DOI: 10.1161/circ.148.suppl_1.13339 ISSN: 0009-7322

Abstract 13339: Proteomic Associations of Cardiac Function Trajectories Through Mid- to Late Life

Anne Marie Reimer Jensen, James Ross, Victoria Lamberson, Michael E Hall, Kunihiro Matsushita, Brandon Lennep, Pamela L Lutsey, Tor Biering-Soerensen, Amil M Shah
  • Physiology (medical)
  • Cardiology and Cardiovascular Medicine

Introduction: Using a novel Bayesian nonparametric modeling approach based jointly on LVEF and E/A ratio, we previously identified 3 trajectories of cardiac function through mid- to late life that associated with heighted risk of incident heart failure (HF): trajectory 1 demonstrated failure to increase LVEF with age; trajectory 2 had progressive LVEF decline; and trajectory 3 had steeper than expected rise in E/A ratio. Trajectories 1 and 2 associated with higher risk of HFrEF while trajectory 3 associated with HFpEF. Whether circulating proteins differ between trajectories is unknown.

Methods: We assigned 4,419 HF-free participants with available proteomic data (SomaScan v4) at study Visit 5 of the Atherosclerosis Risk in Communities study to their most likely trajectory (2,201 trajectory 1, 162 trajectory 2, 348 to trajectory 3). Associations of plasma proteins with trajectory membership were assessed using multinominal logistic regression analyses adjusted for age and sex at a false discovery rate <0.05. Two-sample Mendelian randomization (MR) was performed to assess potential causal effects between trajectory-associated proteins and LV volume and LVEF.

Results: Mean age was 75±5 years and 58% were women. Several unique and shared proteins associated with each high-risk trajectory (Figure). MR identified potential causal effects of 13 trajectory-associated proteins on LV volume and LVEF. These included trajectory 3-associated proteins (B3GNT2, MFAP4) with genetically higher levels associated with smaller LV size and higher LVEF, consistent with the observed association with HFpEF. One trajectory 1-associated protein (HADH) had MR association with lower LVEF, consistent with the observed association with HFrEF.

Conclusions: A Bayesian nonparametric model using multidimensional imaging data identifies trajectories of cardiac function with distinct proteomic associations, a subset of which have potential causal effects on cardiac function.

More from our Archive