Randomized placebo-controlled trial of baclofen for cocaine dependence: preliminary effects for individuals with chronic patterns of cocaine use

Randomized Placebo-Controlled Trial of
Baclofen for Cocaine Dependence: Preliminary Effects
for Individuals With Chronic Patterns of Cocaine Use
Steven Shoptaw, Ph.D.; Xiaowei Yang, Ph.D.; Erin J. Rotheram-Fuller, M.A.;
Ya-Ching M. Hsieh, M.S.; Prudencia C. Kintaudi, M.D.;
V. C. Charuvastra, M.D.; and Walter Ling, M.D.
Received Jan. 2, 2003; accepted May 7, 2003. From the UCLA Integrated Substance Abuse Program, Los Angeles, Calif. (Drs. Shoptaw,Yang, and Ling and Mss. Rotheram-Fuller and Hsieh), and the Friends Background: This screening trial evaluated
Research Institute, Inc., Los Angeles, Calif. (Drs. Shoptaw, Kintaudi, whether the GABA agonist baclofen demonstrated and Charuvastra and Ms. Rotheram-Fuller). sufficient clinical efficacy to recommend an The authors gratefully acknowledge support from the National adequately powered trial of the medication Institute on Drug Abuse (Bethesda, Md.) for assistance in the conduct as a pharmacotherapy for cocaine dependence.
of this study and the analysis of these data from grants: 1P50DA12755 Method: Participants with cocaine dependence
and N43DA2-5513. We also acknowledge C. James Klett, Ph.D., for hisguidance in the design and execution of clinical trials of medications for verified by the Structured Clinical Interview for DSM-IV were randomly assigned to baclofen Corresponding author and reprints: Steven Shoptaw, Ph.D., Friends (N = 35; 20 mg t.i.d.) or placebo conditions Research Institute, Inc., UCLA Integrated Substance Abuse Programs, (N = 35; identical in appearance and dosage rate) 11075 Santa Monica Blvd., Suite 200, Los Angeles, CA 90025 using a 2-group, experimental, 16-week double- (e-mail: Shoptaw@friendsresearch.org). blind design featuring thrice-weekly cognitive-behavioral drug counseling groups. Outcomeswere retention, cocaine use, cocaine craving, espite intensive investment of resources and the Dcompletion of many clinical trials, no medication
Results: A generalized estimating equation
(GEE) model showed that participants assigned to has demonstrated clear evidence of efficacy for the treat- receive baclofen demonstrated statistically signifi- ment of cocaine dependence.1 The search for effective cant reductions in cocaine use over those assigned pharmacotherapies has focused largely on dopaminergic to receive placebo as indicated by urine drug agents, due to the central role of that neurotransmitter sys- screening results (χ2 = 5.34, df = 1, p = .021).
tem in the reinforcing effects of cocaine and other stimu- Confirming the GEE model, longitudinal analysesshowed that participants assigned to receive baclo- lants.2,3 Having come up largely empty-handed from this fen demonstrated significant and stepwise increases search, investigators are increasingly looking to other in the probability of providing benzoylecgonine- brain mechanisms that modulate the behavioral effects of free urine samples throughout the trial as the num- ber of benzoylecgonine-positive samples increased There is preclinical evidence that medications that act during baseline (χ2 = 10.63, df = 1, p = .001). Par-ticipants assigned to placebo demonstrated no such on the inhibitory neurotransmitter system, γ-aminobutyric association. Univariate analyses of aggregates of acid (GABA), may represent viable candidates as treat- urine drug screening showed generally favorable ments for cocaine dependence.5 Both inhibitory and exci- outcomes for baclofen, but not at statistically sig- tatory amino acid systems may be involved in the rein- nificant levels. There was no statistical significance forcing effect of cocaine.6–13 It is understood that the bulk observed for retention, cocaine craving, or inci-dence of reported adverse events by treatment of GABA activities in the central nervous system are in- hibitory in nature, exerting a generally dampening effect Conclusions: Project findings demonstrated
on the overall activities of the reward system.14,15 Specifi- initial clinical efficacy of baclofen over placebo in cally, GABA has been shown to exert an inhibiting effect reducing cocaine use when delivered concurrent on the tonic activity of dopamine neurons in the ventral with thrice-weekly drug abuse counseling sessions.
