A randomized, double-blind, placebo-controlled, crossover design was used to study 38 healthy control subjects who were recruited from local and national resources as volunteers for the ‘CBDB/NIMH Sibling Study' (). Written informed consent was obtained from the subjects after complete description of the study, to be part of the drug protocol (# 03-M-0143) approved by the National Institute of Mental Health Institutional Review Board for administration of oral modafinil. All participants underwent a structured clinical interview to rule out an active axis I or axis II diagnosis (DSM-IV) (APA, 1994
) that could potentially bias the results. Exclusion criteria are reported in Supplementary Materials and Methods. Four subjects did not complete the second arm of the study. Several other subjects—varying according to the task—were excluded because of excessive movement or technical problems during either the drug or placebo sessions. The fMRI analysis, therefore, was limited to data from subjects with fMRI data from both days that passed a rigorous quality control check (all fMRI data were individually examined for motion artifacts and we excluded from further analysis data from subjects with excessive inter-scan motion: >2
mm translation, >1.5° rotation on either the placebo or drug session). This resulted in 19 subjects with usable BOLD fMRI data for the face-matching task (FMT), 23 subjects with usable data during the 2-back task, and 11 subjects with usable data for the variable attentional control (VAC) task on both days (the VAC task was added to the protocol half-way through the study) (). Paired t
-tests between placebo and modafinil sessions showed no significant difference in the inter-scan movement parameters (mean and max values all p
>0.1) and the signal-to-noise ratio values for the fMRI time series images (all p
>0.5) for each of the three tasks. There were no differences in behavioral measures as evaluated by Hamilton anxiety scale (HAM-A) (Hamilton, 1959
) and the profile of mood state (POMS) questionnaire (McNair et al, 1992
) between the subjects included in the three fMRI analyses and the subjects excluded (all p
Subjects with Usable Data for fMRI Analyses
A table of random numbers was used to prepare the randomization. Both modafinil and placebo were coded. Coded modafinil (100
mg once daily) was administered orally every morning for 7 days. After a 1-week wash out period following the first arm, subjects who had received coded modafinil during the first arm received coded placebo, while those who started on coded placebo during the first arm received 100
mg of coded modafinil. The order of drug for each task is reported in . The capsules of modafinil and placebo were identical in appearance (pink color) and taste. Side effects were minimal or absent at the dose used and no subject discontinued the protocol because of side effects.
Functional assessment was performed on days 7 and 21. fMRI was started ≈180
min after drug administration (placebo or modafinil), and completed within 4
h after administration. Timing of testing was based on pharmacokinetic data indicating that plasma levels of modafinil peak 2–4
h after oral administration (Robertson and Hellriegel, 2003
). Three hours after drug administration, just before the scan, a blood sample was obtained for serum modafinil levels (measured at Cephalon Inc., Frazer, PA, using high-performance liquid chromatography—Gorman, 2002
). A clinical assessment including blood pressure and heart rate was performed before the scan. Each subject also completed the HAM-A scale and the POMS questionnaire to determine mood, anxiety, and energy on each test day.
Tasks and Data Acquisition
The blocked fMRI paradigm consists of two experimental conditions, an emotional face-matching condition and a sensorimotor control task. The task consisted of five blocks of 30-s duration each. Blocks 1, 3, and 5 were sensorimotor blocks and blocks 2 and 4 were emotion blocks. Each block of either type consisted of six trials, each of 5-s duration. Each trial consisted of the presentation of three images, two in the lower panel and one in the upper panel. In the six trials of each sensorimotor block, the two lower images were of shapes, and the upper panel image was identical to one of the shapes in the lower panel. Subjects responded with button presses (left or right) to indicate which of the two lower panel images matched the upper panel image. In the six trials of each emotion block, the lower panel consisted of two faces, one angry and one fearful, derived from a standard set of pictures of facial affect (Ekman and Friesen, 1976
). The upper panel consisted of one of the two faces shown in the lower panel. Subject responded with button presses (left or right) to indicate which lower panel face matched the face in the upper panel.
