- Email: [email protected]
Type 2 Diabetes Genetic Variants and Risk of Diabetic Retinopathy
Type 2 Diabetes Genetic Variants and Risk of Diabetic Retinopathy Yong He Chong, MSc,1,2 Qiao Fan, PhD,2 Yih Chung Tham, PhD,1 Alfred Gan, MSc,1 Shu Pei Tan, BSc,1 Gavin Tan, MMed,1 Jie Jin Wang, PhD,3 Paul Mitchell, MD, PhD,3 Tien Yin Wong, FRCS, PhD,1,2,4,* Ching-Yu Cheng, MD, PhD1,2,4,* Purpose: Genetic association studies to date have not identified any robust risk loci for diabetic retinopathy (DR). We hypothesized that individuals with more diabetes genetic risk alleles have a higher risk of developing DR. Design: Case-control genetic association study. Participants: We evaluated the aggregate effects of multiple type 2 diabeteseassociated genetic variants on the risk of DR among 1528 participants with diabetes from the Singapore Epidemiology of Eye Diseases Study, of whom 547 (35.8%) had DR. Methods: Participants underwent a comprehensive ocular examination, including dilated fundus photography. Retinal photographs were graded using the modified Airlie House classification system to assess the presence and severity of DR following a standardized protocol. We identified 76 previously discovered type 2 diabeteseassociated single nucleotide polymorphisms (SNPs) and constructed multilocus genetic risk scores (GRSs) for each individual by summing the number of risk alleles for each SNP weighted by the respective effect estimates on DR. Two GRSs were generated: an overall GRS that included all 76 discovered type 2 diabeteseassociated SNPs, and an Asian-specific GRS that included a subset of 55 SNPs previously found to be associated with type 2 diabetes in East and/or South Asian ancestry populations. Associations between the GRSs with DR were determined using logistic regression analyses. Discriminating ability of the GRSs was determined by the area under the receiver operating characteristic curve (AUC). Main Outcome Measures: Odds ratios on DR. Results: Participants in the top tertile of the overall GRS were 2.56-fold more likely to have DR compared with participants in the lowest tertile. Participants in the top tertile of the Asian-specific GRS were 2.00-fold more likely to have DR compared with participants in the bottom tertile. Both GRSs were associated with higher DR severity levels. However, addition of the GRSs to traditional risk factors improved the AUC only modestly by 3% to 4%. Conclusions: Type 2 diabeteseassociated genetic loci were significantly associated with higher risks of DR, independent of traditional risk factors. Our findings may provide new insights to further our understanding of the genetic pathogenesis of DR. Ophthalmology 2016;-:1e7 ª 2016 by the American Academy of Ophthalmology Supplemental material is available at www.aaojournal.org.
Diabetic retinopathy (DR) is the most common microvascular complication of diabetes mellitus1 and is a leading cause of preventable blindness in working-aged adults worldwide.2 The global prevalence of DR, proliferative DR (PDR), and vision-threatening DR among individuals with diabetes is estimated to be 35%, 7%, and 12%, respectively.3 DR has also been estimated to be the cause of 2.6% of cases of blindness worldwide.4 Our understanding of the pathophysiology of DR is incomplete and constantly evolving with research. Risk factors such as hyperglycemia, hypertension, and prolonged diabetes duration are well established but explain less than 50% of the risk of DR.3,5e9 Genetic susceptibility to DR has also been suspected.10 Previous studies have shown racial and ethnic differences11 and familial aggregation in DR,12,13 suggesting a role for genetic factors in DR development. Several candidate gene studies and genome-wide association studies (GWAS) have also identified potential ª 2016 by the American Academy of Ophthalmology Published by Elsevier Inc.
genetic loci associated with DR.14e22 However, few results have been consistently replicated across different populations. Because DR is the most common microvascular complication of type 2 diabetes,2 it is reasonable to postulate that both diseases share a common genetic background. Type 2 diabetes is a multifactorial disease influenced by many different genetic variants with heritability estimated to be between 40% and 80%.23 Over the past decade, GWAS have identified over 70 susceptibility loci for type 2 diabetes.24e29 Of note, 3 of the type 2 diabetes risk loci (near TCF7L2, PPARG, and KCNJ11) have previously been shown to be associated with DR,30e32 further suggesting that type 2 diabetes susceptibility genes may have an influence on DR development. In this study, we hypothesized that individuals with more risk alleles of type 2 diabetes are more likely to have DR. We aimed to evaluate the association between the aggregate http://dx.doi.org/10.1016/j.ophtha.2016.11.016 ISSN 0161-6420/16
1
Ophthalmology Volume -, Number -, Month 2016 effects of multiple type 2 diabetes genetic variants with DR, through the construct of multilocus genetic risk scores (GRSs), in a population-based, multiethnic study.
Methods Study Population The Singapore Epidemiology of Eye Diseases (SEED) Study is a population-based cross-sectional study of 3 major ethnic groups in Singapore: Malays (2004e2006), Indians (2007e2009), and Chinese (2009e2011). The detailed study methodology has been previously described.33,34 In brief, 4168 Malays, 4497 Indians, and 4605 Chinese aged 40 to 80 years were selected using an agestratified random sampling strategy and invited to participate in the study. From these, a total of 10 033 participated, including 3280 Malays (response rate of 78.7%), 3400 Indians (75.6%), and 3353 Chinese (72.8%), giving an overall response rate of 75.6%. For this analysis, we focused on all participants with diabetes, defined as persons having a random glucose of 11.1 mmol/L or higher, using diabetic medication, or having a self-reported history of diabetes. Ethics approval was obtained from the SingHealth Centralised Institutional Review Board. Written informed consent was obtained from all participants and the study was conducted in accordance with the Declaration of Helsinki.
Diabetic Retinopathy and Risk Factor Assessments DR was assessed through standardized retinal photographs using a digital retinal camera (Canon CR-DGi with a 10-D SLR back; Canon, Tokyo, Japan) at the Singapore Eye Research Institute. After pupil dilation, 2 retinal photographs, centered at the optic disc and macula, were taken from both eyes. Photographs were sent to the University of Sydney and graded for retinopathy by masked, trained graders. DR was considered present if any characteristic lesion as defined by the Early Treatment Diabetic Retinopathy Study severity scale (i.e., microaneurysms, hemorrhages, cotton wool spots, intraretinal microvascular abnormalities, hard exudates, venous beading, and new vessels) was present in either eye.11 DR severity was based on the worse eye and was graded according to the modified Airlie House classification system35 as follows: level 10, DR absent; levels 14e15, questionable DR; level 20, minimal non-PDR (NPDR); level 35, mild NPDR; level 43, moderate NPDR; level 47, moderately severe NPDR; level 53, severe NPDR; level 61, mild PDR; level 65, moderate PDR; level 71, severe PDR; and levels 81 and 85, advanced PDR. All participants underwent a standardized interview for collection of demographic data, lifestyle risk factors, and medical history (e.g., diabetes duration). Blood pressure was measured according to the protocol used in the Multi-Ethnic Study of Atherosclerosis11 and was taken with the study participants seated and after 5 minutes of rest. Systolic and diastolic blood pressure were measured with a digital automatic blood pressure monitor. Blood pressure was measured on 2 occasions 5 minutes apart. If the blood pressures differed by more than 10 mmHg systolic and 5 mmHg diastolic, a third measurement was made. The blood pressure of the individual was then taken as the mean between the 2 closest readings. Hypertension was defined as systolic blood pressure of 140 mmHg or more, diastolic blood pressure of 90 mmHg or more, or use of antihypertensive medication. Nonfasting venous blood samples were collected to measure serum glycosylated hemoglobin (HbA1c) levels and for DNA extraction.
