The neurobiology of addiction
Uhl et al.
Neurobiology of addiction
New Mexico VA Healthcare System Albuquerque New Mexico
NIH/NIAAA Bethesda Maryland
*Address for correspondence: George Uhl, New Mexico VA Healthcare System, Albuquerque, NM 87108. George.Uhl@va.govAbstract
Substance and alcohol use disorders impose large health and economic burdens on individuals, families, communities, and society. Neither prevention nor treatment efforts are effective in all individuals. Results are often modest. Advances in neuroscience and addiction research have helped to describe the neurobiological changes that occur when a person transitions from recreational substance use to a substance use disorder or addiction. Understanding both the drivers and consequences of substance use in vulnerable populations, including those whose brains are still maturing, has revealed behavioral and biological characteristics that can increase risks of addiction. These findings are particularly timely, as law‐ and policymakers are tasked to reverse the ongoing opioid epidemic, as more states legalize marijuana, as new products including electronic cigarettes and newly designed abused substances enter the legal and illegal markets, and as “deaths of despair” from alcohol and drug misuse continue.
Addiction neurobiology is superbly situated to benefit from many neuroscience advances. Advanced imaging that reflects neuronal activity and neurochemistry in humans and experimental animals provides substantial insights into meso‐scale brain changes that are highly relevant for addictions. Addiction researchers’ early adoption of optogenetic and chemogenetic approaches has provided elegant support for and refinement of hypotheses about roles for specific circuits in addiction‐related behaviors and physiology.
Much progress in the neurobiology of addiction can be placed into a heuristic three‐stage addiction cycle framework: binge/intoxication, withdrawal/negative affect, and preoccupation/anticipation. This framework is supported by multiple neuroadaptations in three corresponding domains: (1) increased incentive salience, (2) decreased brain reward and increased stress, and (3) compromised executive function; and in three major neurocircuits: basal ganglia, extended amygdala, and prefrontal cortex (Fig. 1). The focus in the neurobiology of addiction has changed with emphasis on the mechanisms of acute reward in the binge/intoxication stage broadened to include neuroadaptations that are consequent to drug exposure. These include mechanisms driving incentive salience, compulsive habits, deficits in reward and recruitment of stress during the withdrawal/negative affect stage, and modulation of executive function systems and mnemonic systems (and being modulated by mnemonic processes) in the preoccupation/anticipation stages of substance use disorders.
Addiction science is also well poised to use results from a number of the changes in the addictions landscape.1 Legalization of cannabis use by states provides opportunities to examine effects of reduced penalties for cannabis production and use; neighboring states that do not legalize provide control environments. Restrictions on opioid prescribing now provide opportunities to examine how reduced availability of pharmaceutically prepared opioids influences patterns of distribution and the use of illicitly prepared opioids, as well as the treatment of pain for which pharmaceutical opioids were prescribed. Illicit designer substances provide challenges in understanding the actions of novel pharmacological products in humans even before laboratory animal and in vitro testing. Differences, by region and over time, in availability of behavioral and pharmacological addiction therapeutics provide opportunities to assess their worth in new ways.
By definition, drugs form a vibrant part of the neuropharmacology of addiction. Since addictive substances themselves are central to addiction pathogenesis, our etiologic understanding of addictions can advance at greater rates than in the neurobiology of brain illnesses whose etiologic agents are less well understood.
Addictions have usurpation of motivation at their cores. Many have underlined the ways in which substance use, established by rewarding processes, can be maintained by altering motivation, including driving incentive salience, establishing compulsive‐like habits, engaging negative reinforcement, and facilitating impulsivity.
In 2017, 19.7 million people age 12 or older in the United States were estimated to have a substance abuse disorder related to alcohol or illicit drug use. This value includes 14.5 million people with an alcohol use disorder (AUD) and 7.5 million people with an illicit drug use disorder, the most common illicit drug being marijuana. Tobacco use also remains prevalent, with 48.7 million current cigarette smokers, of whom 27.8 million smoke daily, and 11.4 million smoke at least a pack per day.1 Substance abuse disorders exert not only a significant public health burden—individuals with substance use disorders are more likely to suffer from chronic pain, hypertension, injuries, poisonings, and overdose2—but they also impose significant economic burdens. Costs associated with substance abuse disorder exceed US$700 billion annually due to crime, lost work productivity, and health care;3 US$250 billion due to alcohol; and US$300 billion due to tobacco. Adolescents and young adults are particularly at risk for developing substance use disorders; areas of their brains responsible for evaluating risk, weighing consequences, and making decisions are not fully developed until the mid‐20‐year‐old age range.4 Individuals who begin using illicit addictive substances earlier in life ultimately consume more addictive substances more frequently and have higher rates of substance use disorders.5
While preventive and treatment strategies can reduce substance use and substance use disorders, effects of available prevention and treatment strategies are often modest and short term. New research is elucidating the neurobiological changes, genetic markers, and epigenetic changes associated with addictions. These developments are identifying new targets for treatments and should facilitate personalized/tailored preventive and treatment approaches to maximize effectiveness. However, advances in our understanding of addiction biology can only provide benefit if they are adopted by law and policymakers as evidence‐based policies and programs.
In May 2016, the Aspen Brain Forum and the New York Academy of Sciences brought together leaders in neuroscience, addiction medicine, drug and alcohol abuse, and science advocacy and policy to discuss and update topics in the neurobiology of addiction at the 2.5‐day conference “The Addicted Brain and New Treatment Frontiers: Sixth Annual Aspen Brain Forum.” The conference supported the mission of the Aspen Brain Forum to produce, host, and fund an annual meeting on innovative topics in neuroscience to advance global collaborations and scientific breakthroughs. The meeting also highlighted substantial progress and challenges in addictions and in the neuroscience of understanding these substance use disorders. The Aspen Brain Forum neurobiology work also reflects the striking intersections between policy and science. In few other fields would the contributions of such a talented and tireless advocate for addiction and mental health as Patrick Kennedy (see below) seem so appropriate and natural for a “neurobiology” program.
This report presents a synthetic report of individual presentations at the Sixth Annual Aspen Brain Forum and ends with some reflections on the current state of the field.
