Study Design

The present study is a 3-month randomized double-blind study using brain imaging and modafinil 400 mg/day vs placebo (Figure 1). Enrolled cocaine-dependent patients were hospitalized from day 0 to day 17. They were examined twice, before and after 2 weeks of treatment (on days 3 and 17). The participants went through magnetic resonance imaging (MRI)/PET imaging, neuropsychological tests, and clinical and biological assessments (craving and depression rating, blood tests, and urinary toxicology screening tests). From day 4 to day 90, modafinil 100 mg tablets or placebo tablets were orally administered in the morning as follows: four tablets for 26 days, then three tablets for 30 days, and finally two tablets for 31 days. A cocaine abstinence follow-up was carried out from day 17 to day 90 using urinary benzoylecgonine (BE) dosage. Urine samples were collected on days 29, 45, 59, 75, and 90 to assess abstinence after hospitalization. Thereafter the treatment was discontinued. Compliance with medication was measured by pill count using the treatment blisters. Each group was blind to modafinil or placebo until the end of the study. The study was registered as the Cocaine Addiction Imaging Medications and Neurotransmitters (CAIMAN) study, with ClinicalTrials.gov identifier NCT00701532.

Figure 1 Study design and patients’ clinical course. During hospitalization, neuroimaging session, neuropsychological tests (N-back; Iowa Gambling task), clinical assessments (craving and depression ratings), and urinary toxicology screening tests were performed twice. After hospitalization, patients came back every 2 weeks for medical consultation, and urine samples were collected to define abstinence or relapse. The decrease of modafinil dosage occurred on day 30 (400–300 mg) and day 60 (300–200 mg). The horizontal bars represent the clinical course of each participant throughout the study. Full size image Download PowerPoint slide

Participants

The study was approved by the local ethics committee (Comité de Protection des Personnes Ile-de-France 7). Oral and written consent were obtained after complete description of the study. Participants were randomized by the Department of Clinical Research of Paris Sud University.

Twenty-nine cocaine-dependent male patients aged 26–52 years were recruited in the Cocaine Reference Center of Paul Brousse University Hospital, Villejuif (France) over a period of 2 months. All patients met DSM-IV criteria for cocaine dependence, had a positive urine toxicology screen for cocaine in the weeks prior to hospitalization, were seeking treatment, and were willing to give informed consent prior to participation. Exclusion criteria were: a known hypersensitivity to modafinil or modafinil contraindications, any substance-related Axis I disorder (except tobacco dependence), any treatment that interferes with the DAT or modafinil, and MRI contraindications.

Clinical Assessments

Depression was assessed using the Montgomery–Asberg Depression Rating Scale (MADRS) (Montgomery and Asberg, 1979) and the Beck Depression Inventory (BDI) (Beck et al, 1961). Craving was evaluated using a seven-point visual analog scale, the brief version of the Cocaine Craving Questionnaire (10-item CCQ brief) measuring the current craving (Karila et al, 2011b). Cognitive deficits were measured using two neuropsychological tests: Bechara’s Iowa Gambling Task (IGT) (Bechara et al, 1994) assessing decision-making, and a visuospatial N-Back task assessing working memory (Carlson et al, 1998). These tests were carried out in session by a senior neuropsychologist.

Pharmacological Treatments

Modafinil (100 mg tablets) is chemically and pharmacologically distinct from other central nervous system stimulants and is approved for the treatment of sleep disorders. The pharmacokinetics of modafinil is linear, independent of the administered dose. Peak plasma concentrations are reached 2–3 h after its use. The elimination half-life of modafinil is 10–12 h. The pharmacodynamic activity does not seem to affect the autonomic nervous system and the cardiovascular system. The usual dosage for this treatment is 200–400 mg per day as a single dose in the morning. It has numerous side effects and some contraindications (FDA, 2007). The placebo tablets were identical to the modafinil ones. In case of withdrawal symptoms, hydroxyzine 25 mg were administered three times per day.

Image Acquisition and Processing

PET images were acquired on a Siemens ECAT HRRT 3D-PET scanner (CPS innovations Services, Knoxville, TN, USA). Previous studies using [11C]-PE2I and HRRT in healthy subjects reported good suitability and test/retest reproducibility, and better accuracy in quantifying DAT binding, as compared with conventional PET scanners (Hirvonen et al, 2008). [11C]-PE2I was prepared using a TRACERlab FX-C Pro synthesizer (Gems, Velisy, France). The acquisition started with the bolus injection of 300 MBq of [11C]-PE2I and lasted 60 min (20 sequential frames from 1 to 5 min were acquired). Images were reconstructed using the statistical algorithm ordinary Poisson-ordered subset expectation maximization. The voxel size was 1.2 × 1.2 × 1.2 mm3. The injected radioactivity for [11C]-PE2I was 306.3±56.0 MBq in scan 1 (day 3) and 319.46±38.35 MBq in scan 2 (day 17). The specific radioactivity in the two scans was 29.3±12.8 and 28.6±14.0 MBq/μmol, respectively. There was no difference between the groups on days 3 and 17 in injected radioactivity (F(1,26)=0.01, P=0.91 and F(1,21)= 1.33, P=0.26, respectively) or in specific radioactivity (F(1,26)= 0.69, P=0.41 and F(1,21)=1.16, P=0.29, respectively).

