IC‐P‐031: Fornix fractional anisotropy and hippocampal volume, alone and in combination, predict conversion from MCI to AD

Diffusion tensor imaging (DTI) is a method of detecting the integrity of white matter fiber bundles within the brain. The fornix is a predominant outflow tract of the hippocampus. A recent study from our group has reported important differences in fornix integrity between normal controls, amnestic mild cognitive impairment (aMCI), and Alzheimer disease (AD) cases. Fornix integrity is also reduced in presymptomatic PS1 mutation carriers compared to non-carriers. Here we examine whether fornix white matter integrity, as measured via fractional anisotropy (FA), predicts conversion to AD among patients with aMCI, and compare this to the predictive value of hippocampal volume measurement. Twenty-three aMCI patients were enrolled in a longitudinal study and followed for up to 2.5 years. Six participants converted to AD during this time. Baseline DTI data were analyzed with DTIStudio, followed by manual identification of the fornix by skilled operators using a standardized protocol with high reliability. Hippocampus volumes were calculated on binary hippocampus images delineated with a template based segmentation method using the large deformation diffeomorphic metric mapping (LDDMM) algorithm. The predictive value of each neuroimaging measure was examined using ROC analyses and logistic regression. Baseline fornix FA levels discriminated between which MCI cases would convert to AD with lower fornix FA being highly predictive of converting to AD (OR = 80; p = 0.029). Analyses for fornix FA revealed a mean area under the curve (AUC) of 0.946 (95% CI: 0.851-1.00) and a cutoff of fornix FA = 0.45 yielding sensitivity of 83.3% and specificity of 94.1%. This was comparable to the predictive value of both the left (AUC = 0.931; 95% CI: 0.813-1.00) and right (AUC = 0.990; 95% CI: 0.963-1.00) hippocampal volumes, with lower volumes being highly predictive of conversion to AD. A combination of both the right hippocampus and fornix measures resulted in an AUC of 1.00. These findings suggest that fornix FA may be may be a useful, discriminate marker of conversion to AD among aMCI cases, especially when combined with hippocampal volume information. Replication in a larger sample is clearly warranted.