Results of a Randomized, Double-Blind, Placebo-Controlled Trial of Lorcaserin in Cocaine Use Disorder

Abstract

Background:

Cocaine use disorder (CUD) is a major public health problem for which there is no approved pharmacotherapy. The primary purpose of this study was to evaluate the ability of lorcaserin, a 5-hydroxytryptamine_2C (5-HT_2C) receptor agonist, to facilitate abstinence in individuals seeking treatment for CUD.

Methods:

This was a 12-site, randomized, parallel arm study with a 13-week Treatment Phase that included a 1-week, single-blind run-in period when all participants received twice daily 15 mg acetazolamide capsules (a medication adherence marker), followed by randomization to either twice daily 10 mg lorcaserin or placebo capsules for the remaining 12 weeks. Pre-randomization data were utilized in an enrichment strategy aimed at achieving high levels of medication adherence and low placebo response rates in a subgroup of participants that qualified for the “efficacy population.” For lorcaserin vs. placebo, the primary efficacy endpoint was the proportion of participants in the efficacy population achieving abstinence during the last three weeks of treatment, as evidenced by self-report of no cocaine use, confirmed by urine testing.

Results:

Within the efficacy population, 1.1% of 91 participants receiving lorcaserin and 4.3% of 92 receiving placebo achieved abstinence during the last 3 weeks of treatment. Among all randomized participants, 2.5% of 118 receiving lorcaserin and 5.6% of 124 receiving placebo achieved similar abstinence. Study participants receiving lorcaserin exhibited significantly greater reductions in body weight and BMI, indicating that medication adherence was sufficient to produce a pharmacological effect.

Conclusions:

Twice daily 10 mg lorcaserin failed to demonstrate efficacy in the treatment of CUD.

1. Introduction

In recent years, morbidity and mortality associated with cocaine use have increased within the United States (Becker et al., 2022), and the treatment of CUD represents a growing challenge. There are no FDA-approved medications for the treatment of CUD, and psychotherapeutic approaches other than contingency management (i.e., providing financial rewards for periods of cocaine abstinence) show little evidence of benefit (Farrell et al., 2019). Thus, there is an urgent need for medication development efforts in this area.

Preclinical data from multiple laboratories suggested that 5-HT_2C agonists such as lorcaserin, marketed in the U.S. from 2012 to 2020 for weight loss, might have efficacy in the treatment of CUD and other substance use disorders. Extensive basic research efforts have established that 5-HT_2C receptors mediate an inhibitory effect of serotonin on dopamine release in many brain regions, including the mesolimbic dopamine pathway (Alex and Pehek, 2007), and it has been proposed that a selective 5-HT_2C agonist may both blunt the subjective effects of cocaine and decrease impulsivity that can lead to relapse in CUD patients (Howell and Cunningham, 2015). Cocaine self-administration studies yielded positive results in rats (Cunningham et al., 2011; Harvey-Lewis et al., 2016) and monkeys (Collins et al., 2016; Gerak et al., 2016). Given that alcohol use disorder may be common among people seeking treatment for CUD (Carroll and Rounsaville 1992), it is noteworthy that lorcaserin was also found to decrease alcohol self-administration in rats (Rezvani et al., 2014). Encouragingly, similar nicotine self-administration studies (Levin et al., 2011; Higgins et al., 2012) translated to a successful smoking cessation trial (Shanahan et al., 2017).

While the present study was ongoing, two relevant outpatient studies of lorcaserin were initiated. In a placebo-controlled feasibility study (Santos et al., 2021), a total of 22 participants with CUD, all men who have sex with men, were randomized to receive either once daily extended-release 20 mg lorcaserin (N = 16) or placebo (N = 6) for 12 weeks. While urine testing revealed no evidence of decreased cocaine use in participants receiving lorcaserin, an association between lorcaserin treatment and decreased self-report of cocaine use was reported. In an open-label pilot study (Campbell et al., 2021), 10 participants with AUD (as well as eight participants with methamphetamine use disorder) were enrolled to receive 10 mg lorcaserin daily for four days, followed by twice daily dosing for one month. Both alcohol consumption and participant ratings of alcohol craving were reported to decrease over time. Firm conclusions regarding lorcaserin efficacy cannot be drawn from the two studies due to their small sample size and/or the lack of a placebo control group. To our knowledge, the present study was the only large, multi-site, placebo-controlled trial to evaluate lorcaserin efficacy in CUD. Lorcaserin efficacy in treating AUD was also evaluated in a subset of participants with comorbid AUD/CUD.

