Table 2

Summary of prospective epidemiological studies examining associations between metabolites and CVD

ReferenceStudy designCases (N)Metabolomics platformMain results
Ganna et al9Prospective multicohort: TwinGene,
ULSAM and PIVUS
131 CHD events in discovery, 282 events in replicationLC-MS untargeted approachLPC 18:1, LPC 18:2, sphingomyelin 28:1 and MG 18:2 associated with CHD
MG 18:2 showed weak evidence for a causal effect on CHD risk using Mendelian randomisation analysis
Stegemann et al27Prospective cohort study: Bruneck study90 CVD eventsLC-MS shotgun lipidomics,
135 lipid species
Individual species of cholesterol esters, LPCs, phosphatidylcholines, phosphatidylethanolamines, sphingomyelins and triacylglycerols were associated with CVD.
Small improvement in discrimination over standard CVD risk factors for disease prediction
Würtz et al30Prospective multicohort:
FINRISK study, SABRE study and British Women’s Health and Heart Study
800 CVD events in discovery,
573 and 368 events in replication
NMR and targeted LC-MS with Metabolon
68 NMR metabolites
Phenylalanine, monounsaturated fatty acids, omega-6 fatty acids, and docosahexaenoic acids were associated with CVD events
Minimal improvement in discrimination over standard CVD risk factors for disease prediction
Ward-Caviness et al8Prospective multicohort: KORA S4, KORA S2 and AGES-REFINEn=67, 112, 87 myocardial infarction cases, respectivelyTargeted Biocrates AbsoluteIDQ p150 Kit LC -MS
188 metabolites
Arginine and LPC (LPC 17:0 and LPC 18:2) associated with myocardial infarction
Metabolites increased the predictive value of the Framingham Risk Score in KORA S4 (from 0.70 to 0.78).
Holmes et al29Nested case–control study: China Kadoorie Biobank912 myocardial infarctions, 1146 ischaemic strokes, 1138 intracerebral haemorrhageNMR
225 metabolites
Triglycerides within all lipoproteins showed similar associations with myocardial infarction and ischaemic stroke, but not with intracerebral haemorrhage.
Glycoprotein acetyls, ketone bodies, glucose and docosahexaenoic acid were associated with all three diseases.
Lind et al22Prospective multicohort: TwinGene, ULSAM and PIVUS177 cases stroke events in discovery,
194 stroke events in replication
LC-MS untargeted approachSphingomyelin 32:1 was associated with incident stroke.
Tzoulaki et al5Cross-sectional and prospective multicohort: MESA, Rotterdam Study and LOLIPOP630 CVD eventsUntargeted NMR25 metabolites of atherosclerosis also associated with CVD, associations attenuated after adjustments for risk factors
  • AGES-REFINE, Age, Gene/Environment Susceptibility (AGES) and the Risk Evaluation For Infarct Estimates (REFINE)-Reykjavik studies; CHD, coronary heart disease; CVD, cardiovascular disease; FINRISK, Finland Cardiovascular Risk Study; KORA, Collaborative Health Research in the Region of Augsburg; LC-MS, liquid chromatography–mass spectrometry; LOLIPOP, London Life Sciences Prospective Population Study; LPC, lysophosphatidylcholines; MESA, Multi-Ethnic Study of Atherosclerosis; MG, monoglyceride; NMR, nuclear magnetic resonance; PIVUS, Prospective Investigation of the Vasculature in Uppsala Seniors; SABRE, Southall And Brent Revisited Study; ULSAM, Uppsala Longitudinal Study of Adult Men.