The effects of baclofen were particularly apparent tegmental area16 and in the nucleus accumbens,17 and this for those participants with chronic levels of cocaine may attenuate the reinforcing effects of cocaine through use at baseline and provide support for a full-scale efficacy trial for baclofen, especially among this Vigabatrin, a GABA transaminase inhibitor that in- creases GABA neurotransmission, reduced cocaine- (J Clin Psychiatry 2003;64:1440–1448) induced dopamine release by 25% or more in laboratoryanimals.19 Although vigabatrin has exceptional preclinical COPYRIGHT 2003 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
support for reducing consumption of alcohol and co- current (SCID-verified); (2) current psychiatric disorder caine20 and acceptable cardiovascular and hepatotoxicity requiring treatment; (3) active medical conditions that in- profiles in the presence of cocaine,21 the medication terfere with participation (e.g., uncontrolled diabetes); (4) causes unacceptable visual field defects in approximately history of seizures; (5) unstable behavior during screening one third of those exposed to the medication for long period (e.g., transportation problems, erratic attendance); and (6) asthma (current or lifetime ) requiring treatment.
Another GABAergic medication, the GABA agonist, Baseline demographic variables indicated that the bac- baclofen, also modulates cocaine self-administration in lofen and placebo groups were fundamentally similar laboratory animals.23 Baclofen consistently reduces co- (Table 1). Compared to those assigned to placebo, partici- caine self-administration at low doses using a fixed ratio- pants assigned to receive baclofen were significantly 1 schedule24 and dramatically reduces response behaviors more likely to report possession of a valid drivers license, at all cocaine dose levels when tested using a progressive fewer days of work, and fewer days of cannabis use in the ratio schedule.25 Effects for baclofen on cocaine self- 30 days prior to baseline. Baclofen participants also administration in laboratory animals show sensitivity scored significantly higher at baseline on subscales of the to levels of cocaine dose and response requirements.26 Addiction Severity Index,28 indicating greater disturbance Baclofen may also show differential effects in humans in employment compared to placebo-treated participants.
that depend upon such factors as pattern and level of co-caine exposure, although this is as yet untested.
Our initial experience with baclofen as a treatment This 2-group, randomized, placebo-controlled, double- agent involved a 10-patient, open-label study, which found blind trial of baclofen (20 mg t.i.d.) was conducted from that the medication was safe when used in conjunction October 1997 to May 1999 as part of a National Institute with cognitive-behavioral counseling 3 times a week.27 on Drug Abuse (NIDA)–funded project to screen theoreti- Data from preclinical and human experiences using bac- cally promising medications for cocaine dependence. A lofen led our group to mount a randomized, placebo- placebo identical in appearance to the baclofen pill was controlled screening trial of baclofen (20 mg t.i.d.) admin- delivered at the same rate as the active condition. The istered in the context of thrice weekly cognitive-behavioral study design was conducted at the lower ranges of statisti- drug counseling groups for reducing cocaine use. This cal power for detecting medication effects and used an screening trial was undertaken to determine whether bac- alpha level of p < .05, power equal to 0.80, and a moder- lofen produces effects significant enough to warrant a ate medication effect size between conditions using indi- full-scale efficacy trial. We predicted that in measures of cocaine use, treatment retention, and cocaine craving, par-ticipants assigned to receive baclofen would significantly outperform participants assigned to receive placebo.
Participant characteristics at baseline were assessed using the Addiction Severity Index28 and the SCID.29 Four domains of treatment outcome were measured: retention,urine drug testing, cocaine craving, and adverse events.
Compliance with medication taking was monitored using A total of 131 treatment-seeking, cocaine-dependent volunteers began screening procedures for this study. Fol- Addiction Severity Index. The Addiction Severity In-
lowing a 2-week, nonmedication screening period to col- dex, a standardized 40-minute clinical interview used in lect baseline measures, 70 participants were randomly as- addiction research to quantify problem areas experienced signed to receive baclofen (N = 35) or placebo (N = 35).
by substance abusers, was collected at baseline. The mea- Analysis comparing whether the 61 participants who ter- sure has excellent interrater and test-retest reliability and minated prior to randomization differed systematically discriminant and concurrent validity.28 Composite scales from those randomized to the trial showed that there were measure medical status, employment, drug use, alcohol no statistically significant differences in any of the mea- use, legal status, family/social status, and psychiatric sured demographic or drug use characteristics.
