models Age, gender, duration of diabetes, glycosylated hemoglobin level, education level, systolic and diastolic blood pressure, body mass index and serum lipid level

Each diopter decrease in SE

Each millimeter increase in AL

Population-based cross-sectional studies 0.64 (0.47, 0.88) Lim et al. (2010) [2]

0.90 (0.84, 0.96)

0.86 (0.75, 0.99)

0.63 (0.40, 0.99)

Pan et al. (2013) [13]

0.68 (0.46, 0.98)

1.14 (1.05, 1.23)

0.73 (0.63, 0.86)

0.73 (0.49, 1.09)

Ganesan et al. (2012) [14]

1.13 (0.70, 1.84)

–

0.93 (0.70, 1.24)

0.96 (0.49, 1.86)

Jee et al. (2013) [12]

1.15 (0.70, 1.88)

–

–

–

1.12 (1.00, 1.26)

– 0.48 (0.33, 0.71)

– –

– Age, gender, smoking, body weight, duration of diabetes mellitus, glycosylated hemoglobin and serum lipid and protein levels

0.76 (0.63, 0.90)

0.71 (0.57, 0.90)

Age, gender, education level, glycosylated hemoglobin level, systolic blood pressure, and duration of diabetes mellitus Age, gender, hypertension, and duration of diabetes Sex, smoking status, alcohol consumption, and blood sugar level

Population-based cohort studies 1.15 (0.69, 1.92) Moss et al. (1994) [3] Xu et al. (2014) [15]

Clinic-based cross-sectional studies 0.66 (0.40, 1.09) Man et al. (2012) [5]

Yang et al. (2012) [16] Rand et al. (1985) [17]

–

–

0.93 (0.71,1.21)

0.61 (0.39,0.94)

0.53 (0.12–2.44)

–

–

–

Abbreviations: DR = diabetic retinopathy; VTDR = vision-threatening diabetic retinopathy; SE = spherical equivalent; AL = axial length.

Fig. 3 – Forest plot of the estimated effects of the association between myopia (each millimeter increase in AL) and DR. CI = confidence interval; OR = odds ratio; AL = axial length; DR = diabetic retinopathy. Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9 diabetes research and clinical practice xxx (2015) xxx–xxx

7

Fig. 4 – Forest plot of estimated effects of the association between myopia (each millimeter increase in AL) and VTDR. CI = confidence interval; OR = odds ratio; AL = axial length; VTDR = vision-threatening diabetic retinopathy.

associated with a shorter AL (P < 0.001); however, the OR of this result was lower than the ORs of other studies. Cohort studies exhibit many advantages in meta-analyses; however, they also have limitations. Specifically, cohort studies potentially suffer from survival bias, which can mask an actual association. Compared with other 10-year follow-up studies in ophthalmic epidemiology, the response rate in the Beijing Eye Study (66.4% of survivors) was lower than the response rate in both the Blue Mountains Eye Study (75.6% of survivors) and the Beaver Dam Eye Study (82.9% of survivors) because of the participants’ higher mobility and loss to follow-up in the Beijing Eye Study. Another potential limitation is that only two non-stereoscopic fundus photographs were used to detect diabetic changes in the retinas in the Beijing Eye Study, whereas the ETDRS criteria used 7-field stereo images. All of these differences may have led to an inaccurate effect estimate and caused the study to contribute to the observed heterogeneity. Compared with population-based studies, clinic-based studies are more likely to contribute to selection bias. To avoid possible selection bias, we (i) investigated the associations between DR and both myopic SE and each millimeter increase in AL using the pooled population- and clinic-based studies and (ii) investigated the same association using only population-based studies. The results of both analyses indicated that myopia was significantly associated with a decreased risk of DR and VTDR and that axial elongation is the primary contributor to the protective effect of myopia against DR [9]. Myopic eyes are elongated relative to emmetropic eyes [16]. Axial elongation corresponds to the progression of myopia [17]. We assume that myopia may serve as a surrogate marker of a long AL and thereby influence the risk for DR. In the past few decades, significant efforts have been made to explore the mechanism underlying the association between DR and myopia. However, the biological mechanism underlying the observed association has not been elucidated. We offer

