Subphenotyping Obesity in Pursuit of Personalized Medicine: The Devil is in the Details

Thousands of years ago, when originally recognized in Egypt, China, India, and Greece, diabetes mellitus was a disorder with a single phenotype, sugar in the urine.With the discovery of insulin, Sir Harold Himsworth noted distinct populations that were sensitive and relatively insensitive to insulin, ultimately resulting in the recognition of Types 1 and 2 Diabetes Mellitus (T1DM and T2DM).These subtypes have obvious and major implications for screening, prognosis, prevention, and treatment, as observed in the 1950s in terms of response to sulfonylureas.Phenotyping also led to the recognition that many adults with "adult onset diabetes" in fact had Type 1 diabetes (latent autoimmune diabetes in adults or LADA), recognition of which also had clear therapeutic implications.Since then, it has become clear that the broad category of T2DM is heterogeneous.Distinct features are observed when T2DM-like syndrome occurs during pregnancy, for example (gestational diabetes).The lack of homogeneity is underlined by the fact that over 600 genetic loci contribute to diabetes risk, possibly implying the presence of thousands of subtypes.Currently, the widely recognized subtypes include monogenic diabetes (also known as maturity-onset diabetes of the young or MODY), pancreatic diabetes (such as that seen in chronic pancreatitis or cystic fibrosis), glucocorticoid-induced diabetes in Cushing's syndrome or iatrogenic cases, and iron-related diabetes (as seen in hereditary hemochromatosis or transfusional iron overload).1 The categories listed above, however, account for a relatively small proportion of the large T2DM population, and treatment algorithms still have a significant component of "trial and error."The ideal strategy would be to identify the key pathophysiological factors in a given patient's diabetes and deliver targeted therapy to that factor or pathway.In the absence of this information, however, others have taken a more empirical approach, for example, looking for key phenotypic features or clusters of such features to identify distinct populations that might better respond to a given therapy or may have a different natural history.One group identified subgroups of patients with prediabetes that differed in the risk of progression to T2DM and response to metformin.2 The current field of obesity medicine is in many ways similar to that of T2DM.A few subtypes of obesity have been identified based on fat distribution (eg, visceral vs subcutaneous), and these features predict comorbidities.Rare forms of monogenic obesity have illuminated and/or validated the pathophysiological pathways involved in this disorder.In parallel with the story for diabetes, there are new and effective treatments of obesity, but they are most often used in the same "shotgun" or "trial and error" fashion as antidiabetic drugs.Furthermore, these treatments are expensive and have serious side effects; therefore, better targeting strategies are required.As is the case with diabetes, further subphenotyping of obesity (based on biochemical, anthropomorphic, demographic, and other criteria) is a promising tool that allows us to better target weight loss interventions to those more at risk for complications such as major cardiovascular events (MACE).As two of the many examples, identifying obese subjects with distinct clusters of cardiometabolic risk markers identifies individuals with the most and least chances of decreasing the risk of MACE with intervention, 3 and the same is true for quantifying different patterns of fat deposition by MRI. 4