Describing the effects of iron dyshomeostasis on cognitive function and dementia in UK Biobank

Abstract Background Iron dyshomeostasis is a feature of neurodegenerative disease. In circulating blood, changes in iron are associated with dementia disease phenotypes, while in brain tissue increased iron is associated with dementia progression. In UK Biobank, whole body imaging includes variables derived from the brain swMRI data which are indicative of iron in brain tissue. Complete blood counts per participant include hemoglobin concentration indicative of iron in oxygen transportation. Using the cognitive assessment and participants with all‐cause dementia we aimed to understand how measured changes in iron would affect dementia in a large UK‐based cohort. Method Using cross–sectional analysis at shared time points in UK Biobank data collection we compared both the iron phenotypes with cognitive measures. Firstly, the baseline values at the primary assessment centre visit where 478,082 participants were tested for cognitive function and red blood cell measures. Secondly, at Visit 2 (n = 5861), where brain imaging derived T2* measures and complete blood count (hemoglobin concentration) were captured alongside the repeated cognitive measures. Each linear regression model includes cognitive function (outcome) and iron measures (exposure) with correction for age and sex. Adjusted models include correction for education years, assessment centre and ethnicity. Result We found association between low hemoglobin and all‐cause dementia (beta = ‐0.125, p <0.001). This association was replicated with poorer cognitive function and low haemoglobin analysing cognitive function tests (including verbal and numerical reasoning and reaction time) and mean corpuscular hemoglobin, (p<0.05). Using quantile sets of hemoglobin concentration we found a non‐linear effect, whereby both high and low iron had a negative association on cognitive outcome (verbal reasoning ‐0.029,‐0.019, p’s = 0.002,0.049). Using MRI T2* we found associations between newer cognitive function tests, matrix pattern puzzles and symbol digit substitution (p <0.0001) as well as verbal reasoning (p = 0.015). These cognitive function measures were strongly associated with the T2* hippocampus region. Conclusion We have shown an association between dementia and low hemoglobin concentration. We have demonstrated the complexities of this iron homeostasis relationship, showing how lower cognitive function is associated with both increased and reduced iron concentration in blood and how iron changes in brain effect different cognitive traits.