2
Genotyping and Genetic Ancestry Inference Genome-wide genotyping was performed using Illumina Human 610 Quad BeadChips (Illumina Inc, San Diego, CA) based on the manufacturer’s protocols in 7584 of the SEED participants. The detailed data quality control procedure has been previously described.36 Genotype imputation was carried out using the Markov Chain Haplotyping software package,37 using 1000 Genomes Project as reference panels. Individual genetic ancestry was inferred using principal component (PC) analysis to account for spurious associations owing to ancestral differences of individual single nucleotide polymorphisms (SNPs). This was carried out using the smartPCA program (EIGENSTRAT software version 4.2).38 The details of the PC analysis have been previously described.39
Statistical Analysis A 2-stage approach was adopted for analysis. First, we adopted a candidate gene approach by identifying and selecting 76 type 2 diabeteseassociated SNPs identified in the most recent and largestto-date meta-analysis of type 2 diabetes GWAS by the DIAbetes Genetics Replication And Meta-analysis (DIAGRAM) Consortium.24 To determine the association between DR and each type 2 diabetes susceptibility locus in our study population, we performed logistic regression analyses between the individual SNPs with DR under additive genetic models adjusting for age, gender, and the first 3 PCs in each ethnic group. We then obtained the combined effect estimates of individual SNPs by performing random-effect meta-analysis using individual-level data across the 3 ethnic groups. Next, we evaluated the aggregate effects of the type 2 diabetes susceptibility loci by constructing 2 GRSs: (1) an overall GRS that included all 76 SNPs identified by the DIAGRAM Consortium, and (2) an Asian-specific GRS that included a subset of 55 SNPs showing nominally significant association (P < 0.05) in East and/or South Asian ancestry groups from the DIAGRAM Consortium’s aggregated meta-analysis. This was achieved by summing the number of risk alleles for each of the type 2 diabeteseassociated SNPs for each GRS, weighted by the estimated individual SNP effect size on DR (logarithm of the odds ratio).40 Multivariable logistic and ordinal logistic regression analyses were performed to determine the association between GRSs with DR and DR severity levels, respectively, adjusting for diabetes duration, HbA1c, and hypertension (collectively termed “traditional DR risk factors”). In the ordinal logistic regression analyses, the dependent outcome variable DR severity level was coded accordingly on an ordinal scale as 0 (DR absent), 1 (questionable DR), 2 (minimal NPDR), 3 (mild NPDR), 4 (moderate NPDR), 5 (moderately severe NPDR), 6 (severe NPDR), 7 (mild PDR), 8 (moderate PDR), 9 (severe PDR), or 10 (advanced PDR). To evaluate the discriminating ability of the GRSs for DR compared with traditional risk factors, we calculated the area under the receiver operating characteristic curve (AUC) for 2 different models: the first model consisted of traditional DR risk factors and the second model consisted of both the traditional risk factors plus the GRSs. The difference in AUCs between models was compared using C-statistic. All statistical analyses were performed using Stata 14 (StataCorp LP, College Station, TX).
Results Of the 7584 study participants in the SEED study with genomewide genotype information, after excluding participants without diabetes (n ¼ 5805) and participants with diabetes with missing
Chong et al
Effects of Diabetic Genes on DR
Table 1. Summary of Demographic and Clinical Characteristics of the Diabetic Study Participants (N ¼ 1528)
Age, years Gender, female Ethnicity Malay Indian Chinese Diabetes duration, years Age at diabetes onset, years HbA1c, % Systolic BP, mmHg Diastolic BP, mmHg
With DR (N [ 547)
Without DR (N [ 981)
P Value
62.6 (8.8) 267 (48.8)
62.5 (9.6) 477 (48.6)
0.811 0.944
172 279 96 13.7 49.0 8.1 149.3 77.3
(31) (51) (18) (9.8) (11.0) (1.7) (22.9) (10.8)
308 471 202 8.2 54.3 7.5 142.9 77.2
(31) (48) (21) (7.6) (10.6) (1.5) (20.9) (9.6)
0.316 <0.001 <0.001 <0.001 <0.001 0.938
BP ¼ blood pressure; DR ¼ diabetic retinopathy; HbA1c ¼ serum glycosylated hemoglobin. Data are presented as means (standard deviation) or number (%), as appropriate.
relevant clinical data (n ¼ 251), a total of 1528 participants with diabetes (480 Malays, 750 Indians, and 298 Chinese) were included in the analysis. Of these, 547 participants (35.8%) were assessed to have DR. The baseline demographic and clinical characteristics of the included participants are shown in Table 1. Overall, participants with DR tended to be older and female. They also had significantly longer diabetes duration as well as higher HbA1c and systolic blood pressure levels (all P < 0.001). A summary of the 76 type 2 diabeteseassociated SNPs previously reported by the DIAGRAM Consortium used for the construction of the overall GRS is shown in Table 2 (available at http:// www.aaojournal.org). These 76 SNPs’ associations with DR among the study participants are shown in Table 3 (available at http://www.aaojournal.org). The frequencies of both the overall GRS and the Asian-specific GRS were approximately normally distributed (Fig 1A, B). The mean, standard deviation (SD), and range of the GRSs are shown in Table 4. In general, the mean (SD) of the overall GRS was 1.11 (0.45) among participants with DR and 0.93 (0.45) among controls; that of the Asianspecific GRS in participants with DR was 0.33 (0.40) and 0.47 (0.39) in controls. Table 5 shows the associations of the GRSs with DR. The mean SD number of risk alleles carried by the individuals in the highest tertile of the overall GRS was 92.62.6, compared with 81.22.6 carried by those in the lowest tertile (P < 0.001). Similarly, the mean SD number of risk alleles carried by individuals in the highest tertile of the Asian-specific GRS was 66.42.4, compared with 56.42.3 in the lowest tertile (P < 0.001). In general, participants with higher scores of both GRSs were significantly associated with increased odds of having DR, after adjusting for ethnicity, diabetes duration, HbA1c, and hypertension (all P trend < 0.001) (Fig 2). When adjusting for ethnicity and the traditional DR risk factors, participants in the top tertile of the overall GRS were 2.56 times (95% confidence interval [CI]: 1.92e3.40; P ¼ 1.5 1010) more likely to have DR compared with participants in the lowest tertile. Similarly, participants in the top tertile of the Asian-specific GRS were 2.00 times (95% CI: 1.51e2.65; P ¼ 1.3 106) more likely to have DR compared
Figure 1. Distribution of genetic risk scores. A, Overall genetic risk score. B, Asian-specific genetic risk score. DR ¼ diabetic retinopathy.
with participants in the lowest tertile. Similar trends were consistently observed across the 3 ethnic groups of Malay, Indian, and Chinese (Table 6 [available at http://www.aaojournal.org]). The associations of the GRSs with DR severity are shown in Table 7. Both GRSs were associated with higher DR severity levels after adjusting for ethnicity, diabetes duration, HbA1c, and hypertension (all P < 0.001). This association between GRSs and DR severity was similarly observed across the 3 ethnic groups (Table 8 [available at http://www.aaojournal.org]). A sensitivity analysis was conducted that included only controls with a diabetes duration of at least 5 years (n ¼ 547 cases, 548 controls). The mean diabetes duration was 13.7 and 12.8 years in cases and controls, respectively. Similar to the original finding, this sensitivity analysis showed that higher scores of both GRSs were significantly associated with increased odds of having DR, and that both GRSs were associated with higher DR severity levels (Tables 9 and 10 [available at http://www.aaojournal.org]). To determine the discriminatory accuracy of the GRSs compared with the traditional DR risk factors, we compared the difference in AUC estimates in the models consisting of traditional DR risk factors (age, gender, ethnicity, diabetes duration, HbA1c, and hypertension) and the model consisting of traditional DR risk factors and the GRSs (Table 11). The AUC in the model consisting
3
Ophthalmology Volume -, Number -, Month 2016 Table 4. Summary of Genetic Risk Score Characteristics Across Ethnic Groups Total (N [ 1528; 547 Cases, 981 Controls) Mean (Range)
GRS
Overall GRS Case 1.11 (0.29 Control 0.93 (0.48 Asian-specific GRS Case 0.33 (1.45 Control 0.47 (1.78
Malay (N [ 480; 172 Cases, 208 Controls)
SD
to 2.39) 0.45 to 2.39) 0.45
Mean (Range)
Indian (N [ 750; 279 Cases, 471 Controls)
SD
Mean (Range)
Chinese (N [ 298; 96 Cases, 202 Controls)
SD
Mean (Range)
SD
1.13 (0.11 to 2.29) 0.42 0.93 (0.17 to 2.20) 0.41
1.03 (0.29 to 2.39) 0.86 (0.48 to 2.39)
0.47 0.46
1.32 (0.28e2.23) 1.12 (0.18 to 2.33)
0.40 0.42
to 0.94) 0.40 0.36 (1.45 to 0.74) 0.39 to 0.64) 0.39 0.52 (1.43 to 0.54) 0.39
0.36 (1.42 to 0.94) 0.49 (1.78 to 0.53)
0.40 0.39
0.21 (1.07 to 0.69) 0.38 (1.52 to 0.64)
0.37 0.37
GRS ¼ genetic risk score; SD ¼ standard deviation. Cases are individuals with diabetic retinopathy (DR); controls are those without DR.
of the traditional DR risk factors only was 0.71 (95% CI: 0.68e0.74). In the model consisting of both the traditional DR risk factors and the GRSs, the AUC increased by 0.03 (P value for AUC difference of overall GRS ¼ 2.0 104) and 0.02 (P value for AUC difference of Asian-specific GRS ¼ 2.2 103).
