Medication Development for Addictive Disorders: The State of the Science

The National Institute on Drug Abuse (NIDA), a component of the National Institutes of Health, is responsible for research into the causes of and treatments for drug addiction. As part of this effort, the Medications Development Program was established in 1989 with a primary mission to develop medications for opiate and cocaine dependence. For a decade, the Medications Development Program concentrated on these disorders. This review will report the progress made in developing medications for these disorders. More recently, it has increased its focus to include developing medications for methamphetamine and cannabis dependence. Because of space limitations, we will not review medication development for methamphetamine and cannabis disorders.

Opiate Dependence

According to the Office of National Drug Control Policy, there are almost a million long-term opiate users in the United States. Opiate dependence is associated with significant all-causes mortality and significant morbidity. The major pharmacotherapeutic agent for the management of opiate dependence is methadone. Although methadone is not a new medication developed under the NIDA Medications Development Program, a brief review of its effects will serve to illustrate several points: 1) that dose effects for opiates can be seen in both clinical pharmacology studies and clinical trials and 2) that reduction of opiate use and increased retention in clinical trials may be surrogates for the reduction in mortality and morbidity seen in epidemiological studies.

Methadone is an orally active, long-acting synthetic opiate that was recognized in the 1960s as having the potential to treat opiate dependence (1, 2). A clinical pharmacology study demonstrated that increasing doses of methadone affected abstinence signs and symptoms and reduced drug craving. These investigators also reported that doses of methadone ranging from 80–120 mg/day produced a blockade of the effects of intravenously administered heroin, hydromorphone, and methadone. The term “agonist blockade” was coined to describe this phenomenon.

Initial clinical trials using a variety of designs showed a benefit of methadone to reduce opiate use and retain patients in treatment (3–7). These studies also showed a collateral benefit to reduce recidivism (3), criminal activity (4), and mortality (7). Further clinical trials showed a dose effect to reduce opiate use and increase retention in treatment (8–12). An orderly dose effect has been reported for comparisons of 50 mg/day versus 20 mg/day doses of methadone (8), 60 mg/day versus 20 mg/day (9), 80–100 mg/day versus lower doses (10), 80 mg/day versus 30 mg/day (11), and 60–100 mg/day versus 20 mg/day (12). In the 1990s, an increase in average methadone dose in opiate treatment programs was noted (13). Notwithstanding the higher dose trend, patient dosing should be individualized.

Methadone has been shown to reduce opiate-related mortality. A U.S. study (14) reported that untreated heroin addicts had a yearly mortality rate of 8.3%, whereas those in methadone treatment had a mortality rate of 0.8%. A Swedish study (7) reported a sixfold reduction in mortality in methadone-treated opiate addicts. Finally, an Australian study (15) reported a fourfold reduction in the risk of dying for patients in methadone programs.

Methadone has been shown to reduce morbidity insofar as it reduces the risk of infectious disease. In a retrospective study (16), the duration of methadone maintenance therapy was inversely correlated with the prevalence of HIV. In a prospective study (17), a sevenfold reduction in the incidence of HIV was seen in a methadone-treated group versus an untreated cohort that agreed to be a part of the study but refused treatment.

Levomethadyl acetate (LAAM) is the alpha-acetyl congener of methadone. Its principal difference from methadone is its conversion to the active metabolites norLAAM and dinorLAAM (18–21). Initial studies were performed in the 1970s with a dosing regimen of three times a week (22, 23). LAAM was originally thought to be a pro drug, but recent evidence suggests that LAAM has agonist properties (24). A large phase III safety trial led to its approval by the Food and Drug Administration (FDA) in 1993 (25). Dosing recommendations on the product label suggested that induction doses could range from 20 to 40 mg/day and that maximum thrice-weekly dosing regimens of 30/130/180 mg or 140/140/140 mg were permissible. Alternate-day dosing was also added to the labeling. Postmarketing studies (26) noted that a rapid-induction regimen of 25 to 100 mg/day in 17 days yielded the greatest reduction in illicit opiate use, but a higher side effect profile was encountered. A dose effect of LAAM on illicit opiate use was reported (27), with the 100/100/140 mg thrice-weekly regimen giving the greatest reduction in opiate use. The FDA issued a “black box” warning for LAAM because of postmarketing surveillance reports of QTc prolongation in ECGs, with several reports of torsades de pointes, a polymorphic life-threatening ventricular arrhythmia (28).

