Drugs such as cocaine, amphetamine, nicotine, alcohol, and marijuana are commonly used for their mood- and mind-altering properties. These substances also have the potential to be addictive. In some people, regular use leads to “addiction” or “dependence,” i.e., compulsive and repetitive drug-seeking behavior despite negative health and social consequences. However, this type of behavior does not occur in all users (see ). Many people who experiment with drugs do not find the effects rewarding and avoid them in the future. Some people enjoy the effects of the drugs and use them recreationally without ever becoming dependent. For others, however, the drugs gain powerful control over their lives and may replace all other healthy pursuits (see ). The majority of people who self-administer drugs of abuse begin during adolescence. Epidemiological studies have shown that earlier onset of drug intake is associated with greater likelihood of development of substance use problems. However, there is debate about whether early onset uniquely affects brain development in such a way as to promote pathological behavior or whether the same genetic and environmental factors that make an individual likely to develop drug problems also make them likely to initiate early. This review summarizes results from animal models in which the effect of age of onset has been examined.
Fig. 1 Percentages of the US population over the age of 12 years who have ever tried the indicated drug (top number, light gray circle); who used the indicated drug in the past month (middle number, darker gray circle); who meet criteria for dependence on the (more ...)
The terms “addiction,” “drug abuse,” and “drug dependence” are used interchangeably in the vernacular and have varying definitions in psychological, sociological, and neuroscience literature. For the sake of clarity, we will refer to the two substance use disorders (SUDs), drug dependence and drug abuse, as they are defined by the Diagnostic and Statistical Manual of Mental Disorders version IV (DSM-IV 1994).
For a diagnosis of drug abuse, a patient must present at least one of the following four characteristics:
- Recurrent substance use resulting in the failure to fulfill major role obligations at work, school, or home
- Recurrent substance use in situations in which it is physically hazardous
- Recurrent substance-related legal problems
- Continued substance use despite having persistent or recurrent social or interpersonal problems caused or exacerbated by the effects of the substance
For a diagnosis of drug dependence, a patient must present three of the following seven characteristics:
- The substance is taken in larger amounts or over a longer period than was intended
- There is a persistent desire or unsuccessful efforts to cut down or control substance use
- A great deal of time is spent in activities necessary to obtain the substance, use the substance, or recover from its effects
- Important social, occupational, or recreational activities are given up or reduced because of substance use
- The substance use is continued despite knowledge of having a persistent or recurrent physical or psychological problem that is likely to have been caused or exacerbated by the substance
Two of the criteria for drug dependence, withdrawal and tolerance, relate to physiological phenomena that ensue from repeated drug taking and are relatively easy to measure in animal models. New behavioral methods are approaching success at modeling increased intake, intake despite negative consequences, and the choice between drug intake and other activities, as described below.
The DSM-IV criteria provide a “snapshot” that clinicians can use when a patient requires diagnosis or treatment. However, drug dependence is actually a progressive disease, with several defined stages that often overlap with adolescence (Kreek et al. 2005
; see and ). Drug dependence necessarily begins as experimental drug use; no person can become dependent without first taking a drug. Most people try drugs (at least alcohol or tobacco) at some point in their lives, typically experimenting during the late teenage years and early 20s (Chen and Kandel 1995
). Some users repeat drug use under recreational circumstances. Recreational drug use can vary widely but is defined by the fact that the user has control over it. Recreational users seek drugs for their rewarding properties and not out of compulsion (Kalivas and Volkow 2005
). Drug abuse and dependence begin to emerge when use becomes compulsive. The likelihood of progression from experimentation to recreational use to dependence varies by drug. provides a visual interpretation of this point by depicting the percentage of the population of the USA over age 12 that has ever taken a particular drug, uses regularly, or is dependent. Although the percentage that develops dependence varies by drug and is likely influenced by cultural and legal factors, the dependent population represents a small subset of those who have experimented with a drug. A key research question, therefore, is why do some drug users develop SUDs, while others can remain purely recreational?
Stages in the progression to drug dependence (rectangles) and animal models related to each stage (ovals; self-admin, self-administration)
It is crucial at this point to define what we mean by “young.” Experimentation with alcohol, tobacco, and marijuana typically begins during the teenage years (SAMHSA 2008
). Use of alcohol peaks around age 18–20 and declines into adulthood (Chen and Kandel 1995
). Marijuana and tobacco use peaks slightly later, between ages 19 and 22 (Chen and Kandel 1995
). Cocaine use peaks in the early to mid-20s and also declines into adulthood (Chen and Kandel 1995
). The typical age-related pattern of drug use involves experimentation in the late teens and early 20s, so those who experiment before these typical times (alcohol and cigarettes in late childhood or the early teens or illegal drugs in the teens) are the most at risk. While many studies use an age of onset before 15 years as the cutoff for “early onset,” there is, in general, an inverse correlation: younger users are more likely to develop SUDs.