Participants met all of the following inclusion criteria: Structured Clinical Interview for DSM-IV (SCID).
(1) cocaine dependence, current, diagnosed using the The SCID29 was administered to verify cocaine de- Structured Clinical Interview for DSM-IV (SCID); (2) pendence, to exclude participants with concomitant alco- aged 18 to 65 years; (3) English literacy sufficient to com- hol or other illicit drug dependence, and to assess comor- plete measures and scales; (4) understand risks/benefits to bid psychiatric conditions that would preclude study participation; and (5) provide voluntary informed consent.
Participants also met none of the following exclusion Retention. Retention was the number of days that par-
criteria: (1) dependence on alcohol or other substances, ticipants received medication from the first observed dose COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
tion to benzoylecgonine 150,000 ng/mL (urinary metabo- Table 1. Demographic, Drug Use, and Psychiatric
Characteristics by Assignment to Treatment Condition
lite of cocaine) were available. The qualitative level for determining samples positive for cocaine metabolite was benzoylecgonine 300 ng/mL . New use criteria30 were ap- plied to quantitative values to define qualitative results (0 = no new use; 1 = new use) such that an index sample was interpreted to indicate no new use of cocaine if the quantitative benzoylecgonine value of the index sample was at least 50% less than the immediately prior sample.
If the previous sample was missing and the quantitative value exceeded benzoylecgonine 300 ng/mL, the sample was considered a new use. Qualitative results were com- piled using 4 typical aggregates: joint probability index,31 percentage of urine samples negative for benzoylecgo- nine, longest period of continuous abstinence verified by urine drug testing (in days), and percentage of partici- pants that provided 3 consecutive weeks of urine samples Cocaine craving. Cocaine craving ratings were re-
corded once per week using a visual analogue scale (0 = “not at all” to 100 = “strongest ever”) that asked par- ticipants to indicate on a 100-mm line their “most intense craving for cocaine that occurred at any time during the Adverse events. Adverse events were reported by par-
ticipants or observed by medical staff members. Those adverse events rated in severity as moderate or higher that occurred during the trial were tallied, and these adverse events were collapsed into categories. Moderate adverse events were reported as the incidence of each event cat- egory per condition. Serious adverse events were de- Compliance. Pill counts served as the primary marker
of medication adherence and were calculated as the total number of pills dispensed minus the number of pills re- turned divided by the total number of pills dispensed.
Only pills that were returned were included in this pill count. For those who terminated their participation early by not returning to the clinic, only those pills that were verified from returned medication bottles were included in the pill counts. No inferences were made regarding pills taken from bottles that were not returned to clinic.
Self-report of pill taking was used to explain discrepan- cies that occasionally occurred between actual pill counts Abbreviations: ASI = Addiction Severity Index, SCID = Structured and self-report of pill taking. To test whether fatigue, acommon side effect of baclofen, might aid baclofen-treated participants in decoding their condition assign- of study medication to the last clinic visit. Early termina- ment, participants were polled during the second week on tion occurred when participants failed to return to clinic their belief of whether they received baclofen or placebo Urine drug testing. Urine samples for drug testing
were collected on Mondays, Wednesdays, and Fridays during the baseline and treatment periods and analyzed All activities involved in this study were overseen by using radioimmunoassay methods by Northwest Toxicol- the Friends Research Institute West Coast Institutional ogy (Salt Lake City, Utah). Quantitative values with dilu- Review Board (Los Angeles, Calif.). Participants re- COPYRIGHT 2003 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
trial, blood panels (complete blood counts with differen- Figure 1. Study Design Flow Chart
tial), urinalyses, and ECG were conducted at baseline andat weeks 4, 8, 12, and 16. No incentives were provided for Cognitive-behavioral group counseling sessions were held at each clinic. The counseling program is manualdriven33 and has been shown to be an excellent platform when conducting cocaine pharmacotherapy trials.33–35 The cardinal analysis of project data involved evalua- tion of urine samples for benzoylecgonine to determine the effects of baclofen on cocaine use. Univariate tests of aggregate indices of urine sample results provided clini- cally useful information. Longitudinal models were ap- plied to more accurately evaluate the effects of baclofen and maximize the information contained in the data ma-trix, i.e., thrice weekly urine samples (6 samples duringbaseline, 48 samples during treatment). Missing samples were not imputed for univariate or multivariate analyses.
