Opioids

George E. Bigelow and Kenzie L. Preston

Drug-abuse-related research with opioids pursues two primary clinical goals: (i) the development of effective analgesics with reduced abuse liability and (ii) the development of medications for treatment of opioid abuse and dependence. The purpose of the present chapter is to review selected recent human research developments in these two areas. These are closely related, often overlapping, areas. A major goal of medications development research is to identify medications that reduce the abuse liability of opioids. Recent research pursuing these two goals has produced an extensive body of knowledge concerning the clinical and behavioral pharmacology of opioids and of opioid abuse and dependence.

Because opioid abuse is both illegal and often over represented in marginalized segments of society, its prevalence is difficult to determine accurately. It is commonly estimated that there are 600,000 opioid addicts in the United States, but estimates of the number of opioid abusers approach 2 million. Though the population prevalence of opioid dependence is relatively low, the prevalence of vulnerability to dependence appears substantially higher. Historical experience prior to the legal restriction of opioid availability is that substantial portions of the population became regular users when opioids were readily available (39). Similarly, 15-20% of American military enlisted men serving in Vietnam, where opioids were widely and cheaply available, reported having become addicted (54). The lower current population prevalence of opioid dependence is presumably attributable to education, social and legal sanctions, and restrictions on availability, as well as to research that identified drugs with significant abuse potential prior to their introduction into general use.

There is a risk that antidrug education, sanctions, and restrictions, by emphasizing the risk side of opioid use, may deter physicians and patients from using opioids when medically appropriate and may thereby result in inadequate treatment of clinical pain. In fact, the risk that the behavioral disorder of opioid abuse will result from appropriate medical use of opioid analgesics is quite small. A study of 11,882 hospitalized patients who received opioids found only four cases of subsequent new addictions—an incidence of 0.03% (41). Most opioid abuse develops with illicit opioids such as heroin. Of course, medically used opioids have high abuse liability, and they will be sought by abusers, who may feign need; therefore caution in medical use is essential. In addition, opioids pose a significant occupational hazard for abuse among health professionals who may have convenient access and be tempted either to self-medicate or to experiment.

The sections below describe and review the methods that have been developed for assessing opioid abuse liability in humans and summarize the data these methods have generated concerning the clinical pharmacology of opioid agonists, antagonists, and mixed agonist- antagonists. Drug-abuse-related research interests have been the dominant force in studying and characterizing the clinical pharmacology of opioids. The utility of these methods has been broad. They have served an important descriptive function in simply analyzing and quantifying aspects of opioid action and opioid dependence. They have served an important basic science function in developing methods and indices that have permitted examination in humans of the specific behavioral and physiological functions controlled by the increasingly complex neurobiological processes and actions involved in opioid pharmacology, and they have served as a stimulus to advanced basic science conceptualization of how opioids act in brain. They have served an important practical and public health function in assessing the potential abuse liability of new compounds and thereby insuring proper prior professional education about abuse risks and/or proper prior societal restriction or regulation of drug availability. Finally, they have more recently served an important applied research and therapeutic function as they have been adapted and utilized in the task of assessing and developing new pharmacotherapies for opioid abuse and dependence.

Much of the developmental work in human opioid psychopharmacology was conducted at the Addiction Research Center (ARC) in Lexington, Kentucky (2). Many of the methods and principles used today to assess abuse liability in humans are derived from this ARC work. While the emphasis of ARC research was on opioids, the methods developed have served as the model from which adaptations have been derived for assessment of other drug classes.

The major modern methods include: assessment of subjective effect profiles; assessment of discriminative drug effects; assessment of behavioral reinforcing properties and self-administration; assessment of degree of antagonist activity; and assessment of physical dependence capacity. Inclusion of placebo as a negative control and the use of double-blind procedures are critical elements of these procedures. To provide positive reference values against which to interpret the effects of the test drug, it is also critical to include appropriate positive control drugs; these would normally be prototypic drugs of known high and/or low abuse liability, such as morphine or naloxone. This section reviews methods used to evaluate drugs on these dimensions and summarizes results for the clinically available mixed agonist-antagonist opioids.

The most thoroughly developed and most frequently used psychopharmacological technique for evaluating opioids in humans is assessment of subjective effects through questionnaires. The purpose of subjective effect indices is to provide qualitative and quantitative characterization of changes produced by test drugs in subjects' mood, feelings, perceptions, and symptoms. The five basic types of information most frequently collected in subjective effect questionnaires are: whether and/or how much the subject feels a drug effect; whether and/or how much the subject likes the drug effect; what symptoms are produced by the drug; what pharmacological class is the drug most like; and what mood effects does the drug produce. The questionnaires most commonly used to study opioids include the ARC Single-Dose Questionnaire, the Morphine-Benzedrine Group (MBG), Pentobarbital-Chlorpromazine-Alcohol Group (PCAG), LSD scales of the Addiction Research Center Inventory (ARCI) (widely used as indices of "euphoria," "apathetic sedation," and "dysphoria," respectively), global drug effect scales (e.g., magnitudes of drug effect, liking of the drug, good and bad effects, and withdrawal sickness), adjective rating scales, and drug class identification scales (e.g., see refs. 8, 21, 44, and 48). The adjectives used have depended on the class of drugs being studied and their expected effects. In opioid studies, symptoms associated with opioid agonist effects (such as itching, nodding, and talkativeness), agonist-antagonist effects (such as floating, confused, and numb), and opioid abstinence (such as watery eyes, chills, and gooseflesh) have been used. To evaluate the type of drug they have received, subjects identify the pharmacological class of the test drug from a list of eight to ten choices. In most studies, subjective effect questionnaires are completed before drug administration to determine a baseline and then again at specified intervals after administration to determine the time course of effects.