2. Materials and Methods

2.1. Study Design

This was a 12-site, randomized, parallel-arm study that allowed up to 3 weeks for screening, followed by a 13-week treatment phase and a 3-week follow-up period. The protocol was approved by a central institutional review board, and all participants provided written informed consent. Participating sites included established addiction treatment centers, clinics associated with academic institutions, and contract research sites that typically participate in trials sponsored by pharmaceutical companies (see acknowledgements for specific site locations). In recruiting participants, all sites used advertisements to reach individuals who regarded their own cocaine use as problematic and who might be interested in treatment. The advertisements mentioned that all participants would receive counseling sessions in addition to study drug, and contact information was provided so that treatment-seeking individuals could engage sites for screening. The study was conducted in compliance with the Declaration of Helsinki and ethical principles of Good Clinical Practice, as defined by the International Conferences on Harmonization, and it was registered on clinicaltrials.gov (NCT 03007394).

During the screening period, medical and psychiatric assessments were conducted to determine if participants met criteria for enrollment. Urine samples were also collected and assayed on-site for the presence of commonly misused drugs and/or their metabolites, including benzoylecgonine (BE) and ethyl glucuronide (EtG), metabolites of cocaine and ethanol, respectively. A clinical trial subject registry (Verified Clinical Trials) was utilized during screening to prevent duplicate enrolment of study participants at multiple sites within the study, to screen out subjects with exclusionary conditions being treated in other trials, and to avoid enrollment of individuals already participating in other/conflicting studies that were covered by the registry.

Participants meeting enrollment criteria advanced to the 13-week treatment phase, which began with a 1-week, single-blind run-in period when all participants received twice daily 15 mg acetazolamide capsules (a medication adherence marker), followed by randomization to either twice daily 10 mg lorcaserin or placebo capsules (double-blind) for the remaining 12 weeks. The lorcaserin dosing regimen was selected for study because it had received regulatory approval for the treatment of obesity; the established safety profile of the regimen and the commercial availability of 10 mg lorcaserin tablets were important considerations. Throughout the treatment phase, participants visited the clinic once weekly for assessments, to provide urine samples, and to participate in behavioral therapy sessions. Urine samples were assayed at a central laboratory (Laboratory Corporation of America, Research Triangle Park, NC) for BE using a validated liquid chromatography-tandem mass spectrometry method. The samples were also assayed on-site for EtG using commercially available dipsticks. Plasma samples taken on Study Weeks 3 and 7 were assayed for lorcaserin at a central laboratory (XenoBiotic Laboratories, Inc., Plainsboro, NJ) using a validated liquid chromatography-tandem mass spectrometry method. All participants were provided 1-hour, individual cognitive behavioral therapy sessions each week to facilitate avoidance of cocaine use. The sessions were conducted as described in the NIDA therapy manual entitled “A Cognitive-Behavioral Approach: Treating Cocaine Addiction” (Carroll, 1998). In addition, optional sessions were available to facilitate avoidance of alcohol use. The alcohol-focused sessions utilized a 4-module computer-based behavioral platform (“Quit to Quit”) that was a slight modification of alcohol intervention “Take Control” (Devine et al., 2016). This intervention includes alcohol education, motivational enhancement, coping with urges, recreational counseling, coping with social pressure, and recovering from a slip. The first module focused on alcohol education and building motivation for quitting drinking, and it was viewed by all subjects on Study Day 1 (the first day of the 1-week run-in period). On Study Day 8, prior to randomization, participants were asked if alcohol abstinence would be one of their goals during the trial and they were given the option to view the second module. The decision (yes/no) on whether to continue the alcohol-focused behavioral platform was recorded and used as one of three stratification factors for randomization, with placebo or lorcaserin assigned in approximately a 1:1 ratio. The other two stratification factors were the presence of alcohol and/or benzodiazepine dependence (yes/no) and the number of participant-reported cocaine use days during the 30 days prior to screening (< 8 vs. 8 or more). Randomization was performed using an interactive web response system and a randomized block design within each site. All study participants, site staff, sponsor (NIDA) staff, and data management personnel at the VA Cooperative Studies Program were blinded to treatment assignments.

