Association between proteomic biomarkers and myocardial fibrosis measured by MRI: the multi-ethnic study of atherosclerosis

Figures

• 

  Fig. 1. Volcano plots for the proteins of interest in interstitial (a) and replacement (b) fibrosis respectively. In the analysis of proteins of interstitial fibrosis, 17 proteins of interest were identified, including plasminogen activator inhibitor 1 (PAI), Cathepsin D (CTSD), Low-density lipoprotein (LDL) receptor, Fatty acid-binding protein 4 (FABP4), Bleomycin (BLM) hydrolase, Proprotein convertase subtilisin/kexin type 9 (PCSK9), Tissue-type plasminogen activator (t-PA), E-selectin (SELE), Retinoic acid receptor responder protein 2 (RARRES2), Cathepsin Z (CTSZ), Interleukin-1 receptor type 2 (IL-1RT2), Integrin b2 (ITGB2), and Scavenger receptor cysteine-rich type 1 protein M130 (CD163). Proteins with higher levels included Insulin-like growth factor-binding protein (IGFBP)-1, IGFBP-2, N-terminal pro-B-type natriuretic peptide (NT-proBNP), and Serum paraoxonase/lactonase 3 (PON3). The full names of proteins that were statistically significant only prior to adjustment for false discovery rate can be found in Supplemental Table S1. Conversely, no proteins were identified as significantly different in levels based on the presence or absence of replacement fibrosis as shown in the…

• 

  Fig. 2. Importance plots of the top 20 proteins associated with interstitial (a) and replacement (b) fibrosis. In the analysis of interstitial fibrosis, only three of the 20 proteins were found to be statistically significant predictors of disease after accounting for demographics and traditional cardiovascular disease risk factors (see statistical methods for variables). These proteins include plasminogen activator inhibitor (PAI), insulin-like growth factor binding protein 1 (IGFBP-1), and N-terminal pro-B-type natriuretic peptide (NT-proBNP). Similar to the volcano plot analysis, no proteins were determined to be of statistical importance when accounting for demographics and cardiovascular disease risk factors with replacement fibrosis.

• 

  Fig. 3. Hierarchical clustering results of protein groupings for interstitial fibrosis for those without (a, ECVq1) and with (b, ECVq4) interstitial fibrosis. The clusters for interstitial fibrosis included proteins that were also found to be statistically significant in the volcano plot analysis noted with a red asterisk and 3 proteins that remained statistically significant using Bayesian additive regression tree noted with a blue asterisk. A total of 5 clusters were created for ECVq1 and shown as the color bar at the bottom of Fig. 3a. Redistribution of the 5 clusters for ECVq4 are shown in the color bar at the bottom of Fig. 3b. In the Sankey plot (c), the same cluster color scheme is used and a shift or absence of a shift in proteins to different clusters is qualitatively shown between those with and without interstitial fibrosis.