The major recent change in subjective effects assessment has been their increased incorporation as concurrent measures in behavioral studies such as drug discrimination and self-administration. Use of subjective effects indices to predict abuse potential has been criticized because questionnaire responses are quite indirect predictors of reinforcing efficacy. Better understanding of the relationship between subjective effects and behavioral indices may improve our ability to interpret subjective effect data as they relate to abuse liability assessment.

In drug discrimination studies, subjects are typically trained to emit one response in the presence of a training drug and to emit an alternate response in the absence of the training drug. The method has been used extensively in nonhumans to study the pharmacology of opioids and to assess abuse liability of new drugs. In the last decade, investigators have adapted these paradigms to study opioid pharmacology in humans (for review see ref. 43). These studies have combined the collection of behavioral drug discrimination data with the traditional subjective effect questionnaire data. Drug discrimination itself provides a novel procedure for characterizing human opioid pharmacology, and concomitant collection of subjective effect data permits study of the relationship between subjective and discriminative effects of opioids.

Human opioid drug discrimination studies have used either two or three training drugs, and they have used participants with extensive histories of opioid abuse (43). In the discrimination training phase, each training drug administration is paired with an identifying label (frequently an arbitrary letter code). An extrinsic reinforcer (money) is provided contingent on correct discrimination performance. Acquisition of the discrimination is verified by re-exposing subjects to the training drugs and determining whether their correct drug labels (letter codes) are identified. Generalization testing sessions are then conducted to assess the dose-effect function of one or more test drugs. Drugs are not identifiable by appearance or volume and are given under double-blind conditions. In addition to discrimination responses, subjective, physiological, and psychomotor task performance measures can be collected.

A series of studies has been conducted in opioid abusers trained to discriminate between intramuscularly administered placebo and various opioid agonists and mixed agonist-antagonists (45, 50, 53); Preston et al., submitted). The training conditions (e.g., number and types of training drugs and dependence level of the subjects) were systematically varied across the studies, and they included two-choice (hydromorphone/placebo) and three-choice (saline/hydromorphone/pentazocine or saline/hydromorphone/butorphanol) discriminations in nonphysiologically dependent opioid abusers and three-choice saline/hydromorphone/naloxone discriminations in opioid-dependent subjects. Hydromorphone, pentazocine, butorphanol, nalbuphine, and buprenorphine dose-response curves were determined in each study to assess cross-generalization. One significant finding was that opioid abusers had no difficulty discriminating between a mu agonist and a mixed agonist-antagonist with both mu and kappa agonist activity (pentazocine or butorphanol). Not unexpectedly, methadone-dependent subjects had no difficulty discriminating an opioid agonist from an opioid antagonist.

A second major finding was that the results of generalization tests of the agonist-antagonists were highly dependent on the specific discriminations that had been trained. For example, under some conditions (nondependent subjects in a two-choice hydromorphone/placebo discrimination), hydromorphone, pentazocine, butorphanol, nalbuphine, and buprenorphine were all discriminated as hydromorphone-like at one or more doses (53); in contrast, under other conditions (a three-choice saline-hydromorphone-pentazocine discrimination), none of the mixed agonist-antagonists were discriminated as hydromorphone-like (45). It was concluded that three-choice discrimination procedures permit a more precise and refined differentiation among test drugs than do two-choice procedures. Selection of specific discrimination training procedures permits one to focus the experimental question and to assess and detect differences among drugs which also possess similarities. Thus, drug discrimination can be a valuable tool for studying the agonist-antagonist opioids, but assessment under multiple procedural/training conditions may be necessary to characterize fully their pharmacological properties.

Generalization across routes of administration and across drug class has been evaluated by testing oral doses in volunteers trained to discriminate among intramuscular saline, hydromorphone, and pentazocine (4). Results are consistent with the view that drug discrimination performance is controlled by the central pharmacological actions of study drugs and not by peripheral effects that might be related to the route of administration.

Studies of opioids have provided extensive data on the covariation of subjective and discriminative responses over a range of test doses (43). These studies suggest a rather strong relationship between subjective and discriminative effects. Opioids discriminated as similar had similar subjective profiles, and opioids with similar subjective profiles were discriminated as similar. Novel drugs had subjective effects that overlapped with those of the drugs to which they were discriminated as similar. Thus, these human data are supportive of the view that drug discrimination effects in animals may be analogous to subjective effects in human. However, the data also make clear that there are circumstances where subjective and discriminative effects may diverge. This phenomenon clearly supports the value of using multiple methods for assessing abuse liability and the value of incorporating subjective effect assessment as a concurrent and complementary procedure along with other methods.

The utility of human drug discrimination procedures in abuse liability assessment has been reviewed by Preston (42). Drug stimulus similarity, like subjective effect profile, is an indirect predictor of relative reinforcing efficacy and is not a direct index of abuse liability. However, if appropriate discriminations are trained and if appropriate reference drugs of known abuse liability are included, drug discrimination methods can yield quite specific information about relative abuse potential. This is a method that should prove quite valuable in assessing new opioids as they are developed.

There has been extensive development and use of drug self-administration technology as a tool for assessing opioid abuse liability in animals. In contrast, there has been little opioid self-administration research in humans. This now promises to be a fertile new research area.

Studies in the 1970s and 1980s showed that opioid addicts would complete operant work requirements to obtain doses of opioids (38). This self-administration methodology was used in the human laboratory to document the therapeutic efficacy of methadone, naltrexone, and buprenorphine by showing that pretreatment with these agonists, antagonists, and agonist-antagonists decreased the self-administration of opioid agonists (27, 37, 38).

More recently, Lamb et al. (30) examined the self-administration of different doses of morphine and concurrently evaluated subjective effects following drug administration. Subjects could respond on a fixed-ratio, second-order operant schedule to obtain a single daily intramuscular injection of placebo or morphine (3.75, 7.5, 15, and 30 mg). Placebo was not self-administered, but all active doses were self-administered by the majority of subjects. Interestingly, there was a dissociation between self-administration behavior and subjective report of drug effects; only the highest morphine dose reliably increased scores on subjective effects scales. These data suggest that self-administration procedures may be more sensitive indices of abuse liability than are the traditional subjective effects procedures.

The drug self-administration method provides direct assessment of drug-taking behavior and drug reinforcement, and thus it has substantial face validity as an index of abuse liability. Self-administration methods for opioid abuse liability assessment have not yet received sufficient use to develop the necessary validation of their predictive accuracy. Much work is needed to understand the relationships among self-administration, discriminative stimuli, and subjective effects and to develop practical self-administration methods that reliably predict abuse liability.

Far more opioid self-administration work has been done in recent years in the context of patient-controlled analgesia (19). This procedure permits patients with clinical pain to control (within limits) the frequency and timing of opioid analgesic administration. Safe and effective use of patient-controlled analgesia depends on knowledge of dose effects, onset latency, potency, intrinsic activity, duration of action, elimination rate, and relative receptor activity. These same parameters are critical to the understanding of opioid self-administration in the drug abuse context. Thus, increased interactions between researchers across these two areas may significantly enhance our knowledge of the principles of human opioid self-administration (see Animal Models in Drug Addiction, for related discussion.)

Opioids are well known for their capacity to produce tolerance and physical dependence, and these are important dimensions to assess in characterizing the complete abuse liability profile of any opioid. Methods for studying the development of tolerance and physical dependence and their physiological and behavioral consequences in humans have been reviewed by Jasinski (21).

Historically, the method used at the ARC (the direct addiction procedure) involved administration of increasing doses of a test drug to subjects over time until a predetermined maximum stabilization level was reached, then testing with known agonists and/or known antagonists, and then examining the effects of abrupt discontinuation of the maintenance test drug. The dose escalation period provided information about tolerance development, as did the response to supplemental opioid challenges during the maintenance period. Antagonist challenges (the abstinence precipitation procedure) and abrupt substitution of placebo for the maintenance test drug (the spontaneous abstinence procedure) provided information about physical dependence and about the nature of the abstinence syndrome. Both physiological and subjective effect data were assessed throughout these studies. In a modification of the direct addiction procedures, the ability of test drugs to substitute for morphine and suppress abstinence in morphine-dependent humans was tested (the substitution procedure). The series of direct addiction and substitution studies conducted at the ARC from the 1930s through the 1970s was critical to establishing the existence and character of opioid physical dependence and the abstinence syndrome, defining differences between the effects of chronic administration and abstinence syndromes of mu and kappa agonists, and determining effective treatments of abstinence signs and symptoms. These studies were also integral parts of the abuse liability assessment of novel opioids. For a variety of reasons, direct addiction studies are no longer being conducted.

Studies of opioid physical dependence, tolerance, and abstinence in humans have continued in the last decade by combining methods adapted and modified from the ARC direct addiction and substitution procedures with newer research techniques. For example, to evaluate the cross-tolerance to the subjective and physiological effects of supplemental opioids conferred by methadone maintenance treatment, McCaul et al. (36) compared the acute subjective and physiological effects of challenge doses of intravenously administered hydromorphone in methadone-maintained and in non-physiologically dependent opioid abusers. The primary innovation has been to use volunteer methadone maintenance patients or morphine-maintained opioid abusers as the dependent subjects and to conduct substitution studies assessing whether test drugs suppress the opioid abstinence syndrome. Thus, human laboratory studies of physical dependence show good promise for screening of potential treatment agents and for selection of doses for clinical trials of new pharmacological treatments of opioid abuse. (See Adaptive Processes Regulating Tolerance to Behavioural Effects of Drugs., for related discussion.)

Administration of an opioid antagonist to individuals who are physically dependent on opioids produces the signs and symptoms of abstinence. Methods for quantifying opiate abstinence precipitated by antagonists were developed at the ARC (for review see ref. 21). The methods involve administering placebo and one or more antagonists or agonist/antagonists to subjects physically dependent on an opioid agonist such as morphine or methadone. Although the opiate abstinence scoring system developed by Himmelsbach still serves as a basis for the quantifying physiological abstinence signs, subjective measures of abstinence are now included in most studies (51). Precipitation studies can sometimes be complicated by the fact that precipitation of abstinence, while not life-threatening, does involve some physical discomfort to opioid dependent volunteers. Care must be used in designing these studies to minimize this discomfort.

Opioid antagonists have proven to be useful tools in human opioid pharmacology research. Antagonist-precipitated abstinence is similar in profile, but more rapid in onset and shorter in duration than that produced by abrupt discontinuation of the maintenance agonist. Naloxone is the most frequently used antagonist in abstinence precipitation studies. At doses of 0.1-0.2 mg i.m., naloxone produces dose-related abstinence-like effects that last 30-60 min (48, 49)). Naloxone-precipitated abstinence has been shown to serve as a discriminative stimulus in opioid-dependent humans (50). In studies comparing the antagonist activity of the agonist-antagonist opioids, butorphanol, nalbuphine, and pentazocine were shown to precipitate abstinence syndromes that differed somewhat from that produced by naloxone (48, 49, 62). Information about the antagonist potency of these drugs may be used in conjunction with analgesic potency to calculate an antagonist/analgesic potency ratio (the dose precipitating abstinence divided by the analgesic dose). This ratio provides a useful index for comparing among the agonist-antagonists and provides a basis for predicting abuse liability.

