Astrid Nehlig

Caffeine (1,3,7-trimethylxanthine) is the most widely used psychoactive substance in the world. Most of the caffeine consumed comes from dietary sources such as coffee, tea, cola drinks, and chocolate. Caffeine is also present in many nonprescription medications, such as cold remedies, analgesics, and weight loss products. The most notable behavioral effects of caffeine--increased alertness, energy, and ability to concentrate--occur after consumption of low to moderate doses (50-300 mg). Moderate caffeine consumption very rarely leads to health risks (1). Higher doses of caffeine induce negative effects that include anxiety, restlessness, insomnia, and tachycardia; these effects are seen primarily in a small subset of caffeine-sensitive individuals. Even so, caffeine was described in one study as a potential drug of abuse (2) and in a more recent article as "a model drug of abuse" (3). Finally, on the basis of a review of scientific and clinical data, it has been considered that caffeine withdrawal, but not abuse and dependence, should be added to diagnostic manuals in the United States (4).

In this article, I review the available data about caffeine consumption; describe the known effects of caffeine on locomotion, mood, and sleep; and present new data about caffeine dependence that show how caffeine differs from drugs of abuse such as amphetamines and cocaine.

The caffeine content in dietary sources ranges from 4 mg/150 mL for cocoa to 30-50 mg/150 mL for tea, 32-70 mg/330 mL for cola drinks, and 70-220 mg/150 mL for coffee (5). Individual caffeine consumption from all sources can be estimated as 76 mg/day worldwide but reaches 210-240 mg/day in the United States and Canada and >400 mg/day in Sweden and Finland, where 80-100% of caffeine intake comes from coffee alone (5). In the United Kingdom, the consumption is as high as in Sweden and Finland, but 72% is consumed as tea. According to one survey (6), the daily intake of caffeine from all sources in the United States amounts to 2.4-4.0 mg/kg body wt (170-300 mg in a 60-70 kg-individual). In individuals older than 10 years old, two-thirds of this amount comes from coffee. In children, soft drinks represent 55% of the total caffeine intake, chocolate foods and beverages 35-40%, and tea 6-10% (7).

Cerebral energy metabolism
The effects of caffeine on cerebral metabolic function can be explored by using a quantitative autoradiographic imaging technique (8). This method is based on the fact that glucose is the sole metabolic substrate of the adult brain in healthy individuals. Thus, by using an analogue of glucose, [¹⁴C] 2-deoxyglucose (which enters the brain like glucose does but is not metabolized beyond the first phosphorylation step of glycolysis), it is possible to quantify the local cerebral metabolic rate for glucose (LCMR_glc) simultaneously in all areas of the brain in conscious animals or humans. This technique permits the identification of neuronal pathways affected by a pharmacological agent and is very useful for relating behavioral effects to the central action of a drug. Because glucose is the sole metabolic substrate of the brain, the LCMR_glc value reflects cerebral energy metabolism and, hence, functional activity. Therefore, we measured LCMR_glc after the administration of increasing doses of caffeine (1-10 mg/kg body wt) that reflected normal range of human consumption. This study was performed in adult male rats, and we included five or six animals per group. The purpose of the study was first to determine whether we could confirm the effects of caffeine on locomotion and sleep, and then to study the effects of this methylxanthine on the brain structures involved in addiction and reward.

Brain systems affected
The nigrostriatal dopaminergic system involved in the control of locomotion originates in neurons located in the substantia nigra, mainly in the pars compacta (see Figure 1 for locations of cerebral components). These neurons project to the globus pallidus and terminate in the caudate nucleus, which is connected to the sensorimotor cortex.

