Differential short- and long-term effects of illness severity on brain functional connectivity in anorexia nervosa Graham W. Redgrave 1 , Suresh Joel 2 , Lauren Moran 3 , Paul G. Unschuld 1 , Nicholas T. Bello 1 , Janelle W. Coughlin 1 , Angela S. Guarda 1 , Ellen Ladenheim 1 , Julie McEntee 1 , Guillermo Verduzco 1 , Timothy H. Moran 1 1 Department of Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine, Baltimore, MD; FM Kirby Center for Functional Brain Imaging, Baltimore, MD; 3 Maryland Psychiatric Research Center, Baltimore, MD Purpose: Disordered functional connectivity of brain networks in anorexia nervosa (AN) may underlie behavioral differences during the performance of relevant cognitive tasks. We sought to detect differences in brain functional connectivity between women with AN and healthy controls (HC) during the performance of an emotional Stroop paradigm, and further to correlate measures of illness severity with functional connectivity in AN . Method: Nine women with AN and 11 HC underwent functional magnetic resonance imaging (fMRI) while performing a previously described emotional Stroop paradigm [1]. The conn toolbox [2] was used to conduct functional connectivity magnetic resonance imaging (fcMRI) analyses comparing network connectivity using seeds from the default mode network (DMN) seed [3] and regions shown to differ between AN and HC in other studies. The DMN consists of large regions that oscillate together when the brain is not occupied with a task. We first compared connectivity between AN and HC. We then examined the effect of body mass index (BMI) and illness duration on network connectivity within the AN group. Results: In the first analysis, 2 of the 4 seed regions demonstrated altered connectivity in patients compared with controls. In the second analysis, 18 of 84 regions exhibited connectivity correlating differently with illness duration than with BMI, suggesting differential effects of acute starvation and long-term illness on brain functional connectivity. Conclusions: In our analyses fcMRI detected group differences and correlated with short- and long-term measures of illness severity. This method may prove useful to detect true differences between AN subtypes, measure treatment response and predict outcome. The results of the within-group analysis help justify early treatment to reduce long-term functional brain changes that may be a result of chronic AN. Participants: All participants provided written consent for this IRB-approved study. Inclusion criteria were women 18-45 yrs able to undergo MRI without history of neurological illness. Patients were SCID diagnosed with restricting (AN-R) or purging (AN-P) subtypes of AN, and medically stable. Controls had a BMI of 20-25, were not dieting or depressed, and had no psychiatric history. Participants completed the Eating Attitudes Test (EAT26)[4], the Beck Depression Inventory (BDI)[5], and other self report measures. Imaging: Images were acquired on a 3.0T Philips MRI scanner. Functional image acquisition occurred while participants performed the task depicted below. Thirty-four contiguous 3 mm thick transaxial image slices were obtained with no gap, parallel to the intercomissural plane. After acquisition of a Magnetization Prepared RApid Gradient Echo (MP-RAGE) structural image, a gradient echo, echo-planar T2-weighted sequence with the following settings was used to obtain functional images: repetition time=2000 ms, echo time-35 ms, flip angle=70 o , field of view=240mm 2 , echo-planar imaging matrix=80X80 (reconstructed to 128X128). 408 functional images were acquired. Image Processing: All analyses were performed using Statistical Parametric Mapping (SPM8, Wellcome Department of Imaging Neuroscience, University College, London, UK) running under MATLAB (The Mathworks, Sherborn, Massachusetts, USA). The structural images were segmented and bias corrected, then normalized to the EPI template. Slice-timing and motion correction were performed on the functional scans, which were then coregistered and normalized to the structural scans. Smoothing was performed on the functional scans with a 6 mm 3 Gaussian kernel at FWHM. Patients (n=9) Controls (n=11) P Mean age, yrs (SD) 26 (5.0) 24 (2.5) NS Number Caucasian 8 9 NS Education, yrs (SD) 15 (2.0) 17 (1.7) NS Mean EAT26 (SD) 41.7 (7.1) 2.0 (2.3) <.001 Mean BDI (SD) 24.1 (8.7) 1.4 (1.7) <.001 Mean BMI, kg/m 2 (SD) 16.4 (1.5) 22.2 (1.5) <.001 Mean Illness Duration, yrs (SD) 9 (5.8) Number with Purging AN 6 In the first analysis functional connectivity in patients versus controls, controlling for body mass index , was measured over all four conditions of interest contrast using the conn toolbox. Analyses