Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Drug discrimination is a behavioral paradigm that tests and classifies test drugs according to their similarity or dissimilarity to previously experienced training drugs (Stolerman and Shine, 1985). Differences among compounds in drug discrimination performance may be related to differences in receptor activity (Young, 1991). Drug discrimination studies distinguished opioids from other drug classes and have distinguished among opioid agonists acting on different opioid receptor systems (Herling and Woods, 1981; Young et al., 1984). Drug discrimination studies with animals have shown that the opioid mixed agonist-antagonists pentazocine, nalbuphine, buprenorphine, and butorphanol act on µ- and -opioid receptors. Their characterization in drug discrimination testing may vary depending on the species used (e.g., Herling and Woods, 1981), the compounds trained (e.g., Picker and Dykstra, 1987), the discrimination trained (e.g., drug versus drug or drug versus vehicle; Colpaert and Janssen, 1986), and the training dose used (e.g., Young et al., 1992). For example, the mixed-action opioid cyclazocine substitutes completely, partially, and not at all for a low, moderate, and high training dose of morphine, respectively (Koek and Woods, 1989). In contrast, butorphanol, a mixed-action opioid with a greater intrinsic efficacy at the µ receptor relative to cyclazocine, substitutes for both moderate and high morphine training doses (Shannon and Holtzman, 1977; Picker et al., 1990).

The relationship between the discriminative stimulus and subjective effects of drugs is best explored in humans because only humans can provide discriminative subjective-state results. To explore this relationship, drug discrimination studies have examined the pharmacology of the mixed agonist-antagonist opioids and the procedural variables that impact drug discrimination results (see review by Preston and Bigelow, 1991).

Similar to animal models, the discrimination that is trained is inextricably linked to effects observed with mixed-action opioids in human drug discrimination. For example, in a two-choice procedure in which opioid-abusing volunteers were trained to discriminate between 3 mg of hydromorphone (a morphine-like opioid µ agonist) i.m. and saline, it was observed that pentazocine, butorphanol, nalbuphine, and buprenorphine all produced full generalization from hydromorphone (Preston et al., 1992). Contrasting results were observed in three-choice studies that tested the same mixed agonist-antagonists in opiate-abusing volunteers trained to discriminate among i.m. saline, hydromorphone (3 mg), and the mixed µ- and -opioid agonist pentazocine (45 mg/70 kg; Preston et al., 1989) or butorphanol (6 mg/70 kg; Preston and Bigelow, 1994). Overall, these data suggested butorphanol and nalbuphine to be low-efficacy agonists at both µ and  receptors; however, butorphanol revealed more  activity than did nalbuphine in its subjective effects.

The studies previously described suggest that drug discrimination results are dependent upon the specific discrimination procedures used. The present study was designed to assess the discriminative and subjective effects of several opioid mixed agonist-antagonists by manipulating the training dose, another critical factor known to influence discrimination test results. In studies with animals, the importance of the training dose has been demonstrated across numerous drug classes, including opiate (e.g., Picker et al., 1990), stimulant (Mumford and Holtzman, 1991), antidepressant (Zhang and Barrett, 1991), and benzodiazepine (Sannerud and Ator, 1995). For instance, in pigeons using a three-choice discrimination (two morphine doses and saline), the generalization of morphine to the low or high dose varied as an orderly function of the morphine test dose. However, nalbuphine produced generalization only to the low morphine dose, suggesting only partial agonist activity (Young et al., 1989). A similar pattern of effects was reported for nalbuphine in pigeons trained to discriminate between saline and either a low (0.056 mg/kg) or high (0.18 mg/kg) fentanyl dose (Picker et al., 1993). The finding that a test drug substitutes for the stimulus effects produced by a low training dose but not a high training dose may indicate potential agonist activity only with drugs with an equal or greater intrinsic efficacy. The present study was designed to examine the subjective and discriminative effects of mixed agonist-antagonist drugs including hydromorphone, pentazocine, butorphanol, nalbuphine, and buprenorphine in a three-choice hydromorphone dose discrimination (high dose-low dose-saline) and to assess partial agonist activity. Previous discrimination studies with humans suggest that hydromorphone and buprenorphine have high µ and low  activity, that butorphanol and pentazocine have low µ and high  activity, and that nalbuphine has low µ and low  activity (Dykstra et al., 1997). Thus, it was hypothesized that drugs with a greater µ activity would evoke generalization and similar subjective effects to both hydromorphone doses, whereas drugs with less µ activity would show more similarities to the low hydromorphone dose than to the higher dose.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects. Participants were 11 adult, nonphysically dependent opioid abusers who gave written consent and were paid for their participation. The six subjects who completed the study and whose data are presented here ranged in age from 30 to 40 years (mean 36); all were male and all were African-American. They reported illicit opioid use of 8 to 20 years duration (mean, 15 years) and were currently using them two to seven times a week (mean, four times a week). In addition to their opiate use, all of the subjects reported current or past use of marijuana, alcohol, cocaine, hypnotics, and benzodiazepines. All of the subjects reported current use of nicotine. Participants underwent routine medical screening that included reporting their medical history, a physical examination, an EKG, and chemistry, hematology, and urinalysis testing. Results were reviewed by medical staff not involved in the study as investigators. Participants were in good health and free of significant psychiatric disturbance other than their drug abuse.

