Journal of Ophthalmology (Aug 2017)

Predicting the development of diabetic retinopathy based on identification of rs759853 and rs9640883 in the AKR1B1 gene

  • S.Iu. Mogilevskyy,
  • O. V. Bushuieva

DOI
https://doi.org/10.31288/oftalmolzh2017438
Journal volume & issue
no. 4
pp. 3 – 8

Abstract

Read online

Background: Currently, up to 50% of the risk for developing diabetic retinopathy (DR) is attributed to genetics. Identification of patients predisposed to develop DR could facilitate the development of individualized approaches to prevention and treatment of the disease. Associations of the AKR1B1 SNPs rs759853 and rs9640883 with DR have been reported previously. Purpose: To predict the development of DR based on identification of these SNPs in the AKR1B1 gene. Materials and Methods: Four hundred and nine patients were involved in the study. Group 1 comprised 281 patients without DR, and Group 2 comprised 128 patients with DR. TaqMan Mutation Detection Assays (Thermo Fisher Scientific) were used to analyze polymorphic DNA loci in a Real-Time PCR System (7500 system; Applied Biosystems, Foster City, CA). A multiple logistic regression model was built and analyzed to analyze the relationship of the genotype with the risk for developing DT. Statistical analysis was performed using Statistica 10 (StatSoft, Inc., USA) and SPSS Statistics 22 (IBM Corp., USA). Results: A multiple regression model for predicting the development of DR demonstrated sufficient reliability of the effect of independent variables on the estimated index: -2 ? log(likelihood) = 354.467 (?2=42.877; p<0.001), AUC=0.70±0.03 (95%CI: 0.62-0.76), p=2.6E-09. The highest probability for developing DR was observed in A/A rs759853*G/G rs9640883 haplotype (PDR = 0.610), followed by G/A rs759853*G/G rs9640883 haplotype (PDR = 0.407) and А/A rs759853*G/А rs9640883 haplotype (PDR = 0.389). The lowest probability for developing DR was noted in G/G rs759853*A/A rs9640883 haplotype (P (DR) = 0.047). The estimated probability exceeding a threshold of 0.231 identified the presence of DR, and that below or equal to 0.231 identified the absence of disease with a total prognostic accuracy of 66%. Conclusion: The AKR1B1 rs759853 and rs9640883 can be used for predicting the development of DR. A model for estimating the probability of the disease was proposed.

Keywords