Discussion In this multiethnic, population-based study, we showed that persons with a greater number of genetic risk loci for type 2 diabetes were more likely to have DR, using multilocus GRSs constructed from 76 type 2 diabeteseassociated genetic variants. This association was independent of traditional risk factors for DR and may be clinically relevant in predicting the risk of developing DR among patients with diabetes and the disease progression upon diagnosis of diabetes. Overall, our findings showed that participants in the top tertile of the overall GRS had a 2.5-fold increased risk of having DR compared with those in the lowest tertile. Similarly, participants in the top tertile of the Asian-specific GRSs also had a 2.0-fold increased risk of having DR compared with those in the lowest tertile.
However, the effect of individual type 2 diabetese associated SNPs on DR susceptibility was modest. Of note, there was no significant association found between the SNPs in the loci of TCF7L2, PPARG, and KCNJ11 and DR, in contrast to the findings from previous studies. This highlights the challenges in identifying genetic risk factors of DR, as reasons such as small sample sizes and the lack of uniform assessment across studies preclude the replication of results across populations. However, our findings showed that the cumulative effect of the type 2 diabeteseassociated SNPs had significantly stronger associations with DR. Such association consistently mirrors the polygenic, complex nature of DR, which was previously postulated. On the other hand, duration of diabetes and HbA1c are strong predictors for DR development. In our study, duration of diabetes was self-reported and HbA1c was based on a 1-time measurement only. As such, although we adjusted for duration of diabetes and HbA1c in our analysis, our observed association may reflect a possible effect of diabetes severity, mediated though longer duration of diabetes and higher levels of HbA1c, leading to the development of DR. Our findings showed that the GRSs were associated with higher DR severity levels. This suggests that genetic factors may play a prominent role in the development of DR and the
Table 5. Association Between Genetic Risk Scores and Diabetic Retinopathy Model 2y
Model 1* GRS Overall GRS (range) 1st tertile (0.45 to 0.81) 2nd tertile (0.81 to 1.21) 3rd tertile (1.21 to 2.39) Asian-specific GRS (range) 1st tertile (1.78 to 0.58) 2nd tertile (0.58 to 0.25) 3rd tertile (0.25 to 0.94)
OR (95% CI)
P Value
P Value
1.00 (Reference) 1.55 (1.18e2.03) 2.64 (2.01e3.45) P for trend ¼ 1.4 1012
e 1.6 103 1.8 1012
1.00 (Reference) 1.48 (1.10e1.97) 2.56 (1.92e3.40) P for trend ¼ 1.1 1010
e 8.4 103 1.5 1010
1.00 (Reference) 1.49 (1.14e1.95) 2.17 (1.66e2.82) P for trend ¼ 9.1 109
e 3.2 103 9.4 109
1.00 (Reference) 1.43 (1.07e1.89) 2.00 (1.51e2.65) P for trend ¼ 1.2 106
e 0.014 1.3 106
CI ¼ confidence interval; GRS ¼ genetic risk score; OR ¼ odds ratio. *Adjusted for ethnicity. y Adjusted for ethnicity, diabetes duration, serum glycosylated hemoglobin, and hypertension.
4
OR (95% CI)
Chong et al
Effects of Diabetic Genes on DR Table 11. Area-Under-the-Curve Estimates for Diabetic Retinopathy Using Traditional and Genetic Risk Score Prediction Models
Model including traditional risk factors* Model including traditional risk factors and GRS P difference
Overall GRS
Asian-Specific GRS
0.71 (0.68e0.74)
0.71 (0.68e0.74)
0.74 (0.71e0.76)
0.73 (0.70e0.76)
4
2.2 103
2.0 10
GRS ¼ genetic risk score. *Traditional risk factors consist of age, gender, ethnicity, diabetes duration, serum glycosylated hemoglobin, and hypertension. Data are area under the curve (95% confidence interval) unless otherwise indicated.
Figure 2. Proportions and odds ratios (OR) of diabetic retinopathy (DR) in sextile groups of genetic risk scores, with the first sextile as the reference group. A, Overall genetic risk score; B, Asian-specific genetic risk score.
eventual severity of the disease. Although the absolute AUC estimates for the GRS models are possibly overestimated owing to the concurrent use of study population in score construction, our purpose in performing the AUC analysis was to evaluate the discriminating ability of the GRSs for DR compared with traditional risk factors. In this regard, we showed a modest increase in AUC from the addition of the GRSs to traditional risk factors, indicating the robustness of
existing risk factors. This finding of modest risk discrimination improvement using genetic markers has been observed in other multifactorial, complex diseases such as glaucoma,40 which are influenced by numerous genetic variants. Current DR screening methods include regular dilated slit-lamp biomicroscopy eye examinations, mydriatic or nonmydriatic retinal photography, and tele-retinal screening.41 However, studies have shown a high prevalence of undiagnosed DR, ranging from 25% in an American inpatient population42 to 83% in the SEED population,43 suggesting that current screening methods may be inadequate. Early diagnosis and treatment have been shown to reduce the risk of DR progression and blindness.44,45 Clinically, the GRSs may be helpful in predicting patients with diabetes’ risk of developing DR and the disease progression upon diagnosis of diabetes. However, given the polygenic nature of complex diseases such as DR and the cost of genotyping, further discovery of more novel genes for type 2 diabetes or DR and cheaper cost of genotyping will be needed to more robustly improve the clinical usefulness of genetic variants in predicting DR. We showed that there was no significant difference in association between the GRSs and DR among the 3 ethnic groups of Malays, Indians, and Chinese. Combined with previous studies that showed no significant difference in the prevalence of DR among these 3 ethnic groups in the SEED population,43 these results suggest similar genetic susceptibility across the 3 ethnic groups of Malays, Indians, and Chinese.
Table 7. Association between Genetic Risk Score and Diabetic Retinopathy Severity Level Model 2y
Model 1* GRS Overall GRS Asian-specific GRS
b (95% CI) 0.83 (0.60e1.06) 0.87 (0.60e1.12)
P Value 12
2.2 10 9.7 1011
b (95% CI)
P Value
0.74 (0.50e0.98) 0.74 (0.47e1.01)
1.2 109 9.9 108
GRS ¼ genetic risk score. b represents the change in diabetic retinopathy severity level in the ordered log-odds scale for each standard unit increase in GRS. *Adjusted for ethnicity. y Adjusted for ethnicity, diabetes duration, serum glycosylated hemoglobin, and hypertension.
5
Ophthalmology Volume -, Number -, Month 2016 Study Strengths and Limitations The strengths of this study include a large, population-based sample and standardized methodology and assessment of DR based on retinal photographs. Although the type 2 diabetes genetic susceptibility variants had modest individual effects on DR, our study design allowed for the identification of stronger combined effects using the GRS approach. A limitation of this study was that it included only Asian ethnicities. As such, further validation of the GRSs is required in other populations of different ethnicities for further generalization of the current findings. Given that DR regression has been known to occur, especially in patients with minimal or mild NPDR, it is also possible that some controls may have previously had mild forms of the disease that have since regressed. In conclusion, our findings showed that persons with type 2 diabetes with more disease genetic risk alleles were more likely to have DR, independent of traditional DR risk factors. This suggests complex genetic susceptibility to both diabetes and DR, providing new insights into the genetic pathogenesis of microvascular complications in type 2 diabetes.
References 1. Fowler MJ. Microvascular and macrovascular complications of diabetes. Clin Diabetes. 2011;29(3):116-122. 2. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet. 2010;376:124-136. 3. Yau JWY, Rogers SL, Kawasaki R, et al. Global prevalence and major risk factors of diabetic retinopathy. Diabetes Care. 2012;35(3):556-564. 4. Bourne RR, Stevens GA, White RA, et al. Causes of vision loss worldwide, 1990-2010: a systematic analysis. Lancet Glob Health. 2013;1(6):e339-349. 5. Stratton IM, Kohner EM, Aldington SJ, et al. UKPDS 50L risk factors for incidence and progression of retinopathy in Type II diabetes over 6 years from diagnosis. Diabetologia. 2001;44: 156-163. 6. Klein R, Klein BE, Moss SE, Cruickshanks KJ. Relationship of hyperglycemia to the long-term incidence and progression of diabetic retinopathy. Arch Intern Med. 1994;154:21692178. 7. Romero-Aroca P, Baget-Bernaldiz M, Fernandez-Ballart J, et al. Ten-year incidence of diabetic retinopathy and macular edema. Risk factors in a sample of people with type 1 diabetes. Diabetes Res Clin Pract. 2011;94(1):126-132. 8. Xu J, Xu L, Wang YX, et al. Ten-year cumulative incidence of diabetic retinopathy. The Beijing Eye Study 2001/2011. PLoS One. 2014;9(10):e111320. 9. Ding J, Wong TY. Current epidemiology of diabetic retinopathy and diabetic macular edema. Curr Diab Rep. 2012;12(4): 346-354. 10. Liew G, Klein R, Wong TY. The role of genetics in susceptibility to diabetic retinopathy. Int Ophthalmol Clin. 2009;49(2):35-52. 11. Wong TY, Klein R, Islam A, et al. Diabetic retinopathy in a multi-ethnic cohort in the United States. Am J Ophthalmol. 2006;141(3):446-455. 12. Arar NH, Freedman BI, Adler SG, et al. Heritability of the severity of diabetic retinopathy: the FIND-Eye study. Invest Ophthalmol Vis Sci. 2008;49(9):3839-3845.