Aggregate measures of urine drug screens were analyzedseparately: joint probability index scores were comparedbetween conditions at the end of 8 and 16 weeks of treat-ment using differences between uncorrelated proportions sponded to newspaper or radio ads about this treatment research project through a central recruitment number, The primary method for determining intervention which scheduled the individual for an intake interview.
effects using longitudinal analysis involved calculation At this interview, study procedures were explained fully.
of a generalized estimating equation model (GEE).37,38 Following provision of voluntary, signed informed con- Each of these analyses included the baseline variables that sent, participants began a 2-week, nonmedication base- were different between conditions as covariates. In order line period. This baseline period was to identify and ex- to describe medication effects, we calculated the prob- clude those whose cocaine dependence was inappropriate abilities for baclofen and placebo in instilling (i.e., tran- for outpatient intervention, to collect information that sitioning from a benzoylecgonine-positive urine sample documented inclusion and exclusion criteria, and to de- to benzoylecgonine-free urine sample) and maintaining liver interventions that taught early recovery skills. Those cocaine abstinence (i.e., consecutive benzoylecgonine- cleared for study participation were randomly assigned to free urine samples). We calculated the transitional prob- condition at the first clinic visit following the baseline pe- abilities that a specific urine sample was free of ben- riod and received their first week of study medication in zoylecgonine given the result of the previous urine take-home bottles (Figure 1). Ingestion of the first dose of drug screen (e.g., Markov order 1). This also allowed study medication was observed by the research nurse.
calculation of the probabilities of treatment failure Study medication was returned at the end of each research (i.e., transitioning from a benzoylecgonine-free sample week for pill counts, following which participants re- to a benzoylecgonine-positive sample or maintaining con- ceived a new bottle of medication. Clinic was held in the secutive benzoylecgonine-positive samples). This ap- evening hours, which facilitated observed dosing from proach accounts for the non-independence of events in a each new bottle of study medication dispensed. Partici- cocaine clinical trial, i.e., the correlation among different pants continued the treatment protocol until they either measurements of the same individual represents values completed 16 weeks of medication or terminated early.
assumed to influence the index of observation. Transition Participants were assessed by telephone or a clinic visit models retain a dynamic aspect because by describing the 30 days following their last clinic visit to verify their transition from one observation to the next, the history is taken into account.39 Retention was tested using a log- During the trial, participants attended clinic thrice rank analysis. Craving scores were analyzed using GEE.
weekly to complete measures, provide urine samples, and In order to determine whether missing data differen- attend counseling groups. To monitor safety during the tially affected our analysis, we evaluated the pattern of COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
Figure 2. Survival Analysis Depicting the Proportion of
Table 2. Aggregate Measures of Urine Drug Screening for
Participants Retained in Each Condition (Baclofen or
Cocaine Metabolite by Assignment to Treatment Condition
Placebo) Throughout Each of the 16 Weeks of Treatment
aCalculated as the proportion of patients that provided samples negative for cocaine metabolite divided by the number of patientsassigned to the condition (N = 35). Values represent joint probability indexes for the end of week 8, for the end of week 16, and for the average of each of the points between weeks 3–8.
bRepresents the average of the sums of samples negative for cocaine “missingness” (both for dropouts and for intermittent df = 34, p < .001). Figure 3 depicts the favorable perfor- missing values) observed for the baclofen and placebo mance of the baclofen condition over placebo during the conditions by applying a GEE model. We recoded data first 8 weeks of the trial using the percentage of benzoy- into a dichotomous variable of either observed or missing Results of the longitudinal analysis, both GEE and Markov order 1 transition models, evaluating medication effects showed a strongly significant effect for baclofen.
In these analyses, the baseline Addiction Severity Index employment composite and reported cannabis use in the There were similar patterns of missing data observed 30 days prior to baseline were included as covariates.
for both treatment conditions (χ2 = 0.57, df = 1, p = .45).
