Functional alterations in memory networks in early Alzheimer's disease

Abstract

The hallmark clinical symptom of early Alzheimer's disease (AD) is episodic memory impairment. Recent functional imaging studies suggest that memory function is subserved by a set of distributed networks, which include both the medial temporal lobe (MTL) system and the set of cortical regions collectively referred to as the default network. Specific regions of the default network, in particular, the posteromedial cortices, including the precuneus and posterior cingulate, are selectively vulnerable to early amyloid deposition in AD. These regions are also thought to play a key role in both memory encoding and retrieval, and are strongly functionally connected to the MTL. Multiple functional magnetic resonance imaging (fMRI) studies during memory tasks have revealed alterations in these networks in patients with clinical AD. Similar functional abnormalities have been detected in subjects at-risk for AD, including those with genetic risk and older individuals with mild cognitive impairment. Recently, we and other groups have found evidence of functional alterations in these memory networks even among cognitively intact older individuals with occult amyloid pathology, detected by PET amyloid imaging. Taken together, these findings suggest that the pathophysiological process of AD exerts specific deleterious effects on these distributed memory circuits, even prior to clinical manifestations of significant memory impairment. Interestingly, some of the functional alterations seen in prodromal AD subjects have taken the form of increases in activity relative to baseline, rather than a loss of activity. It remains unclear whether these increases in fMRI activity may be compensatory to maintain memory performance in the setting of early AD pathology or instead, represent evidence of excitotoxicity and impending neuronal failure. Recent studies have also revealed disruption of the intrinsic connectivity of these networks observable even during the resting state in early AD and asymptomatic individuals with high amyloid burden. Research is ongoing to determine if these early network alterations will serve as sensitive predictors of clinical decline, and eventually, as markers of pharmacological response to potential disease-modifying treatments for AD.

Fig. 1

Cortical regions in which activity is increased during successful encoding of new items (activation) in prefrontal, medial, and inferior temporal regions, and in which activity is decreased during successful encoding of new items (deactivation) in the medial and lateral parietal regions. Increased and decreased activity is measured with respect to visual fixation. Regions shown in which task-induced deactivations are present represent key nodes within the default mode network (Sperling et al. 2009)

Fig. 2

Both hippocampal activation (top left) and precuneus/posterior cingulate deactivation (top right) are associated with successful memory encoding in young subjects. Activity in these regions is altered in clinically normal older subjects who performed poorly on associative memory tasks. The bottom panels demonstrate MR signal time courses from these regions for four groups of subjects: low-performing elderly (pink), high-performing elderly (blue), low-performing young (green), and high-performing young (red). Low-performing elderly individuals (pink line in bottom graphs) fail to deactivate the precuneus/posterior cingulate (bottom right) during encoding, and demonstrate increased hippocampal (bottom left) and prefrontal activation for successful but not failed encoding trials. This hippocampal hyperactivity may represent a compensatory response to failure of default network activity (Miller et al. 2008b). (Color figure online)

Fig. 3

Group fMRI map demonstrating posteromedial regions of the default network with significant differences in task-induced deactivation comparing ApoE ε4 carriers to non-carriers. Cognitively intact older ApoE ε4 carriers demonstrate failure of normal default network activity during memory encoding, similar to the pattern reported in clinical Alzheimer’s disease patients (Pihlajamaki et al. 2009)

Fig. 4

Surface maps indicating the location of cortical hubs estimated in 127 healthy young subjects (shown on left) and the pattern of amyloid-β deposition on PiB-PET imaging in 10 patients with clinical Alzheimer’s disease compared to 29 healthy older control subjects (Buckner et al. 2009)

Fig. 5

Surface map indicating the overlap between regions of highest PiB-PET retention and aberrant default network fMRI activity in 35 non-demented older individuals. The panel on right demonstrates similar pattern of aberrant default network activity during memory encoding in patients with Alzheimer’s disease (Panel A) and in non-demented older individuals with high amyloid burden (PiB+ in Panel B) (Sperling et al. 2009)