Setting. Subjects participated in the study while residing on a residential behavioral pharmacology research unit. The residential research unit contained a nursing station, participant bedrooms, common living and dining areas, and experimental session rooms. Recreational reading and craft materials and exercise equipment were available at all times other than during experimental sessions.

Drugs. Drugs were administered i.m. under double-blind conditions. The training drugs were saline (4 ml) and hydromorphone (1 and 4 mg). Dose-response generalization testing was conducted with hydromorphone (0.375, 0.75, 1.5, and 3 mg), pentazocine (7.5, 15, 30, and 60 mg), butorphanol (0.75, 1.5, 3, and 6 mg), nalbuphine (3, 6, 12, and 24 mg), and buprenorphine (0.075, 0.15, 0.3, and 0.6 mg). The following commercially available preparations were used: hydromorphone hydrochloride (10 mg/ml; Knoll Pharmaceuticals, Wippany, NJ); pentazocine lactate (30 mg/ml; Winthrop Laboratories, Inc., New York, NY); butorphanol tartrate (2 mg/ml; Bristol Laboratories, Syracuse, NY); nalbuphine hydrochloride (10 mg/ml; DuPont Pharmaceuticals, Wilmington, DE); and buprenorphine hydrochloride (0.3 mg/ml; Reckitt and Colman, Hull, England). Hydromorphone hydrochloride, butorphanol tartrate, and nalbuphine hydrochloride doses were calculated on the basis of the salts, and pentazocine lactate and buprenorphine hydrochloride doses were calculated as the drug base. Drugs were diluted to the appropriate volumes and concentrations with bacteriostatic saline. Doses were injected in a constant volume of 4 ml in two divided doses (2 ml each in the right and left deltoid muscles). Training drugs were identified to the participants only by arbitrary letter codes (such as H, I, and J). Different letters were used for each subject. The letter codes associated with each of the training drugs were determined randomly and varied across subjects but remained unchanged for each subject throughout his participation. For generalization testing, a range of doses for each drug was selected so that the recommended analgesic dose fell midrange.

General Methods. After admission, participants were oriented to the unit, their consent to participate in the study was obtained, and they were oriented to the session room and introduced to the staff. Participants were informed that the purpose of the study was to evaluate the ability of individuals to discriminate one drug from another and to evaluate the subjective, behavioral, and physiological effects of those drugs. It was further explained that during the study they would receive various psychoactive drugs that might have sedative properties, stimulant properties, opioid properties, or opioid-blocking properties. Participants were instructed to attend carefully to the drug effects and to try to discriminate precisely among them. They were informed that correct identification of the administered drug by letter code would result in a monetary bonus. To reduce the possibility that subjects would receive instructions or explanations that might confound the results, staff were explicitly instructed not to discuss the experiment with the subjects except to provide an objective description of the procedures. Before any drug sessions were conducted, participants were given a practice session to familiarize them with the procedures. The study then proceeded in three phases, with one session per day. Except when the letter code information was being given to the subject (either before or after each session), sessions were similar for all three phases.

Discrimination Training (Sessions 1-6). Each participant received, in a random block order, two sessions of exposure to each of the three training conditions (saline and two active drugs). During these training exposures, each drug was identified to the subject by a letter code before drug administration.