Buprenorphine is a mu opiate partial agonist (29) that was originally marketed as an analgesic product. Investigators conducting an abuse liability study in human volunteers reported that subcutaneously administered buprenorphine has less subjective effects than morphine, a lesser withdrawal syndrome, and an ability to block the subjective responses of up to 120 mg doses of morphine (30). Subsequent work established that the sublingual route was preferable to oral dosing because of high first-pass effects (31, 32). The first outpatient treatment study to our knowledge (9) compared 8 mg of sublingual buprenorphine liquid to 20 and 60 mg doses of orally administered methadone in a randomized, double-blind, double-dummy study. Retention and decreased illicit opiate use in the buprenorphine group were superior to the response seen in the group that was receiving 20 mg/day of methadone.

NIDA and Reckitt & Colman (now ReckittBenckiser) received a Cooperative Research and Development Award to develop buprenorphine as a treatment for opiate dependence. Under this agreement, a second study conducted with the liquid formulation was performed in a multisite trial in which opiate-addicted patients were randomly assigned to 1, 4, 8, and 16 mg/day of buprenorphine (33). The a priori comparison was the effects of 1 mg/day versus 8 mg/day on illicit opiate use, retention, and opiate craving. The 8 mg/day dose group had significant reductions in illicit opiate use, reduced craving, and had better retention.

Subsequent to the study by Ling et al. (33), it was decided to develop a sublingual tablet and to add naloxone, a narcotic antagonist, to one of the formulations. The rationale for adding naloxone was to produce a less abusable, less divertable tablet. The dose ratio of buprenorphine to naloxone was chosen from data gathered in clinical pharmacology studies in opiate-dependent subjects maintained with morphine (34), methadone (35), or buprenorphine (36). In the first study (34), the subjects maintained on a dosage of 60 mg/day of morphine sulfate were randomly administered one of six medication treatments intravenously in a counterbalanced order: morphine, buprenorphine, buprenorphine/naloxone at 8:1, buprenorphine/naloxone at 4:1, buprenorphine/naloxone at 2:1, and placebo. Subjective measures of positive and negative effects were assessed for the first hour after dosing. The 4:1 ratio was chosen because it produced significant attenuation of buprenorphine’s effects without producing significant withdrawal signs. The 2:1 ratio was aversive because it produced withdrawal on four measures and was the only dose combination that the subjects reported that they would not pay money for.

A randomized, double-blind comparison of the effects of tablet formulations of 16 mg/day of buprenorphine at 16/4, buprenorphine/naloxone or placebo, was carried out in a multicenter trial (37). The placebo-controlled portion of the trial was carried out for a 1-month treatment duration. Subjects in either buprenorphine dose group had reduced opiate use and reduced craving versus the placebo group. Thereafter, all subjects were given open-label buprenorphine/naloxone for 11 months. Other subjects participating at new sites were given 1 year of open-label buprenorphine/naloxone.

In October 2002, the FDA approved sublingual buprenorphine tablets and buprenorphine/naloxone tablets for the management of opiate dependence (38). Buprenorphine was concurrently changed to a Schedule III drug on the U.S. Controlled Substances Act (39). According to the Drug Abuse Treatment Act of 2000, qualified physicians can prescribe Schedule III, IV, and V opiates for the treatment of opiate dependence, providing that the FDA approves the medications for that purpose.

