Objective To identify clinical phenotypes of type 2 diabetes (T2D) among adults presenting with a first diagnosis of diabetes.

Research design and methods A total of 500 consecutive patients were subject to clinical assessment and laboratory investigations. We used data-driven cluster analysis to identify phenotypes of T2D based on clinical variables and Homeostasis Model Assessment (HOMA2) of insulin sensitivity and beta-cell function estimated from paired fasting blood glucose and specific insulin levels.

Results The cluster analysis identified three statistically different clusters: cluster 1 (high insulin resistance and high beta-cell function group), which included patients with low insulin sensitivity and high beta-cell function; cluster 2 (low insulin resistance and low beta-cell function group), which included patients with high insulin sensitivity but very low beta-cell function; and cluster 3 (high insulin resistance and low beta-cell function group), which included patients with low insulin sensitivity and low beta-cell function. Insulin sensitivity, defined as median HOMA2-S, was progressively increasing from cluster 1 (35.4) to cluster 3 (40.9), to cluster 2 (76) (p<0.001). On the contrary, beta-cell function, defined as median HOMA2-β, was progressively declining from cluster 1 (78.3) to cluster 3 (30), to cluster 2 (22.3) (p<0.001). Clinical and biomarker variables associated with insulin resistance like obesity, abdominal adiposity, fatty liver, and high serum triglycerides were mainly seen in clusters 1 and 3. The highest median hemoglobin A1c value was noted in cluster 2 (88 mmol/mol) and the lowest in cluster 1.

Conclusion Cluster analysis of newly diagnosed T2D in adults has identified three phenotypes based on clinical variables central to the development of diabetes and on specific clinical variables of each phenotype.