A microRNA-based dynamic risk score for type 1 diabetes.

Area of Science:

• Biomarkers and Diagnostics
• Genomics and Bioinformatics
• Endocrinology and Metabolism

Background:

• Early identification of individuals at high risk for type 1 diabetes (T1D) is critical for timely intervention with disease-delaying therapies.
• Functional beta cell loss is a key characteristic of T1D, making it a target for risk assessment.

Purpose of the Study:

• To develop and validate a microRNA (miRNA)-based dynamic risk score (DRS) for T1D risk stratification across diverse populations.
• To assess the predictive capability of the DRS for T1D, future insulin requirements, and treatment efficacy.

Main Methods:

• Discovery analysis identified 50 miRNAs associated with beta cell loss in T1D.
• A four-context, miRNA-based dynamic risk score (DRS) was developed using multicenter, multiethnic cohorts (n=2,204).
• Generative artificial intelligence enhanced the DRS, which was validated on an independent dataset (n=662) and assessed in a clinical trial.

Main Results:

• The miRNA-based DRS effectively stratified individuals with and without T1D.
• The enhanced DRS demonstrated strong predictive power for T1D stratification (AUC=0.84).
• The DRS accurately predicted exogenous insulin requirements post-islet transplantation and distinguished drug responders from nonresponders in a clinical trial.

Conclusions:

• A novel, AI-enhanced, miRNA-based DRS provides a robust tool for T1D risk stratification.
• The DRS shows potential for predicting disease progression and treatment response, enabling personalized medicine approaches.
• This study highlights the utility of machine learning and miRNA signatures in advancing T1D diagnostics and therapeutic strategies.