Cocaine Dependence

According to the Office of National Drug Control Policy, there are over 3 million long-term cocaine users in the United States. Cocaine abuse and dependence are associated with mortality and significant morbidity (40). There are no FDA-approved medications for the treatment of cocaine dependence. The NIDA Medications Development Program has test-marketed medications for their potential efficacy to reduce cocaine use. Several medications have shown efficacy or preliminary evidence of efficacy. These results will be highlighted.

Disulfiram, an inhibitor of sulfhydryl-containing enzymes, is marketed as a treatment for alcohol dependence. Two open-label studies showed reduction in cocaine intake after reductions in alcohol use in disulfiram-medicated patients (41). Clinical pharmacology studies of the interaction of disulfiram and cocaine have been reported (42, 43). Responses to intranasal cocaine (1 or 2 mg/kg) were altered in subjects treated with either 250 or 500 mg/day of disulfiram (42). Subjects receiving disulfiram reported dysphoria, anxiety, and paranoia after cocaine administration.

Several investigators have reported disulfiram-related decreases in cocaine use in outpatient studies. Doses of 250 and 500 mg/day were reported to reduce cocaine use in subjects using cocaine/alcohol (44). Two other studies showed that disulfiram’s efficacy might be direct rather than indirect (through reduction in concurrent alcohol use) because they were performed with cocaine-abusing, opiate-dependent patients (45, 46). Carroll et al. (47) reported the results of a two-by-two design (drug/placebo-by-cognitive behavior therapy/interpersonal psychotherapy). The psychotherapy treatments were administered weekly for 12 weeks. The disulfiram-cognitive behavior therapy group reduced their drug use compared to the placebo-cognitive behavior therapy and the disulfiram-interpersonal psychotherapy groups. This study suggests an interaction between disulfiram and the type of psychotherapy given (47). A multicenter trial is planned as a follow-up to these studies.

Similar results (with a medication interacting with one type of behavioral therapy and not another) have been reported in a relapse-prevention trial with naltrexone (48). Abstinent, formerly dependent, cocaine-addicted patients showed less relapse when their treatment was combined with a relapse-prevention behavioral therapy but not another behavioral therapy.

Baclofen, a γ-aminobutyric acid (GABA) B agonist (60 mg/day), has been reported to reduce the craving for cocaine in a clinical pharmacology setting (49). Reduction of cocaine use has been reported in baclofen-treated patients in a randomized, placebo-controlled trial (50). Of interest, the patients with the highest levels of cocaine use during the baseline period benefited the most from this medication. A follow-up trial is ongoing to attempt to verify this finding. Another GABA-ergic medication, tiagabine, has been reported to reduce cocaine use at a dose of 24 mg/day (51). A follow-up trial has been completed, and the results will be forthcoming.

Investigators at the University of Pennsylvania have devised a scale that captures the intensity of cocaine withdrawal, the Cocaine Selective Severity Scale (52). This scale, used in conjunction with baseline drug urinalysis testing, predicts treatment efficacy, measured by the ability to maintain 3 weeks of continuous abstinence with 87% accuracy (53). Moreover, patients with high Cocaine Selective Severity Scale scores showed a differential positive response to amantadine (54) or propranolol (55) in outpatient studies.

Another use of this scale is to determine whether medications for relapse prevention will differentially affect patients with high or low Cocaine Selective Severity Scale scores. A trial was conducted with topiramate (200 mg/day maximum dose) in cocaine-dependent patients with low Cocaine Selective Severity Scale scores (<21). The patients who were abstinent during the 2-week baseline period had less return to cocaine use that those randomly assigned to placebo or the patients who continued to use cocaine during the baseline period (56). (Topiramate is a marketed antiepileptic medication that has GABA-enhancing and glutamate-inhibiting properties.) This suggests that topiramate may be beneficial, with patients capable of achieving some level of abstinence through behavioral therapy.

