Repository logo

Effects of blood donation and iron homeostasis on medium and long term cardiometabolic outcomes

Change log



Background. Despite several decades of research, the relevance of blood donation and iron homeostasis on cardiometabolic outcomes remains unclear.

Objectives. This thesis aims to reliably characterise the relevance of blood donation and iron homeostasis on cardiometabolic outcomes while overcoming some limitations of previous studies.

Data sources (i) 187,725 UK blood donors drawn from the general blood donor population and linked with hospital records; (ii) up to 40,000 blood donors from the INTERVAL trial, including data on ~30 million genetic variants, 6 iron biomarkers, 7 cardiometabolic markers and serial surveys; (iii) up to 890,000 participants from the UK Biobank and 4 external genetic consortia, with data on 21 cardiometabolic outcomes.

Results. In an analysis of prospective observational data from up to 187,725 donors over a maximum of 11.2 years (median 3.3 years), a higher frequency of whole blood donation was associated with a lower risk of multiple cardiometabolic outcomes. However, after omitting the initial 5 years of follow-up (in an attempt to reduce the healthy donor bias) and after correction for false-discovery rate, the only cardiometabolic association that remained significant was risk of pulmonary embolism (hazard ratio 0.48, 95%CI 0.25 to 0.79 when comparing donors that donated >4 times over their donation career vs those who have donated ⩽4 times). These data confirm the importance of accounting for healthy donor effects in blood donor studies.

Analysis of 28,240 donors recruited in the INTERVAL trial, in which blood donors were randomly allocated to different inter-donation intervals (men: 12, 10, and 8 weeks; women: 16, 14, and 12 weeks) over a 2-year period, showed no significant difference in the concentration of lipids, glycaemic indices and inflammation markers. These results indicate that allocation to higher frequency of whole blood donation is unlikely to affect cardiometabolic traits in the short and medium term.

Additionally, genetic analyses were performed to evaluate the possible causal effects of iron homeostasis on cardiometabolic outcomes in the longer term. In genome-wide association analyses using data from 40,210 participants in the INTERVAL study and supplemented by a meta-analysis involving up to 246,139 participants, 62 variants were associated with 4 iron traits (i.e., ferritin, total iron-binding capacity, transferrin saturation, and serum iron). These variants were used to conduct Mendelian randomisation analyses of 21 cardiometabolic outcomes in up to 898,130 participants while accounting for genetic pleiotropy. No association achieved significance after controlling the false-discovery rate, but several nominally significant (P<0.05) associations were noted. Genetically elevated iron traits were nominally associated with greater risk of deep vein thrombosis (odds ratio [OR] per standard deviation increase in ferritin 5.21, 95%CI 1.39 to 19.61; OR for transferrin saturation 1.14, 95%CI 1.02 to 1.28) and hypertension (OR for transferrin saturation 1.16, 95%CI 1.04 to 1.30; OR for serum iron 1.18, 95%CI 1.02 to 1.36). Less robust associations were found with risk of type-2 diabetes and risk of atrial fibrillation. These findings suggest a potential causal role of iron homeostasis on the development of deep vein thrombosis and hypertension.

Conclusion. Analysis of large-scale observational, randomised and genetic data suggests that blood donation and related alterations of iron homeostasis are unlikely to affect major cardiometabolic diseases except for venous thromboembolism and, possibly, hypertension.





Di Angelantonio, Emanuele


Blood donation, iron, epidemiology, genomics, cohort, observational, RCT, GWAS, Mendelian randomisation


Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge