Research in the Early 1990s

Ferré said that it is well known that the mechanism underlying the motor and reinforcing effects of cocaine and amphetamine are caused by the drugs' stimulation of central dopaminergic transmission, particularly in the striatum. The striatum, the input structure of the basal ganglia, is an area of the brain involved in the elicitation and learning of reward-related behaviors, and it contains the highest concentration of dopamine and dopamine receptors. Cocaine and amphetamine are able to produce psychostimulant effects by binding to what is known as a dopamine transporter and either blocking (e.g., cocaine) or reversing (e.g., amphetamine) its effects. In both cases, the end result is a significant increase of dopamine in the extracellular space, which in turn activates the postsynaptic dopamine D1 and D2 receptors.

In contrast to cocaine and amphetamine, in the early 1990s scientists already knew that the main mechanism underlying caffeine psychostimulation was adenosine receptor antagonism. It was known then that caffeine at brain concentrations obtained after drinking coffee was enough to block the effects of the A1 and A2A receptors, with A2B being involved only in pathological situations and A3 having little affinity for caffeine. (There are four adenosine receptors: A1, A2A, A2B, and A3.) The question then was, How does adenosine modulate the dopaminergic system?

Also in the 1990s, scientists were aware that caffeine does not produce a clear or strong presynaptic dopamine-releasing effect. That is, it does not really increase dopamine in the extracellular space in the brain. Knowing that, Ferré and collaborators investigated the possibility of a postsynaptic interaction between adenosine and dopamine receptor signaling (Ferré et al., 1991a). They used the reserpinized mouse model to test their hypothesis. (Reserpine depletes dopamine and other catecholamines in the brain, resulting in an animal becoming immobile, or cataleptic. The only way to counteract the catatelptic effect is to administer a dopamine receptor agonist, that is, something that stimulates the postsynaptic dopaminergic receptors.) They used bromocriptine (a D2 agonist) to produce locomotor activity in reserpinized mice. They found that the locomotor effect of bromocriptine was counteracted by the adenosine receptor agonists NECA (an A1/A2A agonist) and L-PIA (an A1 agonist) with a potency that suggested predominant involvement of A2A receptors.

Ferré and collaborators (1991a) also found that caffeine (an A1/A2A agonist) and caffeine metabolites theophylline (an A1/A2A agonist) and paraxanthine, but not theobromine, had the opposite effect; that is, they potentiated locomotor activity of bromocriptine. That finding suggested the existence of an antagonistic interaction between the postsynaptic adenosine A2A and dopamine D2 receptors, through which A2A receptor agonists would behave as D2 receptor antagonists, and A2A receptor antagonists would behave as dopamine as D2 receptor agonists. Indeed, in a separate study, Ferré et al. (1991b) demonstrated for the first time that central administration of an A2A receptor agonist would produce catalepsy, as a dopamine D2 receptor antagonist would do. Later, when selective adenosine A2A receptor antagonists became available, others demonstrated the opposite effect: that A2A receptor antagonists elicit motor activation (Karcz-Kubicha et al., 2003).

The findings reported in Ferré et al. (1991a,b) strongly suggested that caffeine produces motor activation by blocking adenosine A2A receptor–mediated inhibition of dopamine D2 receptor activation. Later, through radioligand-binding experimentation, Ferré and his team found evidence for a more direct interaction between the two receptors (Ferré et al., 1991c), with the dopamine D2 receptor antagonist being displaced by dopamine in a dose-dependent manner and with the ability of dopamine to displace the antagonist being modified by the addition of an adenosine A2A receptor agonist (CGS21680). That is, the agonist CGS21680 decreased the affinity of dopamine D2 receptors for dopamine. That experiment also demonstrated that the A2A and D2 receptors should be localized in the same neuron. But which neuron was it?

Subsequent study pointed to the efferent striatal gamma-aminobutyric acid (GABA)-ergic medium spiny neuron, also known as MSN. MSNs are efferent neurons that constitute more than 95 percent of the striatal neuronal population. They receive two main inputs: glutamatergic inputs from the cortical-limbic-thalamic area and mesencephalic dopaminergic inputs from the substantia nigra and ventral tegmental area. There are two subtypes of MSNs, each of which gives rise to a separate efferent pathway connecting the striatum with the output structures of the basal ganglia (i.e., the medial segment of the globus pallidus and the substantia nigra pars reticulate). One of the pathways is direct, the other indirect. Using freely moving rats, Ferré et al. (1993) inserted one probe into the striatum, where cell bodies of the indirect MSN are localized, and another probe into the ipsilateral global pallidus, where the nerve terminals of the indirect MSN are localized and where GABA is released. They found that perfusion of a D2 receptor agonist, pergolide, through the striatal probe resulted in a significant reduction of extracellular levels of GABA in the ipsilateral globus pallidus. The effect was significantly counteracted by the striatal coperfusion of an A2A receptor agonist, CGS21680, and significantly potentiated by the xanthine theophylline.

Two New Concepts

Ferré described what he said were two new concepts being used in pharmacology to help explain the central mechanism of action of caffeine and many other compounds: receptor heteromer and local module. The receptor concept was introduced in 1878; since then, receptors have been considered as single functional units. But that view is changing. A receptor heteromer is defined as a macromolecular complex composed of at least two functional receptor units with biochemical properties that are demonstrably different from those of its individual components (Ferré et al., 2009).

The second concept, local module, relates to the MSN and the convergence of MSN's two main inputs (i.e., the cortical-limbic-thalamic glutamatergic terminal making synaptic contact with the head of the dendritic spine and the mesencephalic dopaminergic terminal making synaptic contact with the neck of the dendritic spine). Together, these various elements—the dendritic spine, the glutamatergic terminal, dopaminergic terminal, and glial processes that wrap around the glutamatergic synapse—constitute a functional unit known as the striatal spine module, a type of local module. A local module is defined as the minimal portion of one or more neurons and/or one or more glial cells that operates as an independent integrative unit (Ferré et al., 2007).