While the phenomenon of physical dependence following chronic administration of opioids is well established, the number of repeated administrations necessary to produce physical dependence is less clear. Evidence now indicates that the physical dependence process begins after even a single opioid agonist administration. Acute physical dependence refers to the abstinence syndrome precipitated by the administration of an opioid antagonist after either a single dose or a short-term administration of an opioid agonist. Acute physical dependence has been demonstrated in humans as well as in mice, rats, dogs, and monkeys (7). The procedure consists of giving a single administration of an opioid to a nondependent subject, followed usually several hours later by administration of an opioid antagonist, and then measuring the response. The measures used to assess the precipitated abstinence are the same as those used in the studies of opioid antagonist administration in subjects receiving chronic administration of opioid agonists.

The doses of agonist used in the paradigm are at those used therapeutically for analgesia and above—for example, 10-30 mg of morphine (7, 17). The doses of opioid antagonist necessary to precipitate abstinence after a single morphine dose are larger than those used to precipitate abstinence after chronic agonist treatment—for example, 10 mg versus 0.2 mg of naloxone, respectively. The signs and symptoms of abstinence produced in the acute physical dependence paradigm are generally similar to those of opioid antagonist-precipitated abstinence in chronically treated subjects, though they tend to be milder. The intensity of the response is directly related to the dose of morphine pretreatment (7) and directly related to dose of the antagonist (17). Precipitated abstinence can be produced as early as 45 min and as late as 24 hr after a single intramuscular injection of morphine (18, 28). Naloxone-precipitated abstinence may be observed even longer (up to 96 hr) after a single dose of the longer-acting agonist methadone, and at much lower doses (approximately 1 mg) of naloxone (60); this suggests that the physical dependence produced by methadone is more intense and protracted than that produced by comparable doses of morphine. Additional studies are needed to understand fully the relationship between duration of agonist action and the time course of acute physical dependence, but it appears that the acute physical dependence paradigm may be quite useful as an index of physical dependence capacity.

The discovery of the agonist-antagonist opioids was a major advance in the development of analgesics with low abuse potential and as tools for pharmacological research, leading to the postulated existence of multiple opioid receptors, the description of the functional consequences of activation of the mu and kappa receptors, and descriptions of the mu and kappa abstinence syndromes (35). A number of agonist-antagonists have been marketed as analgesics and have been extensively studied or have recently become available for therapeutic use. Human behavioral pharmacology abuse liability assessment results and the pharmacological activities of these drugs are summarized in Table 1 and described below.

Pentazocine is an agonist-antagonist with both mu and non-mu agonist activity in humans. It produces both morphine-like and non-morphine-like subjective effects, with the non-morphine-like effects predominantly occurring at higher doses (25, 46). Naltrexone, which is more potent in blocking mu-receptor activity than kappa-receptor activity, antagonized the effects of hydromorphone to a greater extent that those of pentazocine, supporting the suggestion that pentazocine has kappa agonist activity (44). Drug discrimination studies have also shown a mixed mu and non-mu profile of effects. Pentazocine was discriminated as hydromorphone-like in subjects trained in a two-choice discrimination between the effects of intramuscularly administered saline and hydromorphone (53). On the other hand, the stimulus effects of pentazocine were sufficiently different from those of hydromorphone to permit training of a reliable discrimination in subjects trained in a three-choice saline-hydromorphone-pentazocine discrimination (4, 45). Butorphanol, but not nalbuphine or buprenorphine, produced pentazocine-appropriate responding. Repeated administration of pentazocine produces physical dependence characterized by both mu and non-mu characteristics during chronic treatment and on abrupt abstinence (25). Pentazocine appears to have significant antagonist effects as well. Pentazocine precipitated abstinence in morphine-dependent and methadone-dependent subjects at doses of 60-120 mg i.m. (25, 62), and although tested under a number of morphine dose levels, pentazocine did not suppress abstinence in morphine-dependent subjects (25).

In spite of its significant non-mu-like subjective effects and antagonist activity, an outbreak of pentazocine abuse occurred in the mid 1970s in the form of "T's and Blues," intravenously injected combinations with the antihistamine tripelennamine (55). A laboratory study showed that combinations of pentazocine and tripelennamine were identified as opioids more frequently and produced greater euphoria and liking and less dysphoria than either drug alone (31). After pentazocine was moved from unscheduled status to control under Schedule IV of the Controlled Substances Act, and a pentazocine 50 mg/naloxone 0.5 mg combination product was marketed in place of the original pentazocine 50 mg tablets, the incidence of pentazocine abuse significantly declined.

Butorphanol, like the other marketed agonist-antagonists, has similarities to mu agonists as well as differences. Both morphine and butorphanol increased subjects' liking scale scores and increased opiate symptom scale scores, but butorphanol produced other, non-morphine-like effects as well (51). The subjective effects of butorphanol were more similar to those of the agonist-antagonists cyclazocine and pentazocine. Drug discrimination studies have also shown a mixed mu and non-mu profile of effects. Butorphanol was discriminated as hydromorphone-like in nondependent subjects trained in a two-choice saline-hydromorphone discrimination (52). In a three-choice saline-hydromorphone-pentazocine discrimination, butorphanol produced pentazocine-appropriate responding (45). In opioid-dependent individuals, butorphanol acts primarily as an antagonist. The results of precipitation studies with butorphanol have yielded inconsistent results. Butorphanol precipitated abstinence in methadone-dependent subjects (48 but failed to precipitate abstinence in morphine-dependent subjects (51). Interestingly, although butorphanol produced effects generally similar to the effects of naloxone, there were some differences in the abstinence syndromes precipitated by naloxone versus butorphanol. The stimulus properties of butorphanol, however, are clearly abstinence-like in opioid-dependent individuals; butorphanol produced naloxone-appropriate responding in methadone-maintained subjects trained in a three-choice (saline-hydromorphone-naloxone) discrimination (50). Butorphanol also failed to substitute for morphine and to suppress abstinence in humans maintained on a relatively low dose of morphine (60 mg/day) (51). Physical dependence on butorphanol was both morphine-like and nalorphine-like during chronic dosing and more similar to cyclazocine than to morphine after abrupt abstinence (51). Subjects requested morphine for relief of their symptoms; however, when offered butorphanol or a sedative, subjects refused the butorphanol and, instead, chose the sedative.

