The finding of reduced medial orbital frontal cortex (OFC) gray matter (GM) in substance dependent individual (SDI) compared to controls is consistent with previous studies. Franklin et al. were the first to report lower GM in cocaine dependent subjects compared to controls using voxel-based morphometry (VBM) methods (10
). They observed lower GM density in ventral medial OFC, anterior cingulate, and anterior insula. Lyoo et al. found lower GM in bilateral medial OFC in opiate dependent subjects compared to controls (11
). Less GM was also found in superior and middle frontal and anterior temporal lobes. In both of these studies subjects were using drugs close to or at the time of MR scanning. In Franklin et al., the average number of days cocaine was last used prior to imaging was 15. In the second paper, opiate-dependent persons were on methadone maintenance. Thus, a potentially important difference of the current study is the relatively prolonged abstinence. In this cohort of SDI abstinence averaged 2.4 years for amphetamine and longer for other drugs. Reversible effects of drugs on brain structure have been well documented for alcohol. Recovery of brain volume as assessed with MRI methods in alcoholics can be measured within a few weeks and may last months after sobriety (13
). Such recovery appears to be impeded by relapse (13
). While similar studies of reversible tissue loss have not been performed for illicit drugs, PET neuroimaging studies in methamphetamine abusers show a reduction in dopamine transporter availability that reverses with prolonged abstinence (24
). These temporal changes associated with cessation and relapse underscore the importance of studying long-term as well as short-term changes. Thus, the prolonged abstinence in our population could account for relatively specific changes in medial OFC and suggests the possibility that differences in medial OFC reflect more persistent, enduring brain changes.
The orbitofrontal cortex has emerged as a potential neural substrate for an impaired ability to evaluate expected outcomes leading to poor decision making among SDI (8
).Through its connections with the limbic system, OFC integrates associative information to produce a representation of expected outcomes. Chronic drug use results in adaptations in neural morphology and cell signaling that are thought to disrupt cognitive processes such as decision-making (8
). Rats treated with cocaine show deficits in OFC-dependent functions such as reversal learning (4
). In chronic cocaine users, metabolic abnormalities are relatively specific to frontal lobes (7
). As noted above, some changes are transient, but others may persist long after drug exposure (2
Our findings are consistent with behavioral studies showing decision-making deficits on the Iowa Gambling Task (IGT) in patients with ventral medial OFC lesions (16
). Like patients with ventral medial frontal lesions, SDI are impaired on the IGT (27
), although the impairments are less severe (30
). This is consistent with our data suggesting that controls avoid “bad” decks over time more than SDI, but the differences were not significant. The negative correlation between medial OFC GM volume and decision to avoid bad cards is consistent with a role of OFC in evaluating expected outcomes. The correlation seemed mainly driven by controls, not SDI. We subsequently analyzed whether OFC GM correlated with abstinence, as such a relationship could suggest that chronic drug exposure influenced the OFC GM finding. However, there was no relationship between abstinence and morphology. On the other hand, the lack of a relationship does not imply a pre-morbid deficit as a number of other factors including severity of drug dependence, number or type of substances, and environmental factors could contribute to the findings. The possibilities of a pre-morbid condition, post-drug effect, or a combination remain equally likely.
We did not find regions of significantly increased GM in SDI compared to controls. One study using ROI methods found GM increases in striatum, accumbens and parietal cortex (32
). Others have reported increases in striatal volume in cocaine abusers (33
) and in thalamus and pre-central gyrus in marijuana users (34
) compared to controls.
The major methodological difference between our study and previous ones using VBM is the use of the unified model that integrates segmentation, bias correction, and registration(19
). Another technical difference is that MR images were acquired at 3T in this study compared to previous studies at 1.5T (10
). Although this is not expected to significantly impact the results, it is worth noting that studies that have quantified gray matter-white matter contrast-to-noise ratio (CNR) found higher CNR at 3T compared to 1.5 T when parameters are optimized (36
). Higher gray matter-white matter CNR would be expected to result in better tissue segmentation and more accurate VBM results for a given spatial resolution and signal to noise ratio.
There are several limitations to this study. First, the sample size was modest (n=39) although in the range of similar studies. Second, subjects were dependent on multiple substances, precluding inferences about drug-specific effects on brain structure. Third, abstinence was based on self-report. SDI were remanded to residential treatment by the criminal justice system, either on diversion (instead of prison) or following a prison sentence, and before release to community probation. A minimum 2 months treatment compliance was required before they could participate in this study. Thus, the time in diversion or prison plus 2 months at ARTS resulted in relatively long abstinence. SDI were closely supervised and underwent frequent, observed urine drug tests. While self-report may be unreliable, it is highly unlikely that there were acute drug effects. Fourth, the findings of group differences and relationship between behavior and morphology are inconclusive about causality or predisposition. Finally, although a diagnosis of bipolar disorder was exclusionary, we did not specifically screen for major depression which has been shown to be associated with reduced OFC volume (38
In conclusion, we found robust reductions in GM volume limited to bilateral medial OFC in abstinent substance dependent individuals compared to controls. This is the first paper reporting lower GM volume in this population specific to the medial OFC using whole-brain correction for multiple comparisons. Since abstinence was prolonged, the reduced medial OFC GM may reflect long-term adaptations within the reward-learning circuit underlying pathological decision making behavior in substance dependence.