The role of dopamine in addiction
Addictive drugs are inherently rewarding. They highjack the brain's dopamine system to increase dopamine levels in the nucleus accumbens, a key focal point for reward neurocircuitry in the brain.6 While dopamine is critical for the rewarding effects of drugs, its role in substance use disorders is still evolving. Nearly 20 years ago, Nora Volkow (National Institute on Drug Abuse, National Institutes of Health) showed via positron emission tomography imaging that higher dopamine levels correspond with a more intense high in healthy volunteers given intravenous methylphenidate (MPH), a central nervous stimulant also known as Ritalin. There was considerable variability in dopamine levels across subjects; some individuals experienced neither increased dopamine levels nor “high.” Administration of oral MPH, which takes longer to enter the brain, resulted in no high with slower increases in dopamine levels.7
Since the rate of dopamine increase plays a factor in whether a drug will produce a rewarding effect, the different properties and effects of dopamine receptors in the brain are likely to play significant roles. The prefrontal cortex contains both dopamine D1 and D2 receptors. D2 receptors have an approximately 10‐ to 100‐fold greater affinity for dopamine than D1 receptors and are therefore activated at lower dopamine concentrations. Under normal circumstances, the prefrontal cortex receives a low level, stable flow of dopamine owing to relatively slow, tonic firing of dopamine neurons in the ventral tegmental area (VTA) that project to the cortex. However, in response to an unexpected event, such as an extraordinary reward or very aversive event, dopamine neurons fire much more quickly. This phasic firing results in an abrupt, yet transient, increase in dopamine. The high levels of dopamine achieved during phasic firing are able to activate D1 receptors and are thought to be required for dopamine's full rewarding effects.8, 9 Drugs of abuse, particularly psychostimulants, mimic the high dopamine concentrations produced by phasic firing and thus activate both D1 and D2 receptors.10
D1 receptors stimulate both reward, via pathways modulating the striatum and cortex, and conditioning and memory mechanisms that involve the amygdala, medial orbitofrontal cortex (OFC), and hippocampus. The conditioning/memory processes critical to addiction allows individuals to automatically associate a stimulus with a reward or punishment. Perhaps paradoxically, several studies have shown that addictive drugs fail to increase dopamine release in addicted individuals compared with nonaddicted controls. MPH did not significantly increase dopamine levels among active11 or detoxified cocaine addicts.12 Cocaine users also reported less of a high from MPH than controls.12 However, among active addicts shown a video to produce craving, increased dopamine was observed in the dorsal striatum. The magnitude of this dopamine increase was associated with the extent of drug craving.11 These data suggest that in addiction there is thus a switch from the drug itself initiating dopamine release to drug cues and stimuli initiating dopamine release. This shift from reward to conditioning involves dopamine phasic firing leading to drug cravings and compulsive drug use in response to drug and other conditioned cues.6
Normally, D2 receptors modulate the effects of D1 receptors via the striatal indirect pathway;10 however, several studies have shown that addicted subjects have lower expression of dopamine D2 receptors.13 Reductions in D2 receptors among addicted subjects are associated with decreased activity in the OFC, anterior cingulate gyrus, and dorsolateral prefrontal cortex areas of the brain involved in emotion regulation and decision making. Because impairments in the orbitofrontal and anterior cingulate cortices are associated with compulsive behaviors, impaired dopamine signaling in these areas in addicted subjects may be partially responsible for their compulsive behavior and impulsivity.6 In animal studies, increased dopamine D2 receptor expression in the nucleus accumbens reduced drug consumption in models of both alcohol and cocaine dependence.14, 15 In humans, a recent study in methamphetamine users demonstrated that regular aerobic exercise can upregulate striatal dopamine D2 and D3 receptors; whether this results in reduced cravings and drug use remains to be seen.16
Computational modeling of dopamine cells
According to a computational model of dopamine release in response to rewards and expectations, dopamine neurons encode reward prediction errors in their firing rates—they increase their firing rates if results are better than expected and decrease their firing rates if results are worse than expected.17 To test this model and its relationship to behavior, P. Read Montague (Virginia Polytechnic Institute and State University; University College London) has used functional magnetic resonance imaging (fMRI) to monitor the effects of reward prediction error on both brain activity and future behavior in subjects participating in a betting task in a fictitious market. The study found neural signatures associated with reward prediction error and fictive error (how much a person gains versus how much they could have gained if they had bet more). Fictive error was associated with activation in the ventral caudate, ventral putamen, and posterior parietal cortex as well as with behavioral changes. The higher the fictive error, the more likely a person was to change their next bet.18 Fictive error signatures were present in the brains of both smokers and nonsmokers; the magnitude of these signatures did not correlate with a change in behavior in smokers.19
Using electrochemistry to monitor dopamine release in real time while subjects completed such tasks showed that dopamine release roughly correlates with market activity at long timescales. However, at short, millisecond timescales, a different pattern emerged. With high bets, increases in dopamine fluctuations correlated with reward prediction errors; however, as the bet size decreased, the correlation reversed. At low bet sizes, increased dopamine fluctuations were seen with negative errors and vice versa. This behavior suggests two sources of dopamine fluctuation—one that communicates a prediction error and one that communicates a fictive error.20
Endocannabinoids and addiction
We have come a long way in understanding the endocannabinoid system and the potential for therapeutic interventions directed at this system, according to Susan Weiss (National Institute on Drug Abuse). Tetrahydrocannabinol (THC) and cannabidiol mimic aspects of the effects of the endogenous cannabinoids anandamide (AEA) and 2‐arachidonoylglycerol (2‐AG). Under endogenous conditions, AEA and 2‐AG are released into the synapse by postsynaptic neurons. They bind to cannabinoid receptors on presynaptic neurons to dampen their activity, thus participating in a negative feedback loop. The functional effects observed depend significantly on which neural circuits are involved.21
However, much less is known about therapeutic effects of cannabis itself. Part of the complexity stems from the fact that marijuana contains myriad chemicals—hundreds of cannabinoids as well as other chemicals that differ in concentration depending on strain. THC and cannabidiol are currently the two most studied components of marijuana. These two chemicals have profoundly different effects on the brain, with THC producing the high associated with marijuana use. Exogenous administration of cannabinoids is one therapeutic strategy to target the endocannabinoid system. Sativex® (nabiximols), a combination of THC and cannabidiol, has been approved in Europe for spasticity associated with multiple sclerosis22 and was granted fast track designation by the U.S. Food and Drug Administration (FDA) for the treatment of pain in patients with advanced cancer.23 Cannabidiol is being investigated for its anticonvulsant properties.24
Other druggable targets in the endocannabinoid system may offer opportunities for different, even more precise, modulation of these systems. For example, fatty acid amide hydrolase (FAAH) inhibitors, which block the breakdown of AEA, might be expected to have greater effects on activated circuits. In animals and humans, FAAH inhibitors have been shown to reduce anxiety‐ and depression‐like behaviors, to enhance social behavior in autism spectrum disorders, and to reduce nicotine addiction.25, 26, 27, 28
Several studies have also found associations between frequent cannabis use and the risk of psychosis.29, 30, 31 Many questions remain that will help to better understand this connection, including the effect of cannabis on neurodevelopment as well as to better understand the reasons that people with psychotic disorders continue to use cannabis if it exacerbates their psychoses.