MRI was acquired on a 1.5-T Signa scanner (General Electric Healthcare, Milwauke, WI, USA). A T1-weighted sequence was performed using the following MRI parameters: 3D Fourier-transform spoiled-gradient-recalled acquisition with TR=12.5 ms, TE=2.2 ms, 124 contiguous slices of 1.3 mm thickness, a field of view of 24 cm, and a 256 × 256 view matrix, with a voxel size of 0.9 × 0.9 × 1.3 mm3, 16 bits/pixel.

Motion corrections during PET acquisition were carried out using a home-made tool within the BrainVISA/Anatomist software (http://brainvisa.info) consisting in frame by frame co-registration of the PET dynamic series using a mutual information method. Thereafter, in order to process parametric binding potential images, brain regions were determined by T1-MRI automatic parcellation and applied on dynamic co-registered PET images using the PMOD PNEURO tool, version 3.4 (PMOD, Zurich, Switzerland). Time–activity curves obtained from bilateral dorsal caudate and putamen nuclei as high specific binding and crus1 sub-region of the cerebellum as a reference tissue were exported to PMOD’s pixel-wise tool. Parametric maps of the regional [11C]-PE2I non-displaceable binding potential (BP ND ) were generated using Gunn’s basis function method (Gunn et al, 1997), which is closely related to the simplified reference tissue model (Lammertsma and Hume, 1996). Previous studies have confirmed the suitability of specific [11C]-PE2I binding quantification using a compartmental approach with the cerebellum as reference region (Seki et al, 2010). Spatial normalization was applied on the BP ND maps using SPM8 (Statistical Parametric Mapping, Welcome Department of Cognitive Neurology, London, UK) with a ligand-specific [11C]-PE2I template generated according to an MRI-aided procedure (Meyer et al, 1999). The normalized BP ND maps were smoothed using a 10-mm FWHM Gaussian filter. The voxel size was 2 × 2 × 2 mm3.

Statistical Analysis

Among the 29 patients enrolled, two left the study on the first day of hospitalization and were excluded from the analysis. Twenty-seven patients were compared on enrolment on demographic and clinical characteristics using Student’s t-test. In order to assess their course during hospitalization, we performed repeated-measures analyses of variance on clinical and neuropsychological data between day 3 and day 17. These analyses included time (day 3, day 17) as within-subject factor and group (modafinil, placebo) as between-subject factor.

During follow-up, samples containing BE at concentrations 300 ng/ml were considered positive for cocaine. Therapeutic failures were defined by dropouts or at least one cocaine-positive urine sample, and abstinence was defined by negative urine samples. A chi-square test (Fisher’s Exact test) was used to compare the treatment groups in terms of therapeutic failure and abstinence throughout the study for 27 participants. We used post-hoc t-tests on day 3 clinical data to ensure that the therapeutic failures and the abstinent patients were not clinically different at the beginning of the study. JMP 10 (JMP, SAS Institute, Cary, NC, 1989–2007) was used to perform these analyses.

PET images voxel-wise analyses were performed with SPM8. A one-sample t-test of BP ND maps acquired on day 3 was performed to visualize the localization of DAT availability in the brain before treatment and was used to define a mask of analysis. Analyses were conducted within a brain mask defined as all voxels with a minimum value of PET signal, which was 50% superior to the maximum value of the crus1 area (cerebellar reference region). This mask included bilateral striatum, insula, pallidum, claustrum, amygdala, thalamus, midbrain, inferior frontal cortex (gyrus rectus, olfactory cortex and orbitofrontal cortex) and temporal cortex (hippocampus, parahippocampal gyrus).

We compared BP ND maps using a repeated-measures analysis of variance (flexible factorial design in SPM8) with subjects and time as within-subject factors and group as between-subject factor. This analysis was conducted to examine the global group effect, the time effect and the group by time interaction for the patients who were present to the two imaging sessions (N=22). Post-hoc two-sample t-tests were used to specify the differences between treatment groups on day 3 and day 17 separately, with age as confounding covariate. Likewise, post-hoc paired t-tests were used to assess time effect between day 3 and day 17 within the two groups separately.

Correlation analyses were performed between BP ND images and clinical or neuropsychological variables (craving and depression rates, working memory and decision-making scores, and urinary BE amounts) using multiple regression analyses, with age as confounding covariate. Analyses were conducted across all patients on day 3 (n=27) and the two groups on day 17 (n=9 and n=13 for the modafinil and placebo groups, respectively).