The goal of the current report was to validate these results using self-report personality measures of inhibitory control. Tellegen’s Multidimensional Personality Questionnaire (MPQ) [6] includes a three-dimensional constraint superfactor, defined as a threshold variable that modulates stimulus elicitation of behavior, affect, and cognition, providing a measure of individual differences in behavioral inhibition [7]. Following the suggestion that a neurobiological basis for personality traits can be identified most clearly by analyzing more homogeneous as compared to heterogeneous traits [7], and to contain Type I (experimenter-wise) error probability, we chose to target the scale that has the highest loadings on the constraint superfactor, the harm avoidance scale. This scale measures the motivational trait that underlies specific approach and avoidance tendencies, linked to basic individual differences in fearfulness of physically dangerous situations; high-scoring persons report restricting themselves to situations that are relatively safe and avoiding physically dangerous activities and situations [6].

Detailed medical and drug use histories, physical, neurological, and psychiatric evaluations were performed for all subjects, including blood test screenings and HIV testing (HIV serology was only performed for the methamphetamine-dependent subjects). Subjects were excluded for history of psychiatric or neurological disorder, history of head trauma, medical illness, and current or past history of dependence on drugs other than nicotine, caffeine, and methamphetamine (dependence on methamphetamine was exclusionary for the comparison subjects). No subject was taking medication at the time of the study, and a prescan urine test was done to ensure the absence of psychoactive drug use in all subjects. Written informed consent was obtained from the subjects after the procedures had been fully explained. The study was approved by the Institutional Review Boards at both Brookhaven National Laboratory and Harbor-UCLA Medical Center.

Personality assessment: The psychometric properties of the MPQ, a factor-analytically developed 240-item instrument [6], are within the well acceptable range. Alpha coefficients for the MPQ harm avoidance scale averaged 0.83 across four samples of college and community men and women. A 30-day test–retest stability coefficient was 0.88 for the harm avoidance scale and 0.89 for the constraint superfactor. External correlations were also high: 0.56 and 0.53 with trait ratings and 0.55 and 0.58 with heritability data, for the harm avoidance scale and constraint superfactor, respectively [6].

The harm avoidance scale and constraint superfactor were highly correlated in both study groups (r = 0.82, p < 0.0001 for the comparison subjects and r = 0.75, p < 0.01 for the methamphetamine-dependent subjects). The harm avoidance scale and constraint superfactor also had similar correlations with the negative emotion (NEM) superfactor (r = −0.53, p < 0.0001 for harm avoidance with NEM; r = −0.54, p < 0.0001 for constraint with NEM across all subjects; these correlations were similar within each study group). There were no significant correlations between the harm avoidance scale or constraint superfactor and the positive emotion superfactor for any of the groups or across all subjects.

PET scanning: PET scans were performed with a CTI 931 scanner (Siemens, Knoxville, TN; 15 slices, spatial resolution: 6 × 6 × 6.5 mm full width at half maximum) in all but two subjects, who were scanned with a Siemens Exact HR+ scanner. Details on procedures for positioning, arterial and venous catheterization, quantification of radiotracer, and transmission and emission scans have been published [9]. Briefly, one 20-min emission scan was taken 35 min after an i.v. injection of 4–6 mCi of FDG. During the study, subjects were kept lying in the PET camera with their eyes open; the room was dimly lit and noise was kept to a minimum. A nurse remained with the subjects throughout the procedure to ensure that the subjects did not fall asleep during the study.

Regions of interest were selected by using a previously published template that locates 115 non-overlapping regions of interest (ROIs) [9]. In brief, we used small ROIs to minimize the contribution of partial volume effects on the metabolic values. The orbitofrontal gyrus averaged 0.7 cm³. The size and orientation of the ROIs were the same in all subjects. Placement of the regions was determined by reference to an atlas of axial tomographic anatomy [10] by an experienced investigator (G.J.W). Three ROIs in the frontal lobe 12 mm below an oblique plane, parallel to the canthomeatal line, were averaged to obtain measures for the orbitofrontal gyrus (Fig. 1). Global metabolism (average metabolism in the 15 planes scanned) was also obtained. Due to the large variability of the absolute regional metabolic measures, we computed ratios of the absolute regional to the global metabolic measures, thus obtaining relative regional measures of metabolism.

