Fig. 1.

Workflow of the study. Top of figure: disease diagnoses from the Danish Patient Registry spanned 1977–2022, while blood samples were collected in 2007–2011. Diagnoses were divided into prevalent cases (i.e., before blood sampling, left side of figure) and incident cases (i.e., after blood sampling, right side of figure), the latter with cases not having a diagnosis before blood sampling in the disease group of interest. For both prevalent and incident cases, statistical analyses were performed of the epigenetic data (EWAS) and of the gene expression data (TWAS), both at the individual-level and at the twin pair-level (the latter reducing confounding induced by genetics and shared environment). The individual-level analyses are depicted in the figure as individuals in pink and white, while the twin pair-level analyses are depicted as two individuals. Bottom of figure: for each of the four disease groups the following statistical analyses were performed: a1–a4: first, the genes with a p value below 0.05 in both analysis of the epigenetic data (EWAS) and in analysis of the gene expression data (TWAS) were found within each test (e.g., LMER depicted in light blue), b1 + b2: then of these overlapping genes, the genes found both at the individual-level and at the twin pair-level were identified (depicted in beige), and lastly, (c1) then of these overlapping genes, the genes found both for prevalent and incident cases were identified (depicted in purple). The b1, b2, and c1 overlaps for each disease group were investigated by gene set enrichment and network interaction analyses, while the overlaps a1–a4 were reported as lists of genes. Abbreviations: (1) disease diagnoses: CD: cerebrovascular diseases, CAD; coronary artery disease, AOCD: arterial and other cardiovascular diseases and DVL: diseases of the veins and lymphatic system (see Sect. “Cardiovascular disease diagnoses from the Danish National Patient Registry” for details), (2) statistical analyses: LMER: linear mixed effect regression analysis (individual-level analysis of prevalent cases), LM: linear regression analysis (twin pair-level analysis of prevalent cases), COX: Cox proportional hazards regression analysis (individual-level analysis of incident cases) and COXSTRATA: stratified Cox proportional hazards regression analysis (twin pair-level analysis of incident cases) (see Sect. “Statistical analyses” for details) and (3) EWAS: epigenome-wide association analysis, and TWAS: transcriptome-wide association analysis