Background. There is evidence of the participation of adipose tissue hormones leptin, adiponectin and resistin in the formation of metabolic disorders in the retina, retinal neovascularization, and diabetic microangiopathy. The development of methods for the mathematical evaluation of the prognosis of diabetic retinopathy (DR) formation with the participation of adipokines is a relevant problem in modern diabetology. Aim. Elaboration of a mathematical model for assessing the prognostic significance of serum leptin, adiponectin and resistin to study the likelihood of developing and progressing DR in patients with type 2 diabetes mellitus (DM). Materials and methods. An open observational single-center one-stage selective study was conducted among patients with type 2 DM and DR. The blood serum concentration of leptin, adiponectin and resistin, HbA1с, lipid metabolism findings were determined, the results of an instrumental examination of the fundus were analyzed. The diagnostic predictive value of serum leptin, adiponectin and resistin was assessed using discriminant analysis. Statistical analyses were conducted using Statistica 9.0 (StatSoft, Tulsa, OK, USA) software. The differences were considered statistically significant at p < 0.05. A model with linear combinations of the serum leptin, adiponectin and resistin, triglyceride (TG), HbA1с, type of antihyperglycemic therapy (oral anti-hyperglycemic medication or insulin therapy) were developed, and, subsequently, formulas for classification-relevant discriminant functions were derived. Results. Fifty-nine patients (107 eyes) with type 2 DM and DR (men and women; mean age, 58.20±0.18 years; mean diabetes duration, 9.19±0.46 years; mean HbA1с 9.10±0.17%) were assigned to the basic group and underwent the study. They were divided into three DR groups based on the stage of DR. When performing the ranking of patients for discriminant analysis, the stage 2 DR group was aggregated with the stage 3 DR group for convenience to form the stage 2 + 3 DR group based on the pathognomonic sign (portents of proliferation or actual proliferation). Anti-diabetic therapy (ADT) included metformin, either alone (type 1 ADT) or in combination with oral anti-hyperglycemic medication (metformin + OAHGM, type 2 ADT) or insulin therapy (metformin + IT, type 3 ADT). Inclusion criteria were informed consent, age above 18 years, presence of T2DM and DR. Exclusion criteria were endocrine or body system disorders leading to obesity (Cushing’s syndrome, hypothyroidism, hypogonadism, polycystic ovarian syndrome, or other endocrine disorders, including hereditary disorders, and hypothalamic obesity), type 1 DM, acute infectious disorders, history of or current cancer, decompensation of comorbidities, mental disorders, treatment with neuroleptics or antidepressants, proteinuria, clinically significant maculopathy, glaucoma or cataract. The study followed the ethical standards stated in the Declaration of Helsinki and was approved by the Local Ethics Committee. The formulas for classification-relevant discriminant functions were derived based on the results of physical examination, imaging and laboratory tests, and subsequent assessment of clinical signs of DM (HbA1с), DR stage and serum leptin, adiponectin, resistin, TG concentrations and taking into account the type of antihyperglycemic therapy. The classification functions (CF) computed based on the variables found from the above developed models provided the basis for predicting the development of DR. The formulas for CF from model are as follows: CF1=0.29•TG + 1.55•HbA1С + 1.81•ADT_Type + 0.04•Leptin + 0,34•Adiponectin + 0,91•Resistin– 13,82. CF2= 0.05•TG + 1.36•HbA1С + 3.01•ADT_Type + 0.08•Leptin + 0,35•Adiponectin + 1,01•Resistin – 15.95. Astep-by-step approach to a diagnostic decision should be used. First, blood samples are tested for serum leptin, adiponectin and resistin, TG, blood HbA1c, and the patient is assigned a code for ADT Type (metformin only, 1; metformin + OAHGM, 2; or metformin + IT, 3). Second, CF1 and CF2 values are calculated based on clinical and laboratory data. Finally, the two values are compared to determine which is greater. The predictive decision is made by selecting the classification function with the greater value. Thus, if CF1 > CF2, the process can be stabilized at this stage given adequate glycemic control (through compensation of carbohydrate metabolism) and body mass control as well as patient compliance. If CF1 < CF2, the pathological process may progress to the next stage or even within stage 3, and there is an urgent need to reduce BMI, and to correct the ADT and the blood lipid profile. Conclusions. The informative value and statistical significance of the model were 71.4% and p=0.040, respectively. Using the formulas, one can determine the probability of progression of DR.
Read full abstract