The caudate nucleus appears to be very sensitive to the effects of caffeine because the LCMR_glc increases significantly in this structure after the administration of even the lowest dose of caffeine (1 mg/kg body wt) to adult male rats (Figures 2 and 3). The functional activity of this nucleus is further increased at 2.5 mg/kg body wt (>40% over control levels) and remains activated after 5 and 10 mg caffeine/kg body wt. In the substantia nigra pars compacta and globus pallidus, the LCMR_glc increases after 2.5-10 mg/kg body wt of caffeine, whereas the sensorimotor cortex shows an increase in LCMR_glc only after 5 mg/kg body wt of caffeine. There is a good correlation between caffeine-induced functional activation of structures that belong to the nigrostriatal pathway and the well-known stimulant effects of caffeine on locomotion. This effect is dose-dependent; the minimal dose of caffeine necessary to affect locomotion is 1.5 mg/kg body wt, which correlates well with the increase in LCMR_glc recorded in the caudate nucleus after 1 mg/kg body wt caffeine. The stimulant effect of caffeine on locomotion increases with doses up to 20 mg/kg body wt and decreases with doses >40 mg/kg body wt (9). In agreement with the present data, a direct administration of caffeine into the caudate nucleus modifies the spontaneous electrical activity of neurons in that structure, and caffeine is able to induce dopamine release (10) and the expression of immediate early genes in the caudate nucleus of the rat (11).

The serotoninergic cell groupings (the medial and dorsal raphe nuclei) and the noradrenergic cell grouping (the locus coeruleus) (Figure 1) are very sensitive to caffeine. These structures are involved in the control of sleep, mood, and well-being. In these three structures, LCMR_glc values are already activated after 1 mg/kg body wt and remain increased at the higher doses (2.5-10 mg/kg body wt) of caffeine (Figures 3 and 4). These data correlate well with the known sensitivity of sleep and mood to caffeine (9).

Caffeine and addiction
Drug dependence has been defined as "a pattern of behavior focused on the repetitive and compulsive seeking and taking of a psychoactive drug" (12). However, it is necessary to demonstrate psychoactive effects to differentiate drug dependence from other habitual or controlled behaviors, such as the daily ingestion of some kinds of medication, such as aspirin or vitamins. Moreover, it is necessary to demonstrate that the drug is positively reinforcing its own ingestion. First, the criteria for the definition of "dependence" must be established.

Diagnostic manuals from the World Health Organization (WHO) (13) and the American Psychiatric Association (APA) (14) proposed a new set of criteria for dependence. The diagnosis of dependence requires the fulfillment of any three of the six WHO or seven APA criteria. The seven criteria of dependence as proposed by the APA are as follows:
(1) Tolerance (not specified for severity)
(2) Substance-specific withdrawal syndrome (psychic or physiological, not specified for severity)
(3) Substance often taken in larger amounts or over a longer period than intended
(4) Persistent desire or unsuccessful efforts to cut down or control use
(5) A great deal of time spent in activities necessary to obtain, use, or recover from the effects of the substance
(6) Important social, occupational, or recreational activities given up or reduced because of substance use
(7) Continued use despite knowledge of a persistent or recurrent physical or psychological problem that is likely to have been caused or exacerbated by the substance

The six WHO criteria differ only slightly from those proposed by the APA, mainly by a different sequence, slightly different formulations, and the combination of criteria 5 and 6 into a single criterion. The only possible way to differentiate among substances that can lead to dependence is to classify them according to the number of criteria met, the severity of symptoms, and the frequency of occurrence.

The possibility of caffeine dependence has been considered by several groups for more than a decade (15), and, as noted earlier, Holtzman has postulated that caffeine is "a model of a drug of abuse" (3). In two studies that followed the APA criteria, caffeine dependence was shown in a subset of the general population. As a result of a random telephone survey in Vermont, Hughes and co-workers showed that 14 and 3% of the 166 caffeine users interviewed met the criteria for moderate and severe caffeine dependence, respectively (16). After telephone screening of 99 subjects in the United States, Strain and co-workers found 16 individuals who fulfilled criteria 1, 2, 4, and 7 of the APA and thus were considered dependent on caffiene (17). (Criteria 3, 5, and 6 were excluded because they do not apply to a substance widely avaliable and culturally accepted.)