Modafinil is another medication of interest. It is a newer stimulant that increases daytime alertness in narcoleptic patients (57). Preclinical assessment suggests that the stimulatory action is nondopaminergic (58–60). Initial assessments of abuse liability in human subjects suggest that the abuse liability is minimal (61–63). Moreover, an initial report of cases of stimulant abusers treated with modafinil noted that one user described less craving for amphetamines, and another reported decreased craving and less use of cocaine (64). Clinical laboratory testing of the potential interactive effects of 200 and 400 mg/day doses of modafinil and cocaine in cocaine-experienced subjects was performed (65). No additive effects on cardiovascular parameters were noted. Additionally, the subjects reported a dampening of the subjective effects of cocaine on the Addiction Research Center Inventory (ARCI) Amphetamine Scale (a measure of drug euphoria).

Modafinil also has effects on neuropsychological tests that may be germane to the treatment of substance abuse disorders. In a neuropsychological test battery, modafinil selectively improved performance in normal volunteers (66). Notably, a positive effect on prepotent inhibition was seen in a go/no-go test. This suggests that impulsive responding was reduced. The authors suggested that modafinil might possibly benefit patients with attention deficit hyperactivity disorder (ADHD). A subsequent follow-up study in an adult ADHD group (67) verified that increased accuracy of responding with concomitant response delays was indicative of an inhibition of impulsive responding. Such response delays may allow for other cognitive processes to enter into the decision process.

In summary, modafinil has three effects that provide a rationale for outpatient testing in a cocaine-addicted population:

Outpatient studies assessing the safety and efficacy of modafinil in cocaine-dependent patients are ongoing.

Neurobiological Mechanisms and Pharmacotherapy for Addictions

Multiple neurobiological mechanisms are thought to be involved in the addiction to cocaine. Five mechanisms will be highlighted: 1) the effects of cocaine on the dopamine system, 2) conditioned cueing, 3) cocaine-induced priming, 4) cocaine and stress, and 5) frontal cortex inhibition mechanisms.

Appetitive processes linked to dopamine and mesolimbic-mesocortical pathways have been implicated in the reinforcing actions of cocaine (68–70). Cocaine is a dopamine transporter inhibitor that is capable of increasing the extracellular concentration of dopamine severalfold (71–73). Supraphysiological concentrations of dopamine may influence the propensity for repeating the cocaine-taking experience. Moreover, there is laboratory evidence that prolonged cocaine administration may result in a hypodopaminergic state (74).

The “agonist” therapy approach, i.e., the use of a similar drug to mimic certain aspects of cocaine, was first tried in NIDA’s attempt to develop a laboratory-based medication for cocaine dependence. Desirable compounds of this type would have reduced abuse liability, such as transporter inhibitors with a slower onset and a longer duration of action. One of the first compounds of this type to come to our attention was GBR 12909. GBR 12909 was reported to increase extracellular dopamine concentrations in the rhesus monkey, with a peak effect at 1 hour and a sustained duration (71). Similar results were seen in squirrel monkeys (73). GBR 12909 has been evaluated in the monkey and found to block cocaine self-administration (75–77), especially under conditions in which the unit dose of cocaine was low or the response requirements for cocaine were high (78).

Studies in human subjects have shown that GBR 12909 (vanoxerine) is not a psychomotor stimulant (79). NIDA has recently stopped its phase I studies in cocaine-experienced subjects because of the appearance of rate-dependent QTc prolongation in five of five subjects given 75 mg/day of GBR 12909 for 11 days (Cantilena, personal communication).