Test of Acquisition (Sessions 7-12). Acquisition of the discrimination was tested by exposing each participant to each of the three training doses twice in randomized block order. For these and all remaining exposures to the three training conditions, the participant received feedback about the code of the administered training drug at the end of the session. The criterion for acquisition of the discrimination was a correct response for five of the six test-of-acquisition sessions.

Test Sessions (Beginning with Session 13). During this phase, dose-response curves for hydromorphone, pentazocine, butorphanol, nalbuphine, and buprenorphine were determined in that order. Doses in each dose-response curve were tested in a random order. After each test session, the participant did not receive feedback about the correct drug identification but was informed that it had been a test session and that the drug code could not be revealed. Test sessions were interspersed with test-of-acquisition sessions. Of the 30 sessions in this phase, approximately 30% were tests of acquisition and 70% were test sessions.

Apparatus. Microcomputers presented all questionnaires and performance tasks in a prearranged and timed sequence and printed and stored the data for each session. Participants indicated responses on a numeric keypad.

Experimental Session. Daily sessions were conducted with methods similar to those reported previously (Preston et al., 1989). At the start of each experimental session, respiration rate, pulse, temperature, blood pressure, and pupil diameter were recorded, and the subject completed baseline self-report questionnaires and psychomotor performance tests in the experimental room. The scheduled drug or saline was then injected. During the discrimination training phase, the participant was told the correct identifying letter code at the time of injection. The subject remained under observation for 20 min and then was returned to the experimental room to complete the postdrug discrimination, subjective effect, and performance testing. Subjective effect assessments and psychomotor performance tests were conducted 20, 40, 60, and 80 min after drug administration. Drug discrimination performance was assessed twice, 60 and 80 min after drug administration. These various assessment procedures are described in detail below. At the end of the session, the staff again recorded pupil diameter and other physiological measures. During the test of acquisition phase, a sealed envelope was opened and the staff informed the volunteer of the letter code identity of the administered drug. During the test session phase, the card stated that the session had been a test session and that the identity of the drug could not be revealed. The participant was then told his earnings for the session.

Discrimination Procedures. Drug discrimination data were collected in three ways. In each of these three procedures only correct responses were converted to monetary reinforcement for the subject. In Discrete Choice, the volunteer named by letter code (e.g., H, I, or J) the drug he thought he received. In Point Distribution, the subject distributed 50 points among one or more of the three training drug choices depending on how certain he was of the identity of the administered drug. This required typing in the numbers. For the Operant Responding measure, the subject earned points by responding on computer keys designated with drug letter codes on a fixed-interval 1-s schedule for 3 min. Points (displayed on the computer screen) could be earned (at a maximum rate of 1 per s) for each of the three training drugs by pressing the key corresponding to that drug.

Subject-Rated Measures. Four questionnaires were completed: 1) visual analog scales (VAS); 2) a pharmacological class questionnaire; 3) an adjective rating scale; and 4) a short form of the Addiction Research Center Inventory (ARCI). On the VAS, the participant rated the degree of "Drug Effects," "High," "Liking," "Good Effects," "Bad Effects," and "Sick" produced by the drug and rated the similarity of the drug to the training drugs on questions asking "How much is the drug like (each of the training drugs by letter code)?" by placing an arrow along a 100-point line anchored at either end with "none" and "extremely." On the pharmacological class questionnaire, the subject categorized the drug effect as being most similar to one of 10 classes of psychoactive drugs. The questionnaire provided descriptive labels and examples for each of the following classes: placebos, opiates, phenothiazines, barbiturates and sleeping medications, antidepressants, opiate antagonists, hallucinogens, benzodiazepines, stimulants, and phencyclidine. The adjective rating list consisted of 32 items that the subject rated on a 5-point scale from 0 ("no effect") to 4 ("maximum effect"). The items in the adjective list were scored on three subscales (Preston et al., 1989): a 13-item opioid agonist scale (itchy skin, turning of stomach, nodding, relaxed, pleasant sick, talkative, heavy or sluggish feeling, dry mouth, active, carefree, drunken, good mood, and energetic), a 10-item opioid antagonist scale (sleepy, flushing, sweating, watery eyes, runny nose, chills, shaky, goose-flesh, restless, and agitated), and a 9-item mixed agonist-antagonist opioid scale (tired, tingling, coasting or spaced out, headache, floating, confused, light-headed, depressed, and numb). The ratings of individual items were summed up for a total score for each scale. The short form of the ARCI consisted of 49-items (true/false) and contained five subscales: morphine-benzedrine group (MBG) (euphoria); pentobarbital-chloropromazine-alcohol group (PCAG) (sedation); lysergic acid diethylamide (LSD) group (somatic and dysphoric changes); benzedrine group (BG); and amphetamine scales (Martin et al., 1971).