The 2-back task consisted of presentation of visual stimuli in which a series of numbers (1–4) were presented randomly every 2000
ms for 500
ms at set locations at the points of a diamond-shaped box. Subjects were asked to encode the currently observed number and simultaneously recall the number observed two times previously, and respond through a MRI compatible button box, which had four buttons arranged in the same configuration as the stimuli presented on the screen. The task was presented as four blocks of control condition (0-back) alternating with four blocks of the 2-back condition.
Each stimulus was composed of arrows of three different sizes pointing either to the right or to the left. Subjects were instructed by a cue word (BIG, MEDIUM, or SMALL) showed above each stimulus to press the right or left button corresponding to the direction of the large, medium, or small arrows. There were three different levels of attentional control: (1) low level of attentional control (LOW): all three sizes of arrows were congruent in direction with each other, and the stimuli were cued with the word BIG. (2) Intermediate level (INT): the big arrow was incongruent in direction to the small and the medium arrows; the cue was either BIG or SMALL. (3) High level (HIGH): the medium-sized arrows were incongruent in direction to the big and the small arrows; the cue was either SMALL or MEDIUM. In addition, a simple bold arrow pointing to either the left or right was used as a sensorimotor control condition of no conflict.
Each stimulus was presented for 800
ms, and the order of the stimuli was randomly distributed across the session (Friston et al, 1999
). The total number of stimuli was 241: 50 HIGH, 68 INT, 57 LOW, and 66 simple bold arrows. A fixation cross-hair was presented during the interstimulus interval, which ranged from 2000 to 6000
Analysis of Imaging Data
All results are reported in MNI coordinates and at a threshold of p
<0.05 corrected for multiple comparisons based on family-wise error (FWE) within region of interests (ROIs) appropriately chosen according to the task (bilateral amygdala ROI for FMT; bilateral PFC ROI—encompassing BAs 46 and 47- for 2-back task, and anterior cingulate—ACC—for VAC task, defined using the Wake Forest University PICKATLAS toolbox, version 2.0 http://www.fmri.wfubmc.edu
) (Supplementary Figure).
We analyzed amygdala/supragenual-subgenual cingulate coupling that is hypothesized to modulate amygdala reactivity through top–down control (Pezawas et al, 2005
). A complete description of the method used to assess functional connectivity is reported in Supplementary Materials and Methods.
Statistical Analysis on Clinical, Behavioral, and Performance Data
Paired t-test (heart rate, FMT performances, 2-back task performance, HAM-A, and POMS) and analysis of variance for repeated measures (blood pressure, VAC task performance) were performed using STATISTICA software (Statsoft Corp., Tulska, OK). Owing to a computer glitch, reaction time (RT) could not be collected for one subject during one of the two sessions for the 2-back task. Similarly, HAM-A scale and POMS questionnaire could not be completed on one of the 2 days in five subjects, therefore analyses were limited on data from subjects that had data available on both days. Modafinil levels were not available for three subjects included in 2-back and VAC task analyses and for five subjects included in FMT analysis. Bonferroni correction was performed for multiple testing.
Correlation Between Serum Modafinil Levels and Behavioral and Neuroimaging Data
To examine the relationship between serum modafinil levels and changes in behavioral measures (HAM-A scale and POMS questionnaire) and brain activity, we performed linear correlation analysis; Pearson's r was used for neuroimaging data, and a Spearman's rho correlation method was used for behavioral data as they did not follow a normal distribution. For behavioral data, correlation was carried out between serum modafinil levels and the difference scores between placebo and modafinil conditions on the HAM-A scale and POMS subscales.
For neuroimaging data, correlation analysis was carried out between serum modafinil levels and the mean signal extracted from a 10-mm radius spherical ROI around the peak voxel obtained from the contrast looking for a main effect of drug for each of the task conditions (amygdala for FMT, PFC, and ACC for 2-back and VAC task; ROI created using MarsBar).