Predictive models for conversion of prediabetes to diabetes.

Area of Science:

• Endocrinology
• Metabolic Diseases
• Preventive Medicine

Background:

• Prediabetes affects a significant portion of the adult population.
• Understanding the natural progression of prediabetes is crucial for timely intervention.
• Identifying individuals at high risk for conversion to type 2 diabetes is a public health priority.

Purpose of the Study:

• To investigate the natural course of prediabetes.
• To develop and validate predictive models for type 2 diabetes conversion.
• To assess the impact of weight changes on diabetes progression.

Main Methods:

• Retrospective longitudinal study of 2105 adults with prediabetes.
• Multivariate logistic regression and 10-fold cross-validation for model development.
• Analysis of risk factors including family history, sex, blood pressure, glucose levels, HbA1c, and BMI changes (δBMI).

Main Results:

• Diabetes developed in 2.5% of individuals annually.
• Significant predictors of conversion included family history, male sex, elevated blood pressure, plasma glucose, HbA1c, and alanine aminotransferase.
• A risk score model achieved an AUC of 0.80 (0.70-0.87), with 61.8% specificity at 80% sensitivity.
• Conversion rates varied significantly by risk group: 40.1% (high), 18.5% (intermediate), and 5.9% (low).
• Lower conversion rates were observed in individuals with weight loss (δBMI < -0.31kg/m²).

Conclusions:

• Accurate prediction of prediabetes conversion to type 2 diabetes is feasible.
• Weight reduction is associated with a decreased rate of conversion from prediabetes to diabetes.
• The developed risk score can aid in stratifying individuals based on their risk of progressing to diabetes.