Abstract 16931: Network Analysis Identifies Partially Distinct Plasma Protein Associations Underlying HFpEF and HFrEF

Introduction: Underlying biological differences predisposing to heart failure with preserved (HFpEF) or reduced (HFrEF) ejection fraction are unclear. We used a network analysis technique with plasma proteomics to uncover protein relationships differing between HFpEF and HFrEF. Methods: Among 4485 HF-free participants in the Atherosclerosis Risk in Communities study at Visit 5 (2011-2013), we used aptamer-based proteomics values (SomaScan v4; 4955 proteins) as inputs for the LIONESS network approach and reconstructed individual protein-protein interaction networks for each participant (4485 networks). Edge weights across these networks were regressed against incident HF to identify protein correlations statistically different between HFrEF vs no HF and HFpEF vs no HF. This resulted in two differential networks, one for HFpEF and one for HFrEF. Protein node pairs with an absolute difference in correlation of 0.45 were visualized. Two-sample Mendelian Randomization (MR) was used to identify potential causal associations between protein nodes and HFpEF or HFrEF using summary statistics from the Million Veterans Program. Results: Mean age was 75±5 years, 58% were women, and 193 HFpEF and 157 HFrEF events occurred at a median follow-up of 7 [IQR 6-8] years. The HFrEF network had 1205 protein nodes, 839 of which were in the largest connected component (LCC, Figure). In contrast, the HFpEF network had 592 proteins, 166 of which comprised the LCC. Only 257 proteins overlap between the two networks. MR identified 10 HFrEF node proteins with potential causal effects on HFrEF with 7 in the LCC, including 2 highly connected nodes (KIAA1549L, ZNF843) previously implicated in cardiac phenotypes. No causal associations were seen for HFpEF. Conclusions: Single-sample proteomics-based networks suggest different protein interactions characterize HFpEF and HFrEF. The HFrEF network was larger, more connected, and identified proteins with potential causal effects.