Quantitative susceptibility mapping MRI reveals a relationship between iron accumulation, CDR score and cognition across the spectrum from healthy aging to Alzheimer’s disease

Abstract

AbstractBackgroundThe objective of this study was to determine if there is a relationship between iron accumulation in the various white and gray matter regions of the brain, as assessed by susceptibility changes, measured with Quantitative Susceptibility MRI (QSM‐MRI), Clinical Dementia Rating (CDR) Score and a Cognitive Measure of Delayed Recall. Iron is an essential element for normal neuronal functioning. The accumulation of brain iron may signal a reduction in neuronal metabolism and/or trigger a neuroinflammatory response with microglia transporting iron to regions where pathology is taking place. This has led to the hypothesis that disruption of brain iron homeostasis contributes to Alzheimer’s disease symptoms.MethodThis study used a cross‐sectional design. MRI scans (T1 MPRAGE, QSM‐MRI with multi‐echo acquisition) were acquired on 57 participants from the Boston University Alzheimer’s Disease Center. Freesurfer 6.0 was used to generate regional brain volumes. The T1 maps from Freesurfer were co‐registered with the QSM scans to get regional brain QSM intensity values for each of the Desikan‐Killiany gray matter regions and Salat white matter regions. Values from the two hemispheres were combined. Associations between QSM values and clinical outcomes were assessed using stepwise ordinal logistic regression (CDR) or multiple regression (Neuropsychological Assessment Battery list learning long delay) in JMP 15 with age, sex and whole brain volume forced into the analyses as control variables.ResultThe ordinal logistic analysis produced a significant model (R Squared = 0.76). Following FDR correction, the volume of the hippocampus combined with susceptibility regions from subcortical gray matter and parietal white matter regions were important contributors to the logistic model. The multiple regression also produced a significant model (R Squared = 0.73). Following FDR correction, susceptibility values from medial temporal gray matter and parietal white matter regions were selected as important contributors to the regression model. Little overlap was found amongst the predictors chosen in these models.ConclusionThese findings support to the notion that a disruption of brain iron homeostasis is related to clinical status and memory function. As such, it may help us to better understand the clinical syndrome of Alzheimer’s disease.