DiaBar: Predicting type 2 diabetes remission post-metabolic surgery utilizing mRNA expression profiles from...

Area of Science:

• Metabolic research
• Genomics
• Diabetes management

Background:

• Subcutaneous adipose tissue (SAT) is a metabolic organ implicated in type 2 diabetes (T2D) pathogenesis.
• Current methods for predicting diabetes remission after metabolic surgery require improvement in accuracy.
• Gene expression profiles in SAT are hypothesized to offer enhanced predictive capabilities for diabetes remission.

Purpose of the Study:

• To develop and validate a predictive model for diabetes remission following metabolic surgery using SAT gene expression.
• To compare the accuracy of the novel predictive model against existing methods, specifically the DiaRem Score.
• To investigate the potential of specific gene expression profiles (HMGA2, PPARG, ADIPOQ, IL6) in SAT for predicting remission.

Main Methods:

• Retrospective cohort study including patients undergoing metabolic surgery.
• Analysis of gene expression profiles from SAT biopsies (HMGA2, PPARG, ADIPOQ, IL6).
• Development of a classifier (DiaBar) using machine learning models, including multilayer perceptron networks, to predict no, partial, or complete remission based on American Diabetes Association criteria.

Main Results:

• The study included 106 patients (66 with T2D, 40 with prediabetes).
• High rates of remission were observed: 65.1% complete and 18.9% partial remission.
• The DiaBar test, utilizing a multilayer perceptron, achieved 98.0% overall accuracy in predicting remission status, significantly outperforming the DiaRem Score (74.7% accuracy).

Conclusions:

• Gene expression profiles from SAT can be effectively used to predict diabetes remission after metabolic surgery.
• The developed DiaBar test demonstrates superior accuracy compared to the DiaRem score for predicting remission.
• This finding suggests a promising new diagnostic tool for personalized diabetes management post-bariatric surgery.