While the inverse correlation between age of onset and SUD liability is well established in humans, it does not tell us whether early use is causal. Epidemiological studies to test causality require twin or longitudinal studies which are difficult and rare. Two twin studies have resulted in conflicting results, albeit with different substances. One large study examining the risk for alcohol abuse and dependence reported that age of onset was correlated with but not causal in development of alcohol use disorders (Prescott and Kendler 1999
). In contrast, a smaller study of twins who were discordant for early-onset marijuana use reported that age of onset was causal in development of later drug use and abuse problems (Lynskey et al. 2003
). Thus, there is sparse evidence and lingering debate within the epidemiological literature regarding the causality of early-onset drug use as it relates to later drug problems. Human studies show that family history and psychopathology both increase the likelihood of early initiation (Tarter et al. 1999
; Franken and Hendriks 2000
; McGue et al. 2001a
). Do these biological and environmental effects, therefore, operate through early initiation to increase vulnerability to SUDs? Or would users with a family history and/or psychopathology develop SUDs no matter when they initiate? These questions are difficult to address in human studies. To fully address the causality of early drug exposure on later SUD, animal models are necessary.
Animal models have the distinct advantage of experimental control. Experimenters can randomly assign the age of initial exposure, as well as the drug, dose, duration, and timing of exposure, whereas, in human studies, these conditions are determined by the user. For this reason, animal models have provided much valuable information. However, one drawback to animal use is that no model completely recapitulates the stages in development of SUD. For this reason, we must integrate results from multiple behavioral and neurobiological models to achieve a full understanding.
Rodent behavioral models of substance use disorders
Rodent behavioral tasks model basic processes that are components of SUD pathology but cannot completely mimic the disease. Multiple models have been used which vary in validity and relevance to the human condition and are summarized below and in .
Conditioned place preference
Conditioned place preference (CPP) is designed to assess whether a drug is rewarding. The animal is trained to associate a place with the rewarding effects of experimenter-injected drug-induced sensations. If the animal later freely approaches the drug-associated place, then the drug is deemed rewarding (Carr et al. 1989
; Bardo and Bevins 2000
). Rewarding drugs, it is assumed, are more likely to be sought than nonrewarding drugs. This test is useful in measuring the level and persistence of drug-induced reward. It is not a useful model of pathological drug seeking or taking. This test is also highly dose sensitive: drugs of abuse are typically rewarding at low to moderate doses and aversive at high doses.
Conditioned place and taste aversion
These tests are designed to assess aversive effects of drugs of abuse. It is assumed that aversive effects discourage intake. In these tasks, the animals are trained to associate a place or an otherwise palatable flavor with the sensations ensuing from an experimenter-injected drug (Welzl et al. 2001
). Subsequent avoidance of the place or flavor indicates aversive effects. These tests measure the use-limiting effects of drugs of abuse but do not model pathological drug seeking or taking.
Withdrawal is a constellation of affective and physiological changes that occurs after cessation of intake of some drugs of abuse. Symptoms vary based on the drug consumed, duration, and extent of exposure and generally reflect the reversal of initial drug effects. Many of these behaviors are easily quantified in animal models. For example, ethanol withdrawal is marked by signs of autonomic arousal and behavioral activation such as piloerection, locomotor activation, tremor, and seizures (Majchrowicz 1975
). Withdrawal from opiates elicits both behavioral and autonomic activation as indicated by ptosis, teeth chatter, lacrimation, wet dog shakes, and jumping (Rasmussen et al. 1990
). Withdrawal from nicotine includes autonomic and behavioral signs such as body shakes, tremors, writhing, escape attempts, chewing, gasping, ptosis, teeth chattering, and yawning (O’Dell et al. 2007b
). All of these signs are analogous to effects in humans (DSM-IV 1994). For psychostimulants such as cocaine and amphetamine, physiological withdrawal signs such as these are rarely observed (DSM-IV 1994). Withdrawal from the psychostimulants and most other drugs of abuse elicits a generalized “negative motivational state” characterized by elevated reward threshold which can be assessed using intracranial self-stimulation (O’Dell et al. 2007b
). Withdrawal also elicits an anxiety-like state which can be assessed using multiple models such as the social interaction test, elevated plus maze, light–dark task, and others (see below).
