Association of Social Isolation and Plasma Metabolites With the Risk of Venous Thromboembolism

BACKGROUND: To explore the association of social isolation with venous thromboembolism (VTE) risk and the mediating role of metabolomics on this association. METHODS: Overall, 236 026 participants free of VTE and with complete metabolomics data at baseline from the UK Biobank were included. Social isolation was constructed based on the frequency of friend/family visits, weekly group activities, and living alone. The study outcome was incident VTE, including deep vein thrombosis or pulmonary embolism. Least absolute shrinkage and selection operator regression was applied to create a social isolation–associated metabolic signature. We estimated hazard ratios of study outcomes in relation to social isolation/metabolic signature and calculated the proportion of social isolation–VTE association mediated by the metabolic signature. RESULTS: During a median follow-up of 12.5 years, 6681 VTE cases were recorded. Social isolation was positively associated with VTE risk (versus without social isolation; hazard ratio, 1.17 [95% CI, 1.09–1.27]). A metabolic signature for social isolation was constructed using 28 metabolites. This metabolic signature showed significant association with VTE risk (per SD increment; hazard ratio, 1.16 [95% CI, 1.13–1.19]) and mediated 11.0% (95% CI, 7.0–19.8) of the increased VTE risk associated with social isolation. 11 metabolites in the metabolic signature also significantly mediated the association between social isolation and VTE risk, with mediation proportions ranging from 0.5% to 2.6%. Similar results were found for incident deep vein thrombosis and pulmonary embolism, respectively. CONCLUSIONS: Social isolation was positively associated with VTE risk. The metabolic signature for social isolation, especially the 11 major metabolites included, significantly mediated the association between social isolation and VTE risk.