Advances in neurobiological research related to interventions in adolescents with substance use disorders: Research to practice

1. Development of workshop: background and significance

On June 19, 2006, The Society for Adolescent Substance Abuse Treatment Effectiveness (SASATE) sponsored a workshop at the 68th Annual Meeting of the College of Problems on Drug Dependence (CPDD) on advances in neurobiological research and its relevance and application in the treatment of adolescents with substance use disorders.

Three new research presentations were given: (1) “Differences in fMRI Brain Activation Patterns on Decision Making Task Predicts One Year Outcomes in Methamphetamine Dependence”; (2) “Randomized Controlled Trial of Fluoxetine + Cognitive Behavioral Therapy (CBT) in Adolescents with major depressive disorder (MDD), conduct disorder (CD), substance use disorders (SUD): Impact of Treatment on Cortisol and Serotonin”; (3) “Preliminary fMRI Study of Brain Activation Patterns in Response to Drug Related Stimuli in Adolescents Pre-Post Treatment”. Each of the studies illustrated innovative approaches to incorporating basic science applications in clinical research design and highlighted a number of the key challenges of conducting translational research.

Translational research has become an increasingly important NIH research priority to address growing concerns that the gap between research and practice may be widening instead of narrowing, despite unprecedented research advances in the past decade. The increasing complexity and methodological rigor of basic and clinical research has made translation of research to practice more difficult than ever. Narrower patient inclusion criteria to reduce uncontrolled sources of variability have raised questions about the generalizability of research findings to “real world” patients in “real world” settings. Translational research initiatives have been developed to address research-to-practice barriers and to accelerate the pace of application of new research findings to improve patient care.

The NIH will provide considerable resources necessary to build a sustainable infrastructure to support an entirely restructured clinical research enterprise as a new field or discipline of Clinical and Translational Science. The research teams of the future will be comprised of multi-disciplinary teams of well-trained basic and clinical investigators as well as community-based clinicians and practice network leadership (http://www.NIHRoadmap.gov). An important goal is facilitation of meaningful bi-directional interaction and communication between clinicians and researchers in order for clinicians to inform the research mission and to overcome the longstanding research-to-practice barriers between them—a process often referred to as “bench to bedside and back to the bench” (Rogers, 1995; Sobell, 1996).

Perhaps nowhere have the research to practice gaps and barriers been more apparent or formidable than in the area of adolescent addiction treatment research. This growing patient population carries a tremendous burden of illness. Despite the significant impact adolescent addiction on public health, there is a paucity of research to inform treatment. Although the majority of adolescents with SUD also have co-occurring psychiatric disorders, substance abusing adolescents have been systematically excluded from clinical trials evaluating the efficacy of medications or other treatments for pediatric psychiatric illnesses. Conversely, psychiatric comorbidity has generally been ignored or has been an exclusion criterion in clinical research evaluating the efficacy of interventions for adolescent SUD (Whitmore and Riggs, 2006). Even fewer studies have addressed the challenges of including scientific aims that require basic science applications in clinical research design in this patient population. The new research presented in this workshop begins to address these challenges and ways of overcoming them, because, in the end, the success of the broader translational research mission will be measured by its ability to overcome similar barriers and challenges.

2. Summary of presentations

The first presentation of the workshop focused on differences in functional magnetic resonance imaging (fMRI) brain activation patterns on decision-making. Dr. Susan Tapert reviewed recent evidence relating neurocognition to substance use initiation, escalation, relapse, and long-term outcomes in young people, in the context of the dramatic increases in substance use (Johnston et al., 2006) and related disorders (SAMSHA, 2005) during the course of adolescent neuromaturation (Gogtay et al., 2004; Tapert and Schweinsburg, 2005). In a prospective study of 66 adolescents, attentional and executive dysfunction at age 15 predicted a greater likelihood of alcohol or drug use initiation by age 23, above and beyond effects attributable to substance involvement, gender, education, conduct disorder, family history, and learning disabilities (Tapert et al., 2002). Among adolescents whose alcohol use had escalated to problem levels (n = 15), exaggerated brain response to alcohol-related stimuli was apparent upon fMRI during a cue reactivity task, whereas adolescents without histories of problem drinking showed reduced response to alcohol cues (n = 15) (Tapert et al., 2003a,b). The dramatic dorsolateral prefrontal activation observed in this and other (Wilson et al., 2004) studies of non-treatment seeking users suggests context-dependent processing and possible limitations in inhibition and control of goal-directed behaviors. Longitudinal studies are needed to understand if drinking patterns drive brain response, or if the converse is true.

