Examination of pancreatic beta cell reserve and insulin resistance with FAM19A5, fibroblast growth factor, growth differentiation factor 15, mitochondrial open reading frame of 12s RRNA-C (MOTS-c) biomarkers in patients classified by K-MEANS clustering method

Aim: Obesity can cause insulin resistance and pancreatic beta cell insufficiency. The aim of the present is to classify the obese cases into the groups according to insulin resistance and insulin secretion indexes using K-means clustering method and to investigate the relationship of K-means clustering method with insulin resistance and insulin secretion indices. The secondary aim is to evaluate adipokine (FAM19A5) and mitokine (FGF-21, GDF-15, MOTS-c) levels in K-means clustering group and WHO classification groups. Material and Method: Thirty healthy cases who applied to the Endocrinology outpatient clinic of our hospital and 130 obese or overweight cases with suspected type 2 diabetes (T2DM) were included in the study. The cases were classified into 4 cluster groups according to their insulin resistance and insulin release indices by the K-means clustering method. In addition, the cases were divided into 3 groups as T2DM, prediabetes and normal glucose tolerance (NGT) according to the World Health Organization (WHO) diabetes mellitus classification criteria. FAM19A5, FGF-21, GDF-15 and MOTS-c levels were analyzed in all cases and relationships were evaluated both in between cluster groups and WHO groups. Results: There was no significant difference between the FAM19A5, FGF-21, GDF-15 and MOTS-c levels in K-means cluster groups. Serum MOTS-c was decreased in obese group compared to controls (p=0.05); FGF-21 was increased in obese with T2DM than in obese with NGT (p<0.05), and decreased in obese with NGT than in healthy controls (p<0.05); GDF-15 was found to be increased in T2DM compared to the healthy group (p<0.05). There was no significant difference in FAM19A5 between healthy controls and obese groups. Conclusion: We believe that K-means clustering method is superior to WHO diabetes classification because it gives detailed metabolic data such as insulin resistance and pancreatic beta cell reserve in clarifying the pathogenesis of obesity to T2DM. K-means clustering method enables the creation of personalized treatment protocols by elucidating the individual metabolism data via the insulin resistance and insulin secretion levels of the individuals. Key words: K-Means Clustering, Obesity, Insulin Resistance