Most abused drugs stimulate locomotor behavior through activation of the dopaminergic circuits that contribute to their reinforcing effects (Wise 1987
; Di Chiara 1995
). Cocaine and amphetamine typically increase motor activity in two ways. At lower doses, ambulatory activity is increased, which is most often measured as an increase in matrix crossings or distance traveled. At higher doses, locomotion falls and stereotypic behavior can emerge, which is manifested as an increase in sniffing, grooming, head bobbing, or other repetitive behaviors and a consequent decrease in distance traveled. Ethanol, in humans, tends to be activating at low doses (which may result from reduced inhibitions) and sedating at high doses (DSM-IV 1994). In rats, ethanol has been reported to either increase or decrease locomotion, but dose effects do not consistently parallel the human pattern (see below). Similarly, nicotine can either increase or decrease locomotion in rodents (see below). Opiates can also cause locomotor activation (Buxbaum et al. 1973
; Pert and Sivit 1977
; Kalivas et al. 1983
). Mu opioid agonists cause locomotor stimulation in both mice and rats, and repeated treatment causes sensitization (Rethy et al. 1971
; Babbini and Davis 1972
; Stinus et al. 1980
; Kalivas and Stewart 1991
; Gaiardi et al. 1991
). In summary, acute motor responses are one indicator of drug sensitivity but are highly variable.
Repeated exposure to any of these drugs can lead to a phenomenon called sensitization, in which the ambulatory or stereotypic response to a repeated low dose is augmented (Shuster et al. 1975a
; Aizenstein et al. 1990
; Segal and Kuczenski 1992a
). Sensitization is a manifestation of neuroplastic changes in response to repeated exposure, and some researchers have hypothesized that it is a behavioral correlate of increased drug craving and development of dependence (Robinson and Berridge 1993
), although others debate this assertion (Di Chiara 1995
). Clearly, sensitization represents a lasting neuroplastic change that is easily measured. Its relevance to drug dependence is still debated.
Since humans who become drug addicts voluntarily consume drugs, it is important to examine animal models in which drugs are voluntarily administered (or “self-administered”) by the animal. For drugs such as cocaine and nicotine, self-administration (SA) in rodents is achieved via intravenous administration through indwelling jugular catheters (since rodents will not reliably snort or smoke these compounds). Admittedly, while most adolescent humans do not use the intravenous route to administer cocaine and nicotine, they do utilize routes which result in rapid absorption of the drugs into the blood (insufflation of cocaine, smoking of crack cocaine and nicotine). Ethanol presents a much simpler model because oral ingestion is easily performed in rodents, as long as taste, caloric intake, and fluid balance are properly controlled. Both oral and intravenous approaches have been used to assess the age dependence of voluntary intake in animal models.
Animals that acquire drug-seeking behavior more quickly or perform it more frequently are thought to resemble human drug addicts. However, drug taking, even when it is acquired quickly, is not equivalent to drug dependence. Experimental animals will also work to obtain food and other environmental conditions in the absence of any abuse liability. Dependence-like behaviors require more complex testing, and there are several more sophisticated operant conditioning methods currently in use that provide better models of SUDs. One such example is progressive ratio responding, in which each successive infusion requires more lever presses than the previous one. This schedule is designed to assess motivation to seek the drug (Hodos 1961
; Roberts et al. 1989
; Depoortere et al. 1993
). Extinction and reinstatement paradigms are used to model relapse (de Wit and Stewart 1981
; Shaham et al. 2003
). Timeout and punished responding are used to model compulsive use (Vanderschuren and Everitt 2004
; Deroche-Gamonet et al. 2004
). Extended access or long-access (LgA) training schedules are used to model high-level or binge use (Knackstedt and Kalivas 2007
; O’Dell et al. 2007a
; George et al. 2008
; Mantsch et al. 2008
). In a comprehensive self-administration model, Deroche-Gamonet et al. (2004)
combined several of these measures and observed that a small percentage of self-administering rats exhibited multiple dependence-like behaviors, similar to the results obtained in the human population shown in . These models are just beginning to appear in studies comparing adolescents and adults.