Drug cue–induced neuroplasticity
Insights into the physiological processes behind the overwhelming drive in individuals with addiction to seek out a drug and forgo other competing choices were discussed by Peter W. Kalivas (Medical University of South Carolina). When an individual with addiction encounters an external cue or stimulus associated with a drug, such as a call from a friend to meet them at a bar or, in the case of a laboratory animal, a light associated with a drug‐delivering lever, cells in the nucleus accumbens are activated, resulting in a cue‐specific engram that results in drug‐seeking behavior. In individuals without addiction, competing thoughts or cues can alter that response. However, drug cues leave behind long‐term potentiation of activity of the nucleus accumbens that blunts the effects of competing stimuli.70
Michael Scofield (Medical University of South Carolina) described the mechanism behind this overpotentiation. Normally, when a cue comes to the prefrontal cortex, glutamate is released into the nucleus accumbens, activating a small percentage of neurons, resulting in a stimulus‐specific memory trace or engram. Excess glutamate is removed from the synaptic cleft by transporters including GLT‐1, a glutamate transporter found on astroglial cells. However, on the basis of animal models of drug addiction, the hypothesis is that GLT‐1 is downregulated, and there are fewer astroglial cells in the synaptic cleft.71 Thus, drug cues cause accumulation of glutamate in the synaptic cleft; subsequent activation of mGluR5, a receptor found on interneurons that express neuronal nitric oxide synthase; release of nitric oxide into the extracellular space; nitrosylation; and activation of matrix metalloproteases (MMPs), especially MMP9, that cause local degradation of the extracellular matrix. This cascade of events provides transient plasticity that contributes to drug‐seeking behavior. In withdrawn animals, drug cues result in significant increases in MMP9. Inhibiting MMP9 inhibits drug‐seeking behavior in response to drug‐associated cues.72 MMP activity also creates an RGD‐binding ligand that activates β3 integrins on spiny neurons, resulting in an increase in spine head diameter and expression of AMPA receptors.73
Downregulation of GLT‐1 can be found with administration of several classes of addictive substances to animals, including cocaine,74 nicotine,70 heroin,75 and alcohol.76 An accumulation of glutamate has been observed in animal models of cocaine,77 nicotine,70 alcohol,78 and methamphetamine addictions.79 Drugs that enhance GLT‐1 function, including N‐acetylcysteine (NAC), ceftriaxone, and propentofylline, have shown positive results in animal models of addiction to cocaine,80, 81, 82, 83 nicotine,84, 85 and alcohol86 use disorders. NAC has also shown improved behavior in human disorders characterized by intrusive thoughts, such as pathological gambling,87 trichotillomania,88 obsessive compulsive disorder (OCD),89 and depression,90, 91 though there are failures to show improvements in pediatric trichotillomania92 or methamphetamine addiction.93 In a recent double‐blind, placebo‐controlled trial of NAC in veterans with post‐traumatic stress disorder (PTSD) and substance abuse, NAC reduced cravings by week 8. This effect persisted for 4 weeks after stopping NAC. Subjects also reported improvements in CAPS scores of PTSD symptoms and CAPS intrusive thoughts score.94
The role of serotonin in anxiety and addiction
Increasing synaptic serotonin levels through the use of selective serotonin reuptake inhibitors (SSRIs), such as Prozac® (fluoxetine) and Zoloft® (sertraline), is a common strategy to relieve anxiety and depression, but the role of serotonin in the brain is complicated.95, 96 First, one of the primary sources of serotonin, the dorsal raphe, projects to areas of the brain involved in impulsivity, reward, stress, anxiety, and feeding. Second, there are many types of serotonin receptors, which can have different effects on behavior. Thomas L. Kash (University of North Carolina School of Medicine) discussed the role of serotonin in increasing anxiety.
Several lines of evidence suggest that increased activation of at least some serotoninergic systems can be highly aversive. SSRI treatment can lead to anxiety, panic, and suicidal ideation in some patients.97, 98, 99 Serotonin has also been shown to play a role in alcohol‐induced anxiety. In individuals with AUD, the serotonin agonist meta‐chlorophenylpiperazine has been shown to induce cravings.100, 101 SSRI treatment has been shown to increase anxiety and alcohol consumption in some individuals with AUD.102, 103
Work from Kash's laboratory has helped to elucidate the neural networks underlying the role of serotonin in alcohol‐induced anxiety. In a mouse model of alcohol dependence in which mice were exposed to alcohol vapor and evaluated 24 h after withdrawal, mice displayed increased anxiety‐related behaviors.104 This effect is dependent on serotonin, since injecting the mice with a serotonin receptor antagonist reduced these behaviors.105 At a neural level, alcohol induced hyperexcitability in both the dorsal raphe, a key source of the serotonin projections to the forebrain, and the bed nucleus of the stria terminalis (BNST), part of the extended amygdala located between the nucleus accumbens and central amygdala and well known for its role in aversive behaviors.