Butorphanol was recently marketed in a novel transnasal formulation. A study comparing the effects of butorphanol given by the transnasal and intramuscular routes showed that, while the overall profile of effects were not substantially different, transnasal butorphanol had a slower onset and a decreased potency relative to i.m. butorphanol (53). Dysphoric sedation was prominent after administration of 4 mg by the i.m., but not the transnasal, route. The results suggest that there is a ceiling on the effects of transnasal butorphanol, perhaps due to limited absorption.

Nalbuphine produced (a) a profile of subjective effects different from that of the prototypic agonist morphine and (b) a very shallow dose-response curve on acute administration (24). In drug discrimination studies, nalbuphine was discriminated as hydromorphone-like in subjects trained in a two-choice discrimination between i.m. saline and hydromorphone (52). When tested in non-physically dependent subjects trained in a three-choice saline-hydromorphone-pentazocine discrimination, nalbuphine was consistently discriminated from placebo but produced mixed hydromorphone- and pentazocine-appropriate responding, not consistently discriminated as either (50).

Nalbuphine has strong antagonist activity relative to its agonist activity. It precipitated abstinence in morphine-dependent subjects at doses of 6 and 12 mg/70 kg (24) and in doses of 2 mg/70 kg and higher in methadone-dependent subjects (49, 50), doses less than its therapeutic dose for analgesia (10 mg). In opioid-dependent humans trained in a three-choice drug discrimination between i.m. saline, hydromorphone, and naloxone, nalbuphine was discriminated as naloxone-like (50). The ability of nalbuphine to substitute for morphine and suppress abstinence in humans has been tested in subjects maintained on morphine 30 mg/day using the 24-hr substitution technique (Jasinski, personal communication). Nalbuphine failed to suppress the intensity of the abstinence syndrome. In direct addiction studies, nalbuphine administered up to 51 days in increasing doses to 147-240 mg/day produced morphine-like effects at low doses but produced disturbing side effects at higher doses (24). Abrupt abstinence from nalbuphine was followed by a mild abstinence syndrome that was characterized by both morphine-like and nalorphine-like signs and symptoms. Subjects requested opiates for relief from the abstinence.

Buprenorphine, administered parenterally in doses to 2 mg, produced morphine-like effects including increases in drug liking, opiate symptoms, and MBG scale scores (26). Administered sublingually to doses of 32 mg, buprenorphine produced long-lasting, morphine-like subjective effects, without serious side effects (65). In drug discrimination studies, buprenorphine was discriminated as hydromorphone-like in nondependent subjects trained to discriminate saline and hydromorphone. In nondependent subjects trained to discriminate among saline, hydromorphone, and pentazocine, there was not complete generalization to either hydromorphone or pentazocine at any dose tested, though buprenorphine was clearly discriminated from saline (45). Buprenorphine does not have strong antagonist effects. In subjects maintained on relatively low doses of methadone, sublingual buprenorphine to 4 mg (22) and intramuscular buprenorphine to 8 mg (61) failed to produce naloxone-like subjective effects, though reports from circumstances of higher methadone maintenance doses indicate that under some conditions buprenorphine does precipitate abstinence-like effects in dependent patients. The physical dependence capacity of buprenorphine was tested in a direct addiction study in doses increasing to 8 mg over 45-52 days (26). Buprenorphine produced morphine-like effects including liking of the drug effect and was identified as "dope." On abrupt withdrawal the intensity of abstinence from buprenorphine was greater than that produced by placebo but milder than that produced by morphine, cyclazocine, nalorphine, nalbuphine, pentazocine, butorphanol, profadol, and propiram. Overall, buprenorphine appears to have the most morphine-like effects of the marketed agonist-antagonists. Buprenorphine-naloxone combinations have been studied as part of an effort to develop a buprenorphine formulation with low abuse potential. Buprenorphine administered to methadone-maintained subjects in combination with naloxone slightly attenuated the antagonist effects of naloxone (47). Concurrent naloxone administration attenuated the opioid effects of buprenorphine in nondependent subjects (67).

The majority of human behavioral pharmacology research on opioids has been conducted in subjects with histories of opioid abuse. Reports of experienced opioid abusers have been argued to be the best predictors of abuse liability because abusers have histories of self-administration of opioids and presumably are sensitive to their positive mood effects. Until recently, there has been only limited information on the subjective effects of opioids in individuals without substance abuse. Zacny et al. (68, 69, 70, 71) have recently begun to publish a series of studies in which opioids have been tested in healthy, nonabuser volunteers using the methods developed for assessing abuse potential. Intravenously administered morphine, fentanyl, meperidine, and dezocine have been fairly consistent in producing constellations of effects in nonabuser volunteers that are similar to those produced in experienced opioid abusers.

A great deal of research attention in recent years has been directed toward the goal of developing new and/or improved pharmacological treatments for drug abuse. Consequently, there have been significant advances in pharmacotherapies for opioid abuse and dependence.

The purposes of short-term pharmacotherapies for opioid abuse or dependence are as follows: (a) live-saving reversal of acute opioid intoxication and respiratory depression and (b) palliative care to ease the process of opioid withdrawal and detoxification and to suppress the opioid abstinence syndrome. Naloxone remains the treatment of choice for the first of these. Described below are recent developments in opioid detoxification.