As described by Ferré, the concept of a local module provides a framework for understanding the functional roles of extrasynaptic transmission. Dopamine is released not only intrasynaptically, but also extra-synaptically, which allows activation of extrasynaptic receptors localized at dopamine and glutamate synapses and modulation of glutamatergic neurotransmission. The same is true of glutamate. It is not only released intrasynaptically but also spills over and stimulates extrasynaptic glutamate receptors localized at glutamate and dopaminergic synapses and modulates dopaminergic neurotransmission. Extrasynaptic transmission and extrasynaptic localization of receptors, in turn, provide a framework for understanding the existence and possible functional role of receptor heteromers.

According to Ferré, much work has been done using artificial systems and resonance energy transfer techniques (BRET and FRET), as well as mass spectrometry analysis of peptide-peptide interactions, to demonstrate the formation of A2A-D2 receptor heteromers (Canals et al., 2003; Woods and Ferré, 2005; Navarro et al., 2010). Ferré and his colleagues have used patch-clamp experiments (i.e., with transgenic mice that express green fluorescent protein and show fluorescence in the D2 receptor–containing neuron) to gain an understanding of these interactions at the cellular level. Specifically, they have shown that the N-methyl-D-aspartate (NMDA) receptor induces strong activation, an effect that is completely inhibited by the D2 receptor agonist N-1-naphthylphthalamic acid (NPA) and that the A2A receptor agonist CGS21680, which by itself does not produce any effect, completely counteracts the D2 receptor–mediated inhibition (Azdad et al., 2009). Furthermore, Azdad et al. (2009) found that infusing a peptide corresponding to an A2A receptor epitope involved in A2A-D2 receptor heteromerization interrupts the antagonistic interaction between the A2A and D2 receptors.

Other Mechanisms of Caffeine Psychostimulant Effects

In Ferré's opinion, scientists have reached a high level of understanding of at least one mechanism of action of caffeine: the A2A-D2 antagonistic interaction mediated by the A2A-D2 receptor heteromer localized in the indirect MSN. The mechanism explains not only the motor-depressant effects of A2A receptor agonists but also the motoractivating effects of caffeine and other A2A receptor antagonists (Orrú et al., 2011). On the basis of this knowledge, researchers have been testing the efficacy of A2A receptor antagonists in the treatment of Parkinson's disease.

Not all caffeine effects are mediated by A2A, according to Ferré. Some motor effects are mediated by the A1 receptor (Karcz-Kubicha et al., 2003). Ferré did not elaborate, but he did remark that the same methods were used to identify an antagonistic A1-D1 receptor interaction in the direct MSN that also mediates the postsynaptic effects of caffeine (Ferré et al., 1996).

In addition to postsynaptic mechanisms, presynaptic mechanism could also be involved in caffeine's locomotor-activating effects. Although no evidence indicates that caffeine releases dopamine like cocaine and amphetamine do, Solinas et al. (2002) showed that it does release dopamine in the very ventral part of the striatum, in an area called the shell of the nucleus accumbens, by acting on adenosine A1 receptors localized in glutamatergic and dopamatergic terminals.

A final mechanism for the motor and probably reinforcing effects of caffeine was recently described in the literature (Ferré et al., 2013; Orrú et al., 2013). It involves paraxanthine, the main metabolite of caffeine in humans, which has a very strong psychostimulant effect in rats and is correlated with a significant dopamine release in striatal areas of the brain where caffeine is ineffective. Ferré and his team learned that paraxanthine has a unique pharmacological profile. In addition to being an A1 and A2A receptor antagonist, it is also a selective inhibitor of cGMP-preferring phosphodiesterase (PDE) and thus plays a role in potentiating nitrous oxide transmission.

Most of the mechanisms that Ferré discussed were relevant to the motor and reinforcing effects of caffeine. Arousal is another central effect of caffeine that, according to Ferré, seems to be related to multiple interconnected ascending arousal systems moderated by adenosine A1 receptors (Ferré, 2010).

Conclusions About the Neurological Effects of Caffeine

Ferré concluded with four main summary points:

1.

Two new concepts, “receptor heteromer” and “local module,” facilitate the understanding of the functional role of interactions between neurotransmitters and receptor heteromers in the central nervous system and of the mechanisms of caffeine and other central-acting drugs.

2.

The motor and rewarding effects of caffeine depend on its ability to release the pre- and post-synaptic brakes that adenosine imposes on dopaminergic neurotransmission by acting on different adenosine A2A and A1 receptor heteromers localized in different elements of the striatal spine module.

3.

The arousal effects of caffeine depend on its ability to release the A1 receptor-mediated inhibitory modulation of the highly interconnected multiple ascending arousal systems.

4.

Paraxanthine, the main metabolite of caffeine in humans, displays a strong psychostimulant profile that depends on its selective ability to potentiate nitric oxide neurotransmission.

Three Vulnerable Populations

From the perspective of caffeine use, Temple identified three vulnerable populations: (1) pregnant women, with some evidence that excessive caffeine may increase the risk of miscarriage but with little known about the effects of caffeine use during pregnancy on offspring later in life; (2) children, because of their exposure to high doses in terms of milligrams of caffeine per kilogram of body weight and because caffeine may be a gateway to other substances; and (3) adolescents, because of escalating use during adolescence and the combining of energy drinks and alcohol.