The dependence was not related to the daily intake of caffeine, which ranged from 129 to 2548 mg/day. The median daily caffeine intake for the caffeine-dependent individuals was 360 mg, and >40% of the individuals ingested 300 mg or less each day (17). However, despite the absence of current psychiatric disorders at the time of the study in most of the individuals (14 of 16), 11 of the 16 individuals diagnosed with caffeine dependence had a history of psychiatric disorders, mainly substance abuse disorders (10 individuals) and mood disorders (7 individuals). The prevalence of these disorders is higher than that encountered in the general population (i.e., 50%). Moreover, the tendency to associate caffeine, alcohol, and nicotine consumption as well as mood disorders and nicotine dependence has been reported previously (15 ).

Among the seven APA criteria for drug dependence, withdrawal, tolerance, reinforcement, and dependence are described in greater detail in the following sections.

Withdrawal and tolerance
A small percentage of the population may experience withdrawal symptoms following sudden cessation of caffeine consumption, while gradual cessation over 2-3 days has not been shown to result in such symptoms. The most commonly reported symptoms are headache, weakness and drowsiness, impaired concentration, fatigue and work difficulty, depression, anxiety, and irrritability. Withdrawal symptoms generally begin ~12-24 h after sudden cessation of caffeine consumption and reach a peak after 20-48 h. They do not seem to be related to the quantity of caffeine ingested daily (15) and disappear soon after caffeine absorption. This effect is strongly linked to the psychological satisfaction related to the ingestion of caffeine, especially for the first caffeinated beverage of the day. The reversal of caffeine-withdrawal-induced headaches and other symptoms by the absorption of caffeine alone has been shown repeatedly (15).

Signs of caffeine withdrawal also have been observed in rats, cats, and monkeys. The signs include decreased locomotor activity and operant behavior. The severity of caffeine withdrawal symptoms, the length of the decrease in locomotor activity, and the duration of the treatment before the substitution with water depend on the dose. The latency until the onset of caffeine withdrawal effects occurred as in humans (usually within 24 h) and peaked around 24-48 h. Caffeine-withdrawal-induced behavioral changes usually last a few days (15).

Tolerance to a drug refers to an acquired change in responsiveness of a subject repeatedly exposed to the drug and can be considered in two ways. First, tolerance might indicate that the dose necessary to achieve the desired euphoric or reinforcing effects will increase with time, thus influencing people to gradually consume more of the drug. Second, tolerance to the adverse effects of high doses of the drug may occur, leading people to consume higher doses of the drug over time.

Mice, cats, and squirrel monkeys chronically treated with the methylxanthine developed tolerance to many behavioral effects of caffeine. Thus, tolerance to caffeine-induced locomotor stimulation, schedule-controlled responding maintained by presentation of food and electric shock, and thresholds for seizures have been described (9,15). The development of caffeine tolerance in animals is rapid and usually quite insurmountable (9). The exact mechanism underlying the development of tolerance to caffeine remains unclear.

In humans, tolerance to some physiological actions of caffeine has been demonstrated. Such is the case for the effect of caffeine on blood pressure, heart rate, diuresis, plasma adrenaline and noradrenaline levels, and renin activity, which usually develops within a few days of regular consumption. Tolerance to some subjective effects of caffeine--increased tension or anxiety; jitteriness or nervousness; and activity, stimulation, or "energy" --was recently described (9, 15). Conversely, there is only limited evidence for tolerance to caffeine-induced alertness and wakefulness. In agreement with these data, we showed that cerebral energy metabolism does not become tolerant to caffeine; an acute administration of 10 mg caffeine/kg body wt induces quite similar metabolic increases whether the rats had been exposed to a daily, chronic treatment with caffeine or with saline solution for the previous 15 days. Thus, every single exposure to caffeine produces cerebral stimulant effects, and this response is especially true in the areas of the brain that control locomotor activity and in the structures involved in the sleep-wake cycle (18 ).