These baseline variables were not statistically significant We concluded that there were no systematic biases for our predictors in the models and were dropped from the findings by differential rates or patterns of missing data analyses presented. The GEE solution showed a signifi- cant effect of baclofen over placebo (χ2 = 5.34, df = 1,p = .021) in reducing cocaine use as measured using urine drug screening results. The transitional analysis con- Survival analyses indicated that there were no sta- firmed the GEE solution and identified a strongly signifi- tistically significant differences as tested by log rank cant interaction between baclofen and baseline level of (χ2 = 0.54, df = 1, p = .46) in retention by assignment to cocaine use (χ2 = 10.63, df = 1, p = .001). The transitional condition (Figure 2). Baclofen-treated subjects were re- model also found that the treatment effects increased as tained in the protocol a mean (SD) of 56.87 (± 43.41) the number of benzoylecgonine-positive samples in- days, whereas those in the placebo condition averaged creased over the baseline period (Figure 4). By contrast, 48.27 (± 40.57) days. Similar percentages of participants those assigned to the placebo condition demonstrated completed the 16-week treatment period in the baclofen no such association. Placebo-treated participants who (N = 9; 25.7%) and placebo (N = 8; 22.9%) conditions.
provided more benzoylecgonine-positive urine samplesduring the baseline period showed stepwise increases in the likelihood of providing fewer metabolite-free Univariate analyses of aggregate variables showed that urine samples during the treatment period. A post hoc participants assigned to the baclofen condition performed strategy to describe the clinical significance of this effect better than those assigned to placebo, although not at sta- calculated the number of benzoylecgonine-free urine tistically significant levels (Table 2). Post hoc analyses of samples during the trial for participants who provided 3 or the joint probability showed baclofen-treated participants more samples during baseline that were positive for were significantly more likely to provide urine samples cocaine metabolite. Participants who met this criterion that were cocaine metabolite–free between weeks 3 to 8 and who were assigned to the placebo condition (N = 16) than were participants treated with placebo (t = 5.98, averaged 3.38 benzoylecgonine-free urine samples, com- COPYRIGHT 2003 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
Figure 3. Percentage of Urine Samples Provided by Participants in Each of the 2 Treatment Conditions That Tested Free of
Cocaine Metabolite (Benzoylecgonine[BE]) at Each Point During the Triala
t9 t11 t13 t15 t17 t19 t21 t23 t25 t27 t29 t31 t33 t35 t37 t39 t41 t43 t45 t47 aThe first 6 points (b1 to b6) represent the percentage of BE-free urine samples by condition during the baseline period. The line through t1 denotes randomization into treatment condition and the initiation of study medication. Points t1 to t48 represent the percentage of BE-free urine samples bycondition during the treatment period. The upward drift for both lines over the treatment period represents the effect of attrition on theseproportions.
pared with 11.42 samples for those assigned to the baclo- took a mean of 72.50% (± 22.82%) of their study medica- tion doses, a nonsignificant difference. Participants werepolled as to which medication they believed they re- ceived at the beginning of the second study week; a total There was no statistically significant difference be- of 18 (60.0%) of 30 participants assigned to receive bac- tween participant ratings of cocaine craving for the 24 lofen correctly guessed their condition, compared to 10 hours prior to the clinic visit by medication condition as (62.5%) of 16 participants assigned to receive placebo, a Over the course of the trial, there were no clinically significant changes observed in cardiovascular function- This study tested the effectiveness of baclofen as a ing, serology, vital signs, or physical examinations for pharmacotherapy for cocaine dependence when adminis- any study participants. Three serious adverse events oc- tered in the context of thrice weekly cognitive behavioral curred during the study, all of which occurred to partici- drug counseling. Standard, univariate measures of treat- pants receiving baclofen, and none of which was judged ment response using aggregates of urine drug screening to be related to study medication. All 3 serious adverse results showed no statistically significant differences for events involved a worsening of the participants’ cocaine the baclofen over the placebo condition, although all in- problems; each required overnight hospitalization, and 1 dices were in the predicted direction. Post hoc analyses required concomitant treatment for depression. The inci- using the joint probability index evaluated during weeks dence, by condition, of reported adverse events rated 3 and 8 showed statistically significant differences for moderate in severity is shown in Table 3. Participants in baclofen over placebo in reducing cocaine use. Admit- the baclofen condition were generally more likely to ex- tedly, this finding provides only a suggestion of a treat- perience headaches, whereas those in the placebo condi- ment effect. Aggregate indices are particularly sensitive tion were more likely to experience colds and flu, al- to the effect of attrition, especially in screening trials though there were no statistically significant differences utilizing relatively small sample sizes. Identification of between conditions for incidence or percentage of total medication effects is made more difficult beyond 8 weeks in this and other trials of cocaine pharmacotherapies,likely due to reductions in power associated with early Participants in the placebo condition took a mean of Significant medication effects for baclofen over pla- 69.85% (± 22.82%) of their study medication doses. By cebo were detected when using longitudinal models comparison, participants assigned to receive baclofen (GEE), including transitional models calculating the tran- COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
Figure 4. Transitional Probability That a Participant Will
Table 3. Incidence and Percentage of Total of Adverse Events
Provide a Benzoylecgonine (BE)-Negative Urine Sample
Rated Moderate in Severity by Assignment to Treatment
Given the Result of the Immediately Previous Urine Sample
Condition
as a Function of the Number of Urine Tests Provided at
Baseline That Were BE-Positivea
Abbreviation: GI = gastrointestinal.