Optogenetic stimulation of serotonin from the dorsal raphe to the BNST also increased anxiety‐like behaviors and fear learning in mice. Animals were placed into a chamber where they received a small shock in response to a tone. Stimulating serotonin release optogenetically during the tone resulted in increased freezing behavior, suggesting that serotonin can increase fear recall (data unpublished). A similar study from researchers at Columbia University showed that increasing serotonin levels by injecting fluoxetine into the BNST also enhances fear learning.106 There is a population of serotonin‐responsive neurons in the BNST, which express both CRF and the serotonin 5HT2C receptor. Upon activation, these neurons inhibit neurons that project into the VTA and lateral hypothalamus, thus inhibiting reward‐promoting outputs and driving aversive states. Silencing CRF neurons in the BNST blocks the effects of fluoxetine in enhancing fear memory and anxiety.107
Transcriptional and epigenetic markers of addiction and implications for treatment
Eric Nestler (Icahn School of Medicine, Mount Sinai) is identifying genes that display changed transcriptional regulation in settings combining social isolation with drug exposure and genes involved in neuronal structure following chronic drug exposure. To measure histone modifications and other chromatin changes, he used RNA sequencing (RNA‐seq) to identify changes in RNA levels and chromatin immunoprecipitation (CHiP) followed by deep sequencing (ChiP‐seq).
Prior studies have examined the effects of social isolation exposure to addictive drugs administered immediately following the isolation.108, 109, 110 However, discrete periods of stress can lead to long‐lasting changes in behavior that can persist even after the stress has been removed. Mice subjected to early‐life social isolation followed by weeks of normal, group housing showed altered, sex‐dependent cocaine‐conditioned place preference compared with controls. Among males, socially isolated mice showed higher place preference than controls, whereas socially isolated female mice showed lower place preference. Since female control mice have higher place preference than male controls, the effects of social isolation appear to equalize behavioral differences between males and females. RNA‐seq analysis revealed both sex‐specific and housing‐dependent differences in gene expression, particularly in the prefrontal cortex and medial amygdala, but not in the nucleus accumbens or VTA.
Addictive drugs cause structural changes in the neurons of the nucleus accumbens. Chronic cocaine exposure results in longer, thinner, and less functional dendritic spines. During withdrawal, there are increases in more mature, large head spines.111 In work seeking to identify connections between gene transcription and regulation of the actin cytoskeleton, which mediate spine growth, PDZ‐RhoGEF, a RhoA‐activating protein, was induced in the nuclei of nucleus accumbens neurons. RhoA increases actin polymerization, decreases the pool of G‐actin, and activates serum response factor, a transcription factor that induces transcription of Rap1b in nucleus accumbens neurons. Rap1b is both necessary and sufficient for cocaine‐induced formation of thin spines. In behavioral studies, Rap1b knockout blocked the ability of cocaine to produce a strong place preference. Rap1b overexpression increased behavioral responses to cocaine. While increased PDZ‐RhoGEF and subsequently increased Rap1b activity is observed 24 h after cocaine withdrawal, this pattern is reversed 3 weeks after withdrawal. These differences help to explain the bidirectional effects of cocaine withdrawal on spiny neurons, though the factors that promote this switch remain unknown.112
Yasmin Hurd (Icahn School of Medicine at Mount Sinai Hospital) presented additional data on transcriptional and epigenetic profiles associated with drug use, largely in postmortem human brain specimens. There were significant gene expression differences between heroin users and controls, particularly among genes associated with the glutamatergic system.113 In addition, there were significant epigenetic differences between the two groups with correlations between epigenetic markers and changes in glutamatergic gene expression. In general, genes associated with the glutamatergic system were hyperacetylated in ways that were likely to indicate greater accessibility and higher transcriptional rates. Expression of the histone acetyltransferase NCOA1 correlated with increased mGluA1 expression among heroin users, but not controls. Epigenetic changes also correlated with years of use, with higher acetylation levels in individuals who had longer histories of heroin use.
These epigenetic changes could be reproduced in a rat model of heroin dependence, suggesting a relationship between heroin use and increased acetylation. Since acetylated histones can be recognized by bromodomain proteins that recruit protein complexes involved in gene expression and since reversible epigenetic marks present opportunities as drug targets, it is notable that bromodomain inhibitors are under investigation in cancer clinical trials.114 Of functional relevance, the bromodomain inhibitor JQ1 reduced heroin self‐administration and drug‐seeking behavior in rats.113
Personalized treatment for nicotine addiction
Rachel Tyndale (Centre for Addiction and Mental Health, University of Toronto) described how common pharmacogenomic variation in drug‐metabolizing genes affects individuals’ abilities to quit smoking and/or respond to drug treatments for nicotine dependence. Her hypothesis is that nicotine metabolism plays an important role in smoking behavior. There is a high degree of variability in nicotine metabolism and clearance between individuals, owing primarily to differences in the gene CYP2A6,115, 116, 117 which encodes a liver enzyme that metabolizes and inactivates nicotine.118 Smokers with rapid nicotine metabolism must smoke more to maintain nicotine levels similar to those of slow metabolizers. In one study, smokers with two functioning copies of CYP2A6 smoked almost 10 more cigarettes per day than those with two defective copies.119 Slow metabolizers also smoke less intensely, taking slower, more shallow puffs120 and are more likely to successfully quit smoking, even after controlling for smoking quantity.121, 122 A person's nicotine metabolism rate can also affect how well they respond to smoking cessation therapies. In a trial of bupropion, a dopamine and norepinephrine transport inhibitor, versus placebo, bupropion improved quit rates among fast nicotine metabolizers, but not among slow metabolizers.123 Bupropion is metabolized by the liver enzyme CYP2B6. Animal and clinical studies suggest that hydroxybupropion, a metabolite of bupropion, may also be an active agent for smoking cessation success, as higher levels of hydroxybupropion correlate with higher quit rates.124 Genetic variants in CYP2B6 that affect the rate of bupropion metabolism have been identified. Therapeutic drug monitoring of hydroxybupropion or CYP2B6 genotyping could be useful to guide individual dose adjustments in patients taking bupropion.
