Our findings showed that the estimates of severity of use of heroin, alcohol, MDMA and cannabis were negatively associated with DLPFC and temporal cortex regional metabolism in polysubstance dependent individuals with prolonged abstinence from all these drugs. Alcohol use was further associated with lower metabolism in frontal premotor cortex and putamen. Furthermore, severity of stimulants use (cocaine and MDMA) was uniquely associated with inferior parietal/postcentral cortex metabolism. These associations were established after a prolonged period of abstinence, and after controlling for multiple confounding variables, thus surpassing the limitations of previous functional neuroimaging studies that were not specifically designed to study prolonged abstinence, and were confounded by important variables, such as withdrawal related symptoms or concurrent use of multiple substances. The associations found are in agreement with our initial predictions, and with the neuropsychological deficits typically aligned with the chronic use of these substances; these include reduced competency in executive functions (related to all classes of drugs), episodic memory (mainly observed on heroin, alcohol, MDMA and cannabis users), motor control (alcohol and cannabis users), and visuospatial attention (stimulants users) (see meta-analyses in 
As expected, the estimates of amount or duration of different classes of drugs were negatively associated with overlapping regions within the DLPFC (BA 9, 45, 46). This region has been linked with several of the key neuroadaptations associated with drug addiction, including drug conditioning, loss of self-control, and stimulus-driven compulsive behavior 
. This finding is also in agreement with results from several neuropsychological studies showing cognitive deficits on working memory, planning or inhibition processes, which are associated with DLPFC functioning, across users of several drugs 
. We also found negative associations between estimates of duration of heroin, alcohol, and MDMA use and regional metabolism on partially overlapping sections of the temporal cortex. These associates may substantiate the links between the intensity and duration of involvement with these drugs and the degree of memory attrition, which has been observed in longitudinal studies with users of these substances 
. It is worth mentioning that duration of heroin use was shorter to that of cocaine and alcohol use; however, previous brain structural findings indicate that duration of heroin use is a critical factor leading to brain damage, even in users with periods of use below five years 
. Surprisingly, we failed to find significant associations between the severity of the substances used and overlapping limbic, striatal and cerebellar regions but this result could be explained from the fact that these regions play a greater role during current use and short-term abstinence 
. In addition, we found relatively specific negative correlations between amount of cocaine use and regional metabolism in the inferior parietal cortex, and between alcohol duration and regions involved in motor programming and control (frontal precentral and putamen). The link between cocaine use and parietal dysfunction has been observed on previous functional imaging studies during cognitive challenges of attention and executive control 
. The link between alcohol use and precentral and basal ganglia dysfunction is in agreement with structural imaging findings showing that function of these regions is significantly reduced in alcohol users compared to other forms of addiction 
. Additional evidence comes from volumetric studies which associated volumetric measures of these regions and deficient motor skills in alcoholics 
. These more specific findings could be used to design tailored interventions aimed to address specific aspects of frontal-subcortical executive functions in different profiles of substance abusers.
An important implication of this study is the fact that the link between severity of drug use and BM is still observable in substance abusers with prolonged periods of abstinence (an average of eight months in this sample). Therefore, quantity and duration of drug use may still be affect brain functioning after several months of successful cessation of drug use. This is particularly relevant in view of recent findings indicating that certain patterns of brain dysfunction can reliably predict mid-term and long-term relapse. Permanent brain alterations may also have important repercussions for the rehabilitation of SDI, since these alterations can affect the ability of SDI to assimilate the contents and participate in the activities of the rehabilitation programs 
. In fact, preliminary evidence suggests that cognitive rehabilitation techniques, such as errorless learning may be successful in compensating chronic cognitive deficits in substance abusers 
. Thus, the brain networks impacted by severity of drug abuse can play a central role in clinical outcome and relapse and should become priority targets for therapeutic interventions.
Several possible limitations of our study should be considered and addressed by future research. Our findings were not obtained in “pure” users of each of the substances but in mostly polysubstance users. Although this limitation is inherent to the clinical literature on addiction (i.e., “pure” users of one single drug are quite rare among substance abusers enrolled in treatment programs), we attempted to control for the co-abuse of other drugs using statistics. A related confounding variable was the use of nicotine, which is ubiquitous among drug users but was not specifically assessed or controlled for. Furthermore, due to the use of a correlational design, this study cannot yield conclusions about cause-effect relationships between the use of drug and resting BM. However, given the agreement of our results with both resting state PET, activation PET/fMRI and addiction neuropsychology studies, we could hypothesize that, at least in part, the alterations observed in this population could be due to the quantity and duration related effects of the consumption of the substances under consideration in this study.
One last point to consider is that our findings could be due to premorbid brain alterations or the results of the interaction between the premorbid alterations and the neurotoxic effects of drug use. This question should be addressed through longitudinal designs. A related limitation is that no comparison with a healthy control database was performed. As we argued before, a healthy control group was not needed to test the main predictions of our study; however, the lack of controls has stopped us from making any assumption about the direction of these findings in relation to the healthy population.