Number of Baseline BE-Positive Urine Samples The GEE solution applied current biostatistical meth- aThe numbers of participants by condition that provided each level of ods, while the application of a transitional probability BE-positive urine samples during baseline are represented across the analysis, complete with calculation of the probability of top of the figure. No participants provided 6 BE-positive urinesamples during baseline; hence, the range was 0–5 BE-positive urine treatment response at differing levels of cocaine use at samples. These transitional probabilities depict the effect of baseline, represents a solid contribution to the task of successful treatment at any point during the medication periodand are represented by the probability of providing 2 consecutive identifying subgroups of patients that might respond posi- BE-free urine samples (“0,0”—maintaining cocaine abstinence) and tively to medications. The probabilities depicted provide of providing a BE-free urine sample following a BE-positive urinesample (“1,0”—initiating cocaine abstinence).
useful information to clinicians who might considerwhich patients may benefit from treatment with baclofenwhen administered in the context of cognitive-behavioral sitional probabilities of treatment response at different group drug counseling. Descriptive information showing levels of cocaine use observed at baseline. The agreement reductions in cocaine use for those treated with baclofen of these solutions yields confidence in the results that bac- who were heavy users of cocaine during the baseline pe- lofen treatment provided a differential response compared riod indicates the clinical relevance of this medication.
with placebo. The finding that baclofen effects are ob- The findings appear clear: Patients who present for treat- served in stepwise fashion with the extent of cocaine use ment with chronic levels of cocaine use may benefit from detected at baseline implies that those who have a more using baclofen in the early phases of recovery from co- chronic and severe form of cocaine dependence are more caine dependence. Those who present for treatment with likely to show a response to baclofen. These individuals more episodic patterns of cocaine use would likely re- engage in consistent exposure to cocaine, which re- spond positively to behavioral drug counseling ap- sembles the cocaine exposure frequency delivered in laboratory studies. This similarity in administration may Despite 15 years of medication trials, none have dem- be the theoretical link between study findings and pre- onstrated significant effects as cocaine pharmacothera- clinical investigations that show that effects of baclofen pies when evaluated using placebo-controlled, double- are dependent upon elements related to the delivery of co- blind procedures. Hence, outcome measures in cocaine caine in the experimental paradigms, i.e., dose level and pharmacotherapies must extend beyond absolute absti- reinforcement schedules.26 The observation that partici- nence and consider reductions in use as a viable measure pants assigned to the placebo condition demonstrated no of medication performance. Measures of partial response such effects provided further support for demonstration of (i.e., reductions in cocaine use) raise questions of how to medication effects and not artifacts of statistical analyses.
quantify those reductions in use, while at the same time Together, these findings provide an exciting suggestion maintaining an emphasis on clinically meaningful differ- for the role of GABAergic medications and the generally ences. This study used a variety of statistical techniques dampening effects for GABA on the reward mechanisms to more precisely quantify the effects of missing data involved during recovery from chronic levels of cocaine and baseline performance in evaluating the strength of the signal for baclofen. The preliminary nature of this study COPYRIGHT 2003 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
and small sample size preclude determining the clinical 10. Seutin V, Johnson SW, North RA. Effect of dopamine and baclofen on significance of the signal measured, a task that remains for N-methyl-D-asparate induced burst firing in rat ventral tegmental neu- an adequately powered replication trial.
