Alterations of the brain functional organization in young-onset Alzheimer’s disease: Is it a matter of “time”?

Abstract

Compared with more typical late-onset Alzheimer’s disease (AD), the mechanisms of young-onset AD (YOAD; age of symptom onset <65 years) remain less understood. Using resting-state functional MRI data and dynamic causal modeling techniques, Sacu et al. demonstrate that individuals with YOAD (amnestic AD or posterior cortical atrophy) exhibit alterations in effective (i.e., directed) connectivity involving key cortical nodes of the brain’s default mode network. Interpretation of the authors’ findings may benefit from considering disease progression as a temporal factor and a framework to conceptualize brain connectivity changes as a dynamic, non-linear process that evolves across disease stages.

Age of symptom onset is one major factor that drives clinical heterogeneity in Alzheimer’s disease (AD).^1,2 Young-onset AD (YOAD) (also commonly referred to as early-onset AD), most typically defined with the age of symptom onset being 65 years or younger, is associated with distinct patterns of molecular pathology, neurodegeneration, and cognitive impairment relative to more typical late-onset AD (LOAD).^3–11 Although individuals with YOAD make up a small fraction of all AD cases (5–10%),¹² they are more likely to experience an aggressive clinical course and delays in diagnosis than individuals with LOAD.¹³ Better characterization of the pathophysiology of YOAD has important implications for diagnosis, prognostication, and treatment of individuals affected by this AD clinical phenotype, which has disproportionately devastating consequences for patients, families, and society.¹⁴

There is now ample evidence suggesting that AD is associated with the disruption of the brain’s intrinsic functional organization.^15–17 Much of the prior studies emphasized the involvement of the default mode network (DMN) of the cerebral cortex, where the earliest accumulation of cerebral amyloid-β is typically observed in AD.^18,19 Greater amyloid deposition was associated with disruption of intrinsic functional connectivity involving key nodes of the DMN (as measured by resting-state functional magnetic resonance imaging [fMRI]) across the symptomatic spectrum of AD.^20–22 Only a few studies have examined alterations of intrinsic functional connectivity in YOAD; these studies yielded mixed results regarding connectivity differences within the DMN between individuals with YOAD and healthy control participants, while also reporting significant effects involving other cortical networks.¹⁶

The results presented by Sacu et al.²³ extend the literature by demonstrating that individuals with YOAD show alterations in effective (i.e., directed) connectivity among key nodes of the DMN relative to healthy controls. One general observation regarding the results of this study is that the directionality of group differences and the nature of influences (i.e., excitatory versus inhibitory) were highly regionally specific, even though all regions of interest were derived from the same independent component representing the dorsal DMN. It is possible that this pattern is related to the localization of AD-related neuropathology and neurodegeneration within the DMN in YOAD. For example, in individuals with early clinical stage YOAD (Clinical Dementia Rating [CDR] ≤ 1), cortical tau and atrophy are prominently localized to posterior nodes of the DMN (e.g., posteromedial cortex, inferior parietal lobule), with minimal involvement of anterior regions including the medial prefrontal cortex, whereas amyloid deposition is more spatially extensive and also observed frontally.^5,10,24,25 This may suggest that greater excitatory connectivity from the medial prefrontal cortex to posterior DMN in amnestic YOAD is driven by amyloid-dependent hyperexcitability rather than mechanisms mediated by tau or neurodegeneration.