Physiological and Psychomotor Performance Measures. Physiological measures included respiration rate, pulse, blood pressure, oral temperature, and pupil diameter. Vital signs were collected manually. Pupil diameter was measured manually from photographs taken in constant ambient lighting with a Polaroid camera with 3× magnification. Psychomotor/cognitive performance was tested with a computerized version of the digit substitution test (DSST) that was developed in our laboratory (McLeod et al., 1982).

Data Analysis. Only data from the six subjects who completed the study were analyzed. Results of each session were summarized to yield a single value for each of the following measures: within-session means for VAS; percentage of saline-, hydromorphone 1 mg-, and hydromorphone 4 mg-appropriate responses for discrimination measures; and mean change from predrug scores for the ARCI scales, adjective rating scales, DSST scores, and physiological measures.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Of the 11 participants enrolled, six subjects completed the study; their data are presented below. Data from the five dropouts were excluded from the analyses. One was discontinued because of a medical problem unrelated to the study drugs. Another was discharged for failure to follow instructions regarding study tasks. A third elected to leave early. Two subjects failed to acquire the saline-hydromorphone 1 mg-hydromorphone 4 mg discrimination and were dropped after three training attempts.

Six subjects reliably discriminated correctly among the training drugs with a greater than 95% accuracy for saline and hydromorphone 1 mg and a greater than 83% accuracy for hydromorphone 4 mg (see Table 1). Table 1 also shows the results for the subjective, physiological, and psychomotor performance effects of the training drugs that achieved statistical significance (sessions 1-12). Significant treatment effects were produced on the subjective measures of Drug Effect, High, Liking, and Good Effects VAS. Post hoc comparisons with Tukey's test showed that Drug Effect, High, Liking, and Good Effects scales were significantly increased by 4 mg of hydromorphone compared with both 1 mg of hydromorphone and saline. No differences on any of the VAS were seen when 1 mg of hydromorphone was compared with saline. Although subjects reliably discriminated correctly among the training doses, no significant differences were observed on the three VAS that asked the participant to rate "How much is the drug like (each of the training drugs)?" None of the training drugs were rated as being significantly more similar to itself than to saline or to the other hydromorphone dose, nor were the two hydromorphone doses rated more similar to each other than to saline. The results of the agonist adjective rating scales showed significant main effects of the drug; post hoc Tukey's analyses showed that 4 mg of hydromorphone produced significant increases in agonist scale scores compared with 1 mg of hydromorphone or saline. Items of the agonist scale that were increased with 4 mg of hydromorphone relative to saline or to 1 mg of hydromorphone were itchy skin and relaxed. Administration of 4 mg of hydromorphone had increased ratings on the agonist scale items of talkative, dry mouth, carefree, and good mood relative to saline only. There were no significant differences between 1 mg of hydromorphone and saline on any of the individual items of the adjective scales. None of the scales on the ARCI showed significant main effects of treatment. Administration of 4 mg of hydromorphone produced significant decreases in pupil diameter compared with saline or 1 mg of hydromorphone. No statistically significant main effects of drugs were seen on other physiological measures or DSST performance.

On the pharmacological class questionnaire, hydromorphone 4 mg was identified as an opiate on 85% of the occasions, as a benzodiazepine on 11.6% of the occasions, as a stimulant on 2% of the occasions, and as a blank on 2% of the occasions. Saline was identified as a placebo on 68% of the occasions, as an opiate antagonist on 30% of the occasions, and as a benzodiazepine on 2% of the occasions. Hydromorphone 1 mg was not identified as consistently as any one type of drug compared with the other two training drugs. It was identified as a benzodiazepine on 36.7% of the occasions, as a stimulant on 26% of the occasions, as an opiate on 12.5% of the occasions, as a blank/placebo on 9% of the occasions, as an antidepressant on 5% of the occasions, as a barbiturate on 4.2% of the occasions, as an opiate antagonist on 4.2% of the occasions, and as a phenothiazine on 2.5% of the occasions.