Ranking the behavioral models
The relevance of each of these rodent models to human SUD can be debated. In the analysis that follows, we assign greater weight to methods which come closest to modeling human SUD and less weight to models which are not clearly linked to pathological drug intake. Therefore, studies using the more complex methods of self-administration (progressive ratio, extinction, reinstatement, punishment, LgA, etc.) are likely to be the most informative regarding vulnerability to SUD. However, these are the newest techniques, are the most difficult to employ in developmental studies, and consequently are the least examined in adolescent vs. adult rodents. Next, we assign approximately equal weight to studies examining the reinforcing, rewarding, aversive, and withdrawal-related effects of these drugs (simple self-administration, conditioned place preference, conditioned taste/place aversion, and withdrawal measures, respectively). All of these measures are related to the phenomena which promote or discourage drug taking and are therefore useful indirect measures of propensity for drug intake. We rank the locomotor effects of drugs of abuse as less compelling in their validity. Acute locomotor effects are useful indicators of drug sensitivity, and sensitization is widely used as a surrogate for reinforcement. However, locomotion and reinforcement involve overlapping but nonidentical processes (Di Chiara 1995
; Robinson and Berridge 2008
; Vezina and Leyton 2009
In addition to varying relevance to human SUD, these models vary in the phase of development of SUD that they model. Self-administration models both early and late phases of the disease. CPP, conditioned place aversion (CPA), conditioned taste aversion (CTA), and acute locomotor effects model early drug intake, sensitization models repeated intake, and withdrawal model long-term use and attempted abstinence. See .
In this review, we will summarize results from animal models in which the effects of age of onset of drug exposure have been examined in adolescent vs. adult rodents. This comparison is crucial: many studies have examined effects in adolescents only or in adults which were preexposed as adolescents, but such studies do not test for age specificity of the findings. The review focuses on nicotine, ethanol, marijuana, and psychostimulants, as there is a significant literature comparing the effects of these drugs in adolescents and adults. Unfortunately, there are only a few studies which examine narcotics in adolescents (for example, see Zhang et al. 2008
). The outline of the review will follow the path from initial use through addiction. We will begin by examining the effects of a single or a few drug administrations in which the rewarding, aversive, and locomotor effects have been examined. We will then discuss the effects of long-term voluntary intake via oral and intravenous self-administration. Finally, we will discuss evidence from withdrawal studies which model the likely consequences of attempts to quit.
In general, the results obtained from rodent models suggest that:
- Adolescents find some addictive drugs more rewarding than adults
- Adolescent rodents are consistently less likely to demonstrate aversive effects of drugs of abuse
- Adolescent rodents may self-administer higher doses of some drugs of abuse under some conditions
- Adolescent rodents consistently experience less severe withdrawal effects
These conclusions, for which we will provide evidence below, suggest that the developmental stage of adolescence itself could increase early drug taking because addictive drugs are on balance more rewarding and less aversive. However, these studies do not provide any support for the possibility that progression to compulsive use is more likely when drug use begins in adolescence: the critical studies have not been done.
The review will begin with a description of adolescent development in rodents as compared to humans. We will then examine results from studies in which each model has been used to compare adolescent and adult exposure.
Adolescent rodents as models of adolescent humans
We have focused on rodent models of addiction-related behavior because of the extensive data published using these models and the relative simplicity of comparing adolescent and adult rodents. While primate models would be very informative, we found only one study directly comparing drug effects in adolescent vs. adult primates (Schwandt et al. 2007
). Based on data that are reviewed extensively elsewhere (Spear 2000
), we will consider the age range of 28–42 days to be “adolescence” in rodents. By hormonal, physical, and social maturation criteria, this phase of development corresponds to age 12–18 years in humans (Spear 2000
). It is critical to mention that animals are not uniform through this time period. In fact, some behavioral measures discussed below differ markedly between 28- and 42-day-old rodents, much as addiction vulnerability differs markedly between 12- and 18-year-old humans.
Growing evidence suggests that adolescent humans and rodents experience many similar structural and functional changes in the brain as they progress to adulthood. For example, forebrain dopamine innervation is still maturing in both humans (Seeman et al. 1987
) and rodents. Dopamine D1 and D2 receptor levels reach a peak and then decline over adolescence (Gelbard et al. 1989
; Teicher et al. 1995
; Andersen and Teicher 2000
). In addition, connections between the amygdala and prefrontal cortex mature during this phase, as demonstrated by microscopy studies in rodents (Cunningham et al. 2002
) and functional magnetic resonance imaging studies in humans (Ernst et al. 2005
; Eshel et al. 2007
). Thus, brain development during adolescence is likely similar in many ways between humans and rodents.