T86. DIFFUSION MAGNETIC RESONANCE IMAGING FIBER CLUSTER ANALYSIS OF THE ANATOMIC ORGANIZATION OF FRONTOSTRIATAL STRUCTURAL CONNECTIVITY IN HEALTHY SUBJECTS
James Levitt,Fan Zhang,Mark Vangel,Marek Kubicki,Martha Shenton,Lauren O’Donnell
DOI: https://doi.org/10.1093/schbul/sbz019.366
2019-01-01
Schizophrenia Bulletin
Abstract:Alterations in brain connectivity may underlie neuropsychiatric conditions such as schizophrenia. We here assess the pattern of structural connectivity between the prefrontal cortex (PFC) and striatum in 100 healthy subjects (HSs) from the Human Connectome Project (HCP); age: 22 to 35; sex: 46 females and 54 males. We propose a novel method, using fiber clustering of whole brain diffusion Magnetic Resonance Imaging (dMRI) tractography to assess the organization of frontostriatal structural connectivity, which allows us to quantify the degree of deviation from a topographic, parallel, arrangement. To enable the identification of fiber tract parcels from the prefrontal cortex (C) and the striatum (S), we used a data-driven fiber clustering atlas (Zhang et al, 2018) that allows for a whole brain tractography parcellation into 2000 fiber clusters according to the white matter (WM) anatomy (i.e., fiber geometric trajectory). Then, fiber clusters of interest (i.e., from C to S) from the whole brain WM were identified according to their connected anatomical brain regions. We studied multiple Freesurfer PFC regions including orbital, lateral and medial PFC regions and the caudate. We identified 17 WM fiber clusters that connect C and S. To quantify the topographical relationship of these fiber clusters, we measured the mean distances between the endpoints of the fiber clusters within the prefrontal cortex (i.e., cortical distance) and the mean distances between the endpoints of the corresponding fiber clusters terminating in the striatum (i.e., striatal distance). We analyzed the data in 2 ways. First, we generated a plot (not shown) based on the 17 fiber clusters (with 136 pairs of fiber clusters, yielding 136 data points), showing the relationship between the cortical distances and the corresponding striatal distances of the obtained fiber cluster pairs that connect the prefrontal cortex and the caudate. An exponential model was fit to the data points which was superior to a linear model. We showed that the PFC-striatal WM streamline projection pattern was non-linear, which was driven by the results from 10 cluster pairs. Of note, certain clusters, e.g., cluster 6, originating in pars orbitalis PFC, were significantly over-represented in these 10 cluster pairs. Second, we generated plots (not shown) for each of the 17 cluster pairs. For each cluster, we fit a least squares line for predicting striatal distance from cortical distance, for the distance from that cluster to each of the other 16. Then we performed two-tailed binomial tests for the smaller of the number of 16 cluster pair distances with striatal < cortical distance and the number with cortical < striatal distance. Adjusting for the 17 comparisons, clusters 6 and 8 (originating in pars orbitalis PFC), and cluster 10 (originating from both medial and lateral orbitofrontal PFC) significantly deviated from chance with striatal < cortical distance, i.e., in a convergent pattern (adjusted p-value 0.009). Further, clusters 1 (originating in rostral middle frontal gyrus PFC) and 7 (originating in lateral orbitofrontal PFC) trended towards significance (adjusted p-value = 0.071), again, in a convergent pattern. Using dMRI fiber cluster topography analysis in HSs, we show that the PFC projection pattern between the PFC and the striatum deviates from a topographic, parallel, organization, due to convergence, driven by regionally specific, focal, anatomic clusters coming from limbic (medial and lateral orbitofrontal cortex) PFC and ventral associative (pars orbitalis cortex) PFC. This approach will allow us to test for variation in the pattern of fronto-striatal structural connectivity in other neuropsychiatric conditions such as schizophrenia.