computed ROI-to-voxel connectivity over the whole brain, thresholded at p=0.001, with a cluster threshold of p=0.05 false discovery rate to control for multiple comparisons. Seed regions chosen included the four regions of the DMN above, plus bilateral hippocampus, insula, precuneus, striatum, and anterior ventral striatum. Seeds were chosen based on PET and fMRI literature in AN: though there is no unifying model for AN, all of the above regions have been found to differ in patients and controls. fcMRI is a promising biomarker of illness severity AN, able to distinguish between groups and detect correlations with relevant clinical variables. This work is preliminary and requires replication, but it appears that acute starvation and chronic illness effect brain functional connectivity differently. This finding is consistent with clinical impressions that patients with chronic illness are harder to treat and helps provide justification for early treatment intervention in order to minimize the effects of chronic illness on the brain. Future work should explore fcMRI’s ability to predict treatment response, detect phenotypic variants (restricting versus purging), and on the development of multisite imaging studies, since fcMRI has been shown to be robust across sites. 1. Redgrave, G. W., A. Bakker, et al. (2008) Neuroreport19(12): 1181-1185. 2. Fox, M. D., A. Z. Snyder, et al. (2005) Proc NatlAcadSci U S A 102(27): 9673-8. 3. Nieto-Castanon, A., Whitfield-Gabrielli, S. The conn toolbox, v. 12p. (downloaded from http://mit.edu/swg/software.htm). 4. Garner, D. M., M. P. Olmsted, et al. (1982) Psychol Med 12(4): 871-8. 5. Beck, A. T., C. H. Ward, et al. (1961) Arch Gen Psychiatry 4: 561-71. This study was funded in part by the Johns Hopkins General Clinical Research Center (FMKGR090421) and by GWR’s 2007 NARSAD Young Investigator Award. Differential effect of BMI and illness duration on ROI->voxel connectivity within the AN group. This analysis covaries body mass index (BMI) and illness duration using a [0 1 -1, patient illness duration BMI] contrast to assess the differential effect of these two markers of illness severity on functional connectivity. Differences in Functional Connectivity Between Patients and Controls, controlling for body mass index. A: networks with greater connectivity in patients than in controls. B: networks with greater connectivity in controls than in patients. B: Greater Connectivity in Controls vs. Patients 2 networks: red=R insula->BA6, BA24; blue=MPFC->MFG/OFC 4 networks : red, LLP->PCC/Precuneus; green, L Precuneus->R lateral frontal; pink, R Precuneus->L cerebellum, L lateral frontal; blue, PCC- >ACC/DLPFC. A: Greater Connectivity in Patients vs. Controls L AVS->Occipital, R BA 6 R AVS->R BA 10, R IPL/Occipital LLP->PCC/Precuneus PCC->Thal/PCC/Precuneus, DLPFC L Prec->ACC/BA 10 R Prec->ACC/BA 10, PCC R Striatum->Occipital, L Insula, R Parietal/IPL MPFC->OFC/SFG/MFG L Striatum->L Insula, R Prec, BA 3/4/6 L AVS->Occipital, R BA 6 L Hippo->MedFG, MFG LLP->R Frontal MPFC->Thalamus In this analysis, functional connectivity within the AN group was assessed while covarying body mass index (BMI) using a [0 1, patient BMI] contrast that assessed effect of short-term starvation on functional connectivity. A: Clusters shown are those whose connectivity with one of the 14 seeds correlate positively with BMI. Of the 14 seeds, the following have positive clusters in the figure: LLP, MPFC, PCC, L AVS, R AVS, L Precuneus, L Striatum, and R Striatum. B: Clusters shown are those whose connectivity with one of the 14 seeds correlate inversely with BMI. Of the 14 seeds, the following have positive clusters in the figure: MPFC, L Hippocampus. B: Greater Connectivity With Lower BMI A: Greater Connectivity With Higher BMI Functional connectivity measured using a [0 1, patient illness duration] contrast. A: Connectivity with R AVS, L Hippo, L Insula, MPFC, R Precuneus, RLP, and L Striatum correlate positively with illness duration. B: Connectivity with MPFC, PCC, and RLP correlate inversely with Illness Duration. B: Greater Connectivity With Shorter Illness A: Greater Connectivity With Longer Illness Networks Whose Connectivity is More Influenced by BMI Network Connectivity More Influenced by Illness Duration 1000 P t protected. F1000 Posters. Copyr ters. Copyright protected. F1000 Posters. Copyright protected d. F1000 Posters. Copyright protected. F1000 Posters. Copyright protected. F1000 Poste right protected. F1000 Posters. Copyright protected. F1000 Posters. Copyright protected. F100 ht protected. F1000 Posters. Copyright protected. F1000 Posters. Copyright pr F1000 Posters. Copyright protected. F1000 Posters . Copyright protected. F