In humans, sleep seems to be the physiological function most sensitive to caffeine. Generally, >200 mg caffeine is needed to affect sleep significantly. Caffeine prolongs sleep latency and shortens total sleep duration but preserves the dream phases of sleep. It is not yet clearly established whether the difference in the sensitivity to the effects of coffee on sleep could be attributable to tolerance (15 ).

Thus, tolerance to some of the effects linked to regular consumption of coffee seems to occur, especially in animals. In humans, the data are less conclusive and may underlie individual differences in the susceptibility and tolerance to caffeine-induced effects. Moreover, mechanisms of tolerance may be overwhelmed by the nonlinear accumulation of caffeine and its main metabolites in the body when caffeine metabolism becomes saturated under multiple dosing conditions.

Discrimination and reinforcement
The reinforcing efficacy of a drug refers to the drug's relative efficacy in establishing or maintaining a behavior on which its delivery is dependent. Venous catheters have been implanted in animals, allowing them to self-administer caffeine by pressing a lever. Assessment of the animals' behavior revealed inconsistent self-injection. Thus, there is a marked difference between caffeine and classical drugs of abuse such as amphetamines and cocaine, which maintain self-administration across species and conditions. Caffeine does not appear to be a very robust reinforcer in animals (15).

Human subjects can discriminate between caffeine and placebo when offered in either capsules or coffee. Doses of 300 mg or higher are usually more easily detected and mainly indicated by the negative effects of jitteriness, anxiety, or nervousness, whereas lower doses are indicated by caffeine withdrawal symptoms or by no effect at all. However, several studies have failed to demonstrate behavioral effects of caffeine at doses <200-300 mg, that is, at amounts corresponding to the ingestion of two to three cups of coffee (15). In humans, the widely recognized behavioral stimulant and mildly reinforcing properties of caffeine are probably responsible for the maintenance of caffeine self-administration, primarily in the form of caffeinated beverages such as coffee, tea, and cola. There is still a debate on whether the choice to consume caffeine is controlled by avoiding withdrawal or by generating positive effects. Most data show that caffeine reinforcement occurs in 100% of heavy caffeine consumers (>1000 mg/day) who also had histories of alcohol or drug abuse. For moderate caffeine users (130-600 mg/day), caffeine reinforcement occurs in a smaller subset of consumers (15).

Caffeine reinforcement varies with dose. Doses of caffeine encountered in tea and coffee are high enough to act as reinforcers because people tend to ingest tea or coffee while experiencing withdrawal symptoms. Indeed, a dose of 25-50 mg caffeine per cup of coffee acts as a reinforcer, whereas increasing doses beyond 50-100 mg tends to decrease the choice of caffeine or the frequency of caffeine self-administration, and high doses of caffeine (400-600 mg in a single dose) cause an aversion to caffeine.

Caffeine reinforcement also relates to withdrawal symptoms that occur after cessation of coffee consumption. Indeed, subjects that consistently suffer from caffeine withdrawal headaches are 2.6 times more likely than others to select caffeinated coffee (containing 100 mg caffeine). The choice of caffeine seems to be more potently influenced by avoiding withdrawal than by desiring its positive effects. The conditions under which caffeine functions as a reinforcer still are not clearly understood. However, the possible reinforcing effects of coffee--unrelated to caffeine but related to the smell and taste of coffee as well as the social environment that usually accompanies coffee consumption--should be considered with respect to the everyday motivations for consuming caffeinated or decaffeinated coffee (15).

Dependence and cerebral energy metabolism
When Self and Nestler reviewed the molecular mechanisms underlying reinforcement and drug dependence, they emphasized the critical role of the mesolimbic dopaminergic system (19), which consists of the dopaminergic neurons that originate in the ventral tegmental area and end in the nucleus accumbens (Figure 1). Rats self-administer amphetamine and dopamine directly into the nucleus accumbens and the ventral tegmental area (19).

