Quantifying Patterns of Resting State fMRI Associated with EF Task Performance in Youths

• In Event: Poster Session 08
  In Poster Session: PS 08 Topic - Biological Processes: Neuroscience and Genetics

Abstract

Executive function (EF) task performance has been associated with predictions of academic (Cortés et al., 2019; Best et al., 2011) and general life success in a range of domains. Previous work has revealed that EF task performance is both highly heritable (Engelhardt et al., 2015) and that the neural architecture required for successful EFs is developed by middle childhood (Engelhardt et al., 2019). Given the impact of EFs on individual success, a better understanding of the relationship between brain dynamics such as resting state functional connectivity (RS-FC) and EF ability is of potential utility to the field. Graph metrics (GMs) quantify specific patterns of RS-FC brain network organization. However, many studies linking GMs to EF task performance primarily focus on adult populations. The extent to which GMs calculated on RS-FC can accurately predict EF task performance during development is not currently well understood. Further, recent work (Kruschwitz et al., 2018) suggests that previous results may not replicate after changing key RS-FC considerations (such as sample size and scan length), highlighting the need for a more complete understanding of the impact on analysis choices on studies involving GMs.

This pre-registered work leverages a large, combined dataset of 568 (285 F) youths ages 8.5-17.2 (M=10.3) years-old, to test the ability of GMs (calculated from RS-FC) to predict EF task performance. Additionally, this work aims to highlight the impact of sample size, length of fMRI scan, and general GM tuning settings on our results. Data were collected at UT Austin (N=68) and for the ABCD study (N=500). We calculated RS-FC using a cortical surface parcellation (Gordon et al., 2016) with 333 regions of interest (Figure 1a), which were then used to calculate four GMs of interest (characteristic path length, efficiency, rich-club coefficient, centrality) previously associated with EF performance or general intelligence in adults. Regression analyses, controlling for age, were used to predict EF task performance from the chosen GMs. Our preregistered hypotheses are, 1) that GMs that quantify network integration are a valid and useful tool for predicting EF performance in youths, but that 2) extant mixed results and replication difficulty are due to analytic discrepancies and a lack of guidelines for using GMs with RS-FC data.

At the time of this submission, we have calculated normalized GMs using connectivity between all surface parcels, using all available resting state data per participant. We then predicted task performance scores from the working memory and cognitive switching EF domains. Our initial models are unable to accurately predict EF task performance across all included GMs (Figure 1b). Ongoing analyses will probe the impact of resting state scan length, GM tuning settings, and functional network membership of nodes on these initial results.

This pre-registered work aims to 1) identify organizational patterns of RS-FC that are associated with EF task performance in youths, and 2) assess the impact of RS-FC methodological decisions on these analyses.

Authors

• Damion Demeter, University of Texas at Austin
  Presenting Author

• Mary Abbe Roe, University of Texas at Austin
  Non-Presenting Author

• Evan M. Gordon
  Non-Presenting Author

• Mackenzie E Mitchell, University of North Carolina at Chapel Hill
  Non-Presenting Author

• Tehila Nugiel, University of Texas at Austin
  Non-Presenting Author

• Tyler L Larguinho, University of Texas at Austin
  Non-Presenting Author

• AnnaCarolina Garza, The University of Texas at Austin
  Non-Presenting Author

• Jessica A Church, University of Texas at Austin
  Non-Presenting Author