While bupropion is more effective among fast nicotine metabolizers, randomized trials have shown that the nicotine patch is more effective among the slow nicotine metabolizers than fast metabolizers. More rapid nicotine clearance would be expected to decrease nicotine levels in people using nicotine replacement therapies, such as the patch, thereby limiting their effectiveness in smoking cessation.125, 126 In a randomized study comparing efficacies of the nicotine patch, varenicline, a nicotinic receptor partial agonist, or placebo in smoking cessation, varenicline treatment was more effective than the patch among normal nicotine metabolizers, while the two treatments were equally effective in slow metabolizers.127 On the basis of these results, it may be possible to personalize smoking cessation treatment based on the CYP2A6 genotype. The nicotine patch may be appropriate for slow metabolizers as it is effective, inexpensive, and has a low side effect profile. Varenicline may be more appropriate for normal metabolizers, as it was more effective than the patch in this group.
One remaining question is how nicotine metabolism affects smoking rates weeks to months after nicotine has cleared the body. Some clues have emerged from brain imaging studies. Cue‐evoked images had larger effects on brain activity in fast metabolizers compared with slow metabolizers among active smokers, suggesting that differences in nicotine metabolism may have long‐lasting effects on the brain.128
Effects of addictive drugs on the developing brain: adolescents and young adults
Diana Fishbein (Pennsylvania State University) described adolescence as a period of significant, rapid brain development during which the adolescent brain provides increased addiction‐related risks. The last region of the brain to develop is the frontal cortex, responsible for executive functions that include impulse control, risk determination, evaluation of consequences, and decision making. During adolescence and into the early 20s, there are significant changes in both gray and white matter in the frontal cortex, including continued myelination, gray matter thinning, and pruning of excess connections established earlier in development.4, 129 The prefrontal cortex also serves to modulate the activities of noncortical systems, such as regulating emotional circuit activity in the limbic system. Imaging studies have revealed that the dopaminergic connectivity to the frontal cortex is weaker in children than in adults. As the prefrontal cortex matures during adolescence, there is a linear increase in inhibitory control. However, there is also an increased activity in the nucleus accumbens, which increases reward sensitivity.130 These differences are likely to contribute to the greater risk‐taking behavior, novelty seeking, and impulsivity that can be observed in adolescence.
Because their brains are still developing, adolescents are thus more likely to both engage in risk‐taking behaviors, including drug self‐administration, and display enhanced vulnerability to the effects of drugs and alcohol. Large epidemiological studies show that adolescents and young adults are more likely to start using drugs than older adults. Adolescents are also more likely to transition from experimenting with drugs to develop substance use disorders.131 In the National Longitudinal Alcohol Epidemiologic Survey, 45% of people who began drinking before the age of 14 grew up to ultimately have an AUD, compared with 10% of people who began drinking after the age of 21.132 The more recent National Epidemiologic Survey on Alcohol and Related Conditions showed that those who started drinking at an early age were more likely to experience alcohol dependence within 10 years of beginning drinking.133
The effects of alcohol on brain development
Alcohol use among adolescents has been associated with differences in and/or changes in brain structure and function.134 Fifteen‐ and 16‐year‐olds enrolled in an alcohol treatment program showed about a 10% deficiency in memory compared with nondrinking, matched controls as well as differences in cerebellar, hippocampal, and prefrontal cortex volume and white matter quality.135
Susan Tapert (University of California, San Diego) described work on the Youth at Risk Study to elucidate how adolescents who use drugs and alcohol differ from those who do not. In this longitudinal study of 300 middle schoolers who had not started drinking at the time of study entry, 40% remained nondrinkers, 30% drank modestly or moderately, and 30% had become heavy drinkers when studied 4 years later. Adolescents who began drinking during the follow‐up period performed worse on several measures of neurocognition compared with those who did not start drinking. Girls who started drinking showed greater impairment in memory tests; boys did worse on tests of visual attention.136 Extreme binge drinking was associated with poorer performance on several measures of verbal learning and memory compared with subjects who engaged in little or no binge drinking.137
Several differences in brain structure and activity were also observed. Adolescents who initiate heavy drinking showed accelerated reduction of gray matter volumes, particularly in temporal and lateral frontal areas, and attenuated growth in white matter. Structural MRI data also reveal greater reductions in brain volume in the left temporal lobe, caudate, thalamus, and brain stem.138
Functional MRI studies reveal that teens who become heavy drinkers display lower levels of brain activity during visual working memory tasks even before they start drinking, without any deficit in performance.139 This suggests that they are not as cognitively involved in the nonrewarding task as those who do not become drinkers. After they become drinkers, however, their brain activity increases during these tasks, likely indicating that their brains are working harder to do the same task than their nondrinking counterparts.
Lindsey Squeglia (Medical University of South Carolina) described baseline characteristics that predicted future heavy drinking. Demographic variables, including male gender and higher socioeconomic status, were associated with higher risk of becoming a heavy drinker, as were behaviors that included conduct disorders. In addition, several neuropsychological and imaging characteristics differed at baseline in future heavy drinkers. Subjects who began drinking had pre‐existing smaller volumes in the inferior frontal cortex, cingulate, and cerebellum. They also had thinner cortex in several brain regions and showed less brain activity during a working memory task. A predictive model consisting of 38 variables—demographic, behavioral, neuropsychiatric, and imaging—could predict future heavy drinking with 74% accuracy. Understanding the risk factors for heavy drinking among teens can help pediatricians and counselors.140
Functional MRI studies have also shed light on the role that the media and advertising may play in substance use initiation. A group of 15‐ to 18‐year‐olds composed of heavy binge drinkers and moderate, nonbinge drinkers was shown a series of advertisements for alcoholic and nonalcoholic products. Heavy drinkers showed greater brain activation when looking at the alcohol ads compared with nonalcohol ads. Conversely, the nonbinge drinkers showed no difference in brain activation for the two types of ads.141 Abstaining from alcohol for 5 weeks was able to attenuate this response, suggesting that the response may be reversible.142
Two large‐scale, longitudinal studies are underway to better understand the effects of alcohol and/or other abused substances on neurodevelopment in adolescents. The National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA) has recruited over 800 adolescents from ages 12 to 21, many of them at risk of substance use, among five sites in the United States by 2016. Subjects are being followed up annually with neuroimaging and neurophysiological testing.143 A second, larger study, the Adolescent Brain Cognitive Development (ABCD), is following over 11,000 9‐ to 10‐year‐olds for 10 years in 19 sites around the United States. Enrollment of the study began in 2016, and as of early 2018, over 11,000 participants have been enrolled, with baseline data available to researchers for 4500 participants.144, 145
In utero exposure to cannabis
Yasmin Hurd (Icahn School of Medicine at Mount Sinai Hospital) noted that cannabis use is relatively common among pregnant women. In a retrospective study of newborn drug testing in the United States, nearly 20% of fetuses tested positive for THC.146 The long‐term implications of this exposure are unclear. In postmortem brain samples of human fetuses, Hurd showed that exposure to cannabis in utero was associated with lower levels of the dopamine D2 receptor in the amygdala and nucleus accumbens. Lower expression correlated with more maternal smoking.147 Similar expression patterns are observed in animal models, which last into adulthood.148
Perhaps most strikingly, cannabis use may even affect future generations even without direct exposure. In rats, exposure to cannabis during adolescence has epigenetic and behavioral effects in unexposed offspring and later generations. Offspring of parents exposed to THC were more likely to self‐administer heroin and had changes in DNA methylation in genes associated with synaptic plasticity, psychiatric disorders, and neurodevelopment.149
Effects of marijuana on the adolescent brain
In 2015, daily marijuana use (6%) surpassed daily cigarette use (5.5%) among high school seniors. At the same time, the perception of risk and harm associated with marijuana among high schoolers fell to an all‐time low (29%).150 Understanding the effects of marijuana on this vulnerable population will continue to be important as legalization efforts continue across the country.
