Amygdala Abnormalities in First-Degree Relatives of Individuals with Schizophrenia Unmasked

SUPPLEMENTARY MATERIAL

Amygdala abnormalities in first-degree relatives of individuals with schizophrenia unmasked by benzodiazepine challenge

Daniel H. Wolf, MD,PhD1, Theodore D. Satterthwaite, MD,MA1, James Loughead, PhD1, Amy Pinkham, PhD5, Eve Overton, BA1, Mark A. Elliott, PhD2, Gersham W. Dent, PhD4, Mark A. Smith, MD4, Ruben C. Gur, PhD,1,2,3 Raquel E. Gur, MD,PhD1,2

1 Department of Psychiatry, University of Pennsylvania, Philadelphia PA 19104, USA

2 Department of Radiology, University of Pennsylvania, Philadelphia PA 19104, USA

3 Philadelphia Veterans Administration Medical Center, Philadelphia PA 19104, USA

4 AstraZeneca Pharmaceuticals LP, Wilmington DE 19850, USA

5 Department of Psychology, Southern Methodist University, Dallas TX 75275, USA

Contents of Supplementary Materials:

Supplementary Methods

Details on participant assessment,image acquisition, image processing and image analysis.

Supplementary Table S1

Behavioral performance: accuracy and reaction time.

Supplementary Table S2

Significant clusters from voxelwise analysis of emotion identification and memory tasks.

Supplementary Fig. S1

Illustration of acquired brain volume and regions of interest.

Supplementary Fig. S2

Illustration of task activation (vs. fixation baseline) for emotion identification and memory tasks.

Supplementary Fig. S3

Illustration of memory task performance relative to expected chance performance levels.

Supplementary Fig. S4

Illustration of inverse correlation between age and emotion identification performance.

Supplementary Fig. S5

Illustration of correlation between drug-induced reduction in hippocampal activation and drug-induced reductions in memory performance.

Supplementary Fig. S6

Whole brain global perfusion results.

Supplementary References
Supplementary methods

Subject assessment

The clinical evaluation included the Diagnostic Interview for Genetic Studies (DIGS)(Nurnberger et al. 1994) and Family Interview for Genetics Studies (FIGS)(Maxwell 1992). Schizotypy was assessed with the Structured Interview for Schizotypy (SIS)(Kendler et al. 1989) included in the DIGS; total SIS scores were calculated as the sum of global items (SIS scores were unavailable for 1 family member). Anxiety was assessed using the State Trait Anxiety Inventory (STAI)(Spielberger et al. 1983); state anxiety was measured three times each study session: prior to drug administration (baseline), just before scanning, and immediately after the scanning session. Neurocognitive function was assessed with the Computerized Neurocognitive Battery (Gur et al. 2001a; Gur et al. 2001b; Gur et al. 2010). Interviews were administered by trained assessors with demonstrated reliability on symptom scales (reliability criterion 0.90 intraclass correlation). Participants had no current Axis I psychiatric disorders, or any history of a psychotic disorder. Axis II disorders were also exclusionary, with the exception of Cluster A personality disorders for family member participants, as Cluster A personality traits might reflect genetic risk for schizophrenia. Family members had a first-degree relative with DSM-IV Schizophrenia as determined through the FIGS interview, supporting medical records, and when possible a DIGS assessment of the ill family member. Control participants had no family history of a psychotic disorder or bipolar disorder in first- or second-degree relatives. Participants had no history of alcohol or other substance use disorders within the past 6 months, no history of benzodiazepine abuse or dependence at any time, and no recent substance use by history and confirmed by urine drug screen on each study day. Use of alcohol in the 24 hours preceding the study was exclusionary, as was use of psychoactive substances or medications within the prior 2 weeks, except for nicotine and caffeine where subjects maintained regular use to avoid withdrawal effects. Participants had no history of neurological or medical disorders or injuries affecting brain function, or which would contraindicate MRI scanning or benzodiazepine administration.