6
13. Hietala K, Forsblom C, Summanen P, et al. Heritability of proliferative diabetic retinopathy. Diabetes. 2008;57(8): 2176-2180. 14. Uhlmann K, Kovacs P, Boettcher Y, et al. Genetics of diabetic retinopathy. Exp Clin Endocrinol Diabetes. 2006;114(6):275-294. 15. Abhary S, Hewitt AW, Burdon KP, Craig JE. A systematic meta-analysis of genetic association studies for diabetic retinopathy. Diabetes. 2009;58(9):2137-2147. 16. Fu YP, Hallman DM, Gonzalez VH, et al. Identification of diabetic retinopathy genes through a genome-wide association study among Mexican-Americans from Starr County, Texas. J Ophthalmol. 2010:2010. 17. Awata T, Yamashita H, Kurihara S, et al. A genome-wide association study for diabetic retinopathy in a Japanese population: potential association with a long intergenic non-coding RNA. PLoS One. 2014;9(11):e111715. 18. Sheu WH, Kuo JZ, Lee IT, et al. Genome-wide association study in a Chinese population with diabetic retinopathy. Hum Mol Genet. 2013;22(15):3165-3173. 19. Huang YC, Lin JM, Lin HJ, et al. Genome-wide association study of diabetic retinopathy in a Taiwanese population. Ophthalmology. 2011;118(4):642-648. 20. Grassi MA, Tikhomirov A, Ramalingam S, et al. Genomewide meta-analysis for severe diabetic retinopathy. Hum Mol Genet. 2011;20(12):2472-2481. 21. Peng D, Wang J, Zhang R, et al. Common variants in or near ZNRF1, COLEC12, SCYL1BP1 and API5 are associated with diabetic retinopathy in Chinese patients with type 2 diabetes. Diabetologia. 2015;58(6):1231-1238. 22. Simoes MJ, Lobo C, Egas C, et al. Genetic variants in ICAM1, PPARGC1A and MTHFR are potentially associated with different phenotypes of diabetic retinopathy. Ophthalmologica. 2014;232(3):156-162. 23. Dorajoo R, Liu J, Boehm BO. Genetics of type 2 diabetes and clinical utility. Genes (Basel). 2015;6(2):372-384. 24. Mahajan A, Go MJ, Zhang W, et al. Genome-wide transancestry meta-analysis provides insight into the genetic architecture of type 2 diabetes susceptibility. Nat Genet. 2014;46(3):234-244. 25. Saxena R, Saleheen D, Been LF, et al. Genome-wide association study identifies a novel locus contributing to type 2 diabetes susceptibility in Sikhs of Punjabi origin from India. Diabetes. 2013;62:1746-1755. 26. Hara K, Fujita H, Johnson TA, et al. Genome-wide association study identifies three novel loci for type 2 diabetes. Hum Mol Genet. 2014;23(1):239-246. 27. Ma RC, Hu C, Tam CH, et al. Genome-wide association study in a Chinese population identifies a susceptibility locus for type 2 diabetes at 7q32 near PAX4. Diabetologia. 2013;56(6): 1291-1305. 28. Williams AL, Jacobs SBR, Moreno-Macías H, et al. Sequence variants in SLC16A11 are a common risk factor for type 2 diabetes in Mexico. Nature. 2013;506(7486):97-101. 29. Go MJ, Hwang JY, Park TJ, et al. Genome-wide association study identifies two novel Loci with sex-specific effects for type 2 diabetes mellitus and glycemic traits in a Korean population. Diabetes Metab J. 2014;38(5):375-387. 30. Melzer D, Murray A, Hurst AJ, et al. Effects of the diabetes linked TCF7L2 polymorphism in a representative older population. BMC Med. 2006;4:34. 31. Malecki MT, Cyganek K, Mirkiewicz-Sieradzka B, et al. Alanine variant of the Pro12Ala polymorphism of the PPARg gene might be associated with decreased risk of diabetic retinopathy in type 2 diabetes. Diabetes Res Clin Pract. 2008;80(1):139-145.
Chong et al
Effects of Diabetic Genes on DR
32. Liu NJ, Wu HH, Li YL, et al. An analysis of the association between a polymorphism of KCNJ11 and diabetic retinopathy in a Chinese Han population. Eur J Med Res. 2015;20:3. 33. Foong AW, Saw SM, Loo JL, et al. Rationale and methodology for a population-based study of eye diseases in Malay people: the Singapore Malay eye study (SiMES). Ophthalmic Epidemiol. 2007;14(1):25-35. 34. Lavanya R, Jeganathan VS, Zheng Y, et al. Methodology of the Singapore Indian Chinese Cohort (SICC) eye study: quantifying ethnic variations in the epidemiology of eye diseases in Asians. Ophthalmic Epidemiol. 2009;16(6):325-336. 35. Early Treatment Diabetic Retinopathy Study Research Group. Grading diabetic retinopathy from stereoscopic color fundus photographsean extension of the modified Airlie House classification. ETDRS report number 10. Ophthalmology. 1991;98(5 Suppl):786-806. 36. Fan Q, Barathi VA, Cheng CY, et al. Genetic variants on chromosome 1q41 influence ocular axial length and high myopia. PLoS Genet. 2012;8(6):e1002753. 37. Li Y, Willer CJ, Ding J, et al. MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. Genet Epidemiol. 2010;34:816-834. 38. Price AL, Patterson N, Plenge RM, et al. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006;38:904-909. 39. Chua J, Tham YC, Liao J, et al. Ethnic differences of intraocular pressure and central corneal thickness: the Singapore
40.
41.
42.
43.
44. 45.
Epidemiology of Eye Diseases study. Ophthalmology. 2014;121(10):2013-2022. Tham YC, Liao J, Vithana EN, et al. Aggregate effects of intraocular pressure and cup-to-disc ratio genetic variants on glaucoma in a multiethnic Asian population. Ophthalmology. 2015;122(6):1149-1157. Ting DS, Cheung GC, Wong TY. Diabetic retinopathy: global prevalence, major risk factors, screening practices and public health challenges: a review. Clin Experiment Ophthalmol. 2016;44(4):260-277. Kovarik JJ, Eller AW, Willard LA, et al. Prevalence of undiagnosed diabetic retinopathy among inpatients with diabetes: the diabetic retinopathy inpatient study (DRIPS). BMJ Open Diabetes Res Care. 2016;4(1):e000164. Huang OS, Tay WT, Ong PG, et al. Prevalence and determinants of undiagnosed diabetic retinopathy and visionthreatening retinopathy in a multiethnic Asian cohort: the Singapore Epidemiology of Eye Diseases (SEED) study. Br J Ophthalmol. 2015;99(12):1614-1621. UK Prospective Diabetes Study Group. Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 38. BMJ. 1998;317(7160):703-713. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med. 1993;329(14):977-986.
Footnotes and Financial Disclosures Originally received: August 17, 2016. Final revision: November 11, 2016. Accepted: November 11, 2016. Available online: ---.
Author Contributions: Conception and design: Wong, Cheng Manuscript no. 2016-106.
1
Singapore Eye Research Institute, Singapore National Eye Centre, Singapore.
2
Duke-NUS Medical School, Singapore.
3
Centre for Vision Research, Westmead Institute for Medical Research, University of Sydney, Sydney, Australia.
4
Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore. *Both Tien Yin Wong and Ching-Yu Cheng contributed equally to this work. Financial Disclosure(s): Supported by the National Medical Research Council (NMRC), Singapore (grants 0796/2003, 1176/2008, 1149/2008, STaR/0003/2008, 1249/2010, CG/SERI/2010, CIRG/1371/2013, and CIRG/1417/2015), and Biomedical Research Council, Singapore (08/1/35/19/550 and 09/1/35/19/616). C.Y.C.: Supported by an award from NMRC (CSA/033/2012). The funding organization had no role in the design or conduct of this research.