The nucleus accumbens, which plays a central role in the mechanism of drug dependence, is functionally and morphologically divided into core and shell parts. The medioventral shell part is related to the limbic "extended amygdala", which we assume plays a role in emotional, motivational, and reward functions, whereas the laterodorsal core part regulates somatomotor functions (20). The specificity of cocaine, amphetamine, morphine, alcohol, and nicotine is to selectively activate the dopaminergic neurotransmission in the shell of the nucleus accumbens (21, 22). This property has been related to the strong addictive properties of these drugs (19). In contrast to commonly abused drugs, caffeine increases dopamine release in the caudate nucleus (10), which relates to the stimulatory properties of caffeine on locomotor activity (15) but does not induce any release of dopamine in the shell of the nucleus accumbens when injected at doses of 0.5-5.0 mg/kg body wt (23). These data are consistent with the low addictive potential of caffeine. Conversely, in this dose range, caffeine stimulates the release of dopamine in the prefrontal cortex (the terminal area of the mesolimbic dopaminergic system), which is consistent with its reinforcing and psychostimulant properties (23).

Our data on the effects of caffeine on cerebral energy metabolism show that the increase in LCMR_glc values recorded in the structures of the mesolimbic dopaminergic system are of lower amplitude than those recorded in the other brain regions studied (Figure 5). Moreover, the significant activation of functional activity appears only at quite high doses: 5 mg caffeine/kg body wt for the area of origin, the ventral tegmental area, and 10 mg/kg body wt for the two subdivisions of the nucleus accumbens and the medial prefrontal cortex. These data show that at the doses consumed daily by most individuals (2-2.5 mg/kg body wt), caffeine does not activate the brain circuitry of dependence and reward that is activated by psychostimulants. Moreover, the activation of functional activity in the shell of the nucleus accumbens occurs only at high doses of caffeine (10 mg/kg body wt; about four to five times the average daily rate of human consumption), at which the methylxanthine also activates the core of the nucleus (Figure 2) and induces widespread nonspecific metabolic increases in a majority of brain regions (9). These widespread effects on brain function probably reflect the numerous adverse side effects of the ingestion of large amounts of caffeine. Conversely, the effects of amphetamines, cocaine, and nicotine on the neural substrates that underlie addiction are rather specific and occur at doses that usually do not lead to the activation of many other brain regions (15).

The difference in the functional consequences of the psychostimulants (cocaine and amphetamines) compared with those of caffeine could relate to their respective mechanism of action. Amphetamines and cocaine induce the release or inhibit the uptake of dopamine, which binds to dopamine D1 and D2 receptors in the striatum. At low doses, caffeine acts preferentially at the level of adenosine A2a receptors (24), which are found mainly in the striatum, where they colocalize with dopamine D2 receptors (25). When the circulating levels of caffeine increase, the methylxanthine also binds to adenosine A1 receptors (24), which are located in the striatum (among other regions), where they colocalize with dopamine D1 receptors (25). In this study, caffeine appeared to mimic the effects of amphetamines and cocaine only at rather high doses (10 mg/kg body wt) at which binding to adenosine A1 receptors is likely.

Conclusion
The areas that control locomotor activity and the sleep-wake cycle appear to be highly sensitive to low concentrations of caffeine, and the structures involved in addiction and reward are activated only after high doses of caffeine. These doses also activate numerous brain regions and probably induce the adverse effects that occur after ingesting large doses of caffeine (9). My data are somewhat in agreement with the reported observation that the effects of caffeine are used consciously or unconsciously to manage mood and to alleviate the adverse effects of caffeine deprivation (26, 27). It must be considered that normal human caffeine consumption is spread over a day, whereas the doses given in the present study were injected as an intravenous bolus.

Thus, the reported data favor the action of caffeine as a positive reinforcer at doses that reflect general human consumption; they do not support the participation of the brain circuitry of addiction and reward in caffeine dependence reported even at very low doses (1 cup/day) in an earlier study on humans (17).

Acknowledgments
This work was suppported by a grant from the Institut National de la Santé et de la Recherche Médicale (INSERM U 272 and 398) and the Institute for Scientific Information on Coffee (ISIC), Paris, France.

References