Clonidine is an alpha-2-adrenergic agonist, marketed for treatment of hypertension, which has been shown to suppress many of the signs and symptoms of opioid withdrawal. Research with clonidine has been quite informative concerning the neurobiological mechanisms involved in opioid physical dependence and withdrawal. Clonidine acts via autoreceptors in the locus coeruleus to suppress adrenergic hyperactivity there that is involved in the expression of the opioid withdrawal syndrome. Its efficacy is limited primarily to suppression of autonomic signs and symptoms such as sweating, diarrhea, intestinal cramping, and nausea; it is relatively ineffective in suppressing insomnia, muscle aches, and drug craving (23). Clonidine-assisted opioid detoxification involves abrupt cessation of opioids and initiation of clonidine; there is greater symptomatic discomfort than with gradual opioid dose reduction, but much less than with abrupt opioid withdrawal alone. Clonidine's major advantage is that it is a non-narcotic and has low abuse liability; thus, it is especially attractive in settings where diversion and abuse might be of concern or where the legal restrictions on narcotic use present a problem.

Opioid detoxification may be hastened and the duration of the opioid abstinence syndrome shortened by administration of opioid antagonists (such as naloxone or naltrexone), which displace the opioid agonist from the receptor. The precise mechanisms involved in the shortening of the abstinence syndrome are not yet known and require further research. Antagonist administration and clonidine administration have been combined to provide a rapid inpatient detoxification procedure that results in patients' being converted, in approximately 4 days, from opioid physical dependence to maintenance on full blocking doses of narcotic antagonist (11). It has been suggested that this procedure might be further improved by first switching patients to maintenance on the partial agonist buprenorphine, because the abstinence syndrome following buprenorphine discontinuation is less intense than that following discontinuation of pure agonists such as heroin, morphine, or methadone (58).

The purpose of longer-term pharmacotherapy of opioid abuse and dependence is to reduce or eliminate illicit opioid self-administration. Medications achieve this goal by reducing or eliminating the reinforcing effects of illicit opioids. This is done via mechanisms of either opioid tolerance and substitution or opioid receptor blockade. Since the mid-1960s the primary medication for long-term treatment of opioid abuse has been methadone. The available pharmacotherapeutic options are now increasing. New medications have been made available for clinical use or are under development and expected to be available in the near future. At the same time, systematic research has provided new guidance about how to use these and the previously available treatment medications most effectively. Relevant characteristics of the major treatment medications are summarized in Table 2.

In July 1993, levo-alpha-acetyl-methadol (LAAM) was approved by the U.S. Food and Drug Administration for marketing as a maintenance treatment for opioid dependence. Though the research supporting this approval was spread over the preceding four decades, it is worthwhile to review LAAM's characteristics here because its clinical introduction may significantly alter the practice of opioid maintenance treatment of addicts.

LAAM, also sometimes called levomethadyl acetate, is a synthetic mu-opioid agonist structurally related to methadone. Its synthesis derived from efforts to develop opioid analgesics of reduced abuse liability (13). It produces typical opioid mu-agonist effects, including analgesia, respiratory depression, miosis, decreased gastrointestinal motility, euphoria, and, in opioid-dependent subjects, suppression of the opioid abstinence syndrome. It produces or sustains opioid physical dependence, and its discontinuation is followed by the characteristic opioid abstinence syndrome, consisting of rhinorrhea, lacrimation, gooseflesh, mydriasis, restlessness, muscle aches, and drug-seeking behavior. After equivalent maintenance doses the abstinence syndromes following LAAM and methadone are of similar intensity and duration, though that following LAAM has a somewhat slower onset (14, 57). Its therapeutic efficacy is through the mechanism of opioid tolerance and substitution.

The important features of LAAM that led to its consideration and development as a maintenance treatment for opioid dependence are its good oral bioavailability, its slow onset and long duration of action, and its relatively low parenteral abuse liability (20). From a treatment perspective, LAAM is a pharmacokinetically complex drug. This pharmacokinetic complexity likely accounts for many of its desirable features, but also introduces a need for special cautions. LAAM is metabolized to two active metabolites, nor-LAAM and di-nor-LAAM, both of which are more potent than LAAM itself and both of which have long half-lives (9). Following oral administration, LAAM's opioid effects appear at about 90 min, with peak effect approximately 4 hr post dosing; in contrast, following parenteral administration the onset of LAAM's opioid effects is delayed 4-6 hr and the effects gradually increase over 12-16 hr (10). This difference in onset as a function of route of administration is likely due to first-pass inactivation of LAAM and gradual accumulation of active metabolites following parenteral administration. With chronic dosing there is gradual metabolite accumulation such that 2-3 weeks are required for plasma level stabilization (10). One desirable consequence of this slow onset and gradual accumulation is that LAAM has relatively low abuse liability, especially by the parenteral route. A second, and less desirable, consequence is that the onset of therapeutic effects is similarly delayed. This less-than-optimal pharmacological substitution early in treatment may increase patient dropout and illicit heroin use during this time (56). Therefore, patients must be informed of LAAM's time course, urged to remain in treatment, and cautioned not to use illicit drugs during the initiation of LAAM therapy in a way that might additively interact with LAAM's delayed onset and produce overdose or other toxicity.

LAAM suppresses the opioid abstinence syndrome and prevents the effects of injected opioids for 72 hr, versus 24 hr for methadone (32). LAAM's long duration of action is its greatest benefit in comparison to methadone. LAAM is typically administered thrice weekly—for example, Monday, Wednesday, Friday, with the Friday dose being increased by 30-40% to compensate for the longer duration to be covered. Daily dosing with LAAM must be avoided because there is a risk of toxic—even fatal—accumulation of active metabolites. LAAM's longer duration of action, less frequent dosing, and less frequent clinic visits are a convenience to patients and to clinicians, may make treatment more comfortable and acceptable to patients, may decrease per-patient treatment costs (or increase treatment availability) and, because U.S. regulations prohibit LAAM take-home medication, may eliminate the problem of diversion and abuse of take-home medications which has sometimes been a problem in methadone treatment.