While several studies have compared the effects of marijuana use on brain structure and function in smokers versus nonsmokers, few have looked at whether the age at which a person starts smoking cannabis comes into play. Staci Gruber (McLean Hospital, Harvard Medical School) investigated whether the age of onset of marijuana use affects neurocognitive performance, brain function, and brain structure. In fMRI studies, late‐onset smokers had activation patterns that were more similar to control, nonsmokers than to early‐onset smokers. Adolescents who began smoking earlier (before 16 years of age) smoked nearly twice as often and more than 2.5 times as much as those who began smoking later. Earlier age of onset was also associated with poorer performance on measures of executive function than late age of onset.5, 151
Marijuana users also exhibit differences from controls in brain volume, mass, and shape. There are regional differences in cerebral cortical thickness compared with nonusers and differences in density and in gyrification, a measure of the folding of the cortex in the gray matter that has been related to poor performance on attentional tasks. Among early‐onset smokers, significant reductions in white matter integrity have been observed by diffusion tensor imaging compared with images from late‐onset smokers and nonsmokers. This difference in white matter integrity was associated with higher self‐reported impulsivity among early‐onset smokers, but not among late‐onset smokers or nonsmoking controls,152 although the causality of this relationship is not fully understood.
Susan Weiss also noted both the importance of determining the effects of marijuana on the developing brain, given its increasing prevalence and availability to adolescents, and the lack of consistency in the field. For example, a large longitudinal study in New Zealand reported that persistent cannabis use was associated with a decline in IQ;153 however, twin studies failed to observe this connection.154, 155 Studies of the effects of marijuana on brain structure are also mixed.156 Gaps in knowledge include whether the effects of marijuana are reversible with abstinence; how varying doses, strains, and potency of cannabis affect outcomes; whether there are gender‐specific effects; and how the age of onset influences cannabis effects.
Electronic cigarettes and adolescents
Thomas Eissenberg (Virginia Commonwealth University) noted that according to National Youth Tobacco Surveys, while cigarette use has declined among high schoolers,157 e‐cigarette use has been steadily increasing.158 E‐cigarettes are now the most popular tobacco product among U.S. adolescents with 16% of high schoolers reporting recent e‐cigarette use.158 E‐cigarettes represent a constantly evolving class of devices for which there are currently few standards and only emerging regulation; several FDA‐announced regulations are being challenged by e‐cigarette companies.159, 160, 161
Eissenberg expressed concern that e‐cigarettes are being marketed to young consumers with flavors like blue cotton candy, applejack, and hard candy. Advertisements geared toward young adults focus on low nicotine products, which may function as gateway products. These devices deliver nicotine poorly, giving new users opportunities and chances to try products and experiencing some positive reinforcement without the acute nicotine toxicity that they might experience with more potent products.162, 163, 164
E‐cigarettes are often marketed as a safer alternative to cigarettes, with a focus on reduced risk of lung cancer. However, Olusegun Owotomo (University of Texas at Austin) stressed that nicotine addiction is an important health risk of e‐cigarettes. He showed that among a nationally representative sample of 8th and 10th graders from the Monitoring the Future Study, adolescent e‐cigarette users endorsed a number of attitudes, perceptions, and characteristics that are risk factors for cigarette smoking compared with nonusers.165
The variability between products in relation to construction, power, and components in the e‐liquid makes it difficult to study e‐cigarettes as a class. In addition, e‐cigarette liquid usually contains flavorants intended for consumption that have not been tested for inhalation safety. Finally, the presence of other potentially toxic compounds, such as aerosolized propylene glycol or formaldehyde, depends on the type of device and liquid used. Unsurprisingly, there are few data on the long‐term health effects of e‐cigarettes. Indeed, such data will likely be difficult to gather given the variability not only between devices but also between users.
Commercial e‐cigarette products vary widely with respect to how much nicotine they deliver, with some devices delivering more nicotine than a conventional combustible cigarette. Factors such as construction, battery power, the liquid used, and user behavior can significantly affect the amount of nicotine that is delivered to the smoker.166, 167, 168, 169 Increasing the battery wattage by a factor of two can quadruple the amount of nicotine delivered.170 Level of experience can also affect nicotine delivery. Cigarette smokers trying e‐cigarettes for the first time were not as efficient as experienced e‐cigarette smokers. Experienced smokers take drags that are twice as long as those of new smokers, keeping the heating element activated for longer time periods and delivering more nicotine.171
Using neuroscience to tailor drug prevention programs
Diana Fishbein (Pennsylvania State University) described some of the efforts to use neuroscience research in drug prevention programs. Exciting developments in neuroscience have the potential to inform the development of preventive antismoking interventions in a more targeted, precision‐based manner. While many evidence‐based prevention interventions have been shown to be effective, the effects are often modest.