Focusing just on children and adolescents, Temple identified three main differences between those two populations and adults that explain why she considers children and adolescents to be vulnerable populations. First, sources of caffeine are different, again with children and adolescents drinking more soda and adults drinking more coffee. Although the caffeine content of coffee can vary on the basis of how it is brewed and where it is purchased, nonetheless caffeine is a natural component of coffee. Soda and energy drinks do not naturally contain caffeine. Rather, those beverages are vehicles for caffeine. A second difference is that the lifetime experience with caffeine is very different in children than in adults. Most adults consume caffeine and have had a history of caffeine use, which affords them some tolerance to the effects of caffeine. In contrast, children, especially young children, are fairly naïve with respect to caffeine use. They tend to consume caffeine at relatively low doses and with less frequency or less regularity than adults do, which may make them particularly vulnerable to the effects of a large amount of caffeine consumed at once. A third difference is that children's and adolescents' brains are still developing, especially in the frontal lobe, with little known about the impact of high levels of caffeine on the brain during this critical period of brain development.

Evidence on the Effects of Caffeine in Children and Adolescents

When Temple and her colleagues first starting studying the effects of caffeine in children and adolescents, about 7 years ago, so little research had been conducted that she felt as though they were starting from scratch. Her research has focused on four main areas: reinforcing properties of caffeine, cardiovascular responses to caffeine, subjective effects of caffeine, and cognitive effects of caffeine. She discussed each in turn.

Reinforcing Properties of Caffeine

Curious about why manufacturers would add caffeine to soda, Temple and her team first conducted studies on the reinforcing properties of caffeine. The claim from beverage manufacturers is that caffeine is added to enhance flavor. But caffeine has an extremely bitter flavor, and at the levels of caffeine added to sodas, studies have shown that few people can taste the difference between caffeinated and noncaffeinated soda. Temple and her colleagues approached this work with the hypothesis that caffeine is added not just to increase the liking of soda but also to increase the reinforcing properties of soda. Specifically, she and her research team designed a study aimed at testing whether caffeinated soda becomes reinforcing over time (Temple et al., 2009).

Temple described the study participants as 12 to 17 years of age, stratified by caffeine use (<25 mg/day; 25–50 mg/day; 50–75 mg/day; and >75 mg/day). The researchers set up an operant response condition in the lab, where participants pressed a mouse button and after so many mouse button presses were reinforced with a portion of soda. Participants were provided both caffeinated and noncaffeinated versions of the same soda and were evaluated for their willingness to work for each type of soda. After the test, participants were sent home with four 2-liter bottles of either caffeinated or noncaffeinated soda, with participants not knowing which type they had, where they consumed the same amount of soda daily (32 oz) for 1 week. At the end of the first week, they were interviewed about how they liked the soda and their mood over the course of the week and were then provided with the opposite type of soda (either noncaffeinated or caffeinated) and asked to again consume the same amount of soda daily (32 oz) for another week. At the end of the second week, participants were again interviewed about how they liked the soda and what their mood had been like. They were also evaluated again for their willingness to work for each type of soda.

The results for willingness to work for a caffeinated soda are illustrated in Figure 6-1, with the panel on the left reflecting baseline results and the panel on the right showing results obtained after the exposure period. The y-axis represents the number of button presses; the x-axis represents the number of times the button had to be pressed in order to receive a soda. Typically, data like these show an increase in the number of button presses (y) as the schedule of reinforcement increases (x) and then a decrease. With these data, at baseline, there was no difference between males and females. But after the exposure period, the reinforcement value in males increased significantly, and the reinforcement value in females decreased slightly. That is, after becoming more familiar with caffeinated soda, the soda became more reinforcing for males and less reinforcing for females. Temple did not show the data, but she said that there was no change in the reinforcing value of the noncaffeinated soda in either males or females. Nor were any differences observed on the basis of use (stratification). In sum, according to Temple, the study showed that adding caffeine to soda can increase the reinforcing value of soda.

FIGURE 6-1

Results from operant response test for caffeinated soda. NOTES: Baseline results in the left graph and results obtained after exposure in the right graph. See text for detailed explanation. SOURCE: Temple et al., 2009.

Next, Temple and her colleagues wanted to see whether caffeine increases subjective liking of soda. Again, they stratified their participants by caffeine use. They provided participants with seven novel sodas on their visit and then picked the beverage ranked fourth by each participant. For each of four subsequent visits, participants were provided with that number four beverage either with or without caffeine (either 1 mg or 2 mg per kg). On the sixth visit, participants were asked to rerate the liking and ranking of that beverage. As described in Temple et al. (2012), individuals in the placebo group did not change their liking of the soda over time. Individuals provided with 1 mg per kg dose of caffeinated soda showed an increase in liking only during the last visit but not before then. Individuals provided with 2 mg per kg showed a steady increase in liking over time. Temple mentioned that similar findings have been observed in adults and with caffeinated yogurt (Panek et al., 2013).

Future Directions

In sum, caffeine definitely has effects in children that are consistent with some findings in adults. The biggest difference, in Temple's opinion, is that there do not seem to be many differences between low versus high caffeine users. In fact, she and her colleagues have not found any significant differences between low versus high caffeine users. Temple attributes the lack of such differences to the fact that even what are considered high users among children are children who still use caffeine relatively infrequently and at relatively smaller doses compared to adults. It is possible that children have not yet developed tolerance for the effects of caffeine.

In the future, Temple said, she would like to understand the relationship between early caffeine use and later drug use. She remarked that there are some good cross-sectional data showing that caffeine enhances the reinforcing value of nicotine in humans (Jones and Griffiths, 2003) and some good experimental data showing that caffeine enhances the reinforcing value of cocaine in rats (Green and Schenk, 2002). Caffeine has also been shown to induce dopamine release in the nucleus accumbens (Acquas et al., 2002) and to condition flavor preferences in adults (Yeomans et al., 2000, 2001; Yeomans, 2004; Panek et al., 2013) and in children (Temple et al., 2012). These findings suggest to Temple that there could be a relationship between caffeine and drug use. She would like to study that relationship in a prospective design where early caffeine use is measured and children are followed over time.