Imaging acquisition procedures and parameters

Subjects were placed in the scanner supine, earplugs were used to muffle scanner noise, head fixation was ensured by a foam-rubber device mounted on the head coil, and pulse and respiration monitors were attached. Stimuli were rear-projected to the center of the visual field using a PowerLite 7300 video projector (Epson America, Inc.; Long Beach, CA) and viewed through a head coil mounted mirror. Stimulus presentation was synchronized with image acquisition using the Presentation software package (Neurobehavioral Systems, Inc., Albany, CA). Subjects provided responses with a non-ferromagnetic response device (fORP, Current Designs, Inc., Philadelphia, PA) using their dominant hand. Button press responses and response latencies were recorded.

Whole-brain structural data were obtained with a 5-minute magnetization-prepared, rapid acquisition gradient-echo T1-weighted image (MPRAGE, TR 1620ms, TE 3.87 ms, FOV 180x240 mm, matrix 192x256, effective voxel resolution of 1 x 1 x 1mm). Pulsed Arterial Spin-Label (PASL) perfusion MRI was used to measure absolute CBF at rest with eyes open viewing a black screen (Wang et al. 2003; Wong et al. 1998). Forty resting label/control PASL image pairs were acquired with the following parameters: FOV=220 mm, matrix=64X64, TR=4s, TE=17ms, flip angle=90, QUIPSS II (TI1 = 700 ms, TI2 = 1900 ms), 20 slices (6mm thick with 1.2mm gap). BOLD fMRI data was obtained as a slab single-shot gradient-echo (GE) echoplanar sequence using the following parameters: TR/TE=3000/32 ms, FOV=240 mm, matrix= 128 X 128, slice thickness/gap=2/0mm, 30 slices, effective voxel resolution of 1.875 x 1.875 x 2mm. Slices were obtained obliquely (axial/coronal) in order to reduce signal distortion in ventral brain regions. Online geometric distortion correction of the echo-planar images was performed using the DiCo sequence of Maxim Zaitsev (Zaitsev et al. 2004). This sequence implements a point-spread-function mapping method (Zeng and Constable 2002) acquired with a one minute reference scan.

ImScribe automated field-of-view (FOV) prescription

An automated FOV-determining algorithm (ImScribe) was developed by M. Elliott in MATLAB (available at utilizing the coregistration modules from SPM2 ( The program performs rigid-body (intra-subject) or affine (inter-subject) coregistration of a scout localizer to a study-specific template, and returns the optimally matched FOV for the functional images. Template structural and functional scans were acquired with a well-chosen FOV for the functional slab. During each subject’s first fMRI session, the subject’s structural scan (MPRAGE) was imported into the ImScribe program for coregistration to the structural template. The computed affine spatial transformation was then applied to the template functional FOV to determine the optimal FOV prescription in subject space. On the second scan day, the same procedure was used except that the subject’s structural image and functional FOV from the first day was used as the template for rigid-body coregistration. The entire process required <= 5 minutes to execute. To permit comparison of the automated performance to the more typical user-driven method, a functional FOV defined by expert manual placement was also determined and a single repetition functional image (GE-EPI) was acquired for each session. Examination of inter-session and inter-subject FOV overlap revealed clear superiority for the automated ImScribe method.

BOLD image processing

BOLD images were slice-time corrected, motion corrected to median time point reference using tri-linear interpolation (Jenkinson et al. 2002), high-pass filtered (100s), spatially smoothed (4mm FWHM, isotropic), and grand-mean scaled using mean-based intensity normalization. Non-brain areas were removed using BET (Smith 2002). The median functional image was transformed by trilinear interpolation into standard anatomical space (Jenkinson and Smith 2001) using the T1 Montreal Neurological Institute (MNI) template with 2x2x2 mm voxel dimensions; this provided transformation parameters that were later applied to subject-level statistical images for group-level analyses.

Perfusion image processing and analysis

PASL perfusion image pairs were first realigned to the mean image to correct for motion. Forty perfusion-weighted images were then generated via surround subtraction and converted to absolute cerebral blood flow (CBF) following the procedure summarized in (Wong et al. 1998). The resulting series of 40 CBF images was averaged to produce one CBF image per participant.