several possible explanations. First, the decrease in blood flow with increasing AL plays a major role in the protective effect against DR [18,19]. Axial myopia is characterized by retinal arterioles that are longer than those in ‘‘normal’’ or emmetropic eyes. Myopic refractive error and longer AL are associated with narrower retinal arterioles and venules, less tortuous arterioles, and increased branching coefficients in both arterioles and venules [19]. In retinal photographs, these myopic vessels are narrowed by 1.3% for each diopter of increasing myopia; [20] i.e., myopic vessels appear narrower in proportion to the AL of the eye, and retinal blood flow decreases. Retinal blood flow is thought to increase with increasing DR severity [21]. According to the laws of Starling and Laplace [18], increased retinal blood flow can produce retinal capillary pressure and subsequently result in capillary wall dilatation, leakage and rupture, all of which are hallmarks of DR. However, Man et al. recently demonstrated that retinal capillary flow (RCF) is not associated with AL or the risk or severity of DR [22]. These results suggest that diminished RCF may not be a major factor underlying the association between axial elongation and a reduced risk for DR. As RCF constitutes only a portion of the retinal microcirculation, further investigation is necessary. Second, a long AL against DR may be attributed to decreased retinal function and O2 consumption in the outer retina [23]. Many studies have suggested a correlation between retinal neuron dysfunction and neurodegeneration with axial elongation [24,25]. Degenerative changes in the outer retina and decreased metabolic demand may help to mitigate the effects of hypoxia in diabetes, leading to a potential decrease in the production of inflammatory or proangiogenic cytokines, e.g., vascular endothelial growth factor (VEGF) [26,27]. As a result of an elongated eye, increased ocular volume may also lead to decreased concentrations of VEGF due to dilution. We therefore suspect that longer eyes with decreased VEGF levels may delay or prevent the occurrence of DR. This hypothesis should be evaluated in future studies.

Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9

8

diabetes research and clinical practice xxx (2015) xxx–xxx

Third, posterior vitreous detachment (PVD) has been noted in many myopic eyes [28], especially in pathological myopia [29,30]. Several studies have demonstrated that complete PVD is associated with a reduced risk for progression to neovascularization and PDR [31,32]. The protective effect of PVD has been attributed to the removal of the vitreous scaffold for neovascular proliferation and improved oxygen diffusion across the liquefied vitreous. These results may partially explain the mechanism by which myopia protects against DR. There are several strengths of this meta-analysis. Myopia may be related to axial dimensions or to other refractive components of myopia. As data regarding myopia differed among the included studies, we analyzed the association between DR and each of myopic SE, each diopter decrease in SE, and each millimeter increase in AL. This effort validates the impact of myopia on DR. In addition, we exclusively included studies that used a standardized fundus photographic grading system for DR. However, there are limitations to our meta-analysis. First, according to the inclusion and exclusion criteria, few population-based studies were included in our meta-analysis. Therefore, we included not only populationbased design studies but also several clinic-based studies. We then performed the meta-analysis using data pooled from the population-based and clinic-based studies. We also performed an analysis using only the population-based studies. These procedures minimized selection bias and significantly increased the statistical power of the analysis. Second, only English-language articles were included. In addition, the number of included population-based studies was small, limiting the amount of data included in our meta-analysis. Third, potential biases resulting from study differences in methodologies and strategies for adjusting for confounders may have affected the results of this meta-analysis. The observed cross-sectional association may have been confounded by differences in age ranges or other unadjusted factors such as income or education level; these factors are associated with myopia [33] and/or the progression of DR [34,35]. Fourth, most of the included studies in our metaanalysis were cross-sectional, and the number of available cohort studies was small. Selection bias may have occurred in our meta-analysis. Fifth, the different studies used different thresholds for myopia (Table 1), potentially affecting the estimates of the prevalence of myopia. These differences may also have contributed to the observed heterogeneity across studies. Finally, publication bias may have distorted our findings because studies that report significant results are more likely to be published than are studies that report nonsignificant results. Although the Egger test and Begg’s test indicated little evidence of publication bias in this metaanalysis, these estimations may not be sufficiently accurate given that the number of included studies, especially cohort studies, was low. In conclusion, this systematic review and meta-analysis suggests that myopia is significantly associated with a decreased risk for DR. When myopia was analyzed as a continuous variable for each millimeter increase in AL, myopia was found to protect against DR and VTDR. These findings indicate that axial elongation is responsible for the protective effect of myopia against DR. Insufficient evidence is available from population-based studies, especially cohort

studies, in support of these findings from cross-sectional studies. Therefore, more well-designed studies assessing longitudinal associations of myopia, especially in the axial dimensions, with DR and VTDR are required. Similarly, further studies are necessary to elucidate the exact underlying mechanisms of the association between AL and DR. Thus, a potential treatment for DR may be envisioned.