Youth treated for substance use disorders relapse at high rates. To see if neurocognitive abilities, particularly problem solving and attentional skills, might influence coping with temptations to relapse, 79 adolescents from inpatient treatment programs were assessed after discharge. Treatment programs were abstinence focused, and follow-up assessments were conducted one and two years following discharge, including a battery of neuropsychological tests, coping questionnaire, and detailed assessment of alcohol and other drug involvement. Teens with poor scores on tests of attention and problem solving abilities used little or no substances in the subsequent year if good coping skills were evidenced, but used more often than high scorers if their coping skills were poor. This association between coping skills and outcome did not exist for teens with better neurocognitive abilities. Thus, neurocognitive abilities, including executive functions, appear to moderate the degree to which relapse risk is attenuated by coping skills, which often involve behavioral responses to the presence of substance cues.

Predicting relapse may help direct-targeted interventions to individuals at risk. In an fMRI study of adults treated for amphetamine dependence, relapse in the first post-treatment year was predicted by brain response to a decision-making task. Those who went on to relapse (n = 18) showed less activation than non-relapsers (n = 22) in structures critical for decision-making, indicating limited neural effort devoted to decision-making during the treatment stay, potentially setting the stage for relapse. In fact, activation levels in the right insula, right posterior cingulate, and right middle temporal gyrus correctly predicted outcome in 17/18 relapsers and 19/22 non-relapsers (Paulus et al., 2005). This study demonstrated that functional neuroimaging may help identify patients at particular risk for relapse. While promising in the adult group, this method requires investigation in an adolescent population.

Expectations of positive effects from substance use is an established risk factor for initiation and escalation of use, but less is known about the role of positive expectation in predicting long-term post-treatment outcomes for youth. In a sample of 139 youth followed 8 years after adolescent substance abuse treatment, positive expectancies of substance effects predicted significantly poorer long-term outcomes, particularly among youth with above-average verbal skills (Tapert et al., 2003a,b). These results suggest that intellectually advanced youth with substance problems may benefit from expectancy challenge interventions.

In summary, neurocognitive functioning appears to be an important predictor or moderator across all stages of the course of substance use disorders in youth. Prospective studies have shown that reduced executive functioning is a risk factor for the acceleration of substance use, but poor outcomes among problem users with executive dysfunction may be mitigated by strong coping skill repertoires. Further, verbal skills may serve to magnify positive expectancies of substance effects and may require focused interventions. Neuroimaging studies suggest exaggerated brain response to cues representing substances of abuse, but longitudinal studies are needed to see if these abnormal responses predict future use and can be modified with appropriate interventions (see presentation of Dr. Thompson, below). Although reduced brain response during decision making has been linked to increased relapse risk in adults, studies are needed to see if such predictions can be made with adolescents, and if dampened activation can be modified to produce more favorable treatment outcomes.

In the second presentation, Dr. Paula Riggs discussed data from a recently completed randomized controlled trial of fluoxetine + CBT for substance use versus placebo + CBT in 126 adolescents with DSM-IV MDD, CD and SUD. MDD, CD and SUD have all separately been associated with abnormalities in the limbic hypothalamic-pituitary adrenocortical (HPA) axis (Nemeroff et al., 1994; Pajer et al., 2001; Plotsky et al., 1998). Preliminary studies have reported that adolescents with MDD and (separately) CDmay have a “flattened” diurnal cortisol range; lower serotonin receptor binding affinity; possibly higher serotonin receptor density (Coccaro, 1996; Goodyer et al., 2000, 2003; Plotsky et al., 1998). We could find no published research on the relationship between clinical characteristics and treatment response in these biomarkers in “real world” adolescents with SUD, who are often comorbid for all three disorders.

The aims of this study were to evaluate the feasibility of sample ascertainment and assay procedures for cortisol and serotonin receptor characteristics in a subsample of 76 adolescents with MDD, CD, and SUD who were participating in aforementioned controlled trial of fluoxetine + CBT versus placebo + CBT. Additional aims were to compare patients and controls (n = 22) at baseline and to evaluate change in cortisol and serotonin receptor characteristics after 16 weeks of treatment in patients.