Quality control assessments and exclusions

All imaging data was subjected to quality control assessments. For BOLD and perfusion data, exclusion for poor-quality images was based on motion (mean relative displacement, MRD) and temporal signal-to-noise ratio (tSNR) of preprocessed data. tSNR was calculated on preprocessed functional data voxelwise as the temporal standard deviation divided by the mean signal, then averaged across voxels. Scans that were outliers (>2SD across all subjects) on both motion and tSNR measures were excluded; this corresponded to tSNR<51 and MRD > 0.25 mm. In all these cases, visual inspection confirmed severe motion-related artifact. As a result, two controls were excluded from all analyses; one additional control was excluded only from the identification task fMRI analysis, and one family member was excluded only from the recognition task fMRI analysis. A different family member lacked complete perfusion data due to technical problems, and was excluded from perfusion analysis.

Region of interest (ROI) definition

All structural ROIs were defined using the probabilistic Harvard-Oxford Structural Atlas in FSL. The bilateral amygdala ROI was thresholded at 75 (% probability); bilateral hippocampus was thresholded at 50. The right fusiform cortex ROI was obtained from the temporal occipital fusiform cortex atlas region thresholded at 25; this fusiform subdivision includes the coordinates reported in the literature for the fusiform face area (Fox et al. 2009) and corresponds closely to the fusiform activation area observed in this study. The strong activation clusters observed in both functional tasks in posterior orbitofrontal cortex extending into anterior insula were not well captured by available anatomical atlas divisions; therefore the bilateral orbitofrontal ROI was defined functionally, by thresholding the group map of activation across all conditions and all subjects at Z>4.26 (p<0.00001) to separate strongly activated clusters.

Supplementary Table S1 Behavioral performance

Error rates at chance performance are 80% for identification task and 66.67% for memory task (see also Fig. S3). Abbreviations: SD standard deviation, s seconds

Supplementary Table S2 Voxelwise image analyses

EMOTION IDENTIFICATION

HEM# VOXZMAXCOORD

Drug > Placebo

Cerebellar vermisB4243.67-2, -72, -16

ThalamusB3943.976, -24, 2

Cerebellar crus R2413.8938, -78, -32

Placebo > Drug----

Control > Family----

Family > Control----

Interaction----

EMOTION MEMORY

HEM# VOXZMAXCOORD

Drug > Placebo----

Placebo > Drug

HippocampusR3943.5622, -6, -24

Lateral occipital cortexL3043.43-34, -88, 8

Inferior frontal gyrusL2083.11-52, 24 18

Control > Family----

Family > Control----

Interaction----

Abbreviations: HEM hemisphere, R right, L left, VOX number of voxels in each significant cluster, ZMAX peak Z statistic value in each cluster, COORD peak voxel x, y, z coordinates in MNI space

Supplementary Fig. S1 Functional imaging data were acquired in an oblique slab that provided high-resolution coverage of ventral brain regions involved in affective processing. Regions of interest included the amygdala (red), hippocampus (blue), orbitofrontal cortex (green), and fusiform cortex (yellow)

Supplementary Fig. S2 Compared to fixation, both identification and memory tasks activated a distributed network including bilateral inferior frontal gyrus, orbitofrontal cortex, anterior insula, thalamus, and visual cortex; bilateral amygdala was robustly task-activated as well

Supplementary Fig. S3 Emotion memory task accuracy is shown by individual subject, separated by group and by drug condition. Lower horizontal dashed line indicates chance accuracy (0.33, or 33% correct responses, 20/60). Upper horizontal dashed line indicates threshold for statistically better-than-chance performance (0.45, or 45% correct responses, 27/60), as determined by one-sample binomial test implemented in STATA, one-tailed p<0.05. Abbreviations: PLC placebo, ALP alprazolam

Supplementary Fig. S4 Inverse relationship between age and emotion identification performance (efficiency = proportion correct responses/median correct reaction time in seconds). Individual subject data is plotted together with a regression line, for each drug condition separately. Corresponding Pearson’s correlations were -0.42 under alprazolam and -0.35 under placebo. Older age related to both lower accuracy and longer reaction times, contributing to the relationship with efficiency (accuracy vs. age: placebo r=-0.34, drug r=-0.27; reaction time vs. age: placebo r=0.32, drug r=0.36). Abbreviations: PLC placebo, ALP alprazolam