Generalization Testing. Tables 2 and 3 show the results of the generalization testing. Selected variables are shown in Figs. 1 to 5. Table 2 shows P values for the measures for which P values  .10 were determined in the ANOVA. Arrows indicate the direction of statistically significant (P  .05) drug effects. Arrows in parentheses indicate the direction of drug effects for P values that are > 0.05 but  0.10. Because all three discrimination measures produced similar results, only the results from the operant discrimination measure are presented.

View larger version (42K): [in this window] [in a new window]   Fig. 1.   Dose-response curves of selected measures produced by test doses of hydromorphone and saline in six heroin addicts trained to discriminate among hydromorphone 1 mg, hydromorphone 4 mg, and saline. Each point is the mean ± S.E.M. based on one administration of each dose; the S.E.M. is not shown when it is less than the radius of the symbol. *, significantly different from saline (P < .05).

View larger version (35K): [in this window] [in a new window]   Fig. 2.   Dose-response curves of selected measures produced by test doses of pentazocine and saline in six heroin addicts trained to discriminate among hydromorphone 1 mg, hydromorphone 4 mg, and saline. Each point is the mean ± S.E.M. based on one administration of each dose; the S.E.M. is not shown when it is less than the radius of the symbol. *, significantly different from saline (P < .05).

View larger version (38K): [in this window] [in a new window]   Fig. 3.   Dose-response curves of selected measures produced by test doses of butorphanol and saline in six heroin addicts trained to discriminate among hydromorphone 1 mg, hydromorphone 4 mg, and saline. Each point is the mean ± S.E.M. based on one administration of each dose; the S.E.M. is not shown when it is less than the radius of the symbol. *, significantly different from saline (P < .05).

View larger version (33K): [in this window] [in a new window]   Fig. 4.   Dose-response curves of selected measures produced by test doses of nalbuphine and saline in six heroin addicts trained to discriminate among hydromorphone 1 mg, hydromorphone 4 mg, and saline. Each point is the mean ± S.E.M. based on one administration of each dose; the S.E.M. is not shown when it is less than the radius of the symbol. *, significantly different from saline (P < .05).

View larger version (35K): [in this window] [in a new window]   Fig. 5.   Dose-response curves of selected measures produced by test doses of buprenorphine and saline in six heroin addicts trained to discriminate among hydromorphone 1 mg, hydromorphone 4 mg, and saline. Each point is the mean ± S.E.M. based on one administration of each dose; the S.E.M. is not shown when it is less than the radius of the symbol. *, significantly different from saline (P < .05).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The purpose of this study was to examine the discriminative and subjective effects of mixed-action opioids in a discrimination that varied the amount of µ-receptor agonist activity represented by the discrimination options. The three trained discrimination options varied only in dose, such as to represent zero, low, and high µ-agonist activity. Differential generalization to different training doses has been used in animal drug discrimination research as an indicator of partial agonist activity. The results showed that human subjects could be successfully trained to discriminate among saline and two hydromorphone doses (1 mg versus 4 mg i.m.) in a paradigm similar to that used with animals (e.g., Young et al., 1989). Similar to a previous study with nondependent subjects (Preston et al., 1989), the reinforced behavioral discrimination measure appeared more sensitive than the nonreinforced self-report VAS measures. This discrepancy may be a result of volunteers not using the full scale of the visual analog line or it may have resulted because the two doses compared have overlapping stimulus effects. Furthermore, generalization testing with novel doses of hydromorphone produced dose-related increases in dose-appropriate responding. As illustrated in Table 1, during training and test-of-acquisition sessions, hydromorphone 4 mg produced a greater magnitude of effects than those found with either hydromorphone 1 mg or saline. On the pharmacological class questionnaire, hydromorphone 4 mg was predominantly identified as an opiate, whereas hydromorphone 1 mg resulted in a greater variability in labeling, as has been previously observed (e.g., Preston et al., 1989). One interpretation is that this labeling variability may reflect that the low dose is a relatively nonspecific stimulus and may not indicate partial agonist activity so much as activity dissimilar to hydromorphone. However, the fact that the low dose was correctly discriminated suggests that it was a distinguishable stimulus. Thus, the paradigm was sensitive to dose, and the two doses of hydromorphone were differentiated by both behavioral discrimination measures and by subjective responses.