The correlation between the MPQ harm avoidance scale and relative orbitofrontal gyrus metabolism was significant for the methamphetamine-dependent subjects (r = 0.57, p < 0.05) but not for the comparison subjects (r = 0.07, p > 0.1). To examine whether our definition of the orbitofrontal gyrus region influenced these correlations, we expanded it to include two more ROIs situated dorsally (2.8 and 3.4 mm above the canthomeatal line, respectively) to the original ROI. The correlation for the methamphetamine-dependent subjects was unchanged (r = 0.50), while for the controls it showed a trend towards a negative association (r = −0.18). When using a different normalization method (to decrease extreme values we divided the absolute metabolism in this expanded region by the metabolism only in gray matter, instead of by the global metabolism, which includes both gray and white matter), the correlation for the methamphetamine-dependent subjects was again unchanged (r = 0.56, p < 0.05). The negative correlation for the comparison subjects increased even further (r = −0.25) although it still did not reach the nominal significance threshold (Fig. 2). The magnitude of this correlation increased even further (r = −0.39, p = 0.087), when selecting the 14 comparison subjects that best matched the methamphetamine subjects on age and education.

To examine whether the two regression lines that best fit the data for the methamphetamine-dependent subjects vs controls are significantly different, we performed a test for their coincidence. Results revealed that the null hypothesis specifying that the two regression lines are coincident, that is, have the same slope and intercept, was rejected with almost 95% confidence (F = 3.3, df = 2,32; p = 0.051).

Similar results were documented for the constraint superfactor; its positive correlation with relative orbitofrontal gyrus metabolism approached significance for the methamphetamine-dependent subjects (r = 0.40, p < 0.1) but not for the comparison subjects (r = −0.02, p > 0.1). These correlations increased for the methamphetamine-dependent subjects (r = 0.46, p = 0.05) and controls (r = −0.29, p = 0.09) when expanding the definition of the orbitofrontal gyrus as described above.

Overall, these results are consistent with the role of the orbitofrontal cortex in evaluating response–reinforcement relationships; patients with orbitofrontal lesions show perseverative touching of a previously rewarded stimulus, unable to change their behavior in response to a change in reward contingencies [11]. Similarly, neuroimaging studies have documented orbitofrontal hypermetabolism in individuals with obsessive–compulsive disorder (for review see [12]) and Tourette syndrome [13], disorders that are characterized by perseverative behavior or the decreased ability to respond appropriately to changing contingencies in the environment. Intriguingly, a recent functional magnetic resonance imaging study demonstrated hypoactivity of the orbitofrontal cortex during aversive conditioning in criminal psychopathy and hyperactivity in social phobia, implicating under-responsiveness of the fear circuitry in the former and over-responsiveness in the latter [14]. A positive correlation between metabolism in the orbitofrontal gyrus at resting baseline and a self-reported trait of behavioral constraint and harm avoidance (fear) in the methamphetamine-dependent subjects in the present report, might indicate a greater disposition or vulnerability to experience periodic lack of control and compulsive drug intake (perseverative behavior) and increased behavioral impulsivity (including over-responsiveness to immediate rewards and decreased avoidance of longer-term punishment) [15].

For the comparison subjects, there was a tendency towards a negative correlation between relative orbitofrontal gyrus metabolism and the MPQ variables, such that higher harm avoidance and constraint scores were associated with lower orbitofrontal metabolism at rest. However, these results were not statistically significant. It is possible that glucose metabolism during activation would be more reliably associated with neuropsychological measures than metabolism at rest, as implicated in a recent PET study where metabolism in the visual and auditory regions was correlated with neurocognitive function (dementia severity) during stimulation but not at rest in 15 Alzheimer’s disease patients [16]. An alternative explanation pertains to the methodology we used to define this ROI, which was based on pre-selection of small regions [9] possibly contributing to a Type II error. Performing correlational analyses on a pixel-by-pixel basis, a significant positive correlation between regional cerebral blood flow in the ventromedial prefrontal cortex (a region that is located in close proximity to where we obtained the ROIs for the orbitofrontal gyrus) with negative affect was demonstrated in two large samples (n = 51 and n = 38) of healthy subjects in a recent PET study [17]. The positive direction of this association is consistent with the observed trends for the comparison subjects in the current report (NEM was negatively associated with harm avoidance and constraint, which in turn were negatively associated with the orbitofrontal gyrus in this group).

There were no significant correlations between the MPQ variables and the absolute orbitofrontal gyrus metabolism for any of the study groups, which may reflect the fact that regional absolute measures are strongly influenced by the overall metabolic activity in the brain of a given subject; in contrast, relative measures reflect much better regional differences between brain areas [18]. In addition, the increased variability in measures of absolute regional metabolism would necessitate a larger sample size to document significant correlations or group differences.