Research on neural networks, genetics, and epigenetics should help to lead to tailored and targeted interventions. One area that has shown to be effective is targeting of stress regulatory systems. Chronic stress can prime the brain for novelty seeking and drug use.172 Interventions that target stress physiology and neural markers have shown some efficacy in behavioral change. Mindfulness programs have the potential to affect brain function and structure across age groups.173, 174, 175, 176 The PATHS curriculum increases social competence and decreases behavioral problems.177 The Early Risers program can promote executive function and reduce conduct problems among homeless youth.178 The Head Start REDI program promotes gains in executive function that partially mediate school readiness among kindergarten children.179 Interventions have even shown to improve physiological markers of stress. A family‐based intervention conducted in young foster children normalized cortisol levels and improved hypothalamic‐pituitary‐adrenal axis functioning.180
Newer approaches to treating addiction: drug vaccines
Ron Crystal (Weill Cornell Medical College) described work in developing vaccines against addictive drugs that would prevent them from entering and thus affecting the brain. Addictive drugs are small molecules that are not highly immunogenic, however. Though the immune system does not readily produce good antibodies directed against addictive drugs, this hurdle is being addressed via two approaches.
Active vaccination strategies conjugate the addictive drug to adenovirus capsid proteins, which are highly immunogenic. Crystal has developed a cocaine vaccine by conjugating the cocaine analog GNE to adenovirus that has been denatured so that it cannot replicate.188 The vaccine, dAd5GNE, can engender high anticocaine antibody titers in both rodent and nonhuman primate models.189 The vaccine prevents cocaine distribution in the brain in rodent models, even with frequent administration and at very high doses.190 It also reduces cocaine self‐administration in nonhuman primates.191
A phase 1 clinical trial is underway in human cocaine users. Participants receive 6 monthly injections of either conjugate vaccine or placebo. The study will investigate the ability of three different vaccine doses to produce anticocaine antibodies and will assess safety.192
In passive immunization approaches, expression of a gene that encodes an anticocaine antibody is delivered to the liver using an adeno‐associated virus vector. Transfection with the AAvrh.10 vector containing this anticocaine antibody gene can produce high, persistent anticocaine antibody titers following a single administration to animals. In mice, AAvrh.10 reduced cocaine levels in the brain and reduced cocaine‐induced hyperactivity for periods of months after the transfection.193
Pharmacologic agents for opioid addiction
Over the past decade, the number of heroin users has increased significantly, while the number of deaths due to heroin and prescription opioids has increased fivefold.194, 195, 196 David Gastfriend (American Society of Addiction Medicine) noted that pharmacotherapy can be an important component of a successful treatment program, that multiple agents approved for opioid dependence provide different advantages to suit patients’ needs, but that there is inadequate use of pharmacotherapy to treat opioid dependence. Of the 2.5 million Americans who abused or were dependent on opioids in 2012, fewer than 1 million received medication‐assisted therapy.197
The three approved agents for opioid dependence—methadone, buprenorphine, and naltrexone—each has different characteristics in practice. Methadone is the most tightly controlled and least accessible of the three, dispensed by the 1300 certified methadone clinics in the United States. In randomized controlled trials, methadone treatment has been shown to stabilize people in recovery and to reduce harms including HIV and HCV transmission.198 After terminating methadone treatment, 82% of patients return to heroin use within a year.199 Because of this, methadone is best used as a long‐term treatment. Because of its tightly controlled access, patients with more chaotic lifestyles who need close, daily supervision and who are prepared for long‐term treatment, including those with psychiatric illness or a high tolerance for opioids, may be suitable candidates for methadone treatment.
For patients with more structured lives who can maintain treatment plans without daily monitoring, buprenorphine or buprenorphine/naloxone combinations may be suitable options. While a physician must be licensed to prescribe buprenorphine, any physician can apply to be a buprenorphine prescriber. Retention can be an issue. In one study of HIV‐infected opioid‐dependent patients, 1‐year retention in buprenorphine/naloxone treatment was only 49%.200 Several studies have shown mean retention rates of 2–3 months.201, 202 Up to 92% of patients relapse within 8 weeks of tapering treatment.203 A meta‐analysis of 31 trials showed that methadone maintenance therapy had higher retention rates than low‐dose or flexible‐dose buprenorphine therapy. However, fixed medium or high buprenorphine doses, though less common in office‐based clinical practice, were equivalent to methadone in rates of retention in treatment and suppression of illicit drug use.204 Buprenorphine may disrupt cognitive function less often than methadone, especially early in treatment. During the first months of treatment, patients on methadone show greater delays in reaction time than controls.205 After maintenance is established, this discrepancy is diminished but is still present for several measures of cognitive function.206, 207, 208, 209
The newest agent available for opioid dependence, the antagonist naltrexone, was recently approved as an extended‐release formulation administered as a monthly injection. Naltrexone XR is a significant improvement over short‐acting oral naltrexone, which failed to improve retention or abstinence210 rates more than placebo and was associated with three‐ to sevenfold higher death rates than methadone.211 In brain imaging studies, naltrexone XR modulates the brain's response to drug cues in abstinent heroin‐dependent patients.212 There were both activity decreases in limbic regions and activity increases in areas involved in self‐reflection and self‐regulation, including the medial frontal gyrus. In a double‐blind, placebo‐controlled trial, naltrexone XR treatment provided significantly higher rates of abstinence, decreases in cravings, and higher retention rates.213 Naltrexone XR has also been shown to be more cost effective than methadone or buprenorphine. Although naltrexone XR is more expensive, overall healthcare costs, including inpatient, outpatient, and addiction care, are significantly lower in patients treated with naltrexone XR (US$8,582 in the first 6 months) compared with methadone (US$16,752).202
Because naltrexone requires that patients be opioid free before beginning treatment, it is most appropriate for motivated patients who are dedicated to undergo detox and who have structure and support systems in place. A theoretical concern with naltrexone treatment is that the abstinence associated with naltrexone ablates opioid tolerance in ways that may predispose to increased overdose risks if treated patients relapse to opioid use. In clinical trials, however, no overdose‐related deaths have been observed up to 18 months.214, 215
Monitoring treatment efficacy for opioid use disorder
Silvia Lopez‐Guzman (New York University) described research detailing how impulsivity can predict relapse in patients undergoing treatment for opioid dependence. In a meta‐analysis of 46 studies, temporal discounting (a measure of the value of a reward that is available now versus one available at different times in the future) was consistently higher in substance users for several drugs of abuse.216 There is also evidence that treatment for opioid dependence may reduce impulsivity.217 In a longitudinal study of patients starting treatment for opioid dependence and matched controls, Lopez‐Guzman measured each subject's impulsivity at several time points using a model of temporal discounting in which the person is asked to decide between receiving a small sum of money now or a larger sum at some point in the future. For individuals with high impulsivity, rewards lose their value more quickly over time, that is, the longer they must wait to receive a reward, the less valuable it is to them. While baseline impulsivity levels did not predict relapse rates, patients who experienced increased impulsivity during the trial often relapsed. Impulsivity‐rate increase correlated with an increase in relapse and may be a marker for patients at risk of relapse. Conversely, a decrease in impulsivity over time was associated with reduced relapse and may be a signal of recovery or resiliency.