Generalization testing with novel doses of hydromorphone produced dose-related increases in hydromorphone 4 mg-appropriate responding and an inverted U shape function for hydromorphone 1 mg-appropriate responding. Hydromorphone was the only drug that produced 100% discrimination as hydromorphone 4 mg. However, all novel drugs were partially discriminated as hydromorphone 1 mg at one or more doses. These results are in agreement with the predictions of the receptor theory (e.g., Holtzman, 1983; Woods et al., 1988), which predicts that generalization from a higher-dose stimulus requires a compound with a greater efficacy than does generalization from a lower-dose stimulus. Thus, the observation that all drugs evoked generalization with the low but not the high hydromorphone training dose may indicate that these drugs are less efficacious µ agonists than hydromorphone. Thus, the three-choice discrimination task appears to be a sensitive bioassay to assess partial µ-agonist activity.

The results of the present and previous investigations in the series demonstrate differences between hydromorphone and the mixed agonist-antagonists that are in agreement with their putative receptor activities. As expected, neither butorphanol nor pentazocine, both of which have low µ- and higher -agonist activity, generalized to the high hydromorphone dose, and butorphanol produced subjective effects similar to those of  antagonists. Nalbuphine, a drug with low - and µ-agonist activity, generalized completely to the low hydromorphone dose. However, based on the current conception of relative µ and  activities of these opioids in humans, it was expected that buprenorphine would produce generalization to both doses of hydromorphone. This was not the case. Buprenorphine did produce subjective effects that were similar to µ-agonist effects.

In the previous studies, hydromorphone, a full µ agonist, has been used as a training drug. Hydromorphone has been discriminated from saline (Preston et al., 1992), pentazocine (Bickel et al., 1989; Preston et al., 1989), butorphanol (Preston and Bigelow, 1994), and naloxone (Preston et al., 1987a; Preston and Bigelow, 1990). This is the first time, to our knowledge, that different opioid training doses have been directly compared in human drug discrimination. Results from previous studies showed that 3 mg of hydromorphone produced opioid-agonist effects, including a profile of subjective effects, similar to that produced for other classical µ agonists and showed no evidence of antagonist activity. In the present study, we observed similar effects, with the most prominent profile of µ-opioid subjective effects being produced by the 4-mg dose of hydromorphone.

Studies of pentazocine have had mixed results in animal models (e.g., Picker and Dykstra, 1989) and in humans (e.g., Preston et al., 1987b). The present results are most similar to another three-choice (hydromorphone-butorphanol-saline) discrimination in which pentazocine was partially discriminated as both active drugs (Preston and Bigelow, 1994). Those results, suggesting that pentazocine has at least partial µ-agonist activity, are in accord with pentazocine having morphine-like subjective effects at low doses and -like effects at higher doses (Jasinski et al., 1970; Preston et al., 1987b).

Similar to previous studies, butorphanol was differentiated from both doses of hydromorphone. Butorphanol was not consistently discriminated as either 1 mg or 4 mg of hydromorphone, although partially as both. Previously, in a three-choice (hydromorphone-saline-pentazocine) discrimination, butorphanol was discriminated as pentazocine-like (Preston et al., 1989). In a two-choice (hydromorphone-saline) paradigm, butorphanol was discriminated as hydromorphone-like (Preston et al., 1992). In physically dependent subjects trained to discriminate among naloxone-saline-hydromorphone, butorphanol was discriminated as naloxone-like. In the present study, the subjective effects of butorphanol were different from those of hydromorphone. Butorphanol failed to increase the subjective ratings of good effects, produced a dose-dependent increase in bad effects, and was frequently identified as a sedative (i.e., benzodiazepine or barbiturate). Differences between butorphanol and hydromorphone were also observed on ARCI scales, with butorphanol producing increased dysphoria (LSD scale). Butorphanol also produced a profile of subjective effects different from that of the other drugs tested, with increases of Bad Effects antagonist adjective ratings and dysphoria becoming more evident with the higher doses. This same profile of effects has been described for the other  agonists examined in humans (Kumor et al., 1986; Pfeiffer et al., 1986). In animal studies, the  activity of butorphanol has been demonstrated with increases in urine output and with partial antagonism of urine output produced by the  agonist bremazocine (Leander, 1983). Less than maximal -opioid effects have been obtained in drug discrimination procedures with animals (e.g., Smith and Picker, 1995). Overall, it appears that butorphanol has both nonmaximal µ-agonist and nonmaximal -agonist activity. Thus, it appears to be a partial agonist at both receptor systems.

