Comparison of diffusion weighted, imaging‐derived, fractional anisotropy with a novel MRI contrast‐weighted ratio method for measuring myelin in older adults at risk for Alzheimer’s disease

Abstract

AbstractBackgroundMRI measures sensitive to myelin integrity have been investigated as potential AD biomarkers. T1‐weighted/T2‐weighted ratio is a simple yet informative measure, but many cohorts have transitioned to only T1‐weighted and T2‐weighted Fluid Attenuated Inversion Recovery (FLAIR) acquisitions. We previously obtained similar results using FLAIR instead of T2 (T1w/T2FLAIRw). Our goal now is to compare T1w/T2FLAIRw to fractional anisotropy (FA) from diffusion weighted imaging (DWI).MethodPearson’s correlations of T1w/T2FLAIRw and FA values of hippocampal white matter (WM) integrity were calculated in two independent cohorts acquired on different scanners: Cohort 1 (21 CN; 7 SCD; 20 MCI; 4 AD) and Cohort 2 (40 CN; 43 SCD; 45 MCI; 20 AD). Image‐processing algorithms were designed in‐house, building on previous methods and implemented using SPM12 and FSL dtifit. Diagnostic group differences were assessed in both cohorts using one‐way ANOVA and independent sample T‐tests with Tukey correction.ResultCohort 1 demonstrated significant correlation of WM integrity values between methods in the left hippocampus (p=.002) and approached significance in the right (p = .074). Between‐group analyses showed significant differences in left hippocampal WM between cognitively unimpaired (CU [CN and SCD]) and cognitively impaired (CI [MCI and AD]) groups, with follow‐up tests showing differences between CN and AD patients in both methods. Right hippocampal T1w/T2FLAIRw ratio values showed significant differences between CU and CI, though no significant diagnostic group differences. Right hippocampal FA values were not significant.Cohort 2 demonstrated significant bilateral correlation between imaging methods (p < .001), with bilateral significant differences between CU and CI groups in both methods. CN and MCI right hippocampal values were significantly different from each other in both methods. Box plots demonstrated clear similarities in patterns of diagnostic profile between methods.ConclusionA similar overall pattern was observed across the T1w/T2FLAIRw and DWI‐FA methods. This may reflect myelin disruption although further studies are needed to address source of signal differences on imaging. Differences in FA effect sizes for the hippocampus across samples and scanners merits further exploration.