428-P: Heart Failure (HF) and Its Risk Factors in Long-Duration Type 1 Diabetes (T1D)

Abstract

To assess HF incidence in childhood onset, T1D, and its risk factors, we used 25 year follow-up data from the Epidemiology of Diabetes Complications Study (n=644, baseline age and T1D duration 27 and 19 years respectively). HF was determined by clinic exams, hospitalization, and/or self-reports. In a separate analysis, serum NT-proBNP (electrochemiluminescence immunoassay) was measured in the earliest available sample (n=572; HF diagnostic cutoff of 450 pg/mL). Hazard ratios (HR) per 1 s.d. increase in risk factor levels were estimated with Cox models. Forty-one (6.3%) HF cases (3 baseline prevalent and 38 incident) were identified during follow-up. The 5.9% (38/642) cumulative incidence of HF yielded an incidence density of 2.97/1000 person-years. In the overall population (and not including NT-proBNP), allowing for a wide range of potential baseline confounders (sex, T1D duration, smoke, BMI, hypertension, HbA1c, lipids, WBC, eGFR and albumin excretion), T1D duration (HR=1.7, p=0.003), ever smoking (HR=2.4, p=0.02) and non-HDLc (HR=1.4, p=0.04) were strongly associated with incident HF after backward elimination. In the sub analysis of 572 participants with NT-proBNP, 21 had values >450 pg/mL, 85.6% (18) of whom had end stage renal disease (ESRD) including 3 with concurrent HF. In those without ESRD at the time of NT-proBNP testing (n=551), NT-proBNP significantly predicted incident HF (32 clinically defined events, HR=1.7, p=0.03). Although in a backward elimination model (removal criterion: p≥ 0.2), allowing for other risk factors, NT-proBNP (HR=1.5, p=0.09) became marginal, it was retained in this final model along with T1D duration (HR=1.8, p=0.003), ever smoking (HR=2.6, p=0.01) and non-HDLc (HR=1.4, p=0.05). HF incidence in this long-duration T1D cohort is increased compared to the general population with similar age (e.g., incidence=0.97/1000 pyrs). Diabetes duration, smoking and non-HDLc are strong risk factors of HF; while in those without ESRD, NT-proBNP is a moderate predictor. Disclosure J. Guo: None. T. Costacou: None. T.J. Orchard: Consultant; Self; Boehringer Ingelheim International GmbH. Funding National Institutes of Health