Subphenotyping of prediabetes with a phenotypic tree model and its correlations with outcomes: Insights from two Chinese cohorts

Background and Objectives: To investigate whether stratifying participants with prediabetes using the Scottish tree model could identify their risks of developing diabetes and related outcomes. Methods: The current study enrolled subjects with prediabetes at baseline from the China Health and Retirement Longitudinal Study and the Chinese follow-up cohort, comprising 10,165 and 4,090 participants, respectively. We assigned individuals to the Scottish tree using the mapping function based on nine phenotypic variables, including HbA1c, BMI, blood lipids, liver enzyme, creatinine, and blood pressure. Cox proportional hazards regression models were utilized to assess the relationships between dimensions on the tree and future diabetes, renal and cardiovascular diseases, and death from all causes. Results: Our findings showed that individuals in the upper-right branch of the tree exhibited the increased risk of progression to subclinical atherosclerosis, cardiovascular disease, microalbuminuria, and all-cause mortality, potentially linked to elevated blood pressure, dyslipidemia, and body fat percentage. Individuals in the right part of the tree were more likely to progress to the onset of type 2 diabetes and chronic kidney disease, which was possibly associated with insulin resistance, triglycerides, and blood pressure. Conclusions: The phenotypic tree model could be incorporated into future risk stratification and treatment plans to better benefit individuals with prediabetes.