In conclusion, Temple identified several data gaps in the literature. First, while preparing her presentation, she found it very difficult to find a current survey of caffeine use in adults and children in the United States. Many of the data she found were old and did not really capture the potential shifts in usage since energy drinks have flooded the market. In addition to prospective studies on relationships between early caffeine use and later substance use, she called for prospective studies examining factors that relate to high caffeine use and risk of high-level caffeine use. Finally, she called for studies on the long-term effects of caffeine use, particularly studies beginning in childhood and progressing to adulthood.

Subjective Effects

Griffiths described the subjective effects as drug-induced changes in an individual's experience or feelings. Numerous studies have shown that the qualitative subjective effects of caffeine are dose dependent, with lower doses (20–200 mg) producing predominately positive subjective effects, such as well-being, energy, and alertness. Higher doses (300–500 mg) produce predominately dysphoric subjective effects.

Reinforcing Effects

Reinforcing effects, which refer to the self-administration of caffeine, have been demonstrated very clearly in both laboratory animals (e.g., baboons) and humans. Griffiths summarized key findings from approximately 20 scientific studies on reinforcing effects of caffeine in humans:

• Caffeine can function as a reinforcer when administered in capsules, coffee, or soft drinks.
• The range of conditions under which caffeine functions as a reinforcer is not as broad as with classic psychomotor stimulants such as amphetamine or cocaine.
• Caffeine reinforcement is an inverted U-shaped function of dose.
• In normal subjects there are wide individual differences in susceptibility to caffeine reinforcement.
• Avoidance of abstinence-associated withdrawal symptoms plays a central role in reinforcement among regular consumers. Nevertheless, such a history is not necessary for demonstrating caffeine reinforcement.

In addition, there is overwhelming circumstantial evidence that caffeine has reinforcing effects: regular daily consumption of pharmacologically active doses is widespread, with caffeine being the most widely used mood-altering drug in the world; historically, caffeine consumption has been long term, relatively stable, and resistant to suppression; and consumption occurs in widely different vehicles and in widely varying cultural and social contexts.

Tolerance

Tolerance, which refers to reduced responsiveness due to drug exposure, has been clearly demonstrated in both laboratory animals and humans. Studies with rats have shown that chronically treated rats show no response to caffeine, compared to untreated rats, who exhibit an inverted U-shaped response. Studies with rats have also shown no cross-tolerance to amphetamine. Complete tolerance also occurs in humans at high doses. For example, Griffiths and colleagues showed that a 300-mg challenge to caffeine-free individuals maintained on placebo caused tension, anxiety, and jitteriness, compared to a total absence of effect among individuals receiving a chronic dose of 900 mg per day (Evans and Griffiths, 1992).

Physical Dependence

Physical dependence, or withdrawal, refers to time-limited disruption of mood or behavior after cessation of chronic dosing. Withdrawal has been very well demonstrated in both animals and humans. Activity in rats has been shown to decrease when switched from chronic caffeine to water, with recovery to normal activity occurring over the course of several days. Similar results have been observed in humans. Griffiths and colleagues demonstrated increased headaches and lethargy and decreased ability to concentrate after abruptly switching individuals from caffeine to placebo, with the effects resolving over the course of several days to a week (Griffiths et al., 1990). In another study in which individuals were blind to the manipulation (Silverman et al., 1992), about 50 percent of individuals who were switched from caffeine to placebo reported moderate or severe headache and about 11 to 12 percent reported substantial increases in depression and fatigue. Individuals switched from caffeine to placebo also demonstrated decreased psychomotor tapping performance and increased unauthorized medication use, mostly for headache. According to Griffiths, in the approximately 75 experimental studies conducted that permit this kind of analysis, about 50 percent of individuals reported headache (Juliano and Griffiths, 2004). So headache is a common symptom of withdrawal, although withdrawal can also occur without headache.

Headache is one of several caffeine withdrawal symptom clusters recognized by the DSM-5. Others are fatigue or drowsiness; dysphoric mood; depressed mood; irritability; difficulty concentrating; and flu-like somatic symptoms (nausea, vomiting, or muscle pain/stiffness). In the literature, the incidence of clinically significant or functional impairment (i.e., people unable to do what they normally do) averages 13 percent in prospective experimental studies and 9 percent in retrospective survey studies (Juliano and Griffiths, 2004). As just one example, Griffiths mentioned the range of functional impairments reported in a double-blind placebo-controlled challenge study: missed work and vomited; could not perform work responsibilities, needed spouse to care for children, and went to bed early; performed multiple costly mistakes at work, left work early, and went to bed early; screamed at children (Strain et al., 1994).

A variety of studies have shown caffeine withdrawal to be what Griffiths described as “a robust parametric phenomenon.” Chronic maintenance dose, duration of caffeine maintenance, and within-day frequency of dosing all impact the probability and severity of withdrawal. Even just three days of chronic exposure and once-a-day administration are sufficient to trigger withdrawal signals. In addition, readministration of caffeine has been shown to reverse abstinence effects in a very rapid and dose-dependent way.

Equally important, in Griffiths's opinion, many studies have demonstrated that avoidance of abstinence-associated withdrawal symptoms plays a central role in the habitual consumption of caffeine. Studies have also demonstrated that withdrawal potentiates the reinforcing effects of caffeine and that withdrawal plays an important role in the development of preferences for flavors paired with caffeine (Juliano and Griffiths, 2004). Regarding the latter point, Griffiths referred workshop participants to some of the data cited by Jennifer Temple during her presentation.

Addiction: DSM Substance Dependence Syndrome

The DSM-5 does not officially recognize caffeine addiction, or dependence syndrome, as a diagnosis, given that too few studies have been completed; they did propose research criteria. Still, Griffiths identified eight studies showing that some people do in fact fulfill DSM-4 or DSM-5 criteria for a diagnosis of substance dependence as applied to caffeine: Strain et al. (1994), Hughes et al. (1998), Bernstein et al. (2002), Jones and Lejuez (2005), Svikis et al. (2005), Ciapparelli et al. (2010), Striley et al. (2011), and Juliano et al. (2012).