Conflict of interest No conflicting relationship exists for any author.

Financial disclosures This research was supported by grants 201302015 from the National Health and Family Planning Commission of the People’s Republic of China. The funding organizations played no role in the design, conduct, analysis or publication of this research.

Acknowledgment The authors would like to thank Prof. Zhengqiu Sun for his methodological advice and help in designing the metaanalysis.

references

[1] Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet 2010;376(9735):124–36. http://dx.doi.org/10.1016/ s0140-6736(09)62124-3. [2] Lim LS, Lamoureux E, Saw SM, Tay WT, Mitchell P, Wong TY. Are myopic eyes less likely to have diabetic retinopathy? Ophthalmology 2010;117(3):524–30. http:// dx.doi.org/10.1016/j.ophtha.2009.07.044. [3] Moss SE, Klein R, Klein BE. Ocular factors in the incidence and progression of diabetic retinopathy. Ophthalmology 1994;101(1):77–83. [4] Xie XW, Xu L, Wang YX, Jonas JB. Prevalence and associated factors of diabetic retinopathy The Beijing Eye Study 2006. Graefes Arch Clin Exp Ophthalmol 2008;246(11):1519–26. http://dx.doi.org/10.1007/s00417-008-0884-6. [5] Man RE, Sasongko MB, Wang JJ, Lamoureux EL. Association between myopia and diabetic retinopathy: a review of observational findings and potential mechanisms. Clin Exp Ophthalmol 2013;41(3):293–301. http://dx.doi.org/10.1111/ j.1442-9071.2012.02872.x. [6] Sanderson S, Tatt ID, Higgins JP. Tools for assessing quality and susceptibility to bias in observational studies in epidemiology: a systematic review and annotated bibliography. Int J Epidemiol 2007;36(3):666–76. http:// dx.doi.org/10.1093/ije/dym018. [7] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327(7414):557–60. http://dx.doi.org/10.1136/bmj.327.7414.557. [8] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21(11):1539–58. http:// dx.doi.org/10.1002/sim.1186. [9] Man RE, Sasongko MB, Sanmugasundram S, Nicolaou T, Jing X, Wang JJ, et al. Longer axial length is protective of diabetic

Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

DIAB-6496; No. of Pages 9 diabetes research and clinical practice xxx (2015) xxx–xxx

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

retinopathy and macular edema. Ophthalmology 2012;119(9):1754–9. http://dx.doi.org/10.1016/ j.ophtha.2012.03.021. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315(7109):629–34. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50(4):1088–101. Jee D, Lee WK, Kang S. Prevalence and risk factors for diabetic retinopathy: the Korea National Health and Nutrition Examination Survey 2008–2011. Invest Ophthalmol Vis Sci 2013;54(10):6827–33. http://dx.doi.org/ 10.1167/iovs.13-12654. Xu J, Xu L, Wang YX, You QS, Jonas JB, Wei WB. Ten-year cumulative incidence of diabetic retinopathy. The Beijing Eye Study 2001/2011. PLoS ONE 2014;9(10):e111320. http:// dx.doi.org/10.1371/journal.pone.0111320. Stefansson E. Ocular oxygenation and the treatment of diabetic retinopathy. Surv Ophthalmol 2006;51(4):364–80. http://dx.doi.org/10.1016/j.survophthal.2006.04.005. Pierro L, Brancato R, Robino X, Lattanzio R, Jansen A, Calori G. Axial length in patients with diabetes. Retina 1999;19(5):401–4. Atchison DA, Jones CE, Schmid KL, Pritchard N, Pope JM, Strugnell WE, et al. Eye shape in emmetropia and myopia. Invest Ophthalmol Vis Sci 2004;45(10):3380–6. http:// dx.doi.org/10.1167/iovs.04-0292. Lin LL, Shih YF, Hsiao CK, Chen CJ, Lee LA, Hung PT. Epidemiologic study of the prevalence and severity of myopia among schoolchildren in Taiwan in 2000. J Formos Med Assoc 2001;100(10):684–91. Quigley M, Cohen S. A new pressure attenuation index to evaluate retinal circulation. A link to protective factors in diabetic retinopathy. Arch Ophthalmol 1999;117(1):84–9. Lim LS, Cheung CY, Lin X, Mitchell P, Wong TY, Mei-Saw S. Influence of refractive error and axial length on retinal vessel geometric characteristics. Invest Ophthalmol Vis Sci 2011;52(2):669–78. http://dx.doi.org/10.1167/iovs.10-6184. Wong TY, Knudtson MD, Klein R, Klein BE, Meuer SM, Hubbard LD. Computer-assisted measurement of retinal vessel diameters in the Beaver Dam Eye Study: methodology, correlation between eyes, and effect of refractive errors. Ophthalmology 2004;111(6):1183–90. http://dx.doi.org/10.1016/j.ophtha.2003.09.039. Dimitrova G, Kato S, Yamashita H, Tamaki Y, Nagahara M, Fukushima H, et al. Relation between retrobulbar circulation and progression of diabetic retinopathy. Br J Ophthalmol 2003;87(5):622–5. Man RE, Sasongko MB, Xie J, Best WJ, Noonan JE, Lo TC, et al. Decreased retinal capillary flow is not a mediator of the protective myopia-diabetic retinopathy relationship. Invest Ophthalmol Vis Sci 2014;55(10):6901–7. http:// dx.doi.org/10.1167/iovs.14-15137.