Analysis and interpretation: Chong, Fan, Tham, Gan, S.P. Tan, G. Tan, Wong, Cheng Data collection: Wang, Mitchell, Wong, Cheng Obtained funding: Not applicable Overall responsibility: Chong, Fan, Tham, Gan, S.P. Tan, G. Tan, Wang, Mitchell, Wong, Cheng Abbreviations and Acronyms: AUC ¼ area under the receiver operating characteristic curve; CI ¼ confidence interval; DIAGRAM ¼ DIAbetes Genetics Replication And Meta-analysis; DR ¼ diabetic retinopathy; GRS ¼ genetic risk score; GWAS ¼ genome-wide association study; HbA1c ¼ serum glycosylated hemoglobin; NPDR ¼ nonproliferative diabetic retinopathy; PC ¼ principal component; PDR ¼ proliferative diabetic retinopathy; SD ¼ standard deviation; SEED ¼ Singapore Epidemiology of Eye Diseases; SNP ¼ single nucleotide polymorphism. Correspondence: Ching-Yu Cheng, MD, PhD, Singapore Eye Research Institute, The Academia, 20 College Road, Discovery Tower Level 6, Singapore, 169856. E-mail: [email protected].
7
Diabetic retinopathy (DR) is the most common microvascular complication of diabetes mellitus1 and is a leading cause of preventable blindness in working-aged adults worldwide.2 The global prevalence of DR, proliferative DR (PDR), and vision-threatening DR among individuals with diabetes is estimated to be 35%, 7%, and 12%, respectively.3 DR has also been estimated to be the cause of 2.6% of cases of blindness worldwide.4 Our understanding of the pathophysiology of DR is incomplete and constantly evolving with research. Risk factors such as hyperglycemia, hypertension, and prolonged diabetes duration are well established but explain less than 50% of the risk of DR.3,5e9 Genetic susceptibility to DR has also been suspected.10 Previous studies have shown racial and ethnic differences11 and familial aggregation in DR,12,13 suggesting a role for genetic factors in DR development. Several candidate gene studies and genome-wide association studies (GWAS) have also identified potential ª 2016 by the American Academy of Ophthalmology Published by Elsevier Inc.
genetic loci associated with DR.14e22 However, few results have been consistently replicated across different populations. Because DR is the most common microvascular complication of type 2 diabetes,2 it is reasonable to postulate that both diseases share a common genetic background. Type 2 diabetes is a multifactorial disease influenced by many different genetic variants with heritability estimated to be between 40% and 80%.23 Over the past decade, GWAS have identified over 70 susceptibility loci for type 2 diabetes.24e29 Of note, 3 of the type 2 diabetes risk loci (near TCF7L2, PPARG, and KCNJ11) have previously been shown to be associated with DR,30e32 further suggesting that type 2 diabetes susceptibility genes may have an influence on DR development. In this study, we hypothesized that individuals with more risk alleles of type 2 diabetes are more likely to have DR. We aimed to evaluate the association between the aggregate http://dx.doi.org/10.1016/j.ophtha.2016.11.016 ISSN 0161-6420/16
1
Ophthalmology Volume -, Number -, Month 2016 effects of multiple type 2 diabetes genetic variants with DR, through the construct of multilocus genetic risk scores (GRSs), in a population-based, multiethnic study.
Methods Study Population The Singapore Epidemiology of Eye Diseases (SEED) Study is a population-based cross-sectional study of 3 major ethnic groups in Singapore: Malays (2004e2006), Indians (2007e2009), and Chinese (2009e2011). The detailed study methodology has been previously described.33,34 In brief, 4168 Malays, 4497 Indians, and 4605 Chinese aged 40 to 80 years were selected using an agestratified random sampling strategy and invited to participate in the study. From these, a total of 10 033 participated, including 3280 Malays (response rate of 78.7%), 3400 Indians (75.6%), and 3353 Chinese (72.8%), giving an overall response rate of 75.6%. For this analysis, we focused on all participants with diabetes, defined as persons having a random glucose of 11.1 mmol/L or higher, using diabetic medication, or having a self-reported history of diabetes. Ethics approval was obtained from the SingHealth Centralised Institutional Review Board. Written informed consent was obtained from all participants and the study was conducted in accordance with the Declaration of Helsinki.
Diabetic Retinopathy and Risk Factor Assessments DR was assessed through standardized retinal photographs using a digital retinal camera (Canon CR-DGi with a 10-D SLR back; Canon, Tokyo, Japan) at the Singapore Eye Research Institute. After pupil dilation, 2 retinal photographs, centered at the optic disc and macula, were taken from both eyes. Photographs were sent to the University of Sydney and graded for retinopathy by masked, trained graders. DR was considered present if any characteristic lesion as defined by the Early Treatment Diabetic Retinopathy Study severity scale (i.e., microaneurysms, hemorrhages, cotton wool spots, intraretinal microvascular abnormalities, hard exudates, venous beading, and new vessels) was present in either eye.11 DR severity was based on the worse eye and was graded according to the modified Airlie House classification system35 as follows: level 10, DR absent; levels 14e15, questionable DR; level 20, minimal non-PDR (NPDR); level 35, mild NPDR; level 43, moderate NPDR; level 47, moderately severe NPDR; level 53, severe NPDR; level 61, mild PDR; level 65, moderate PDR; level 71, severe PDR; and levels 81 and 85, advanced PDR. All participants underwent a standardized interview for collection of demographic data, lifestyle risk factors, and medical history (e.g., diabetes duration). Blood pressure was measured according to the protocol used in the Multi-Ethnic Study of Atherosclerosis11 and was taken with the study participants seated and after 5 minutes of rest. Systolic and diastolic blood pressure were measured with a digital automatic blood pressure monitor. Blood pressure was measured on 2 occasions 5 minutes apart. If the blood pressures differed by more than 10 mmHg systolic and 5 mmHg diastolic, a third measurement was made. The blood pressure of the individual was then taken as the mean between the 2 closest readings. Hypertension was defined as systolic blood pressure of 140 mmHg or more, diastolic blood pressure of 90 mmHg or more, or use of antihypertensive medication. Nonfasting venous blood samples were collected to measure serum glycosylated hemoglobin (HbA1c) levels and for DNA extraction.
2
Genotyping and Genetic Ancestry Inference Genome-wide genotyping was performed using Illumina Human 610 Quad BeadChips (Illumina Inc, San Diego, CA) based on the manufacturer’s protocols in 7584 of the SEED participants. The detailed data quality control procedure has been previously described.36 Genotype imputation was carried out using the Markov Chain Haplotyping software package,37 using 1000 Genomes Project as reference panels. Individual genetic ancestry was inferred using principal component (PC) analysis to account for spurious associations owing to ancestral differences of individual single nucleotide polymorphisms (SNPs). This was carried out using the smartPCA program (EIGENSTRAT software version 4.2).38 The details of the PC analysis have been previously described.39
Statistical Analysis A 2-stage approach was adopted for analysis. First, we adopted a candidate gene approach by identifying and selecting 76 type 2 diabeteseassociated SNPs identified in the most recent and largestto-date meta-analysis of type 2 diabetes GWAS by the DIAbetes Genetics Replication And Meta-analysis (DIAGRAM) Consortium.24 To determine the association between DR and each type 2 diabetes susceptibility locus in our study population, we performed logistic regression analyses between the individual SNPs with DR under additive genetic models adjusting for age, gender, and the first 3 PCs in each ethnic group. We then obtained the combined effect estimates of individual SNPs by performing random-effect meta-analysis using individual-level data across the 3 ethnic groups. Next, we evaluated the aggregate effects of the type 2 diabetes susceptibility loci by constructing 2 GRSs: (1) an overall GRS that included all 76 SNPs identified by the DIAGRAM Consortium, and (2) an Asian-specific GRS that included a subset of 55 SNPs showing nominally significant association (P < 0.05) in East and/or South Asian ancestry groups from the DIAGRAM Consortium’s aggregated meta-analysis. This was achieved by summing the number of risk alleles for each of the type 2 diabeteseassociated SNPs for each GRS, weighted by the estimated individual SNP effect size on DR (logarithm of the odds ratio).40 Multivariable logistic and ordinal logistic regression analyses were performed to determine the association between GRSs with DR and DR severity levels, respectively, adjusting for diabetes duration, HbA1c, and hypertension (collectively termed “traditional DR risk factors”). In the ordinal logistic regression analyses, the dependent outcome variable DR severity level was coded accordingly on an ordinal scale as 0 (DR absent), 1 (questionable DR), 2 (minimal NPDR), 3 (mild NPDR), 4 (moderate NPDR), 5 (moderately severe NPDR), 6 (severe NPDR), 7 (mild PDR), 8 (moderate PDR), 9 (severe PDR), or 10 (advanced PDR). To evaluate the discriminating ability of the GRSs for DR compared with traditional risk factors, we calculated the area under the receiver operating characteristic curve (AUC) for 2 different models: the first model consisted of traditional DR risk factors and the second model consisted of both the traditional risk factors plus the GRSs. The difference in AUCs between models was compared using C-statistic. All statistical analyses were performed using Stata 14 (StataCorp LP, College Station, TX).