For example, in a study of individuals who were sufficiently distressed by their caffeine use to seek outpatient treatment, Juliano et al. (2012) evaluated what they identified as the three DSM-5 criteria most definitional of addiction. Individuals were recruited from the community using advertisements inviting participation in a program for caffeine dependence. In an effort to be very conservative and include only hard cases of pure caffeine dependence, individuals with other current drug dependence excepting nicotine were excluded. The group comprised 94 total participants. Griffiths described them as a high-functioning educated group of adults. Their mean age was 41 years, 55 percent were female, and 86 percent were college or postgraduate educated. Mean caffeine use was 548 mg/day, so it was over the 90th percentile. A clinical psychologist conducted the evaluations.

Among the 94 total participants in Juliano et al. (2012), 89 percent reported persistent desire or unsuccessful efforts to cut down or control substance use; 96 percent reported characteristic withdrawal symptoms or use to relieve or avoid withdrawal symptoms, with 43 percent reporting functional impairment (i.e., severity sufficient to produce an impairment of normal activities, such as being unable to work or sleeping at work); 87 percent reported continued use despite persistent or recurrent physical or psychological problems. Regarding the reports of physical or psychological problems, 83 percent reported physical problems (e.g., stomach problems, cardiovascular problems, complications of pregnancy, sleep problems, urinary problems); 67 percent reported psychological problems (e.g., anxiety, irritability, anger); and 43 percent reported having been told by a physician or other medical professional to modify their caffeine use because of various medical conditions (e.g., pregnancy, headache).

Conclusions with Respect to Caffeine Withdrawal and Addiction

In Griffiths's opinion, with respect to withdrawal, numerous studies, around 75 percent, indicate that cessation of caffeine consumption after a period of daily intake can result in a distressing withdrawal syndrome involving functional impairment. This conclusion is consistent with the DSM-5 committee recognition of caffeine withdrawal as a diagnosis. It is also consistent with a recent survey of 500 addiction professionals—most of whom endorsed the idea that caffeine withdrawal can be of clinical importance (Budney et al., 2013).

Caffeine addiction is a less well-established effect than caffeine withdrawal, which is consistent with the DSM-5 committee recommendation that caffeine use disorder be recommended as a diagnosis for further study. Still, Griffiths pointed out that the majority of addiction professionals surveyed in Budney et al. (2013) endorsed the idea that caffeine use disorder occurs and that some people could benefit from professional help in quitting. Griffiths identified eight studies suggesting that some people become clinically dependent on caffeine, that is, they are unable to quit, they continue to use despite medical problems, and they are sufficiently distressed to seek treatment (Meredith et al., 2013).

Implications for Youth as a Vulnerable Population

Several of these findings have potential implications for youth. First, with respect to tolerance, Griffiths said, individuals who do not use caffeine regularly will likely be substantially more sensitive to the acute effects of caffeine, including its adverse effects. Studies show that tolerance readily occurs, with lower doses leading to partial tolerance and higher doses to complete and insurmountable tolerance. Nevertheless, because most studies characterizing the adverse effects of caffeine have examined those effects in habitual consumers, they are of little relevance in estimating the risk of adverse events in nonusers.

Another implication for youth is that caffeine reinforcement, tolerance, and withdrawal are dose dependent. Individuals who weigh less receive a proportionally greater dose of caffeine for a given serving size, with a 13-year-old boy weighing about 55 percent as much as a 50-year-old man.

Conditioned taste preference also has implications for youth as a vulnerable population. It is well known that consumers often develop strong preferences for specific types and brands of caffeinated beverages. The likely mechanism behind this is that caffeine conditions specific flavor preferences, with initial flavor preferences likely evolving into habitual brand preferences, perhaps lasting a lifetime. Griffiths opined that these facts are not lost on those marketing energy drinks and may incentivize promotion of products to younger and younger populations, much as the tobacco companies were accused of doing until such marketing became more tightly regulated.

Finally, Griffiths noted, with respect to withdrawal and addiction, if physical dependence develops, youth are less likely to have the financial, transportation, or other resources to ensure an uninterrupted supply of caffeine. When their habitual pattern of intake is delayed or disrupted, withdrawal-sensitive individuals experience adverse emotional, cognitive, and behavioral consequences.

Substance Use Disorder: Differences Between DSM-IV and DSM-5

O'Brien served for 7 years as chair of the DSM-5² committee, and he explained some important differences between the DSM-IV and DSM-5. First, for all drugs, the DSM-IV differentiated between use, abuse, and addiction. On reexamining 150,000 diagnostic interviews, the DSM-5 committee realized that the severity of use is a progressive phenomenon, from gradual use to addiction. The DSM-5 committee identified 11 symptoms, with a greater number of symptoms indicating greater severity: tolerance (not counted if prescribed by a physician); withdrawal (not counted if prescribed by a physician); more use than intended; craving for the substance; unsuccessful efforts to cut down; excessive time spent in acquisition; activities given up because of use; use despite negative effects; failure to fulfill major role obligations; recurrent use in hazardous situations; and continued use despite consistent social or interpersonal problems. Generally, exhibiting two symptoms is considered mild, up to four is moderate, and over four is severe. All but one of the 11 symptoms were recognized in DSM-IV. The DSM-5 committee eliminated liver problems as a symptom because it was considered not useful, and they added craving. O'Brien described tolerance as a “normal reaction,” with caffeine being one of several types of drugs that shows very rapid tolerance. Others are antihypertensive drugs, antidepressants, antianxiety drugs, and opioid analgesics.

The DSM-5 committee did not include addiction, or caffeine use disorder, as a diagnosis. But they did include it in the appendix to stimulate research. O'Brien said, “Most of us are not prepared to say that there is such a thing as caffeine addiction, but there is definitely caffeine withdrawal.” According to O'Brien, many committee members resisted adding caffeine withdrawal disorder to DSM-5. But for the committee, it was a trivial issue. The evidence is abundant that caffeine withdrawal exists, ranging from very mild to very severe.