Numerous clinical trials have supported the conclusion that thrice-weekly LAAM treatment is equi-efficacious to daily methadone treatment (15, 20, 33, 34, 56, 57, 72). Outcome indices have included opioid-positive urines, withdrawal symptoms, and treatment retention. It appears that 1.2-1.3 mg LAAM thrice weekly is equivalent to 1 mg methadone daily. Thus, typical thrice-weekly LAAM maintenance doses might be 70/70/100 mg or 100/100/140 mg; these maintenance levels should be attained via 2-4 weeks of gradual dose escalation.

Because the majority of the clinical research supporting LAAM's safety and efficacy was one to two decades old at the time, prior to granting approval for general clinical use the Food and Drug Administration required the gathering and assessment of some modern experience. This so-called "labeling assessment study" was an open-label, multisite assessment of the ability of typical treatment clinics (i.e., not specialized research centers) to provide LAAM treatment guided simply by the instructions in the proposed LAAM package insert. No special problems were encountered in the treatment of females (who had been largely absent from older trials) or in the treatment of patients with substantial concurrent cocaine abuse (as is common now). However, presumably because of symptomatic withdrawal complaints, a substantial proportion of patients received take-home methadone doses as a supplement to help them through the long 72-hr interdose interval. This raises some questions about how successful LAAM treatment will be in eliminating the need for take-home medications.

Buprenorphine is an opioid partial agonist exerting its primary pharmacological effects at the mu receptor and producing a morphine-like profile of effects (26). It is being very actively studied and developed as a treatment for opioid dependence. Over the range of 2-16 mg/day, s.l. buprenorphine produces dose-related attenuation of response to parenteral opioid challenge (6). Several double-blind outpatient clinical trials have reported buprenorphine to be equi-efficacious to methadone treatment (5, 64). The primary features of buprenorphine that make it attractive as a potential treatment medication are its combination of agonist and antagonist actions, and consequent reduced abuse liability, its mild abstinence syndrome following discontinuation, and its improved safety profile relative to full opioid agonists.

As a partial agonist, buprenorphine has the potential to display either agonist-like or antagonist-like effects, depending upon the circumstances. In nondependent subjects buprenorphine acts as an opioid agonist (26); in moderately dependent subjects it may show neither agonist nor antagonist effects (61); and in more highly-dependent subjects it may display antagonist-like abstinence-precipitation effects (June et al., submitted). This potential for antagonist activity is seen as reducing the abuse potential of buprenorphine below that of full agonists; however, further study of this relationship is needed in order to clarify the conditions and parameters under which opioid-dependent patients may be treated with buprenorphine without precipitating the opioid abstinence syndrome. It is not clear at this time whether buprenorphine's clinical efficacy occurs via a mechanism of opioid tolerance and substitution or via opioid receptor blockade. It may be that different aspects of its efficacy derive from each of these mechanisms.

Several studies have reported that discontinuation of chronic buprenorphine treatment results in a minimal-to-mild opioid abstinence syndrome with onset 5-13 days after the last dose of buprenorphine (16, 26). This suggests that buprenorphine may be of significant benefit in opioid detoxification treatment. It also suggests that patients may feel no need to attend treatment regularly. One study has reported buprenorphine attendance/retention rates to be equivalent to those with methadone (64). Another has reported inferior attendance/retention with buprenorphine, but this may have been due to an inadequate dose of buprenorphine (29).

Buprenorphine appears to have quite a favorable safety profile as a consequence of its being a partial agonist. Its limited ability to activate opioid mechanisms results in a ceiling on the magnitude of its effects, and it appears that this ceiling is below that of opioid toxicity. Buprenorphine challenges up to 32 mg s.l.—that is, up to 100 times the analgesic dose—were tolerated by nondependent opioid abusers without adverse effect other than prolonged sedation (65).

Buprenorphine has several potential disadvantages as a treatment medication. Chief among these are its low oral bioavailability and the fact that it has significant morphine-like abuse liability. Buprenorphine's oral bioavailability is about 15%. Because it is expensive to produce, this low oral availability has the consequence that buprenorphine is given sublingually—clinically a somewhat inconvenient procedure, but one with bioavailability of about 50%. This route requires that buprenorphine be provided in a water-soluble dosage form that has the potential for diversion to parenteral administration. Buprenorphine analgesic preparations have been diverted to parenteral abuse by heroin abusers where they have been marketed. This potential for parenteral diversion, combined with the large doses of buprenorphine used in addiction treatment, has resulted in buprenorphine treatment's being designed as a clinic-only procedure—that is, without the opportunity for take-home dosages.

The feasibility and efficacy of less-than-daily dosing with buprenorphine are under active investigation. The slow onset and mild characteristics of the opioid abstinence syndrome in buprenorphine-treated subjects suggest that less-than-daily dosing should be acceptable. However, one study of alternate-day dosing found corresponding fluctuations in symptom reports (16). Another study has reported that doubling the buprenorphine dose prior to omitting a daily dose results in stable symptom profiles across both days (1). These latter data suggest that for a partial agonist, with a ceiling on the magnitude of agonist effects, dosage level may function as a surrogate variable for controlling duration of action. It remains necessary, however, to replicate these data and to determine whether response to opioid challenge is similarly attenuated throughout the 2-day period.

Approximately 90,000 patients are enrolled in methadone maintenance treatment in the United States. The primary pharmacological mechanism of methadone's clinical efficacy is opioid tolerance and substitution. In human laboratory studies methadone produces dose-related attenuation of opioid self-administration (27). Extensive clinical research documents both the efficacy of methadone treatment and the fact that nonpharmacological factors have a substantial influence on that efficacy (3). Controlled double-blind studies have documented the dose-related efficacy of outpatient methadone treatment (63). At the same time, surveys of clinical practice have documented that a large proportion of clinics use dosages that are below optimal (12). Optimal dosage appears to be in the range of 60-100 mg/day orally.

