Early identification of individuals at risk for type 2 diabetes (T2D) is critical for prevention strategies. Investigators evaluated a novel model-derived disposition index without insulin (mDI-woI), calculated using only three glucose measurements from an oral glucose tolerance test (OGTT: 0, 60, and 120 minutes).The study was published in the Diabetes Research and Clinical Practice.

The study included 5,742 healthy Koreans (mean age 51.2 ± 8.6 years; BMI 24.5 ± 3.1 kg/m²) followed biennially for up to 14 years with repeated OGTTs. Predictive performance of baseline mDI-woI was compared with fasting plasma glucose, 1-hour post-load glucose (1h-PG), 2-hour glucose (2h-PG), HbA1c, mean OGTT glucose (mean G), and the oral disposition index (oDI) using area under the receiver operating characteristic curve (AUC-ROC) analyses.

mDI-woI and mean G demonstrated the strongest prediction for incident T2D (AUC = 0.79 each), outperforming fasting plasma glucose (0.67), 1h-PG (0.77), 2h-PG (0.72), HbA1c (0.71), and oDI (0.68; all P < 0.001). Among individuals who progressed to T2D, baseline mDI-woI, mean G, and 1h-PG exceeded thresholds while fasting and 2-hour glucose remained below prediabetes cutoffs. mDI-woI identified risk approximately 4 years earlier than mean G and 4.5 years earlier than 1h-PG.

In this prospective cohort, mDI-woI demonstrated predictive performance for incident type 2 diabetes comparable to mean OGTT glucose and higher than several conventional biomarkers. Elevated mDI-woI values were observed years before traditional glycemic thresholds were exceeded.

As an observational cohort study, findings demonstrate predictive associations rather than causal effects.