A Double-Blind Controlled Study of Caffeine Withdrawal

Impressed with the many placebo-controlled studies they had each conducted over the course of their careers, with individuals in placebo groups reporting many of the same adverse effects reported by individuals in treatment groups, from headache to psychosis, O'Brien and colleague Peter Dews were curious about the “real effects” of caffeine withdrawal. As far as O'Brien was aware, the study they conducted to answer that question, Dews et al. (1999), is the only study of its type where at no point during the study did the researchers tell the participants that they were studying caffeine withdrawal. Starting with a population of about 11,000 people, some of whom consumed caffeinated beverages on a daily basis, the researchers asked participants about problems with stopping caffeine and randomized participants who reported withdrawal into three groups. All three groups received roughly the same 400–500 mg daily dose of caffeine for 1 week to 10 days. After the 10-day period of stabilization, one group continued to receive the same dose, the second group experienced abrupt withdrawal, and the third group received a gradual reduction in dose. Then the researchers asked participants a series of questions about their energy, alertness, leisure time, and other symptoms. To distract participants, the researchers also asked about the smell, appearance, and taste of coffee (i.e., those questions were considered a distracter because the researchers were interested only in withdrawal). The end result was that individuals who continued to receive the same dose showed no withdrawal symptoms; females in the sudden reduction group showed symptoms—for example, they reported being less alert—but males showed no symptoms; and individuals in the gradual withdrawal group reported minimal if any symptoms. According to O'Brien, withdrawal may not be as common as placebo-controlled studies suggest.

Neurodevelopmental Influences on Risk-Taking Behavior During Adolescence

Scientists have learned a great deal during the past 20 years, especially the past 10 years, about the human brain and how the brain undergoes very complex and functional changes during the adolescent years and into the early 20s (Kuhn, 2006; Crews et al., 2007; Steinberg, 2008; Johnson et al., 2009; White, 2009; Casey and Jones, 2010; Gladwin et al., 2011; Pharo et al., 2011; Sturman and Mogghaddam, 2011; Spear, 2013). These changes partially explain why adolescents are more likely than older individuals to engage in risk-taking behavior and perhaps less likely to fully recognize the consequences of such behavior. Moreover, adolescents appear to be more susceptible to the rewarding properties of substances. The evidence also helps to explain the long-established robust finding that early use of substances increases the risk of addiction in adulthood. In short, Arria explained, there is an inherent vulnerability of the developing brain to psychoactive substances.

Energy Drinks: Potential Exacerbation of Health-Risk Behaviors

Several naturalistic and one experimental study have clearly demonstrated that energy drink users are more likely to engage in risk-taking behavior (Miller, 2008; Arria et al., 2010, 2011; Stasio et al., 2011; Velaquez et al., 2012; Peacock et al., 2013; Woolsey et al., 2013). Many forms of risk-taking behavior have been studied, including drug use, sexual risk taking, alcohol use, and the mixing of energy drinks and alcohol. Arria also considers studies on anxiety and sleep quality important factors to consider when evaluating adolescent behavior, even though they are not necessarily considered risk-taking behaviors. The one experimental study, Peacock et al. (2013), involved measuring risk-taking behavior in a laboratory setting using an analog measure called BART (Balloon Analogue Risk Task).

Arria noted that the frequency of energy drink use among the studies she was able to locate that specifically focused on risk-taking behavior were studies on college students and that the prevalence estimates of energy drink consumption among that age group are much higher than was alluded to earlier during the workshop discussion. Recent studies are showing prevalence estimates of up to 83 percent in the past year and 57 percent in the past week (i.e., the year or week prior to collecting data). Her research team's data have shown a 65 percent annual increase in prevalence of use between the second and third years of college. She suggested that snapshot measures of 2-day or 7-day frequency cannot capture past year or past month use and identified the lack of valid assessment methods for energy drink consumption as an important data gap.

According to Arria, contrary to an earlier workshop remark that there are no prospective data on the relationship between energy drink use and subsequent use of other drugs, she and her colleagues have in fact been collecting prospective data on a cohort of more than 1,200 students, with a response rate of 81 percent. The study is now in its 10th year. The researchers have examined the relationship between different types of substances and the subsequent increase in the use of other substances over time. As far as she knows, the data represent the only prospective epidemiologic data on energy drink consumption over time in a large sample of young adults. Specifically, guided by prior research suggesting that caffeine use might exacerbate the underlying vulnerability to the use of other substances, the researchers asked whether energy drink use during the second year of college predicted incident or new use of other drugs during the following year.

After adjusting for sex, demographics, socioeconomic status, sensation seeking (i.e., according to Arria, a variable that measures novelty seeking), and other types of caffeine use, the researchers found that, yes, the use of energy drinks in the second year of college (23 percent of the sample) predicted frequency of tobacco use and incident (new) nonmedical use of prescription stimulants and prescription analgesics in the third year (Arria et al., 2010). The adjusted odds ratio for stimulants was 2.5 (p < 0.001), with 8.2 percent of nonenergy drink users and 18.8 percent of energy drink users starting to use prescription stimulants the following year (see Figure 6-2). The adjusted odds ratio for analgesics was 1.5 (p <0.05).

FIGURE 6-2

Summary of results from prospective study on energy drink use among second-year college students and the use of other substances during the third year of college. NOTE: AOR = adjusted odds ratio. SOURCE: Arria et al., 2010.

Implications of This New Evidence

In Arria's opinion, new evidence from developmental neuroscience underscores the inherent vulnerability of the developing brain to psychoactive substances. In addition, the balance of evidence in the scientific literature supports the argument that the levels of caffeine in today's products, in the way those products are consumed, are associated with increased risk-taking behaviors. Nor has the addition of caffeine to energy drinks at the levels present in most products been demonstrated to be safe with regard to risk-taking behaviors in adolescents and young adults. Until evidence has been presented that demonstrates safety, actions to change current regulations on these products are warranted to protect and promote the health of the public in general and the health of adolescents in particular.