Salivary cortisol samples were collected at approximately 8 a.m. and 8 p.m. over 3 days at baseline in patients and controls, and after 16 weeks of treatment in patients. Adolescents were provided with cell phones during sample collection days to enable study personnel to contact participants at appropriate times as a reminder to collect and store saliva samples in pre-labeled sample storage tubes. Cortisol analyses included all patients who had at least one valid a.m. and p.m. cortisol sample at both baseline and 16 weeks (n = 45). Blood sample isolates collected at baseline and after 16 weeks of treatment were assayed for serotonin receptor density and receptor binding affinity at a Washington University laboratory according to procedures specified by Coccaro (1996). Patients and controls were compared on baseline measures using independent t-tests. Mixed model analyses and ANOVAs were used to evaluate main effects of medication and the interaction between medication treatment and depression remission status on cortisol and serotonin measures.¹

Patients had significantly lower serotonin receptor binding affinity (K_d) at baseline compared to controls (p = .002) and trended toward higher receptor density (B_max; p = .07). Fluoxetine had a significant main effect on increasing serotonin receptor binding affinity (K_d) pre-post treatment compared to those receiving placebo (p = .05). There was no difference between fluoxetine and placebo with regard to pre–post change in B_max. There was no pre–post difference between remitters or non-remitters with regard to change in either B_max or K_d.

Contrary to our prediction that patients would have significantly lower cortisol and ‘flattened’ diurnal cortisol range at baseline compared to controls, there was no difference between patients and controls on mean morning cortisol or diurnal range. The unexpected negative result may have been influenced the smaller sample size (n = 22) of the control group and the lack of gender and age matching between groups. Control females had significantly lower morning cortisol (n = 11; 9.0 nm/L) compared to male controls (n = 11; 14.0 nm/L; p = .023) which lowered controls’ overall baseline mean cortisol. This was in contrast to patients who did not show a baseline difference in cortisol levels between males (n = 45; 11.0 nm/L) and females (n = 22; 10.7 nm/L).

Fluoxetine + CBT treatment trended toward a significant increase in diurnal cortisol range with treatment compared to placebo + CBT treatment (p = .07). There was no difference between medication groups with regard to pre–post treatment change in morning cortisol levels. However, patients whose depressions remitted (regardless of medication group assignment) had a significant increase in morning cortisol (p = .017) and diurnal range compared to non-remitters (p = .007).

Results of this study indicate that biomarker sample ascertainment and assay procedures are feasible in adolescents with SUD and multiple comorbidity when adequate support (cell phones; pre-package; incentives) is provided to support sample ascertainment and procedural adherence.

Patients had significantly lower serotonin receptor binding affinity compared to controls at baseline which was responsive (increased) to treatment with fluoxetine (fluoxetine main effect > remission). Serotonin receptor density trended to be lower than controls at baseline and did not change in response to fluoxetine or remission status. Remission of depression was associated with a significant increase in morning cortisol and broadening of the diurnal cortisol range compared to non-remitters and was a stronger predictor of increased diurnal range than fluoxetine (remission main effect > fluoxetine). If replicated, these findings may have important clinical implications by increasing our understanding of the biological underpinnings and the predictors of treatment response.

In the third and final presentation, Dr. Laetitia Thompson described results from a pilot study comparing functional brain imaging of the final 11 participants in the clinical trial described by Dr. Riggs. This study was a preliminary step toward building appropriate methodology to examine neurobiological changes that may occur with substance abuse treatment. Neurobiological changes have been found with treatment of other disorders such as depression (Goldapple et al., 2004), obsessive compulsive disorder (Nakao et al., 2005), and schizophrenia (Wexler et al., 2000). The ultimate goal of this type of research is to increase our understanding of how substance abuse treatment works by correlating neurobiological change with behavior change and clinical outcome. This research requires development of an appropriate behavioral paradigm that is suitable for fMRI and also sensitive to changes that might occur in the brain regions involved during processing of drug-related stimuli.

Previous research on understanding regional brain activation in reaction to drug-related stimuli in addicts provides a foundation upon which to build a paradigm for examining change. In individuals with substance use disorders, multiple brain regions activate to drug cues, suggesting activation of multiple neural circuits involving reward, salience, motivation/drive, learning and memory, and cognitive control. The neuroanatomic regions that are frequently described as being activated by drug cues include dorsolateral prefrontal cortex, orbital frontal cortex, anterior cingulate cortex, striatum, amygdala, and cerebellum (e.g., David et al., 2005; Franken et al., 1999; Maas et al., 1998). Most of the research has been conducted in adults, but the work by Tapert and her colleagues suggests that adolescents with SUD respond to drug-related stimuli in a similar pattern (Tapert et al., 2003a,b).