A second approach that is derived from the dopamine hypothesis of cocaine’s action is the possible development of dopamine “stabilizers,” which may be partial agonists or “partial” mixed-action antagonists. Compounds of this type are mild stimulants in animals but can reduce the locomotor stimulant effects of amphetamine (80). The interpretation at the time was that autoreceptor antagonists preferentially inhibited presynaptic dopamine receptors, but at higher doses and at higher levels of dopamine stimulation, they also block postsynaptic receptors. Thus, in low-dopaminergic states, such compounds would increase dopamine release but act as antagonists under conditions of higher dopaminergic tone. Other compounds of this type (AJ76 and UH-232) antagonized cocaine depression of ventral-tegmental area firing rates and reduced cocaine-induced locomotor activity when injected simultaneously (UH-232) or 30 minutes after cocaine administration (AJ76) (81). AJ76 also antagonized cocaine self-administration (82). A third potential autoreceptor antagonist, DS-121, was reported to produce a dose-dependent (high dose) antagonism of amphetamine or cocaine discrimination (83) and antagonized cocaine self-administration in the rat (84).

Dopamine D₂ partial agonists can also function as both agonists and antagonists and may also have stabilizing effects. Terguride, a partial agonist at D₂ receptors, reversed amphetamine withdrawal signs in animals (working for a sweetened solution) (85) and also reduced amphetamine self-administration under a progressive ratio schedule (86). Of interest, quinpirole, a full D₂ agonist, did not antagonize amphetamine self-administration. These data suggest that these “stabilizer” compounds may have different properties than full agonists or full antagonists.

The first marketed medication with a similar mechanism that has been dubbed a dopamine “stabilizer” (87, 88) is aripiprazole, which is approved for the treatment of schizophrenia. Its efficacy has been reviewed, and it is believed to have “mixed” actions or different efficacy at different receptor subtypes (89). Neuroimaging studies have show that clinically relevant doses occupy 95% of the dopamine receptors without producing extrapyramidal side effects (90). Given its characterization as a dopamine system stabilizer, NIDA is interested in evaluating its effects in both cocaine- and methamphetamine-dependent patients.

Conditioned cueing can be characterized as an intersection of Pavlovian and instrumental conditioning processes, whereby previously neutral stimuli paired to drug experiences develop both motivational and reinforcing significance. For a review of the behavioral concepts and laboratory approaches to studying this phenomenon, refer to reviews by Everitt et al. and Everitt and Robbins (91, 92). Second-order schedules of drug reinforcement are used to study the behavior, its neural substrates, and its pharmacological modulation. Lesioning studies in animals have differentiated the involvement of nuclei, such as the basolateral amygdala, in responding for cues, as opposed to a role in primary reinforcement. Human imaging studies have reported that conditioned cues activate the amygdala, anterior cingulate, lateral orbitofrontal cortex, rhinal cortex, and right hemispheric dorsolateral prefrontal cortex (93–95). Conditioned cueing may be of relevance to human addiction, and rodent models may be of value in discovering pharmacological modulators of conditioned cues.

Multiple neurotransmitter systems may be involved in behavioral responses to conditioned cues. Diverse neurotransmitter systems, such as GABA B (49), the endocannabinoid system (96), the dopamine D₃ system (97), and ionotropic glutamate antagonists (98, 99) have been implicated. Cannabinoid antagonists, D₃ partial agonists and antagonists, and AMPA antagonists are being sought by NIDA for testing in conditioned-cue paradigms.

Drug priming is another neural mechanism of interest (100). Priming is defined as the response to reintroduction of a drug, in the form of subsequent increased drug intake, in a formerly dependent drug user. This behavior can be modeled in the laboratory by using noncontingent administration of cocaine in animals trained to self-administer cocaine. Noncontingent administration can also provoke reinstatement in animals after extinction of self-administration (101).

There is evidence of both dopaminergic and glutamatergic influences on the effects of drug priming. Systemic administration of a D₁ agonist blocked cocaine priming-induced reinstatement (102), and priming could be blocked by microinjections of the dopamine antagonist flupenthixol into the medial prefrontal cortex. Reinstatement induced by microinjections of cocaine into the medial prefrontal cortex can be blocked by administration of CNQX, an AMPA/kainate antagonist, into the nucleus accumbens, suggesting that the glutamatergic pathway from the medial prefrontal cortex to the nucleus accumbens plays an important role (103). Similarly, Cornish and Kalivas (104) demonstrated a facilitatory role for glutamate acting at AMPA receptors in the effects of priming. Systemic administration of cocaine or the direct infusion of AMPA or dopamine into the nucleus accumbens can produce reinstatement of drug-seeking behavior in rats, and intra-accumbens administration of CNQX blocked the priming effects of AMPA, dopamine, and cocaine. In contrast, intra-accumbens administration of fluphenazine was only successful in blocking the effects of intra-accumbens dopamine on reinstatement, suggesting the importance of glutamate in priming-induced reinstatement.