In addition, Sacu et al.²³ report intriguing differences in DMN effective connectivity across two AD syndromic phenotypes (amnestic AD and posterior cortical atrophy [PCA]). For instance, differences in effective connectivity involving the posterior DMN nodes (versus controls) were largely dissociable across the two phenotypes. One possibility is that, while the posterior DMN is affected in both amnestic YOAD and PCA, this network exhibits differential vulnerability in these AD clinical phenotypes with respect to its timing. Prior neuroimaging studies of (largely amnestic) YOAD identified AD-related neuropathology accumulation and cortical atrophy of the largest magnitude in the posteromedial cortex and inferior parietal lobule,^3,5,8,10,24 suggesting the role of these regions as disease epicenters in this AD phenotype. In contrast, abnormal tau and atrophy are typically maximally observed in occipital and posterior temporo-parietal regions part of the visual network and dorsal attention network of the cerebral cortex in PCA.^26–30 As the disease progresses in PCA, tau and neurodegeneration spread to anterior regions preferentially connected to these networks (e.g., frontal eye fields).^26,29,31 Anterior progression in the topography of neurodegeneration over time within the posterior cortical regions in PCA has also been reported, characterized by greater involvement of lateral temporo-parietal and medial parietal areas,^29,31,32 which may be in part driven by increased functional connectivity between the visual network and DMN.³³ More research is needed to clarify the differential involvement of functional networks over the disease course and its interaction with syndromic phenotypes of AD. Nonetheless, these possibly distinct patterns of network vulnerability and disease progression in amnestic YOAD and PCA may be useful in contextualizing Sacu and colleagues’ findings.

Sacu et al.²³ acknowledge several limitations of their study that warrant additional discussion. For example, one major issue concerns the lack of information about clinical staging (e.g., CDR scores) in the current sample, which makes it challenging to interpret the observed connectivity differences through the lens of disease progression as discussed above. As with any other biomarkers of AD, it is important to conceptualize brain functional or effective connectivity changes as part of a dynamic, non-linear trajectory that evolves throughout disease progression. This is particularly important for YOAD given that the rate of clinical decline in this population can be faster than that in LOAD.^34,35 It is also conceivable that there may be periods of rapid functional network reconfiguration within a single clinical stage in YOAD, suggesting that brain connectivity dynamics may capture subtle pathophysiological changes before they manifest as detectable clinical transitions. This heterogeneity in the timing and pattern of network disruption may account for the inconsistent and seemingly contradictory findings reported across previous functional connectivity studies in YOAD. Future work examining effective connectivity in YOAD needs to consider disease severity as a factor (both cross-sectionally and longitudinally) and clarify how the level of clinical impairment influences the pattern of network involvement as well as the direction of connectivity changes.

Second, the modest sample size limits the our ability to understand the generalizability of the authors’ findings to larger and more diverse samples of individuals with YOAD. Several datasets are now available to qualified investigators that provide opportunities for replication and extension studies using larger cohorts of YOAD patients, including the Longitudinal Early-onset Alzheimer’s Disease Study (LEADS)³⁶ and the Alzheimer’s Disease Neuroimaging Initiative (ADNI).³⁷ Recent studies leveraging these datasets—with some examining more than 300 YOAD patients—have begun to characterize clinical, neuropsychological, and neuroimaging profiles of YOAD phenotypes with unprecedented specificity.^3–6,10,38 Future studies of effective connectivity in YOAD might benefit from these resources in conducting large-scale and integrative analyses.

Finally, it is essential for future research to incorporate multimodal neuroimaging biomarkers to better characterize and interpret the pattern of effective connectivity changes in YOAD. Prior studies combining resting-state fMRI and PET imaging biomarkers provided compelling support to the hypothesis that intrinsic functional network connectivity is a major driver of cortical spread of amyloid and tau pathology in AD.^{26,28,39–41} Extending these studies, it would be interesting to clarify the role of effective connectivity in driving AD-related pathological spread beyond the contributions of functional connectivity. Examining associations between effective connectivity and molecular pathology would also be useful in testing hypotheses about amyloid-driven hyperexcitability in YOAD discussed by Sacu et al.²³

Overall, the results reported by Sacu et al.²³ have major implications for better understanding early changes to brain functional integrity in YOAD. Additional work addressing the limitations discussed above would further help clarify the utility of fMRI-based effective connectivity for facilitating diagnosis, prognostication, and outcomes monitoring in YOAD. Future work should also aim to clarify molecular and cellular bases of the observed macroscale connectivity changes in YOAD, including how specific proteinopathies lead to the disruption of synaptic transmission and neural circuit dynamics. Understanding these “ground truth” mechanisms would not only improve the interpretation of macroscale connectivity findings, but may also inform the design of clinical trials for disease-modifying therapies in AD from which individuals with YOAD are commonly excluded.