Results Of the 7584 study participants in the SEED study with genomewide genotype information, after excluding participants without diabetes (n ¼ 5805) and participants with diabetes with missing
Chong et al
Effects of Diabetic Genes on DR
Table 1. Summary of Demographic and Clinical Characteristics of the Diabetic Study Participants (N ¼ 1528)
Age, years Gender, female Ethnicity Malay Indian Chinese Diabetes duration, years Age at diabetes onset, years HbA1c, % Systolic BP, mmHg Diastolic BP, mmHg
With DR (N [ 547)
Without DR (N [ 981)
P Value
62.6 (8.8) 267 (48.8)
62.5 (9.6) 477 (48.6)
0.811 0.944
172 279 96 13.7 49.0 8.1 149.3 77.3
(31) (51) (18) (9.8) (11.0) (1.7) (22.9) (10.8)
308 471 202 8.2 54.3 7.5 142.9 77.2
(31) (48) (21) (7.6) (10.6) (1.5) (20.9) (9.6)
0.316 <0.001 <0.001 <0.001 <0.001 0.938
BP ¼ blood pressure; DR ¼ diabetic retinopathy; HbA1c ¼ serum glycosylated hemoglobin. Data are presented as means (standard deviation) or number (%), as appropriate.
relevant clinical data (n ¼ 251), a total of 1528 participants with diabetes (480 Malays, 750 Indians, and 298 Chinese) were included in the analysis. Of these, 547 participants (35.8%) were assessed to have DR. The baseline demographic and clinical characteristics of the included participants are shown in Table 1. Overall, participants with DR tended to be older and female. They also had significantly longer diabetes duration as well as higher HbA1c and systolic blood pressure levels (all P < 0.001). A summary of the 76 type 2 diabeteseassociated SNPs previously reported by the DIAGRAM Consortium used for the construction of the overall GRS is shown in Table 2 (available at http:// www.aaojournal.org). These 76 SNPs’ associations with DR among the study participants are shown in Table 3 (available at http://www.aaojournal.org). The frequencies of both the overall GRS and the Asian-specific GRS were approximately normally distributed (Fig 1A, B). The mean, standard deviation (SD), and range of the GRSs are shown in Table 4. In general, the mean (SD) of the overall GRS was 1.11 (0.45) among participants with DR and 0.93 (0.45) among controls; that of the Asianspecific GRS in participants with DR was 0.33 (0.40) and 0.47 (0.39) in controls. Table 5 shows the associations of the GRSs with DR. The mean SD number of risk alleles carried by the individuals in the highest tertile of the overall GRS was 92.62.6, compared with 81.22.6 carried by those in the lowest tertile (P < 0.001). Similarly, the mean SD number of risk alleles carried by individuals in the highest tertile of the Asian-specific GRS was 66.42.4, compared with 56.42.3 in the lowest tertile (P < 0.001). In general, participants with higher scores of both GRSs were significantly associated with increased odds of having DR, after adjusting for ethnicity, diabetes duration, HbA1c, and hypertension (all P trend < 0.001) (Fig 2). When adjusting for ethnicity and the traditional DR risk factors, participants in the top tertile of the overall GRS were 2.56 times (95% confidence interval [CI]: 1.92e3.40; P ¼ 1.5 1010) more likely to have DR compared with participants in the lowest tertile. Similarly, participants in the top tertile of the Asian-specific GRS were 2.00 times (95% CI: 1.51e2.65; P ¼ 1.3 106) more likely to have DR compared
Figure 1. Distribution of genetic risk scores. A, Overall genetic risk score. B, Asian-specific genetic risk score. DR ¼ diabetic retinopathy.
with participants in the lowest tertile. Similar trends were consistently observed across the 3 ethnic groups of Malay, Indian, and Chinese (Table 6 [available at http://www.aaojournal.org]). The associations of the GRSs with DR severity are shown in Table 7. Both GRSs were associated with higher DR severity levels after adjusting for ethnicity, diabetes duration, HbA1c, and hypertension (all P < 0.001). This association between GRSs and DR severity was similarly observed across the 3 ethnic groups (Table 8 [available at http://www.aaojournal.org]). A sensitivity analysis was conducted that included only controls with a diabetes duration of at least 5 years (n ¼ 547 cases, 548 controls). The mean diabetes duration was 13.7 and 12.8 years in cases and controls, respectively. Similar to the original finding, this sensitivity analysis showed that higher scores of both GRSs were significantly associated with increased odds of having DR, and that both GRSs were associated with higher DR severity levels (Tables 9 and 10 [available at http://www.aaojournal.org]). To determine the discriminatory accuracy of the GRSs compared with the traditional DR risk factors, we compared the difference in AUC estimates in the models consisting of traditional DR risk factors (age, gender, ethnicity, diabetes duration, HbA1c, and hypertension) and the model consisting of traditional DR risk factors and the GRSs (Table 11). The AUC in the model consisting
3
Ophthalmology Volume -, Number -, Month 2016 Table 4. Summary of Genetic Risk Score Characteristics Across Ethnic Groups Total (N [ 1528; 547 Cases, 981 Controls) Mean (Range)
GRS
Overall GRS Case 1.11 (0.29 Control 0.93 (0.48 Asian-specific GRS Case 0.33 (1.45 Control 0.47 (1.78
Malay (N [ 480; 172 Cases, 208 Controls)
SD
to 2.39) 0.45 to 2.39) 0.45
Mean (Range)
Indian (N [ 750; 279 Cases, 471 Controls)
SD
Mean (Range)
Chinese (N [ 298; 96 Cases, 202 Controls)
SD
Mean (Range)
SD
1.13 (0.11 to 2.29) 0.42 0.93 (0.17 to 2.20) 0.41
1.03 (0.29 to 2.39) 0.86 (0.48 to 2.39)
0.47 0.46
1.32 (0.28e2.23) 1.12 (0.18 to 2.33)
0.40 0.42
to 0.94) 0.40 0.36 (1.45 to 0.74) 0.39 to 0.64) 0.39 0.52 (1.43 to 0.54) 0.39
0.36 (1.42 to 0.94) 0.49 (1.78 to 0.53)
0.40 0.39
0.21 (1.07 to 0.69) 0.38 (1.52 to 0.64)
0.37 0.37
GRS ¼ genetic risk score; SD ¼ standard deviation. Cases are individuals with diabetic retinopathy (DR); controls are those without DR.
of the traditional DR risk factors only was 0.71 (95% CI: 0.68e0.74). In the model consisting of both the traditional DR risk factors and the GRSs, the AUC increased by 0.03 (P value for AUC difference of overall GRS ¼ 2.0 104) and 0.02 (P value for AUC difference of Asian-specific GRS ¼ 2.2 103).
Discussion In this multiethnic, population-based study, we showed that persons with a greater number of genetic risk loci for type 2 diabetes were more likely to have DR, using multilocus GRSs constructed from 76 type 2 diabeteseassociated genetic variants. This association was independent of traditional risk factors for DR and may be clinically relevant in predicting the risk of developing DR among patients with diabetes and the disease progression upon diagnosis of diabetes. Overall, our findings showed that participants in the top tertile of the overall GRS had a 2.5-fold increased risk of having DR compared with those in the lowest tertile. Similarly, participants in the top tertile of the Asian-specific GRSs also had a 2.0-fold increased risk of having DR compared with those in the lowest tertile.