Practical Implications

According to Smith, the study that arguably demonstrates most clearly the practical implications of all these various findings on the beneficial effects of caffeine exposure is Lieberman et al.'s (2002) study on caffeine and sustained military operations. The researchers examined the effects of caffeine in U.S. Navy Seals during what is known as “hell week,” a very fatiguing and stressful training week where the Seals conduct excessive work on little sleep. The researchers found that a dose of 200 mg of caffeine improved vigilance, learning, memory, and mood and concluded that the administration of caffeine may provide a significant advantage when cognitive performance is critical and must be maintained during exposure to severe stress.

Smith himself has examined the impact of caffeine on real-life work performance in two ways. The first was what Smith described as an “after-effect” technique, which involves obtaining both subjective and objective measurements both before and after work and using the difference between the before- and after-work measurements as an indicator of performance during the work period. So someone who had a very fatiguing work day would show a much larger after-effect of that day compared to someone who had a relatively light work day. Smith (2005) measured reported alertness and simple reaction time among 110 workers both before and after work and found that workers who had consumed caffeine during the day were more alert and had faster reaction times.

Because after-effect measures are only indirect measures of work performance, Smith has also conducted epidemiological research on associations between caffeine consumption and accidents or errors during work. Specifically, Smith (2005) sampled more than 2,500 workers who were in jobs where accident risk was high and found that higher caffeine consumption was associated with half the risk of frequent cognitive failures and accidents. (Cognitive failures are human errors involving problems with memory and attention.)

An Alternative View

Although these findings tell what Smith said is a “very nice story,” he acknowledged that there is an alternative view: that caffeine has no positive effects, that rather it just removes the negative effects of caffeine withdrawal. He referred to earlier workshop discussions on the negative effects of caffeine withdrawal, including headaches, mood changes, and impaired performance (see summaries of Roland Griffiths's and Charles O'Brien's presentations earlier in this chapter). In Smith's opinion, this alternative view is unlikely for three reasons. First, the same (beneficial) effects can be observed in animals and in nonconsumers who by definition cannot be withdrawing. Second, the effects are observed even with repeated doses. If the withdrawal explanation were correct, then one would observe effects after the first dose but not after repeated doses. Third, the effects are observed following “wash-out,” that is, 1 week to 10 days after withdrawal when negative withdrawal effects are no longer present. Again, if the withdrawal explanation were correct, one should not see effects after withdrawal is over.

In addition, said Smith, effects are observed between different personality types (e.g., introverts versus extroverts) even with the same level of withdrawal. Again, the reversal of withdrawal is an unlikely explanation for these differences. The differences are more likely caused by arousal effects. Also, acute cardiovascular effects of caffeine and the effects of caffeine on sleep are usually explained in terms of stimulant effects. It is not clear why another mechanism, that is, withdrawal reversal, is needed to explain such effects.

Conclusions About Caffeine and Performance

In conclusion, Smith reiterated that the levels of caffeine consumed by most people have largely beneficial effects on alertness, attention, and other similar behaviors. He emphasized, however, that excessive consumption can lead to problems, especially in sensitive individuals. For Smith, here “sensitive” means a child. In a pilot study on diet, behavior, and attainment in 200 secondary school children, researchers found several associations between diet and detention (personal communication, Nicholas Milward, Pool Academy, January 2012). For example, students who consumed energy drinks were 60 percent more likely to receive detention.

On the basis of the results of that pilot study, Smith and colleagues conducted a longitudinal study involving 2,000 pupils. They administered two dietary surveys, one at the start and the other at the end of the school year, and collected two sets of measures of attainment and behavior. The researchers are currently analyzing cross-sectional data.³ Thus far, they have shown that those who often consumed energy drinks were more likely to have low attendance, receive a sanction, and receive poorer grades. These findings are true even when controlling for possible confounders, such as socioeconomic status and special educational needs.

Smith acknowledged that he and his colleagues are unable to infer causality. Longitudinal data and dose–response data will provide a clearer view, as will results of a planned intervention study aimed at measuring the effects of reducing energy drink intake. Until such clarity is reached, there are two plausible mechanisms. Either energy drinks are causing the problems among the school children that he and his colleagues are observing, or energy drink consumption may itself be an outcome, with some other factor driving both energy drink consumption and poor attainment, attendance, and behavior. It is a critical distinction, Smith observed, and one that they hope to have an initial answer for soon.

Mechanism of Caffeine's Effect on the Central Nervous System

There was some discussion about conflicting results in the scientific literature on where exactly dopamine is released after exposure to caffeine. Ferré explained that as he mentioned during his talk, caffeine is a weak “dopamine releaser” (although more research needs to be done on the clear dopamine-releasing properties of paraxanthine). Nevertheless, he and his research team found that caffeine in fact induces dopamine release in a specific part of the shell of the nucleus accumbens and that other data suggesting that it occurs not in the nucleus accumbens but in the cortex might be the result of contamination from the shell of the accumbens. He referred workshop participants to a review that he and his team wrote explaining the difference (Ferré, 2008). The take-home message, according to Ferré, is that caffeine is not a very good dopamine releaser when compared to cocaine or amphetamine, because the main mechanism is postsynaptic and results from adenosine-dopamine receptor interactions.

When asked how to reconcile the fact that the mechanism of action for caffeine (which acts on adenosine receptors) is very different from the mechanism of action for cocaine (which acts on dopamine receptors), Ferré responded that the effects are similar because they act in the same brain areas, that is, in the striatum, and that the difference is more quantitative than qualitative.

Most of the panel discussion following Arria's presentation revolved around the interpretation of the evidence presented and the gaps in data.