Despite its demonstrated efficacy, methadone treatment is not without its limitations. Opioid abusers and society at large both have some ambivalence toward treatment with a medication that sustains substantial physical dependence. Some opioid abusers decline methadone treatment, and some communities restrict or prohibit methadone treatment availability. The abstinence syndrome following methadone discontinuation is somewhat less intense but of a longer duration than that following discontinuation of pharmacologically comparable doses of morphine or heroin. However, the relatively high doses of methadone required for optimal efficacy may result in a greater degree of physical dependence during methadone treatment than abusers are able to sustain with illicit drug supplies. These differences in time course and degree of dependence result in patients' reporting that methadone withdrawal is more difficult than heroin withdrawal.

Because methadone's duration of withdrawal suppression is approximately 24 hr, it must be administered daily. To reduce the frequency of required clinic visits, to make treatment more convenient for patients, and to support their employment and social rehabilitation, methadone take-home doses may be provided. Because methadone is an opioid agonist, these take-home doses have abuse potential and may be used inappropriately or diverted to the illicit market. This, of course, produces adverse community consequences for this treatment modality.

Despite its imperfections, methadone treatment remains unsurpassed in efficacy. It has good patient acceptability and treatment retention; it dramatically reduces both illicit opioid use and the criminal activity often associated with acquiring illicit opioids; it substantially improves employment rates, and it substantially improves morbidity and mortality (including HIV infection rates). Pharmacotherapeutic alternatives to methadone are being developed not because methadone is ineffective but because it is not universally delivered, not universally effective, and not universally appropriate. The goal of developing pharmacotherapeutic alternatives is to permit individualized treatment selection in response to the heterogeneous characteristics and goals of patients and of communities.

Naltrexone is an instructive example of what can at times be a great disparity between pharmacological efficacy and clinical efficacy. As a long-acting, orally effective, opioid antagonist that blocks opioid receptors and prevents acute opioid effects as well as the development of physical dependence, naltrexone would appear to be a pharmacological wonder drug ideally suited to addiction treatment; however, it has had relatively little clinical impact. This is due to poor patient compliance with use of the medication (40). A depot dosage form that would be effective for 1 month is under development and may partially overcome this compliance problem. At present, effective clinical use of naltrexone appears to be limited to highly motivated patients, or to circumstances where medication use can be supervised (66). Recent clinical trials indicating efficacy of naltrexone in alcoholism treatment suggest involvement of opioid mechanisms in alcohol abuse.

Drug abuse pharmacotherapies are not in themselves sufficient to treat the wide range of personal and psychosocial problems that so frequently accompany opioid dependence. While pharmacotherapy alone may be beneficial, concurrent behavioral or psychosocial treatment is generally indicated. Different pharmacotherapies may have differing compatibilities with such concurrent psychosocial treatment. Thus, selection of pharmacotherapies should not be guided solely by the pharmacological aspects of treatment. For example, the frequency of required clinic attendance, which may co-vary implicitly with the selection of treatment medication, may itself be therapeutically important. Also, the potential benefit of using medication take-home privileges as contingent behavioral incentives to motivate therapeutic behavior change (59) will be absent with medications, such as LAAM, for which medication take-homes are prohibited. It is not yet known whether these differing compatibilities with concurrent behavioral therapies will have important clinical consequences. Behavioral treatments for drug abuse are discussed in the chapter by Stitzer and Higgins, in this volume.

One of the implications to be drawn from this review is that drug abuse and drug abuse treatment involve the complex interplay of both pharmacological and environmental/behavioral factors. Pharmacology, though critical, is not sufficient to determine whether a drug will be abused, or whether a pharmacotherapy will be clinically useful. We are struck by the overwhelming importance of practical, logistical, and behavioral factors as determinants of the abuse and/or the clinical efficacy of opioids. Such factors as dosage form and route of administration play major roles, as do such factors as fads, patient compliance, and community standards. Basic and systematic clinical pharmacological information is an essential element in predicting abuse liability, in preventing drug abuse, and in developing effective therapies.

Areas in which we expect to see important advances in the next decade include the following:

1. More refined data concerning the specific functions controlled by various opioid receptors. In particular, we would expect advances in the recognition of subtypes of the mu, kappa, and delta opioid receptors, and possibly identification of other opioid receptor classes, and characterization of their behavioral and physiological functions. We would expect to see data characterizing the clinical effects of delta receptor ligands. And we would expect to see advances in our understanding of the interrelationships and interactions among different classes of opioid receptors. We would expect to see greater bridging of the gap between preclinical and clinical research, so that knowledge about the neurobiology of opioid receptors and about the pharmacology of endogenous ligands can be more fully tested and utilized in humans. In particular, we would expect to see advances in our understanding of how various opioidergic systems interact to control and to modulate the development and expression of physical dependence and the opioid abstinence syndrome, as well as the basic behavioral reinforcement process that is at the heart of drug abuse. Such data may lead to identification of new therapeutic uses for opioids—especially for mixed agonist-antagonists and drug combinations.

2. More refined data concerning comparative efficacy of various pharmacotherapies. Now that opioid dependence pharmacotherapy has moved beyond the "one therapy fits all" stage, it is essential that we develop comparative efficacy data and examine efficacy in relation to the heterogeneous characteristics of patients. This should include comparative assessment of the acceptability to patients of various treatments. We will likely see increasing recognition of the role of practical, logistical, and behavioral factors in determining the clinical efficacy of pharmacotherapies. This may involve the development of new dosage forms and the development of pharmacotherapies specifically for their compatibility with concurrent behavioral and psychosocial treatments.

This work was supported, in part, by NIH grants K05-DA00050, R01-DA04089, R18-DA06120, R18-DA06165, and P50-DA05273 from the National Institute on Drug Abuse.

published 2000