Multilocus genetic risk score for diabetic retinopathy in the Han Chinese population of Taiwan

Abstract

The aim of this study is to explore the effect of genetic variation on diabetic retinopathy (DR) risk in a Taiwanese population. The logistic regression model was used to evaluate the relationship between DR status and risk factors, including the conventional parameters and genetic risk score (GRS). Candidate single nucleotide polymorphisms (SNPs) in GRS were selected based on previous reports with a combined P < 10−4 (genome-wide association) and P < 0.05 (meta-analysis). In total, 58 SNPs in 44 susceptibility loci were selected, and four were used to calculate GRS. After adjustment for age, systolic blood pressure, diabetes duration, and HbA1c, the DR risk was 4.95 times higher for patients in the top GRS third tile than for those in the bottom third tile (95% CI = 2.99–8.18; P < 0.001). The addition of genetic information improved DR prediction, increasing the area under the curve (AUC) from 0.72 to 0.77 (P = 0.0024) and improving the sensitivity of the model such that 40 more subjects were reclassified into DR status. The developed multivariate logistic regression model combining conventional risk factors and the multilocus GRS can predict DR, thus enabling timely treatment to reduce blindness in T2D patients.

Causal Relationship between Adiponectin and Diabetic Retinopathy: A Mendelian Randomization Study in an Asian Population

“…The detailed information pertaining to the selection of SNPs for each IV are shown in Table S4 [ 27 ]. The details of GRS calculation are described elsewhere [ 28 , 29 ]. To ensure the strength of genetic methodology, we calculated the F-statistics for each individual genetic instrument, with a F-value above 10 indicating that a causal estimate was unlikely to be biased due to weak instruments.…”

Lack of association of genetic variants for diabetic retinopathy in Taiwanese patients with diabetic nephropathy

“…These 33 SNPs were used to calculate the wGRS. Detailed description of the wGRS calculation has been reported previously26 27 ; the formula is as follows: wGRS=33/77.189×((rs767763×2.025)+(rs16958803×1.992)+(rs4129423×2.036)+(rs1559438×1.287)+(rs62324351×1.926)+(rs3791242×2.587)+(rs77625440×2.289)+(rs56170305×1.769)+(rs10055994×5.019)+(rs28610956×1.457)+(rs12076129×2.795)+(rs4665299×1.950)+(rs6433562×2.253)+(rs12991409×2.191)+(rs11889778×2.160)+(rs75759133×1.362)+(rs7374667×1.665)+(rs34766496 ×2.453)+(rs200796238×1.809)+(rs62328468×2.080)+(rs6841985×1.966)+(rs6554985×3.221)+(rs35019626×1.931)+(rs60421526×3.803)+(rs73357792×2.014)+(rs12680033×2.463)+(rs11318592×1.110)+(rs4618795×1.990)+(rs7940618×1.870)+(rs1263663×3.007)+(rs75631519×4.166)+(rs1894151×5.049)+(rs6065597×1.494)). Furthermore, the DR-related wGRS was calculated for each individual among the DN cases and the ERFD cohort.…”

Deep Learning for Automated Diabetic Retinopathy Screening Fused With Heterogeneous Data From EHRs Can Lead to Earlier Referral Decisions

“…4 Moreover, 80% of those suffering from type 2 diabetes develop DR within 10 years. 5 However, DR-related vision impairment and blindness can be prevented if patients receive regular fundus examinations, which can lead to early diagnosis and treatment. 6,7 Poor adherence to these preventative examinations has been observed in Taiwan nevertheless, and this is believed to be due to two primary reasons.…”

“…Several risk factors for DR, as determined by previous cross-sectional studies, can be retrieved from a patient's EHR, such as age, gender, body mass index (BMI), diabetes history, hypertension history, glycosylated hemoglobin (HbA1c), and systolic blood pressure. 5,[14][15][16][17] Additionally, characteristics associated with different countries and ethnicities should be considered so clinicians may better understand and improve the classification results of deep learning models (e.g., the average HbA1c level may differ by race/ethnicity). 18 This broad scope of information can lead to improvements in the performance and robustness of models when compared with current techniques that do not consider EHR information when focusing on specific segments of a population.…”