Cross-Sectional Versus Prospective Studies for Evaluating Long-Term Effects of Exposure in Children

There was a question about the roles of cross-sectional versus prospective designs in evaluating the long-term effects of caffeine exposure in children and adolescents. Temple remarked that cross-sectional data are confounded in many ways and that there is a strong need for long-term prospective studies.

Cardiovascular Effects of Caffeine Exposure in Children and Adolescents: Sex Differences

Temple was asked whether any of her research involved electrocardiogram monitoring of children and adolescents. Temple explained that her team was not set up to do that and agreed that it would be interesting. Her team measured only heart rate and blood pressure. In this regard, John Higgins expressed intrigue at the blood pressure findings described by Temple, specifically the sex difference found after puberty and the greater responsiveness seen in postpubertal males in comparison to postpubertal females to some of the effects of caffeine. He noted that five of the six deaths reported to be associated with caffeine-containing energy beverages were in males between the ages of 12 and 19. Temple suggested that the difference might be related to circulating steroid hormones. According to Temple, it is well known that steroid hormones affect caffeine metabolism. She and her team are trying to figure out how to test that hypothesis other than by measuring salivary hormone levels. Other data (which she did not present) have shown that blood pressure effects in females are lower when salivary estradiol levels are higher. Temple reiterated that she and her research team have found a greater responsiveness to caffeine among postpubertal males “across the board,” that is, not just with cardiovascular effects but also with reinforcing and subjective effects.

Blinded Studies of Caffeine Withdrawal

Griffiths identified Silverman et al. (1992) as another study on caffeine withdrawal that did not inform participants that caffeine was being tested. Other withdrawal studies have similarly blinded participants (see Juliano and Griffiths, 2004). In Silverman et al. (1992), participants were told only that they were participating in a study on dietary substances. They were provided with misinformation about shellfish, NutraSweet, and so forth, to distract them. In addition, Juliano and Griffiths (2004) have estimated a 13 percent incidence of significant functional impairment, compared to Dews et al.'s (1999) 2.6 percent. Even 2.6 percent is not trivial in a population in which caffeine is consumed by 85 percent of the population, in Griffiths's opinion.

The Association Between Caffeine Use and Other Substance Use in Adolescents and Young Adults

A driver for both caffeine use and the nonmedical use of prescription drugs is the availability of resources needed to acquire those substances, according to a member of the audience. The audience member asked Arria if she and her colleagues had examined the purchasing power of the study participants in the Arria et al. (2010) study and whether possibly the individuals with the ability to purchase caffeinated beverages were, coincidentally, the same individuals with the ability to acquire prescription drugs. Arria explained that she and her team have studied availability and access to nonmedical use of prescription stimulants and have found that, by and large, students obtain them for free from friends, relatives, and acquaintances. The substances are widely accessible. Because all the study participants in Arria et al. (2010) came from the same campus, she thinks it unlikely that some students would have greater access than others.

Arria was also asked about the pattern of use among the students she and her colleagues followed. For example, were they consuming greater doses of energy drinks over time in order to get the same buzz? Were they later substituting analgesics or other substances for the energy drinks because they were no longer getting the same buzz with the energy drinks? Were they using both simultaneously? Arria found it an interesting suggestion that consumption might be related to the likelihood to try something with greater potency. Arria referred workshop participants to a recent study, Woolsey et al. (2013), where the researchers found a great overlap between the substitution of energy drinks and the use of nonmedical prescription stimulants for studying. In addition, the researchers reported that every individual with a prescribed attention deficit hyperactivity disorder (ADHD) medication was using energy drinks, a finding that suggested to Arria that someone should probably be studying the interaction between energy drinks and medical use of prescription stimulants.

Another audience member observed that many people with ADHD self-medicate with caffeine. He asked Arria whether individuals in her study might be substituting the stimulants for caffeine, not necessarily because they were seeking something with greater potency, but as a way to self-medicate. She explained that her study has collected data on the motives of energy drink consumption and has yet to analyze the data.

Arria was also asked whether results were different between female and male participants. She explained that she and her research team controlled for gender in Arria et al. (2010). She noted that she has observed a difference in energy drink use, with a higher proportion of girls drinking coffee and a higher proportion of boys drinking energy drinks.

She was also asked about the nature of the survey. She did not send the survey out to students. Rather, her research team conducted face-to-face interviews. She also clarified that other studies have looked at a variety of risk-taking behaviors but that, for the sake of time, she chose to focus her presentation of the subsequent use of an illicit drug as the behavior of interest. When asked whether she was suggesting that energy drinks were causative of risk-taking behavior, she replied that it will take an accumulation of evidence to infer causality. Arria et al. (2010) was the first of what she hopes will be a series of prospective investigations into the contribution of energy drinks to future illicit drug use. In her opinion, at this point, rather than causality, the focus should be on safety. She said, “I think the burden of proof on whether or not regulations need to occur is really [on] a demonstration of safety rather than on a demonstration of causality.” When the same audience member pressed her further about whether there has been a demonstration of causality between energy drinks and risk-taking behavior, she replied that there are very compelling, consistent data across studies to demonstrate a contributory association but agreed that more data are needed to demonstrate causality. When asked about what her theory was, she referred to her earlier comments about neural development of the adolescent brain.

Withdrawal Suppression

Following Smith's presentation, Roland Griffiths commented about withdrawal suppression and the fact that some experts attribute all observed beneficial effects to caffeine withdrawal suppression. “That seems radical,” Griffiths said. At the same time, he did not think that the withdrawal suppression hypothesis should be so readily dismissed. The right methodology for addressing it would be a balanced design involving chronic caffeine administration compared to chronic placebo administration (e.g., Sigmon et al., 2009). Smith agreed that the hypothesis should not be dismissed and that there certainly are individuals for whom withdrawal is a significant problem. At the same time, he does not think it is a ubiquitous explanation. He agreed that more research along the lines of what Griffiths suggested is necessary and observed that withdrawal is likely more important with mood changes than with performance changes.