Since impulsivity can be monitored by smartphone or tablet apps, testing could be useful to identify patients who are responding well to opioid‐dependence treatment versus those at elevated risk of relapse.
Public policy considerations for addiction: legalized marijuana
Susan Weiss (National Institute on Drug Abuse) summarized thoughts that much of the current research into effects of policy is inconclusive as it fails to take into account heterogeneity between states with regard to how new policy laws are implemented. In addition, the measures being collected, namely prevalence of drug use, do not always correspond to measures of harm, such as hospitalizations. State legislatures legalizing marijuana are not using evidence‐based policy research from tobacco and alcohol control. The legalization process, such as regulations on advertising, pricing, taxes, and potency, can have significant effects on mitigating the harm of legalized marijuana.
Benefits and harms of public health policies
Mark Kleiman (New York University) discussed the societal effects of addiction and public policies aimed to curb addiction. It is clear that addiction and drugs of abuse can have negative consequences on public safety and public health, whether in the form of overdose, sexually transmitted diseases, violence, or drunk driving. What may be less clear, however, is that the public policies put in place to mitigate these harms may have unintended consequences of their own. For example, forbidding the use of illicit drugs often creates illicit markets, which carry risks in terms of drug‐related violence, police enforcement, and incarceration. Therefore, when thinking about public policy, Kleiman stressed that it is important to take a holistic view and realize that drug‐control policies can increase harm, even as they attempt to decrease use.
Kleiman advised that there are several strategies for curtailing drug use, including levying taxes, marketing and antimarketing campaigns, and sobriety programs. These methods can have a significant effect on overall public health. For example, the Sobriety 24/7 program, which establishes zero tolerance for repeat drug and alcohol‐related offenders, was shown to reduce automobile crashes by 12%, domestic violence reports by 8%, and all‐cause mortality by 4%.218
The ongoing cannabis legalization efforts offer an opportunity for policymakers to learn from lessons in alcohol and tobacco regulation. However, Kleiman argues that, owing to the influence of corporate interests, which focus on the small percentage of heavy (and profitable) users, effective measures to moderate cannabis use have not been implemented. Such measures could include limiting sales to government stores, high taxation to prevent price collapse, implementing standardized measures of cannabis intake, regulating the concentrations of THC and cannabidiol, and requiring users to pass a cannabis test and receive a license to become a registered cannabis user, similar to a driver's license.
Improving the state of mental health care
Patrick J. Kennedy (One Mind, former United States representative from Rhode Island) oversaw the passage of the Mental Health Parity and Addiction Equity Act of 2008 (MHPAEA), which requires health insurance companies to cover mental health and substance use disorders in the same manner as physical disorders and diseases. Issues with the current state of mental health and addiction care include a need for more physicians willing to treat addiction, the lack of coordinated and integrated care with a patient's primary provider, and the need to incorporate proven, evidence‐based methods into treatment. Public perception needs to change to view addiction as a chronic disease. Like other chronic diseases, there is no cure for addiction. Therefore, retention in addiction treatment programs is critical. In addition, the care model for chronic diseases, in which patients are continually monitored to assess the effectiveness of treatment, and treatment plans are modified in accordance with measurable markers of disease control, should be adopted for addiction treatment.
There are cautions concerning roles for increasingly large marijuana companies in the current cannabis legalization processes. Similar to what has been seen with the tobacco and alcohol industries, the marijuana industry may be targeting young, poor, and minority customers.219, 220 Legalization has not improved the potential negative impact of marijuana on these populations. In Colorado, for example, marijuana‐related arrests among Black and Hispanic youth are higher after the legalization of cannabis than they were before.221 With the state of mental health care in this country, addiction can be devastating. Kennedy warned that “we cannot stand by and watch a new industry like Big Tobacco take over and addict poor people who don't have access to [comprehensive addiction treatment].”
Conclusions
Attendees at the Sixth Annual Aspen Brain Forum “The Addicted Brain and New Treatment Frontiers” were privileged to hear NIH institute directors and political leaders join with clinical, translational, and basic researchers to discuss their varying perspectives on the neurobiology of addiction. Therapeutic opportunities and increased political and regulatory sophistication have been brought to bear on the challenge of diagnosing, preventing, and treating addictions. The magnitude of the human burden of the disease of addiction and timeliness of the challenges that addiction provides were never far from even the more technical scientific discussions. Hope, excitement, and challenges animate this field and are all found in this exceptional conference.
Competing interests
G.F.K. directs the National Institute on Alcoholism and Alcohol Abuse, which funded significant portions of the work reported herein.
Acknowledgments
This report was written to reflect topics presented at the Sixth Annual Aspen Brain Forum, “The Addicted Brain and New Treatment Frontiers” (https://www.nyas.org/events/2016/the-addicted-brain-and-new-treatment-frontiers-sixth-annual-aspen-brain-forum/?tab=sponsors), which convened at the New York Academy of Sciences May 18–20, 2016. Publication of the report satisfies requirements for disseminating the forum's content. All topics discussed by the authors (J.C., G.R.U., and G.F.K.) reflect their interpretations of the speakers’ presentations at the forum in 2016. Open Access of the report was supported by the National Institute on Drug Abuse of the National Institutes of Health, Award Number R13DA041813.