Pathophysiological Risk Factors Preceding Incidence of Type 2 Diabetes Subtypes: A Pooled Cohort Study in the United States

OBJECTIVE Understanding the pathophysiology of type 2 diabetes subtypes (severe insulin-deficient diabetes, severe insulin-resistant diabetes, mild obesity-related diabetes, and mild age-related diabetes) before diagnosis can enable precision prevention. We estimated the associations of seven biomarkers with type 2 diabetes subtypes prior to diagnosis. RESEARCH DESIGN AND METHODS We analyzed 69,725 observations from 9,661 adults without diabetes (median follow-up 10 years [range 0–17 years]) from four U.S. cohorts. Time-dependent Cox models estimated cause-specific hazard ratios (HRs) of seven pathophysiological biomarkers (BMI [kg/m2], systolic blood pressure [SBP] [per 10 mmHg], HbA1c [%], LDL cholesterol [per 10 mg/dL], homeostatic indices [HOMA2 of β-cell function (HOMA2-%B) (per 10 units) and HOMA2 of insulin resistance (HOMA2-IR)], and estimated glomerular filtration rate [eGFR] [per 10 mL/min/1.73 m2]) with incidence of type 2 diabetes and subtypes, adjusting for demographic and clinical covariates. Model performance was evaluated using participants’ first and random visits before last follow-up. RESULTS Of the pooled cohort, 1,569 individuals developed type 2 diabetes. BMI (HR 1.03 [95% CI 1.02, 1.04]), SBP (HR 1.09 [95% CI 1.05, 1.13]), HbA1c (HR 2.46 [95% CI 1.73, 3.52]), HOMA2-%B (0.89 [95% CI 0.87, 0.91]), HOMA2-IR (HR 1.92 [95% CI 1.78, 2.07]), and LDL cholesterol (HR 1.02 [95% CI 1.00, 1.04]), but not eGFR (HR 1.02 [95% CI 0.98, 1.07]), were independently associated with diabetes. Associations were similar across subtypes. Models based on single time points demonstrated a high C-index (0.81–0.90) but modest F1 scores (0.26–0.51) and varying sensitivity and calibration (slope 0.22–1.24). CONCLUSIONS Pathophysiological biomarkers were prospectively associated with type 2 diabetes and subtypes, but single-time-point measurements before diagnosis could not reliably distinguish subtypes at diagnosis.