[23] Man RE, Lamoureux EL, Taouk Y, Xie J, Sasongko MB, Best WJ, et al. Axial length, retinal function, and oxygen consumption: a potential mechanism for a lower risk of diabetic retinopathy in longer eyes. Invest Ophthalmol Vis Sci 2013;54(12):7691–8. http://dx.doi.org/10.1167/iovs.1312412. [24] Luu CD, Lau AM, Lee SY. Multifocal electroretinogram in adults and children with myopia. Arch Ophthalmol 2006;124(3):328–34. http://dx.doi.org/10.1001/ archopht.124.3.328. [25] Wolsley CJ, Saunders KJ, Silvestri G, Anderson RS. Investigation of changes in the myopic retina using multifocal electroretinograms, optical coherence tomography and peripheral resolution acuity. Vision Res 2008;48(14):1554–61. http://dx.doi.org/10.1016/ j.visres.2008.04.013. [26] Jonas JB, Tao Y, Neumaier M, Findeisen P. VEGF and refractive error. Ophthalmology 2010;117(11):2234. http:// dx.doi.org/10.1016/j.ophtha.2009.12.006. e2231. [27] Sawada O, Kawamura H, Kakinoki M, Sawada T, Ohji M. Vascular endothelial growth factor in the aqueous humour in eyes with myopic choroidal neovascularization. Acta Ophthalmol 2011;89(5):459–62. http://dx.doi.org/10.1111/ j.1755-3768.2009.01717.x. [28] Saw SM, Gazzard G, Shih-Yen EC, Chua WH. Myopia and associated pathological complications. Ophthalmic Physiol Opt 2005;25(5):381–91. http://dx.doi.org/10.1111/j.14751313.2005.00298.x. [29] Grossniklaus HE, Green WR. Pathologic findings in pathologic myopia. Retina 1992;12(2):127–33. [30] Ganesan S, Raman R, Reddy S, Krishnan T, Kulothungan V, Sharma T. Prevalence of myopia and its association with diabetic retinopathy in subjects with type II diabetes mellitus: a population-based study. Oman J Ophthalmol 2012;5(2):91–6. http://dx.doi.org/10.4103/ 0974-620x.99371. [31] Tagawa H, McMeel JW, Trempe CL. Role of the vitreous in diabetic retinopathy. II. Active and inactive vitreous changes. Ophthalmology 1986;93(9):1188–92. [32] Akiba J, Arzabe CW, Trempe CL. Posterior vitreous detachment and neovascularization in diabetic retinopathy. Ophthalmology 1990;97(7):889–91. [33] Vitale S, Sperduto RD, Ferris 3rd FL. Increased prevalence of myopia in the United States between 1971–1972 and 1999– 2004. Arch Ophthalmol 2009;127(12):1632–9. http:// dx.doi.org/10.1001/archophthalmol.2009.303. [34] Klein R, Klein BE, Jensen SC, Moss SE. The relation of socioeconomic factors to the incidence of proliferative diabetic retinopathy and loss of vision. Ophthalmology 1994;101(1):68–76. [35] West SK, Munoz B, Klein R, Broman AT, Sanchez R, Rodriguez J, et al. Risk factors for Type II diabetes and diabetic retinopathy in a Mexican-American population: Proyecto VER. Am J Ophthalmol 2002;134(3):390–8.

Please cite this article in press as: Wang X, et al. Myopia and diabetic retinopathy: A systematic review and meta-analysis. Diabetes Res Clin Pract (2015), http://dx.doi.org/10.1016/j.diabres.2015.10.020

9