A limitation of this study is the small size of the sample 11. Santiago M, Cano J, Westerink BHC. Are bilateral nigrostriatal dopami- nergic pathways functionally linked in the rat brain: a microdialysis study used to calculate these transitional probabilities. Prob- in conscious rats. Brain Res 1993; 628:187–192 abilities calculated at the extreme high end of cocaine use 12. Santiago M, Machado A, Cano J. Regulation of the prefrontal cortical during baseline are based on small samples due to the low dopamine release by GABAA and GABAB receptor agonists and antago-nists. Brain Res 1993;630:8–31 number of participants meeting those criteria. Still, the 13. Santiago M, Machado A, Cano J. In vivo release of dopamine from rat documented effects for baclofen in doubling or tripling the striatum, substantia nigra and prefrontal cortex: differential modulation probability of a positive response to treatment for those by baclofen. Br J Pharmacol 1993;109:814–818 14. Kushner SA, Dewey SL, Kornetsky C. Gamma-vinyl GABA attenuates with chronic cocaine use at treatment entry when com- cocaine-induced lowering of brain stimulation reward thresholds.
pared to placebo provide compelling empirical support for Psychopharmacology (Berl) 1997;133:383–388 using the medication in combination with behavioral drug 15. Kushner SA, Unterwald EM. Chronic cocaine administration decreases the functional coupling of GABA(B) receptors in the rat ventral tegmen- counseling for this subgroup of patients, particularly in the tal area as measured by baclofen-stimulated 35S-GTPgammaS binding.
early phase of recovery from cocaine dependence.
Findings from this study also are limited in that they 16. Wood PL. Actions of GABAergic agents on dopamine metabolism in the nigrostriatal pathway of the rat. J Pharmacol Exp Ther 1982;222: represent a single efficacy trial of baclofen as a cocaine pharmacotherapy. Results may only generalize to that 17. Pycock C, Horton R. Evidence for an accumbens-pallidal pathway in the group of cocaine-dependent individuals who are willing to rat and its possible gabaminergic control. Brain Res 1976;110:629–634 18. Cousins MS, Roberts DC, de Wit H. GABA(B) receptors agonists for the join an intensive outpatient medication research trial. Yet, treatment of drug addiction: a review of recent findings. Drug Alcohol the data indicate strongly that a larger trial of baclofen is warranted. These medication findings in the subgroup of 19. Morgan AE, Dewey SL. Effects of pharmacologic increases in brain GABA levels on cocaine-induced changes in extra cellular dopamine.
participants with chronic cocaine use at baseline also cor- respond with recent findings from a placebo-controlled 20. Stromberg MF, Mackler SA, Volpicelli JR, et al. The effect of gamma- trial of amantadine,40 in which participants with higher vinyl-GABA on the consumption of concurrently available oral cocaineand ethanol in the rat. Pharmacol Biochem Behav 2001;68:291–299 ratings of cocaine severity at baseline showed significant 21. Molina PE, Ahmed N, Ajmal M, et al. Co-administration of gamma-vinyl response to amantadine. Combination strategies that inte- GABA and cocaine: preclinical assessment of safety. Life Sci grate both of these medications may provide new direc- 22. Harding GF, Spencer EL, Wild JM, et al. Field-specific visual-evoked tions for the continuing search for an effective cocaine potentials: identifying field defects in vigabatrin-treated children.
23. Roberts DC, Brebner K. GABA modulation of cocaine self-administra- Drug names: amantadine (Symmetrel and others), baclofen (Lioresal 24. Brebner K, Phelan R, Roberts DC. Effect of baclofen on cocaine self- administration in rats reinforced under fixed-ratio 1 progressive-ratioschedules. Psychopharmacology (Berl) 2000;148:314–321 25. Roberts DC, Andrews MM, Vickers GJ. Baclofen attenuates the reinforc- ing effects of cocaine in rats. Neuropsychopharmacology 1996;14:1–7 1. Ling W, Shoptaw S. Integration of research in pharmacotherapy for 26. Campbell UC, Lac ST, Carroll ME. Effects of baclofen on maintenance addictive disease: where are we? where are we going? In: Miller NS, and reinstatement of intravenous cocaine self-administration in rats.