However, the effect of individual type 2 diabetese associated SNPs on DR susceptibility was modest. Of note, there was no significant association found between the SNPs in the loci of TCF7L2, PPARG, and KCNJ11 and DR, in contrast to the findings from previous studies. This highlights the challenges in identifying genetic risk factors of DR, as reasons such as small sample sizes and the lack of uniform assessment across studies preclude the replication of results across populations. However, our findings showed that the cumulative effect of the type 2 diabeteseassociated SNPs had significantly stronger associations with DR. Such association consistently mirrors the polygenic, complex nature of DR, which was previously postulated. On the other hand, duration of diabetes and HbA1c are strong predictors for DR development. In our study, duration of diabetes was self-reported and HbA1c was based on a 1-time measurement only. As such, although we adjusted for duration of diabetes and HbA1c in our analysis, our observed association may reflect a possible effect of diabetes severity, mediated though longer duration of diabetes and higher levels of HbA1c, leading to the development of DR. Our findings showed that the GRSs were associated with higher DR severity levels. This suggests that genetic factors may play a prominent role in the development of DR and the
Table 5. Association Between Genetic Risk Scores and Diabetic Retinopathy Model 2y
Model 1* GRS Overall GRS (range) 1st tertile (0.45 to 0.81) 2nd tertile (0.81 to 1.21) 3rd tertile (1.21 to 2.39) Asian-specific GRS (range) 1st tertile (1.78 to 0.58) 2nd tertile (0.58 to 0.25) 3rd tertile (0.25 to 0.94)
OR (95% CI)
P Value
P Value
1.00 (Reference) 1.55 (1.18e2.03) 2.64 (2.01e3.45) P for trend ¼ 1.4 1012
e 1.6 103 1.8 1012
1.00 (Reference) 1.48 (1.10e1.97) 2.56 (1.92e3.40) P for trend ¼ 1.1 1010
e 8.4 103 1.5 1010
1.00 (Reference) 1.49 (1.14e1.95) 2.17 (1.66e2.82) P for trend ¼ 9.1 109
e 3.2 103 9.4 109
1.00 (Reference) 1.43 (1.07e1.89) 2.00 (1.51e2.65) P for trend ¼ 1.2 106
e 0.014 1.3 106
CI ¼ confidence interval; GRS ¼ genetic risk score; OR ¼ odds ratio. *Adjusted for ethnicity. y Adjusted for ethnicity, diabetes duration, serum glycosylated hemoglobin, and hypertension.
4
OR (95% CI)
Chong et al
Effects of Diabetic Genes on DR Table 11. Area-Under-the-Curve Estimates for Diabetic Retinopathy Using Traditional and Genetic Risk Score Prediction Models
Model including traditional risk factors* Model including traditional risk factors and GRS P difference
Overall GRS
Asian-Specific GRS
0.71 (0.68e0.74)
0.71 (0.68e0.74)
0.74 (0.71e0.76)
0.73 (0.70e0.76)
4
2.2 103
2.0 10
GRS ¼ genetic risk score. *Traditional risk factors consist of age, gender, ethnicity, diabetes duration, serum glycosylated hemoglobin, and hypertension. Data are area under the curve (95% confidence interval) unless otherwise indicated.
Figure 2. Proportions and odds ratios (OR) of diabetic retinopathy (DR) in sextile groups of genetic risk scores, with the first sextile as the reference group. A, Overall genetic risk score; B, Asian-specific genetic risk score.
eventual severity of the disease. Although the absolute AUC estimates for the GRS models are possibly overestimated owing to the concurrent use of study population in score construction, our purpose in performing the AUC analysis was to evaluate the discriminating ability of the GRSs for DR compared with traditional risk factors. In this regard, we showed a modest increase in AUC from the addition of the GRSs to traditional risk factors, indicating the robustness of
existing risk factors. This finding of modest risk discrimination improvement using genetic markers has been observed in other multifactorial, complex diseases such as glaucoma,40 which are influenced by numerous genetic variants. Current DR screening methods include regular dilated slit-lamp biomicroscopy eye examinations, mydriatic or nonmydriatic retinal photography, and tele-retinal screening.41 However, studies have shown a high prevalence of undiagnosed DR, ranging from 25% in an American inpatient population42 to 83% in the SEED population,43 suggesting that current screening methods may be inadequate. Early diagnosis and treatment have been shown to reduce the risk of DR progression and blindness.44,45 Clinically, the GRSs may be helpful in predicting patients with diabetes’ risk of developing DR and the disease progression upon diagnosis of diabetes. However, given the polygenic nature of complex diseases such as DR and the cost of genotyping, further discovery of more novel genes for type 2 diabetes or DR and cheaper cost of genotyping will be needed to more robustly improve the clinical usefulness of genetic variants in predicting DR. We showed that there was no significant difference in association between the GRSs and DR among the 3 ethnic groups of Malays, Indians, and Chinese. Combined with previous studies that showed no significant difference in the prevalence of DR among these 3 ethnic groups in the SEED population,43 these results suggest similar genetic susceptibility across the 3 ethnic groups of Malays, Indians, and Chinese.
Table 7. Association between Genetic Risk Score and Diabetic Retinopathy Severity Level Model 2y
Model 1* GRS Overall GRS Asian-specific GRS
b (95% CI) 0.83 (0.60e1.06) 0.87 (0.60e1.12)
P Value 12
2.2 10 9.7 1011
b (95% CI)
P Value
0.74 (0.50e0.98) 0.74 (0.47e1.01)
1.2 109 9.9 108
GRS ¼ genetic risk score. b represents the change in diabetic retinopathy severity level in the ordered log-odds scale for each standard unit increase in GRS. *Adjusted for ethnicity. y Adjusted for ethnicity, diabetes duration, serum glycosylated hemoglobin, and hypertension.
5
Ophthalmology Volume -, Number -, Month 2016 Study Strengths and Limitations The strengths of this study include a large, population-based sample and standardized methodology and assessment of DR based on retinal photographs. Although the type 2 diabetes genetic susceptibility variants had modest individual effects on DR, our study design allowed for the identification of stronger combined effects using the GRS approach. A limitation of this study was that it included only Asian ethnicities. As such, further validation of the GRSs is required in other populations of different ethnicities for further generalization of the current findings. Given that DR regression has been known to occur, especially in patients with minimal or mild NPDR, it is also possible that some controls may have previously had mild forms of the disease that have since regressed. In conclusion, our findings showed that persons with type 2 diabetes with more disease genetic risk alleles were more likely to have DR, independent of traditional DR risk factors. This suggests complex genetic susceptibility to both diabetes and DR, providing new insights into the genetic pathogenesis of microvascular complications in type 2 diabetes.
References 1. Fowler MJ. Microvascular and macrovascular complications of diabetes. Clin Diabetes. 2011;29(3):116-122. 2. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet. 2010;376:124-136. 3. Yau JWY, Rogers SL, Kawasaki R, et al. Global prevalence and major risk factors of diabetic retinopathy. Diabetes Care. 2012;35(3):556-564. 4. Bourne RR, Stevens GA, White RA, et al. Causes of vision loss worldwide, 1990-2010: a systematic analysis. Lancet Glob Health. 2013;1(6):e339-349. 5. Stratton IM, Kohner EM, Aldington SJ, et al. UKPDS 50L risk factors for incidence and progression of retinopathy in Type II diabetes over 6 years from diagnosis. Diabetologia. 2001;44: 156-163. 6. Klein R, Klein BE, Moss SE, Cruickshanks KJ. Relationship of hyperglycemia to the long-term incidence and progression of diabetic retinopathy. Arch Intern Med. 1994;154:21692178. 7. Romero-Aroca P, Baget-Bernaldiz M, Fernandez-Ballart J, et al. Ten-year incidence of diabetic retinopathy and macular edema. Risk factors in a sample of people with type 1 diabetes. Diabetes Res Clin Pract. 2011;94(1):126-132. 8. Xu J, Xu L, Wang YX, et al. Ten-year cumulative incidence of diabetic retinopathy. The Beijing Eye Study 2001/2011. PLoS One. 2014;9(10):e111320. 9. Ding J, Wong TY. Current epidemiology of diabetic retinopathy and diabetic macular edema. Curr Diab Rep. 2012;12(4): 346-354. 10. Liew G, Klein R, Wong TY. The role of genetics in susceptibility to diabetic retinopathy. Int Ophthalmol Clin. 2009;49(2):35-52. 11. Wong TY, Klein R, Islam A, et al. Diabetic retinopathy in a multi-ethnic cohort in the United States. Am J Ophthalmol. 2006;141(3):446-455. 12. Arar NH, Freedman BI, Adler SG, et al. Heritability of the severity of diabetic retinopathy: the FIND-Eye study. Invest Ophthalmol Vis Sci. 2008;49(9):3839-3845.
