Differential Network-Based Dietary Structure and Type 2 Diabetes Risk: A Prospective Cohort Study Using Food...

Area of Science:

• Nutritional Epidemiology
• Network Analysis
• Disease Risk Prediction

Background:

• Traditional dietary pattern methods struggle to identify disease-specific structures.
• Food co-consumption networks offer a novel approach to understanding dietary influences on health.
• Type 2 diabetes (T2D) risk is influenced by complex dietary patterns.

Purpose of the Study:

• To develop and validate network-derived dietary scores for predicting incident type 2 diabetes (T2D) risk.
• To identify disease-specific dietary structures using differential food co-consumption networks.
• To compare network-based dietary assessment with traditional methods.

Main Methods:

• Constructed food co-consumption networks from Korean Genome and Epidemiology Study (KoGES) data, stratified by T2D status.
• Generated a differential co-consumption network-derived (D_CCN) score based on network centrality.
• Validated the D_CCN score's association with T2D risk in independent KoGES cohorts (CAVAS and HEXA) using modified Poisson regression.

Main Results:

• Differential network analysis revealed T2D-specific structures, with simpler networks centered on refined flour foods.
• The D_CCN score was significantly associated with increased T2D risk in both the CAVAS (IRR=1.45) and HEXA (IRR=1.58) cohorts.
• Consistent dose-response relationships were observed, with p-trends < 0.0001.

Conclusions:

• Differential network analysis effectively identifies T2D-specific dietary structures.
• The D_CCN score demonstrates consistent predictive power for T2D risk across different populations.
• Network-based dietary assessment offers a promising advancement beyond traditional methods for disease risk prediction.