Synopsis

Connectome predictive analysis was applied to MCI subjects using both resting-state and task data. While a measure of free recall was predicted equally well in rest or task, a measure of total cognition was only predicted succesfully using the task data. This argues for the utility of cognitive "stress tests" to better capture relevant brain biomarkers.

Introduction

The field of fMRI has made remarkable progress in measuring connectivity within and between brain networks, primarily using data collected during the resting (i.e., task-free) state. A recent multivariate technique, Connectome Predictive Modelling (CPM), has shown promise in relating single and multiple modality imaging-derived measures to clinical/behavioral observations in patient populations^1,2,3. However, other work has shown that connectome approaches can be enhanced through application to task instead of resting data^4,5; the theory being that a driven network is more reflective of the associated cognitive/clinical scores.

The object location association (OLA) paradigm was designed to emulate real-world complaints expressed by those with MCI and AD. Prior work revealed hypoactivation of the hippocampus and lateral frontoparietal cognitive control regions in those with MCI relative to cognitively intact older adults⁶. It was recently reported that the continuous measurement of memory was significantly more related to the volume of the entorhinal cortex (i.e., one of the earliest neocortical regions affected by AD) than traditional neuropsychological tests⁷.

In this study we use the ecologically relevant associative memory test to “stress” the memory system as combined with CPM to evaluate the ability to predict behavioral measures of recall and overall cognition. This was compared to using CPM with resting-state data in the same subjects.

Methods

Subjects: 41 participants with MCI were included in this study (age=72.7±7.9, 23M/18F). Behavioral measures included a Free Recall memory for the OLA (i.e., an error score that represented the distance, in cm, between recalled position and actual position), and Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) Total Score.

MRI Acquisition: All functional scans were acquired on a 3T MR750 GE scanner using multiband EPI (MB factor=3). Resting-state scans were acquired with a fixation cross (TR: 900 ms, TE: 30 ms, FOV: 240 mm, 45 axial slices of 3mm thickness, Voxel size = 3.25x3.25x3mm).

OLA task was acquired using published methods⁷: (TR: 1200 ms, TE: 30 ms, FOV: 220 mm, 51 axial slices of 2.5mm thickness, Voxel size = 2.5mm isotropic). Participants complete a total of 2 functional runs (each 6’20” in duration) using a mixed event-related block design (6 active & 7 rest (20”) blocks per run). During active blocks, five stimuli are presented for five seconds each and are separated by an interstimulus interval (ISI) of 1, 2, 3, 4, or 5 seconds, resulting in active block lengths (from 34 – 46 seconds). For this analysis, the first run was analyzed for each subject.

Data analysis: Preprocessing included cardiac and respiration noise removal using RETROICOR⁸, slice timing correction in SPM8, and image registration in FSL. Frames above a 0.5 framewise displacement (FD) threshold⁹ was smoothed between the pre and post frames¹⁰ without any frame deletion (scrubbing). Structural data was co-registered with the functional data in SPM8, then segmented and used for normalization into MNI space using VBM8. The normalization matrix was then applied to the functional time-series image. The resting-state data were then band-passed filtered (0.01-0.10 Hz) to limit the analysis to resting-state frequencies of interest¹¹.

All data was parcellated into 264 regions of interest (ROIs) using the Power atlas¹², with eight additional ROIs selected from the L/R amygdala and hippocampus region. Pearson product-moment correlation coefficients were calculated between average time courses in these 272 spherical ROIs and all other voxels of the brain. These correlation matrices were then transformed to z-scores using a Fisher r-to-z transformation.

CPM analysis was then applied following the protocol prescribed in Shen³. Briefly, the connectivity matrices were thresholded at 0.01 significance, then the raw connectivity values at those locations were summed to give an overall brain score (done separately for positive and negative correlations). A leave-one-out framework was used to predict the composite memory score for each subject, using the rest of the data to generate the linear model. Goodness of fit was measured as the correlation between the predicted and actual measures (Free Recall, RBANS). Significant connections were visualized using the Yale BioImage Suite.

Results

Significant relationships were found between between the CPM brain score and Free Recall for both resting-state and OLA task data (Figure 1).

The CPM brain score was also related to the RBANS metric, but only in the OLA task data (Figure 2).

Discussion

Acknowledgements

References

1. Finn ES, Shen X, Scheinost D, Rosenberg MD, Huang J, Chun MM, Papademetris X, Constable RT. Functional connectome fingerprinting: identifying individuals using patterns of brain connectivity. Nature neuroscience. 2015,18:1664.

2. Lake EM, Finn ES, Noble SM, Vanderwal T, Shen X, Rosenberg MD, Spann MN, Chun MM, Scheinost D, Constable RT. The functional brain organization of an individual predicts measures of social abilities in autism spectrum disorder. bioRxiv. 2018, 290320.

3. Shen X, Finn ES, Scheinost D, Rosenberg MD, Chun MM, Papademetris X, Constable RT. Using connectome-based predictive modeling to predict individual behavior from brain connectivity. nature protocols. 2017, 12:506.

4. Rosenberg MD, Finn ES, Scheinost D, Papademetris X, Shen X, Constable RT, Chun MM. A neuromarker of sustained attention from whole-brain functional connectivity. Nature neuroscience. 2016, 19:165.

5. Greene AS, Gao S, Scheinost D, Constable RT. Task-induced brain state manipulation improves prediction of individual traits. Nature communications. 2018 Jul 18;9(1):2807.

6. Hampstead 2011

7. Hampstead BM, Sathian K, Bikson M, Stringer AY. Combined Mnemonic Strategy Training and High Definition Transcranial Direct Current Stimulation for Memory Deficits in Mild Cognitive Impairment Alzheimer's & Dementia. Translational Research & Clinical Interventions 2017, 3:459.

8. Glover GH, Li TQ, Ress D. Image‐based method for retrospective correction of physiological motion effects in fMRI: RETROICOR. Magnetic Resonance in Medicine. 2000, 44:162.

9. Power JD, Mitra A, Laumann TO, Snyder AZ, Schlaggar BL, Petersen SE. Methods to detect, characterize, and remove motion artifact in resting state fMRI. Neuroimage. 2014, 84:320.

10. Satterthwaite TD, Wolf DH, Loughead J,Ruparel K, Elliott MA, Hakonarson H, Gur RC,Gur RE. Impact of in‐scanner head motion on multiple measures of functional connectivity: Relevance for studies of neurodevelopment in youth. NeuroImage, 2012, 60:623.

11. Cordes D, Haughton VM, Arfanakis K, Wendt GJ, Turski PA, Moritz CH, Quigley MA, Meyerand ME. Mapping functionally related regions of brain with functional connectivity MR imaging. Am J Neuroradiol. 2000, 21:1636.

12. Power JD, Cohen AL, Nelson SM, Wig GS, Barnes KA, Church JA, Vogel AC, Laumann TO, Miezin FM, Schlaggar BL, Petersen SE. Functional network organization of the human brain. Neuron. 2011, 72:665.