The contribution of glutamate to reinstatement was also studied in cocaine self-administering rats and their yoked comparison subjects. Glutamate concentrations increased in the nucleus accumbens only in response to reinstatement of lever pressing for cocaine. In contrast, dopamine concentrations increased in both groups of rats. Reversible inhibition of the frontal cortex by infusion of the GABA B agonist, baclofen, prevented the increase in accumbal glutamate, suggesting that the glutamatergic prefrontal cortex connection to the accumbens is a major mediator of cocaine-primed reinstatement (105). The neuronal pathways thought to be involved in cocaine addiction have been reviewed by Kalivas (106). Reciprocal glutamatergic connections between the amygdala and the medial prefrontal cortex also suggest that some components of drug-associated cueing may be mediated through a medial prefrontal cortex-accumbal pathway. For these reasons, NIDA has great interest in evaluating AMPA antagonists or medications with an AMPA antagonist component for the prevention of relapse.

Stress is another factor thought to increase the propensity of alcohol and drug users to relapse (107–111). There are multiple models of stress in laboratory animals, and to some extent, the effects of drugs on stress are model specific and stressor specific (112). In the intermittent foot shock stress model, stress has been shown to reinstate response in animals trained to self-administer cocaine (113, 114), heroin (115, 116), nicotine (117), alcohol (118, 119), and cannabinoids (120). Central noradrenergic and extrahypothalamic corticotropin-releasing factor (CRF) have been shown to mediate the effects of intermittent foot shock stress on reinstatement. Lofexidine (121) and CRF antagonists (122) have been shown to modulate the noradrenergic and CRF components of the response to stress-induced reinstatement of cocaine self-administration.

Finally, the effects drugs of abuse on frontal cortex inhibitory processes and drug abuse as compulsive behavior are just beginning to be understood and modeled in the laboratory (123). As our understanding of the deleterious effects of long-term drug abuse on frontal lobe function and cognitive decision-making processes evolves (see reference 124 for a review), medications that affect these processes through cognitive enhancement (e.g., modafinil) will be evaluated for their efficacy in the treatment of drug abuse.

In summary, LAAM, buprenorphine, and buprenorphine/naloxone have been developed for the treatment of opiate dependence. The buprenorphine products may be prescribed by qualified physicians, representing a shift toward the medical management of opiate addiction in the office-based setting. Progress in developing medications for cocaine dependence has been realized. The “first generation” of medications for cocaine dependence are in confirmatory testing in clinical trials or in the planning stages of confirmatory trials. Molecular targets derived from neuroscience will form the second generation of products for the treatment of cocaine dependence. Insofar as these products may affect processes that may, to one extent or another, affect all substance dependence disorders, the “second generation” of products may be developed for multiple addictions.

Presented in part at the 157th annual meeting of the American Psychiatric Association, New York, May 1–6, 2004. Received Sept. 3, 2004; revision received Nov. 30, 2004; accepted Dec. 10, 2004. From the Division of Pharmacotherapies and Medical Consequences of Drug Abuse, National Institute on Drug Abuse, NIH, Department of Health and Human Services. Address corresponding and reprint requests to Dr. Vocci, Division of Pharmacotherapies and Medical Consequences of Drug Abuse, 6001 Executive Blvd., Rm. 4133, MSC 9551, Bethesda, MD 20892-9551; [email protected] (e-mail).