Multilayer Network of Cardiovascular Diseases and Depression via Multipartite Projection

TL;DR

A multipartite projection method based on mutual information correlations to construct multilayer disease networks as a novel approach to explore such intricate relationships between CVD and depression, offering a comprehensive, system-level perspective on the biological pathways contributing to comorbidity.

Abstract

Cardiovascular diseases (CVD) and depression exhibit significant comorbidity, which is highly predictive of poor clinical outcomes. Yet, the underlying biological pathways remain challenging to decipher, presumably due to the non-linear associations across multiple mechanisms. In this study, we introduced a multipartite projection method based on mutual information correlations to construct multilayer disease networks as a novel approach to explore such intricate relationships. We applied this method to a cross-sectional dataset from a wave of the Young Finns Study, which includes data on CVD and depression, along with related risk factors and two omics of biomarkers: metabolites and lipids. Rather than directly correlating CVD-related phenotypes and depressive symptoms, we extended the notion of bipartite networks to create a multipartite network, linking these phenotypes and symptoms to intermediate biological variables. Projecting from these intermediate variables results in a weighted multilayer network, where each link between CVD and depression variables is marked by its layer (i.e., metabolome or lipidome). Applying this projection method, we identified potential mediating biomarkers that connect CVD with depression. These biomarkers may therefore play significant roles in the biological pathways underlying CVD-depression comorbidity. Additionally, the network projection highlighted sex and BMI as key risk factors, or confounders, in this comorbidity. Our method is scalable to incorporate any number of omics layers and various disease phenotypes, offering a comprehensive, system-level perspective on the biological pathways contributing to comorbidity.

Figures (5)

• Figure 1: Stylized description of projection method used in the main analysis. The network on the left is a tripartite network, in which blue nodes depict metabolomic variables, green nodes represent lipidomic variables, and nodes in the middle include red nodes representing depressive symptoms, yellow nodes representing CVD-related phenotypes, purple nodes representing risk factors. The figures on the right show the projected multilayer networks on blue and green panels. The blue one shows the metabolomic layer of the projected network. The green one presents the lipidomic layer of the projected network.
• Figure 2: The multipartite projection and weighted projected multilayer disease networks.A: The significant tripartite MI correlation network ($p-value<0.01$), in which variables are partitioned into three groups: metabolites, lipids, and phenotypes (including CVD-related phenotypes, depressive symptoms and related risk factors). B: The metabolomic layer of the projected bipartite network of CVD-related phenotypes and depressive symptoms. C: The lipidomic layer of the projected bipartite network.
• Figure 3: Projected score vs. MI correlation on a log-log scale. The logarithmic projected score and MI correlation are fitted by a linear model. $p$ is the p-value of the slope. $r$ is their Pearson correlation coefficient. This plot shows a roughly linear relationship between the MI correlation and projected score on a log-log scale, confirming the rationality of the multipartite projection method.
• Figure 4: The top 10 significant mediating metabolites and lipids related to CVD-depression comorbidity (all links between CVD phenotypes and depressive symptoms).A: The top ten significant metabolites in terms of mean total contribution score over 20 random imputations. B: The top ten significant lipids in terms of mean total contribution score over 20 random imputations. X-axis is the mean total contribution score and the error bar represents the unbiased standard error of the mean score, over 20 random imputations.
• Figure 5: Projected multilayer network of CVD phenotypes, depressive symptoms, and related risk factors; and the top ten mediating biomarkers related to specific risk-phenotype links.A: The metabolomic layer of the projected network. B: The lipidomic layer of the projected network. Yellow, red, and purple nodes represent CVD phenotypes, depressive symptoms, and risk factors, respectively. C: The top ten mediating metabolites and lipids contributing to the strong projected link between sex ('SP') and changes in appetite ('b18'). Only eight metabolites are in the ranking of metabolites because there are less than 10 metabolites contributing to the link. D: The top 10 mediating metabolites and lipids contributing to the strong projected link between sex ('SP') and the ideal cardiovascular health score ('IdealCVH').