Supplementary Fig. S5 Alprazolam-induced changes in memory efficiency correlate with changes in hippocampal activation in the memory task (values are alprazolam minus placebo, so negative numbers indicate a reduction by alprazolam). Data here plotted by subject with marker shape/color indicating group membership; the regression line is fitted to all subjects and corresponds to the Pearson’s correlation value of 0.37 (p=.014) reported in the main text. Positive correlations are present in both groups, more robustly in controls (r=0.47, p=0.018) than family (r=0.20, p=0.411) but without a statistically significant group difference in the strength of the correlations (p=0.385). Abbreviations: CTR controls, FAM family, HC hippocampus

Supplementary Fig. S6 Global perfusion did not vary across groups or by drug condition (PLC, placebo, ALP, alprazolam; p’s >0.6 for all main and interaction effects). This indicates that regional effects of drug in the BOLD data are not driven by nonspecific cerebral perfusion effects. Error bars indicate standard error of the mean

Supplementary references

Fox CJ, Iaria G, Barton JJ (2009) Defining the face processing network: optimization of the functional localizer in fMRI. Hum Brain Mapp 30:1637-1651

Gur RC, Ragland JD, Moberg PJ, Bilker WB, Kohler C, Siegel SJ, Gur RE (2001a) Computerized neurocognitive scanning: II. The profile of schizophrenia. Neuropsychopharmacology 25:777-788

Gur RC, Ragland JD, Moberg PJ, Turner TH, Bilker WB, Kohler C, Siegel SJ, Gur RE (2001b) Computerized neurocognitive scanning: I. Methodology and validation in healthy people. Neuropsychopharmacology 25:766-776

Gur RC, Richard J, Hughett P, Calkins ME, Macy L, Bilker WB, Brensinger C, Gur RE (2010) A cognitive neuroscience-based computerized battery for efficient measurement of individual differences: standardization and initial construct validation. J Neurosci Methods 187:254-262

Jenkinson M, Bannister P, Brady M, Smith S (2002) Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 17:825-841

Jenkinson M, Smith S (2001) A global optimisation method for robust affine registration of brain images. Med Image Anal 5:143-156

Kendler KS, Lieberman JA, Walsh D (1989) The Structured Interview for Schizotypy (SIS): a preliminary report. Schizophr Bull 15:559-571

Maxwell ME (1992) Manual for the FIGS. Clinical Neurogenetics Branch, Intramural Research Program, National Institute of Mental Health, Clinical Neurogenetics Branch, Intramural Research Program, National Institute of Mental Health, Bethesda

Nurnberger JI, Jr., Blehar MC, Kaufmann CA, York-Cooler C, Simpson SG, Harkavy-Friedman J, Severe JB, Malaspina D, Reich T (1994) Diagnostic interview for genetic studies. Rationale, unique features, and training. NIMH Genetics Initiative. Arch Gen Psychiatry 51:849-859; discussion 863-844

Smith SM (2002) Fast robust automated brain extraction. Hum Brain Mapp 17:143-155

Spielberger CD, Gorsuch RC, Lushene RE, Vagg PR, Jacobs GA (1983) Manual for the State-Trait Anxiety Inventory. Consulting Psychologists Press, Consulting Psychologists Press, Palo Alto

Wang J, Aguirre GK, Kimberg DY, Roc AC, Li L, Detre JA (2003) Arterial spin labeling perfusion fMRI with very low task frequency. Magn Reson Med 49:796-802

Wong EC, Buxton RB, Frank LR (1998) Quantitative imaging of perfusion using a single subtraction (QUIPSS and QUIPSS II). Magn Reson Med 39:702-708

Zaitsev M, Hennig J, Speck O (2004) Point spread function mapping with parallel imaging techniques and high acceleration factors: fast, robust, and flexible method for echo-planar imaging distortion correction. Magn Reson Med 52:1156-1166

Zeng H, Constable RT (2002) Image distortion correction in EPI: comparison of field mapping with point spread function mapping. Magn Reson Med 48:137-146

1