Reduced monoaminergic nuclei MRI signal detectable in pre-symptomatic older adults with future memory decline

Abstract

Evidence from murine models and human post-mortem studies indicates that monoaminergic nuclei undergo degeneration at the pre-symptomatic stage of Alzheimer's disease (AD). Analysing 129 datasets from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and relying on the Clinical Dementia Rating as group-defining instrument, we hypothesised that the MRI signal of monoaminergic nuclei would be a statistically significant predictor of memory decline in participants initially recruited in ADNI as healthy adults. As opposed to a group of cognitively stable participants, participants developing memory decline had reduced signal in the ventral tegmental area at baseline, before any evidence of functional decline emerged. These findings indicate that monoaminergic degeneration predates the onset of memory decline in an AD-centred initiative, with a crucial involvement of very-early changes of a dopaminergic region. This translates into potential informative avenues for pharmacological treatment of pre-symptomatic AD.

Figure 1

Flowchart showing the procedure of sample selection from the ADNI database. The dotted frames identify the three groups of participants included in the study.

Figure 2

Subcortical nuclei investigated in this study and their structural covariance (in alphabetical order), superimposed to the standard Montreal Neurological Institute space.

Figure 3

Signal consistency within each nucleus (A–E) between Timepoint-1 and Timepoint-2. All associations were significant (r > 0.89), supporting test–retest reliability of T1-weighted signal from these small regions. A signal change index was calculated subtracting the signal at Timepoint-2 from the signal at Timepoint-1. This index was plotted (F) to confirm normality of data for longitudinal analyses.

Figure 4

Definition of pontine white matter regions for extraction of T1-weighted signal. (A) Eleven spherical regions (3-mm radius) were created around the following Montreal Neurological Institute coordinates. WM 01: x = − 6, y = − 24, z = − 15; WM 02: x = 6, y = − 24, z = − 15; WM 03: x = − 6, y = − 19, z = − 11; WM 04: x = 6, y = − 19, z = − 11; WM 05: x = − 6, y = − 30, z = − 18; WM 06: x = 6, y = − 30, z = − 18; WM 07: x = 0, y = − 21, z = − 13; WM 08: x = − 9, y = − 11, z = − 13; WM 09: x = 9, y = − 11, z = − 13; WM 10: x = − 9, y = − 20, z = − 8; WM 11: x = 9, y = − 20, z = − 8. These regions were created to keep the overlap to a minimum and cover a large portion of the pons. The DR, LC and VTA are indicated in yellow, light blue and blue, respectively. z slices in the Montreal Neurological Institute space are: − 6, − 8. − 10, − 12, − 14, − 16, − 18, − 20. The white-matter regions shown in each slice are indicated below each image. (B) The graphical representation of the T1-weighted signal from the three monoaminergic nuclei and the eleven white-matter regions (plus the global pons), showing significant differences in intensity.

Figure 5

(A, B) Between-sample comparisons between the group of stable (n = 86) and declining (n = 43) participants at the two timepoints. Graphs indicate arithmetical means and error bars represent one standard deviation. (C) Graphical representation of the mixed ANOVA models. Black and red lines show declining and stable participants, and error bars indicate the standard error of the mean. *p < 0.01 (Bonferroni-corrected threshold of significance).

Figure 6

(A) Area under the receiver-operating characteristic curve quantifying how well Timepoint-1 monoaminergic nuclei’s signal predicts Timepoint-2 group membership. (B, C) Timepoint-1 and Timepoint-2 (respectively) voxel-by-voxel cluster level differences between the two groups (stable participants > declining participants).