Stimmel B. Binghamton NY, eds. The Integration of Pharmacological Psychopharmacology (Berl) 1999;143:209–214 and Nonpharmacological Treatments in Drug/Alcohol Addictions.
27. Ling W, Shoptaw S, Majewska D. Baclofen as a cocaine anti-craving medication: a preliminary clinical study Neuropsychopharmacology 2. Self DW, Nestler EJ. Molecular mechanisms of drug reinforcement and addiction. Annu Rev Neurosci 1995;18:463–495 28. McLellan AT, Kushner H, Metzger D. The fifth edition of the Addiction 3. Jentsch JD, Taylor JR. Impulsivity resulting from frontostriatal Severity Index. J Subst Abuse Treat 1992;9:199–213 dysfunction in drug abuse: implications for the control of behavior 29. Spitzer RL, Williams JB, Gibbon M, et al. The Structured Clinical Inter- by reward-related stimuli. Psychopharmacology (Berl) 1999;146: view for the DSM-IV (SCID). Washington DC: American Psychiatric 4. Kleber HD. Treatment of cocaine abuse: pharmacotherapy.
30. Preston KL, Silverman K, Schuster CR, et al. Use of quantitative CIBA Foundation Symposium 1992;166:195–200 urinalysis in monitoring cocaine use. NIDA Res Monogr 1997;175: 5. Wise RA. D1-and D2-type contributions to psychomotor sensitization and reward: implications for pharmacological treatment strategies.
31. Ling W, Shoptaw S, Wesson D, et al. Treatment effectiveness score as an outcome measure in clinical trials. NIDA Res Monogr 1997;175: 6. Cameron DL, Williams JT. Dopamine D1 receptors facilitate transmitter 32. Gawin FH, Morgan C, Kosten TR, et al. Double-blind evaluation of the 7. Engberg G, Klingpetersen T, Nissbrandt H. GABA( B) receptor activation effect of acute amantadine on cocaine craving. Psychopharmacology alters the firing pattern of dopamine neurons in the rat substantia nigra.
33. Rawson RA, Shoptaw SJ, Obert JL, et al. An intensive outpatient ap- 8. Klitenick MA, DeWitte P, Kalivas PW. Regulation of somatodendritic proach for cocaine abuse treatment. The Matrix model. J Subst Abuse dopamine release in the ventral tegmental area by opioids and GABA: an in vivo microdialysis study. J Neurosci 1992;12:2623–2632 34. Shoptaw SJ, Rawson RA, McCann MJ, et al. The Matrix model of 9. Rick CE, Lacey MG. Rat sustantia nigra pars reticulata neurons are outpatient stimulant abuse treatment: evidence of efficacy. J Addict Dis tonically inhibited via GABBA, but not GABAB, receptors in vitro.
COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.
35. Shoptaw SJ, Kintaudi PC, Charuvastra C, et al. A screening trial of aman- a generalized estimating equation approach. Biometrics 1988; tadine as a medication for cocaine dependence. Drug Alcohol Depend 39. Stram DO, Wei LJ, Ware JH. Analysis of repeated ordered categorical 36. Guilford JP, Fruchter B. Fundamental Statistics in Psychology and outcomes with possibly missing observations and time dependent covar- Education, 6th ed. New York, NY: McGraw-Hill; 1978:159–163 37. Liang KY, Zeger SL. On the use of concordant pairs in matched 40. Kampman KM, Volpicelli JR, Alterman AI, et al. Amantadine in the case-controlled studies. Biometrics 1988;44:1145–1156 treatment of cocaine-dependent patients with severe withdrawal symp- 38. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: toms. Am J Psychiatry 2000;157:2052–2054 COPYRIGHT 2003 PHYSICIANS POSTGRADUA
TE PRESS, INC. COPYRIGHT 2003 PHYSICIANS POSTGRADUATE PRESS, INC.

Source: http://www.josephmatta.com/baclofencocaine.pdf