Nalbuphine, like butorphanol, has been consistently differentiated from hydromorphone under three-choice training conditions. For example, nalbuphine was discriminated as butorphanol under hydromorphone-butorphanol-saline conditions (Preston and Bigelow, 1994). Nalbuphine was also discriminated as antagonist-like under naloxone-hydromorphone-saline conditions in dependent subjects (Preston and Bigelow, 1990). In the present study, nalbuphine was discriminated as 1 mg of hydromorphone, suggesting that it is a low-efficacy µ compound. The present results with nalbuphine are similar to those observed in animals when training dose has been examined; nalbuphine produces low levels of substitution at high doses of µ agonists and substitutes completely at low doses (Young et al., 1989; Picker et al., 1993). The subjective effects of nalbuphine were similar to those of hydromorphone in that both drugs produced increases in the ratings of Good Effects and did not increase ratings of Bad Effects. Additionally, both drugs were generally identified as an opiate. Nalbuphine appears to be a low-efficacy partial agonist at the µ receptor.

Buprenorphine has been discriminated as hydromorphone-like under two-choice (hydromorphone-saline) and three-choice (hydromorphone-butorphanol-saline) discriminations (Preston et al., 1992; Preston and Bigelow, 1994). The present results indicate that buprenorphine was not discriminated clearly as either 1 mg or 4 mg of hydromorphone. However, the subjective effects of buprenorphine were similar to those of hydromorphone in that both drugs increased the ratings of Good Effects on agonist adjective rating scales and failed to elevate Bad Effects. On the ARCI, hydromorphone and buprenorphine tended to increase euphoria (morphine-benzedrine group scale). Additionally, both drugs were generally identified as an opiate. The present data concur with previous reports of buprenorphine being identified as opioid-like, producing opioid like effects (e.g., Jasinski et al., 1978), and having µ-agonist-like discriminative stimulus effects without -agonist effects in humans and animals (e.g., Preston and Bigelow, 1994). Buprenorphine appears to be a partial µ agonist with an extremely shallow dose-effect function. Because the doses tested in the present study did not approach the ceiling of the activity of buprenorphine (Walsh et al., 1994), whether higher buprenorphine doses would have produced greater hydromorphone 4 mg responding is unclear.

Several limitations of the present study should be acknowledged. The study does not prove that the observed effects were via µ- and/or -opioid receptor mechanisms. Some effects may have resulted from activity at different opioid or nonopioid systems (Gauvin and Young, 1989). More conclusive evidence regarding the activity of the mixed agonist-antagonists could have been revealed by comparing their effects to a number of other full µ- and nonopioid agonists. To determine whether the stimulus effects of both training doses were mediated through specific receptor systems, interaction studies with opioid antagonists would be needed. Furthermore, the lack of robust generalization of pentazocine, butorphanol, nalbuphine, and buprenorphine to the hydromorphone 4 mg stimulus may be due to either the order of the drug administration or the dose ranges used. Although it is of scientific interest to study different orders of drug presentation and higher doses, it was not deemed clinically practical.

In conclusion, these results demonstrate that, in humans, the discriminative stimulus and subjective effects of opioid drugs, as assessed by the behavioral tasks used, are not static properties of the particular drug doses examined but are influenced by the training conditions. Not only is the type of discrimination (two- or three-choice) important, but the type of drug(s) and the dose of the drug(s) are critical. The data concur with predictions from receptor theory (e.g., Holtzman, 1983; Woods et al., 1988) that generalization from a higher-dose stimulus requires a compound with a greater efficacy than compounds needed to generalize from a lower-dose stimulus. Thus, the observation that all drugs evoked generalization with the low but not the high hydromorphone training dose may indicate that the drugs are less efficacious than the µ agonist hydromorphone. Thus, the three-choice (hydromorphone 1 mg-hydromorphone 4 mg-saline) discrimination task appears to be a sensitive bioassay for examining partial agonist relationships.