P4663Hyperinsulinemia, hypertension and two clusters of biomarkers predict aortic stenosis in 10,144 Finnish men
Abstract BACKROUND Recent studies show that hypertension predicts aortic stenosis (AS). Other predictors of AS are not established. Purpose To investigate a large panel of biomarkers as predictors of AS in the population-based METSIM cohort. Methods Anthropometric, metabolic and inflammatory biomarkers were measured at baseline in the cohort of 10,197 Finnish men. Subjects with AS at baseline (n=53) were excluded from the analyses. Cases of AS were identified from the medical records. Cox regression analysis was used to identify variables predicting AS. Principal components analysis was applied to investigate clustering of variables predicting AS. Uni- and multivariate logistic regression analysis were used to investigate the clusters of biomarkers as predictors of AS. Results Over a mean follow-up time of 10.8 years, incident AS was diagnosed in 116 (1.1%) men. In Cox regression analysis, fasting plasma insulin (9.8±12.8 in men without AS vs. 14.5±18.2 mU/l in men with incident AS; P=6.13 x 10–6) and systolic blood pressure (138.2±16.7 vs. 146.3±19.4 mmHg; P=3.8 x 10–7) were associated with higher risk of AS (HR 1.44 (95% CI 1.23–1.68); P=4.04 x 10–5 and HR 1.54 (1.30–1.83); P=3.01 x 10–7, respectively). Other biomarkers, which significantly predicted AS were age, body mass index, waist, waist/hip ratio, body fat mass percentage, urine albumin, CRP, blood GHbA1C, fasting plasma glucose and proinsulin, oral glucose tolerance test (OGTT) 30 min plasma insulin and proinsulin, OGTT 120 min plasma insulin and proinsulin, and serum C-peptide. Glucose tolerance (ADA 2003), and insulin resistance and insulin secretion indices based on HOMA were significant predictors of AS. After adjusting for age, the same biomarkers except for OGTT 120 min plasma proinsulin, blood GHbA1C and fasting plasma glucose significantly predicted AS. After exclusion of diabetic subjects, all biomarkers mentioned above except for GHbA1C, fasting plasma glucose and glucose tolerance predicted AS in unadjusted Cox models. Two clusters of risk factors were found in principal components analysis, one consisting of fasting plasma insulin, HOMA insulin resistance index, waist/hip ratio, GHbA1c and CRP, and another consisting of age, systolic blood pressure and urine albumin, explaining 38.33 and 15.37% of the total variance, respectively. In univariate logistic regression analysis both clusters predicted AS (HR 1.35 (1.15–1.59); P=2.47 x 10–4 and HR 1.73 (1.46–2.04); P=1.53 x 10–10, respectively), and were statistically significant when entered in the multivariate model (HR 1.30 (1.11–1.52); P=1.01 x 10–3 and HR 1.69 (1.43–1.99); P=5.84 x 10–10, respectively). Conclusion In the present large-scale population-based study, several biomarkers, particularly hyperinsulinemia and systolic blood pressure, predicted AS. In addition, two clusters of biomarkers, one with high loading on insulin and another on systolic blood pressure, independently predicted AS. Acknowledgement/Funding Kuopio University Hospital ja Academy of Finland