Therefore, a paradigm was developed specifically for adolescents with cannabis dependence or abuse. Dr. Thompson and colleagues chose to study cannabis because it is one of the major drugs of abuse in adolescents. Three types of pictures (cannabis, food, and neutral) were presented in a blocked design while functional images were obtained on a 3 Tesla GE scanner.

Eleven participants (5 males and 6 females) were recruited; all agreed to participate and completed the pretreatment fMRI scan. Participants were polydrug users, but all met clinical criteria for cannabis dependence or abuse. Mean age was 16.9 (range 13–19). After high-resolution anatomic images were obtained two functional runs were presented. Each run included six 90-s epochs comprised of three blocks (one block of each stimulus) plus an 18-s rest period. Each block included 5 pictures of a particular type; each picture was displayed for 4.8 s. Order of blocks and the rest periods within each epoch were pseudorandomized.

Functional imaging processing and analysis were conducted in via Statistical Parameter Mapping-2 software. After motion correction and coregistration, spatial normalization, and smoothing, statistical analysis focused on the contrast between the two types of appetitive (reinforcing) stimuli, specifically when activation to drug pictures significantly exceeded activation to food pictures. Analyses of pre-treatment brain scan data revealed preliminary evidence that these 11 adolescents activated certain cortical and limbic regions to drug stimuli to a greater extent than they activated to food. In addition to posterior visual processing regions, these included lateral orbital frontal cortex, medial frontal cortex, midbrain, cingulate gyrus, cerebellum, ventral tegmental area, nucleus accumbens and other limbic regions such as amygdala and thalamus. Higher activation in these regions may suggest a more sensitive reaction to drugs in the brain reward circuitry.

Ten of the participants completed the post-treatment fMRI scan. Motion was a significantly greater problem in the post-treatment session than in the pretreatment session, such that 4 of the 10 subjects were excluded from analyses. Pre- to post-treatment comparisons of the 6 participants with usable scans at both time periods revealed significantly greater activation to drug compared to food post-treatment than pre-treatment (p < .01, not corrected for multiple comparisons) in 3 cortical areas: medial frontal, lateral inferior frontal, and dorsolateral prefrontal (an area of cognitive control that did not activate more to drug than food pre-treatment).

This pilot study supports the feasibility of this promising area of research. The preliminary finding that certain areas of brain activation increased rather than decreased after treatment when comparing drug stimuli to food stimuli was a bit surprising. Limbic and subcortical areas involved in the “reward circuit” did not change. Rather areas of increased activation were seen in cortical, frontal regions involved in different executive functions. The absence of a control group precludes interpretation that these changes are related to treatment. However, the data are sufficiently promising to support further refinements in the paradigm and a controlled study with a larger sample size. Such research may yield very interesting insights into brain processing and changes that are related to clinical outcomes.

Joseph Frascella, Ph.D., Director of the Division of Clinical Neuroscience and Behavioral Research at NIDA, was the discussant for the session. Dr. Frascella emphasized NIDA’s ongoing commitment to the success of the NIH translational research mission, both within NIDA’s programs as well as within the NIH Roadmap. He also highlighted that both basic and clinical research, particularly their integration, can greatly improve our understanding of the issues involved with adolescents with SUD and their treatment outcomes. These types of studies reflect a clear priority within NIDA’s research agenda and fit well within the NIH efforts to transform how clinical and translational research is conducted to bring forward new treatments more efficiently and quickly to the population.

Dr. Frascella emphasized the importance of translational and transdisciplinary approaches that are focused on the interactions between developmental processes and substance abuse. He also discussed the importance of identifying behavioral markers that can help predict future substance abuse as described by Dr. Susan Tapert, who showed how early behavioral indices such as neurocognitive functioning (attention and executive dysfunction) could be a marker for future substance abuse. He also remarked on another study conducted by Dr. Tapert and colleagues showing how adolescents who were problem drinkers had functional brain changes in response to alcohol cues (Tapert et al., 2003a,b). This study highlights an example of a clear brain–behavior interaction in which a continued behavior (alcohol abuse) is directly related to brain responses (an exaggerated brain response to alcohol cues), and this relationship is markedly different from those of adolescents without a history of problem drinking. It is yet unresolved if the exaggerated brain response is preexistent and “drives” the drinking behavior in these adolescents with high levels of alcohol consumption, or if the brain is changed by the alcohol and the drinking behavior. Further investigation will clarify this issue more specifically; however, an important point this study illustrates is that the salience for the substance is distinctly changed in these adolescents, both from a behavioral as well as a neurobiologic perspective.

