Introduction
Type 2 diabetes mellitus (T2D) is caused by combinations of insulin resistance and β-cell dysfunction.1 Asians with T2D tend to develop diabetes at a lower body mass index (BMI) than Caucasians,2 and a β-cell function defect occurs in the early stage of Asians with T2D.3 Patients with T2D are highly heterogeneous in disease progression, difficulty in glycemic controls, and the risk of developing chronic diabetic complications.4 Untangling the heterogeneity of T2D can improve prediction of clinical outcomes and facilitate precision medicine, thus leading to better care of patients with T2D.5
Ahlqvist et al proposed subgrouping newly diagnosed patients with T2D using six variables in five clusters tested in the All New Diabetics in Scania cohort.6 The parameters used were antiglutamic acid decarboxylase (GAD) antibody, age at diagnosis, baseline BMI, glycated hemoglobin (HbA1c), and homeostatic model evaluation (HOMA) 2-measured insulin resistance and β-cell function calculated by the C peptide. Their five clusters were cluster 1, severe autoimmune diabetes; cluster 2, severe insulin-deficient diabetes (SIDD); cluster 3, severe insulin-resistant diabetes (SIRD); cluster 4, mild obesity-related diabetes (MOD); cluster 5, mild age-related diabetes (MARD). The five clusters had different microvascular complication progression trajectories, thus showing potentially clinically important differences in disease progression and risk of complications between clusters. There was a faster progression of kidney disease and a higher prevalence of non-alcoholic fatty liver disease in the insulin-resistant group (SIRD), while retinopathy was more prevalent in the insulin-deficiency group (SIDD).6 7 In addition, clustering has also been reported to predict treatment response to specific oral antidiabetic drugs.8
This novel clustering was tested in Chinese and US patients9 using data from the China National Diabetes and Metabolic Disorders Study (CNDMDS) and the 1988–1994 National Health and Nutrition Examination Survey (NHANES III) using five variables proposed by Ahlqvist et al but excluding anti-GAD. They were able to subgroup newly diagnosed T2D into four clusters (cluster 2–5 according to the study by Ahlqvist et al). However, the fact that anti-GAD and HOMA are not routinely measured in patients with T2D has limited the generalizability of this method. A recent study using NHANES III data has compared the two clustering methods using five parameters proposed by Ahlqvist et al and using only three simple parameters (age, BMI, HbA1c). The study found that this simple classification, which is accessible in most patients, could be used to identify T2D with several health and mortality risks.10 Even without HOMA measurements, the MARD, MOD and SIDD subgroups can be identified. The SIDD group had significantly higher HbA1c than other subgroups; MARD was older; and MOD had a higher BMI. However, these three simple parameters had difficulty identifying the SIRD group due to the absence of clinical parameters to identify insulin-resistant patients without using HOMA insulin resistance (HOMA2-IR).
Ferrannini et al found that insulin sensitivity declined linearly with BMI. However, the degree of insulin resistance was different among obese subjects.11 The characteristic of dyslipidemia in patients with insulin resistance and metabolic syndrome is hypertriglyceridemia and low plasma high-density lipoprotein cholesterol (HDL-C) levels.12 Furthermore, elevation of plasma triglyceride (TG) to HDL ratio has been observed in patients with T2D who had chronic diabetic complications13 and patients with poorly controlled hypertension.14 Therefore, this study aimed to classify newly diagnosed T2D using variables that are commonly available based on routine clinical parameters and add more parameters (TG, HDL) to help identify patients with insulin resistance, thus resulting in better grouping of patients. The parameters used were age at diagnosis, baseline BMI, HbA1c, TG and plasma HDL-C levels. Furthermore, the response to treatment and the prevalence of chronic complications among different clusters were evaluated. The comparison between the T2D subgroup using five simple parameters and using HOMA2-β and HOMA2-IR was also analyzed.