Functional connectivity reorganization over age and Alzheimer’s disease

282 

Abstract Submission 

Jonathan Rittmo¹, Laura Wisse², Olof Strandberg³, Nicola Spotorno³, Hamid Behjat³, Danielle van Westen⁴, Sebastian Pamqvist³, Niklas Mattsson-Carlgren³, Shorena Janelidze⁵, Erik Stomrud³, Theodore Satterthwaite⁶, Hansson Oskar³, Jacob Vogel⁷

¹Department of Clinical Sciences Malmö, Faculty of Medicine, SciLifeLab, Lund University, Lund, Sweden, ²Department of Diagnostic Radiology, Clinical Sciences, Lund University, Lund, Sweden, ³Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ, Lund, Sweden, ⁴Department of Diagnostic Radiology, Clinical Sciences, Lund University, Lund, Sweden, Lund, Sweden, ⁵Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of MediLund University, Lund, Sweden, ⁶Penn Lifespan Informatics and Neuroimaging Center (PennLINC), Philadelphia, PA, United States, Philadelphia, PA, ⁷Lund UniversityDepartment of Clinical Sciences Malmö, Faculty of Medicine, SciLifeLab, Lund Universi, Lund, Sweden

Jonathan Rittmo  
Department of Clinical Sciences Malmö, Faculty of Medicine, SciLifeLab, Lund University
Lund, Sweden

Laura Wisse  
Department of Diagnostic Radiology, Clinical Sciences, Lund University
Lund, Sweden

Olof Strandberg  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ
Lund, Sweden

Nicola Spotorno  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ
Lund, Sweden

Hamid Behjat  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ
Lund, Sweden

Danielle van Westen  
Department of Diagnostic Radiology, Clinical Sciences, Lund University, Lund, Sweden
Lund, Sweden

Sebastian Pamqvist  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ
Lund, Sweden

Niklas Mattsson-Carlgren  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ
Lund, Sweden

Shorena Janelidze  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of MediLund University
Lund, Sweden

Erik Stomrud  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ
Lund, Sweden

Theodore Satterthwaite  
Penn Lifespan Informatics and Neuroimaging Center (PennLINC), Philadelphia, PA, United States
Philadelphia, PA

Hansson Oskar  
Clinical Memory Research Unit, Department of Clinical Sciences Malmö, Faculty of Medicine, Lund Univ
Lund, Sweden

Jacob Vogel, PhD  
Lund UniversityDepartment of Clinical Sciences Malmö, Faculty of Medicine, SciLifeLab, Lund Universi
Lund, Sweden

Cognitive aging is a phenomenon that eventually affects most elderly individuals. This process is accelerated in neurodegenerative diseases like Alzheimer's disease (AD), which involve clinical impairment and decline in functional activities of daily living. Aging is accompanied by changes in brain functional network organization, with one of the hallmarks being decrease in system segregation (1). Similarly, nonlinear alterations to functional networks have been described in AD, posited as early neuronal responses to (and perhaps drivers of) AD pathophysiology (2). Functional changes in aging and AD have, however, mostly been studied in isolation, and the degree to which these phenomena interrelate is not well understood. Further, little is known about how variable such changes are in the population. In this exploratory study, we investigate how the brain's functional networks are reorganized at the individual level in aging and AD independently.

The present work uses resting-state functional MRI (rsfMRI) data from the BioFINDER-2 study, encompassing 917 individuals (after quality filtering) with a baseline diagnosis of cognitively unimpaired (Normal, n=390), cognitively unimpaired with amyloid-β (Aβ) positivity (Normal+, n=95), mild cognitive impairment irrespective of Aβ (MCI, n=253) or Alzheimer's disease (AD, n=179). The rsfMRI images were acquired with a 3T scanner and preprocessed using a modified CPAC pipeline (3), including slice-timing correction, motion correction, bandpass filtering, frame censoring, and regression of physiological components, motion parameters and WM/CSF. Subjects were excluded based on mean (>0.3mm) and max (>3mm) frame-displacement. Images from the resulting dataset were smoothed (6 mm FWHM) and mean signal parcellated into 1000 regions as defined by the Schaefer atlas (4) to derive individual functional connectomes. To understand heterogeneity in functional connectivity over both age and AD, inter-subject similarity was estimated by averaging the pairwise Pearson correlation coefficients of each parcel's connectivity map between all subjects within each disease group (5). Personalized network atlases were constructed using an iterative approach of parcel reassignment (6), with the Yeo 2011 (7) atlas as prior. For each parcel, the probability of belonging to each network was estimated using logistic regression, with diagnosis, age and sex as independent variables. Network size was defined as the proportion of parcels belonging to that network for each subject and modeled using generalized additive models with penalized thin plate splines, diagnosis and sex as grouping variables and smoothed over age.

Across the majority of parcels, we observed a decrease in inter-subject similarity as both age and disease status progress, suggesting diverging patterns of age- and AD-related network fragmentation (Fig. 1A). Fig. 1B,C and Fig. 2A,D summarize migration of parcels between networks in AD and aging, while Fig. 2B,C describes changes in network size. In MCI and AD, parcels in sensory networks tended to be recruited by adjacent attention networks. Progression along the AD continuum involved consistent increase in the size of association networks; decrease in limbic network size occurred only during the MCI-AD transition. In contrast, aging was associated with substantial reorganization of the dorsal attention and limbic networks, with the former increasing in size with age and the latter decreasing sharply.

These preliminary findings show a complex landscape of network reorganization associated with AD and aging. The differential reorganization observed in aging and AD may highlight the brain's distinct compensatory responses to neuropathology – but such interpretation should be balanced with the possibility that the alterations might themselves be dysfunctional or pathological. Further understanding of these dynamics could open avenues for targeted (e.g., stimulation-based) interventions and therapies.

Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) ¹

Aging ²

Connectivity (eg. functional, effective, structural)

Segmentation and Parcellation

Task-Independent and Resting-State Analysis

Aging

Degenerative Disease

FUNCTIONAL MRI

Other - Individual parcellation