Dr. Frascella noted that from a translational perspective, neuroimaging is a tool that can be effectively applied in behavioral paradigms to provide keener insights into our understanding of brain-behavior interactions, which in turn, can better inform treatment. In another study, Dr. Tapert’s group demonstrated how neuroimaging, combined with a decision-making task, was used to predict relapse in adults dependent on methamphetamine (Paulus et al., 2005). Areas of the brain involved in decision making were less responsive in the relapsers than in the non-relapsers. This study is the first to provide direct evidence of a biomarker predicting treatment success in this population. Importantly, this type of work and these approaches are extremely powerful in that they hold great promise for better defining individuals at risk, for providing specific brain region targets for remediation, and also for monitoring treatment progress and outcome. Although this work was not conducted in adolescents, these types of studies can be easily adapted to incorporate a developmental perspective.

Dr. Frascella also highlighted Dr. Paula Riggs very exciting work on a population of adolescents with co-occurring MDD, CD, and SUD. This study represents a real tour de force as it evaluated the feasibility of biomarker establishment in adolescent populations with co-occurring MDD, CD, and SUD. Dr. Riggs was commended for this extremely clinically relevant research as very few investigators attempt to study such complex adolescent populations with mixed diagnoses. Results from this preliminary investigation established that biological measures such as cortisol levels and serotonin receptor density and binding affinity could be used as biomarkers in this population. The study also identified some very provocative preliminary results showing differences between remitters and nonremitters on these biological measures, which changed as a function of treatment and remission of symptoms. Determining biological measures that can be established as clear and accurate biomarkers to monitor treatment progression and success are essential, particularly in populations with co-occurring disorders. The implications of these types of approaches to better inform treatment are clear.

In the final presentation of the session, Dr. Laetitia Thompson described some preliminary yet intriguing functional neuroimaging data on substance-abusing adolescents before and after treatment. A critical question to be answered in the area of substance abuse is: How does treatment (behavioral and/or pharmacological) change the brain? That is, given that the brain is the main site of action for drugs of abuse, and the brain is changed by the continued use of these drugs, can treatment reverse brain structural and functional alterations? Answers to these questions can now be obtained with the application of brain imaging techniques as exemplified in Dr. Thompson’s study. As she aptly pointed out, to assess functional brain changes, a suitable behavioral paradigm must be designed and implemented that is useful in the context of a magnetic resonance imaging study. In this case, cannabis-dependent individuals were presented with drug-related versus food-related stimuli, and brain responses were obtained before and after treatment. Brain activations were measured in a number of regions. The results of this study were admittedly preliminary and somewhat unexpected, though importantly, they revealed consistent post-treatment changes. Moreover, this study illustrates an important application of brain imaging to probe treatment effects at the neurobiological level. This approach will advance our understanding of brain correlates of treatment and will thereby serve as a critical tool for translational research.

Overall, this panel provided excellent examples of the value of integrating basic science in clinical research, particularly in adolescent populations in the context of substance abuse treatment. Given the complex nature of substance abuse with multidimensional factors contributing to its etiology and progression to addiction, translational efforts, particularly in adolescents, will lead to advancements in our understanding of the relationships between brain and behavior. Through this knowledge, the door to more targeted and effective treatment interventions will be opened.

3. Clinical implications and research recommendations

Each of the studies presented demonstrate the feasibility and promise of integrating basic science and clinical research by advancing our understanding of the biological underpinnings and predictors of treatment response. More research is needed to elucidate the neurobiological antecedents, biological correlates, and developmental differences in addictive processes in children, adolescents and adults. Additional research is also needed to clarify the relationship between decision-making, executive functioning and clinical outcomes. For example, can treatment strengthen executive functioning or other neuorcognitive processes to improve decision-making and over-ride behavioral compulsivity in response to conditioned cues, which are often triggers for drug use? Further advances in these exciting areas of research may have important clinical implications and direct clinical application in the development of more effective addiction treatment interventions, but will require multidisciplinary collaboration to succeed.

Multidisciplinary collaboration is needed between basic scientists and clinical researchers to refine cue-reactivity and decision-making paradigms as well as identification the most relevant biological markers with the most promising potential as predictors of clinical outcomes. Increasing the speed of “real world” application of research advances to improve patient care will require meaningful collaboration and sustainable partnerships between researchers, clinicians and third-party payers, policy-makers and other stakeholders if transformation of the clinical research enterprise is to be successful as envisioned by the NIH translational research mission. This workshop is dedicated to the success of that mission.