Bibliografisk note

Funding Information:
O.K. received research funding from Vinnova for collaboration between Astra Zeneca, Takeda, Pelago, and Labcyte. O.K. is also a board member and a co-founder of Medisapiens and Sartar Therapeutics and has received a royalty on patents licensed by Vysis-Abbot. R.S. is currently employed at Crown CRO.

Funding Information:
Among the strongest correlations of the BIN (|?| > 0.6), we detected distinct measurements of the same molecular entity using different assays (i.e., TSH [clinical-PEA], ? = 0.96; cholesterol [clinical-GCMS] ? = 0.82), or associations between different features such as LDL-receptor protein and triglycerides (? = 0.82), or LEP protein and BMI (? = 0.68). The network consisted of 375 nodes and 570 edges (Table S5). Several clinical variables represented hubs with the largest number of associated connections to metabolites and proteins (Figure 5). These hubs were often inversely correlated with modifiable behavioral risk factors, such as wearable-based measures of physical activity (e.g., insulin with intense physical activities and waist circumference with number of steps). Similarly, measures of physical fitness (lower body, abdominal, and upper body) correlated inversely with plasma LEP levels. Notably, the proteins with common genetic defects in familial hypercholesterolemia appeared in the cardiometabolic subnetwork (LDLR, PCSK9, and ApoB). For instance, PCSK9, a pharmacological target of LDL-lowering therapies, correlated positively to several glycerophospholipids and ApoB, but negatively to CMPF (3-carboxy-4-methyl-5-propyl-2-furanpropionic acid), a metabolite that is formed from the consumption of fish oil and may have positive metabolic effects (Prentice et al., 2018). The whole network included 24 GWAS summary scores (Figure S13), and 11 scores concerned hematological measurements (erythrocytes, leukocytes, and thrombocytes). We observed a direct link between the genetic score and the corresponding trait measured in this study, including hematological traits, but also associations between proteins or metabolites and the genetic susceptibility to a disease/trait with cardiometabolic relevance. For example, we observed an inverse association between the genetic risk score for BMI in physically inactive individuals and several FAEs, carnitine and tyrosine. Furthermore, an inverse association was observed between the genetic risk score for abdominal aortic aneurysm and IL-6RA levels, supporting the contribution of IL-6 signaling and inflammation in the pathogenesis of this disease (Paige et al., 2019). These observations indicate that the between-individuals variability may be explained at least partly by personal genetic differences and that these genetic differences exert their impact via specific molecular pathways.

Funding Information:
We thank all the study subjects for their generous participation and commitment. We express our sincerest thanks to Riina Aaltonen, Annette Evokari, Iiro Hietamäki, Anna Seppänen, and Paula Vartiainen for their expert assistance. The DHR program was coordinated and managed by the Center for Health and Technology, University of Oulu, Finland. Genotyping was performed by the Institute for Molecular Medicine Finland (FIMM) Technology Centre, HiLIFE, University of Helsinki, Finland. This study was supported by Academy of Finland, Tekes/Business Finland, Sigrid Jusélius Foundation, and Knut and Alice Wallenberg Foundation.

Publisher Copyright:
© 2021