6
13. Hietala K, Forsblom C, Summanen P, et al. Heritability of proliferative diabetic retinopathy. Diabetes. 2008;57(8): 2176-2180. 14. Uhlmann K, Kovacs P, Boettcher Y, et al. Genetics of diabetic retinopathy. Exp Clin Endocrinol Diabetes. 2006;114(6):275-294. 15. Abhary S, Hewitt AW, Burdon KP, Craig JE. A systematic meta-analysis of genetic association studies for diabetic retinopathy. Diabetes. 2009;58(9):2137-2147. 16. Fu YP, Hallman DM, Gonzalez VH, et al. Identification of diabetic retinopathy genes through a genome-wide association study among Mexican-Americans from Starr County, Texas. J Ophthalmol. 2010:2010. 17. Awata T, Yamashita H, Kurihara S, et al. A genome-wide association study for diabetic retinopathy in a Japanese population: potential association with a long intergenic non-coding RNA. PLoS One. 2014;9(11):e111715. 18. Sheu WH, Kuo JZ, Lee IT, et al. Genome-wide association study in a Chinese population with diabetic retinopathy. Hum Mol Genet. 2013;22(15):3165-3173. 19. Huang YC, Lin JM, Lin HJ, et al. Genome-wide association study of diabetic retinopathy in a Taiwanese population. Ophthalmology. 2011;118(4):642-648. 20. Grassi MA, Tikhomirov A, Ramalingam S, et al. Genomewide meta-analysis for severe diabetic retinopathy. Hum Mol Genet. 2011;20(12):2472-2481. 21. Peng D, Wang J, Zhang R, et al. Common variants in or near ZNRF1, COLEC12, SCYL1BP1 and API5 are associated with diabetic retinopathy in Chinese patients with type 2 diabetes. Diabetologia. 2015;58(6):1231-1238. 22. Simoes MJ, Lobo C, Egas C, et al. Genetic variants in ICAM1, PPARGC1A and MTHFR are potentially associated with different phenotypes of diabetic retinopathy. Ophthalmologica. 2014;232(3):156-162. 23. Dorajoo R, Liu J, Boehm BO. Genetics of type 2 diabetes and clinical utility. Genes (Basel). 2015;6(2):372-384. 24. Mahajan A, Go MJ, Zhang W, et al. Genome-wide transancestry meta-analysis provides insight into the genetic architecture of type 2 diabetes susceptibility. Nat Genet. 2014;46(3):234-244. 25. Saxena R, Saleheen D, Been LF, et al. Genome-wide association study identifies a novel locus contributing to type 2 diabetes susceptibility in Sikhs of Punjabi origin from India. Diabetes. 2013;62:1746-1755. 26. Hara K, Fujita H, Johnson TA, et al. Genome-wide association study identifies three novel loci for type 2 diabetes. Hum Mol Genet. 2014;23(1):239-246. 27. Ma RC, Hu C, Tam CH, et al. Genome-wide association study in a Chinese population identifies a susceptibility locus for type 2 diabetes at 7q32 near PAX4. Diabetologia. 2013;56(6): 1291-1305. 28. Williams AL, Jacobs SBR, Moreno-Macías H, et al. Sequence variants in SLC16A11 are a common risk factor for type 2 diabetes in Mexico. Nature. 2013;506(7486):97-101. 29. Go MJ, Hwang JY, Park TJ, et al. Genome-wide association study identifies two novel Loci with sex-specific effects for type 2 diabetes mellitus and glycemic traits in a Korean population. Diabetes Metab J. 2014;38(5):375-387. 30. Melzer D, Murray A, Hurst AJ, et al. Effects of the diabetes linked TCF7L2 polymorphism in a representative older population. BMC Med. 2006;4:34. 31. Malecki MT, Cyganek K, Mirkiewicz-Sieradzka B, et al. Alanine variant of the Pro12Ala polymorphism of the PPARg gene might be associated with decreased risk of diabetic retinopathy in type 2 diabetes. Diabetes Res Clin Pract. 2008;80(1):139-145.
Chong et al
Effects of Diabetic Genes on DR
32. Liu NJ, Wu HH, Li YL, et al. An analysis of the association between a polymorphism of KCNJ11 and diabetic retinopathy in a Chinese Han population. Eur J Med Res. 2015;20:3. 33. Foong AW, Saw SM, Loo JL, et al. Rationale and methodology for a population-based study of eye diseases in Malay people: the Singapore Malay eye study (SiMES). Ophthalmic Epidemiol. 2007;14(1):25-35. 34. Lavanya R, Jeganathan VS, Zheng Y, et al. Methodology of the Singapore Indian Chinese Cohort (SICC) eye study: quantifying ethnic variations in the epidemiology of eye diseases in Asians. Ophthalmic Epidemiol. 2009;16(6):325-336. 35. Early Treatment Diabetic Retinopathy Study Research Group. Grading diabetic retinopathy from stereoscopic color fundus photographsean extension of the modified Airlie House classification. ETDRS report number 10. Ophthalmology. 1991;98(5 Suppl):786-806. 36. Fan Q, Barathi VA, Cheng CY, et al. Genetic variants on chromosome 1q41 influence ocular axial length and high myopia. PLoS Genet. 2012;8(6):e1002753. 37. Li Y, Willer CJ, Ding J, et al. MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. Genet Epidemiol. 2010;34:816-834. 38. Price AL, Patterson N, Plenge RM, et al. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006;38:904-909. 39. Chua J, Tham YC, Liao J, et al. Ethnic differences of intraocular pressure and central corneal thickness: the Singapore
40.
41.
42.
43.
44. 45.
Epidemiology of Eye Diseases study. Ophthalmology. 2014;121(10):2013-2022. Tham YC, Liao J, Vithana EN, et al. Aggregate effects of intraocular pressure and cup-to-disc ratio genetic variants on glaucoma in a multiethnic Asian population. Ophthalmology. 2015;122(6):1149-1157. Ting DS, Cheung GC, Wong TY. Diabetic retinopathy: global prevalence, major risk factors, screening practices and public health challenges: a review. Clin Experiment Ophthalmol. 2016;44(4):260-277. Kovarik JJ, Eller AW, Willard LA, et al. Prevalence of undiagnosed diabetic retinopathy among inpatients with diabetes: the diabetic retinopathy inpatient study (DRIPS). BMJ Open Diabetes Res Care. 2016;4(1):e000164. Huang OS, Tay WT, Ong PG, et al. Prevalence and determinants of undiagnosed diabetic retinopathy and visionthreatening retinopathy in a multiethnic Asian cohort: the Singapore Epidemiology of Eye Diseases (SEED) study. Br J Ophthalmol. 2015;99(12):1614-1621. UK Prospective Diabetes Study Group. Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 38. BMJ. 1998;317(7160):703-713. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med. 1993;329(14):977-986.
Footnotes and Financial Disclosures Originally received: August 17, 2016. Final revision: November 11, 2016. Accepted: November 11, 2016. Available online: ---.
Author Contributions: Conception and design: Wong, Cheng Manuscript no. 2016-106.
1
Singapore Eye Research Institute, Singapore National Eye Centre, Singapore.
2
Duke-NUS Medical School, Singapore.
3
Centre for Vision Research, Westmead Institute for Medical Research, University of Sydney, Sydney, Australia.
4
Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore. *Both Tien Yin Wong and Ching-Yu Cheng contributed equally to this work. Financial Disclosure(s): Supported by the National Medical Research Council (NMRC), Singapore (grants 0796/2003, 1176/2008, 1149/2008, STaR/0003/2008, 1249/2010, CG/SERI/2010, CIRG/1371/2013, and CIRG/1417/2015), and Biomedical Research Council, Singapore (08/1/35/19/550 and 09/1/35/19/616). C.Y.C.: Supported by an award from NMRC (CSA/033/2012). The funding organization had no role in the design or conduct of this research.
Analysis and interpretation: Chong, Fan, Tham, Gan, S.P. Tan, G. Tan, Wong, Cheng Data collection: Wang, Mitchell, Wong, Cheng Obtained funding: Not applicable Overall responsibility: Chong, Fan, Tham, Gan, S.P. Tan, G. Tan, Wang, Mitchell, Wong, Cheng Abbreviations and Acronyms: AUC ¼ area under the receiver operating characteristic curve; CI ¼ confidence interval; DIAGRAM ¼ DIAbetes Genetics Replication And Meta-analysis; DR ¼ diabetic retinopathy; GRS ¼ genetic risk score; GWAS ¼ genome-wide association study; HbA1c ¼ serum glycosylated hemoglobin; NPDR ¼ nonproliferative diabetic retinopathy; PC ¼ principal component; PDR ¼ proliferative diabetic retinopathy; SD ¼ standard deviation; SEED ¼ Singapore Epidemiology of Eye Diseases; SNP ¼ single nucleotide polymorphism. Correspondence: Ching-Yu Cheng, MD, PhD, Singapore Eye Research Institute, The Academia, 20 College Road, Discovery Tower Level 6, Singapore, 169856. E-mail: [email protected].
7