Hemochromatosis Mutations, Brain Iron Imaging, and Dementia in the UK Biobank Cohort

Background: Brain iron deposition occurs in dementia. In European ancestry populations, the HFE p.C282Y variant can cause iron overload and hemochromatosis, mostly in homozygous males. Objective: To estimate p.C282Y associations with brain MRI features plus incident dementia diagnoses during follow-up in a large community cohort. Methods: UK Biobank participants with follow-up hospitalization records (mean 10.5 years). MRI in 206 p.C282Y homozygotes versus 23,349 without variants, including T2* measures (lower values indicating more iron). Results: European ancestry participants included 2,890 p.C282Y homozygotes. Male p.C282Y homozygotes had lower T2* measures in areas including the putamen, thalamus, and hippocampus, compared to no HFE mutations. Incident dementia was more common in p.C282Y homozygous men (Hazard Ratio HR = 1.83; 95% CI 1.23 to 2.72, p = 0.003), as was delirium. There were no associations in homozygote women or in heterozygotes. Conclusion: Studies are needed of whether early iron reduction prevents or slows related brain pathologies in male HFE p.C282Y homozygotes.

A small number of participants have since withdrawn from the UK Biobank, leaving the final analysis set of 451,186 for analyses.

Disease ascertainment
Disease ascertainment at baseline was by self-reported doctor diagnoses (verified by a trained nurse during the verbal interview), plus diagnoses recorded in linked ICD10 coded inpatient hospital records covering the period 1996 to the date of participant baseline interview.
Incident hospital admissions data were from Hospital Episode Statistics (HES) for England, the Patient Episode Database for Wales (PEDW) and the Scottish Morbidity Record (SMR) [4]. National death certification was also used. In analysis of incident diagnoses based on HES, the date of censoring for each participant was either March 2020 (for participants in England), October 2016 (for participants with Scotland), or February 2016 (for participants in Wales). If the participant died, then this date was used instead.

Selection and analysis of participants for MRI figure
Two UK Biobank participants were chosen to illustrate the differences between C282Y homozygotes and those without any HFE mutations. In the male C282Y homozygous population with complete MRI data in the first 20,000 data release (n=34 male C282Y homozygotes), six were excluded due to having a diagnosis of haemochromatosis (prevalent or incident, either self-reported, from HES, or from the primary care data, where available).
The remaining participants (n=28) had no recorded diagnoses of dementia, mild cognitive impairment, Parkinson's disease, or strokes and were therefore eligible for inclusion. We selected the participant closest to the mean T2* putamen for that group. We then selected a participant of the same age (60 years) with no HFE C282Y or H63D genotypes (and no haemochromatosis, dementia, mild cognitive impairment, Parkinson's disease, or stroke) who was closest to the mean T2* putamen for this control group.
Sequences acquired included T1-weighted structural images, susceptibility weighted images (SWI) and T2-weighted FLAIR images. Whilst the SWI sequences are most sensitive to iron deposition, and has been utilised for quantitative analysis, the image resolution is such that the anatomy is not clearly demonstrated. Therefore, we examined the T2-FLAIR images to provide an example of signal changes seen in the putamen identified in the analysis of T2* SWI parameters and C282Y genotype. The T2-FLAIR images were acquired in the sagittal plane and reconstructed axially. See the UK Biobank brain MRI documentation for more detail regarding image generation and processing, including "de-facing" for anonymization (http://biobank.ctsu.ox.ac.uk/showcase/showcase/docs/brain_mri.pdf).
T2 FLAIR images in NIFTI format were examined by a diagnostic radiographer (Dr Christine J Heales, who has further qualifications in MRI reporting of the head and spine) using the MRIcron (http://www.mricro.com) software. The software used did not enable the baseline to be adjusted so the axial slices were anatomically matched for both the control and case example as closely as possible using the corpus callosum and the lateral ventricles as reference points: see Supplementary Figure 1.
Horizontal blue line shows the axial slice from which the putamen comparison was made. Images provided by UK Biobank © under licence.