Pre-randomization data were utilized in an enrichment strategy aimed at achieving high levels of medication adherence and low placebo response rates in a subgroup of participants that qualified for the “efficacy population,” and data from participants failing to qualify for this population were excluded from consideration during analysis of the primary and secondary efficacy endpoints. Participants were required to meet two specific criteria for inclusion in the efficacy population. To address medication adherence, plasma samples taken on Study Day 8, prior to randomization, were assayed for acetazolamide at a central laboratory (XenoBiotic Laboratories, Inc.) using a validated liquid chromatography-tandem mass spectrometry method. An extremely low plasma level of acetazolamide following the one-week run-in period was regarded as evidence of low medication adherence during the week and viewed as a predictor of low medication adherence after randomization. Accordingly, a Study Day 8 plasma acetazolamide level ≥15 ng/ml was required for inclusion in the efficacy population. Based on a prior human laboratory study (unpublished; conducted under a NIDA contract with Vince & Associates, Overland Park, KS), plasma acetazolamide concentrations can be expected to stay above 15 ng/ml for 48 hours after a single 15 mg oral dose in approximately 99.9% of people (the lower boundary of the 99.9% CI for plasma samples collected 48 hours post-dose was 15.2 ng/mL). Because participants in the current study were supposed to take 15 mg acetazolamide capsules twice daily during the run-in period, plasma levels < 15 ng/ml would be expected only if they missed at least 3 or more consecutive doses at the end of the run-in period. The second criterion for inclusion in the efficacy population was intended to minimize the placebo response rate. In an unpublished analysis of placebo response rates in previous NIDA trials, a greater probability of achieving abstinence from cocaine was observed in participants who reported < 8 days of cocaine use during the 30 days prior to screening; therefore, in the current study, only participants reporting 8 or more days of cocaine use during this period were eligible for the efficacy population. Participants with a Study Day 8 plasma acetazolamide level < 15 ng/ml and/or reporting < 8 days of cocaine use during the 30 days immediately prior to screening were excluded from the prespecified efficacy population, and their data were not considered during analysis of the primary and secondary efficacy endpoints. For a general discussion of this type of enrichment strategy and related FDA guidance, see McCann et al., 2015.

To promote medication adherence after randomization and to facilitate daily participant reports on cocaine and alcohol use, subjects received a cell phone on day 8 installed with software designed to monitor medication adherence and collect outcome data. This software, developed by AiCure Technologies (New York, NY), was compliant with the Health Insurance Portability and Accountability Act (HIPAA) and provided real-time data to assess medication adherence. The AiCure software used facial recognition and motion-sensing technology to confirm the ingestion of study medication twice daily, and participants were asked to enter information on the previous day’s cocaine and alcohol use at the time of the first dosing event each day. The resulting data (encrypted, with facial images blurred), were transferred to a secure central server in real-time. During weekly clinic visits, AiCure data from the previous week were reviewed by site staff, who had the opportunity to discuss apparent medication adherence lapses, encourage better medication adherence, and collect any missing data on cocaine or alcohol use.

Following the conclusion of the 13-week treatment phase, participants visited the clinic for follow-up on Study Weeks 14 and 16. Adverse events were assessed and recorded throughout the study. Any adverse events at Study Week 16 were recorded and followed to resolution if deemed serious or clinically significant.

Participants received $15 compensation for the time and inconvenience required for each clinical visit, and they could receive additional compensation for completing specific tasks. For example, they received $5 for returning medication bottles and they received $15 if they used the AiCure software to report their cocaine and alcohol use every day since their previous clinic visit. They also received periodic bonuses for attending 4 consecutive weekly clinic visits, with the bonus value escalating to $40 if they visited the clinic during study weeks 11–14 (the last 3 weeks of treatment and the first follow-up visit). In total, participants could receive a maximum of $905 compensation during screening, treatment, and follow-up.

2.2. Participants

Treatment-seeking males and females between 18 and 65 years of age who met DSM 5 criteria for cocaine use disorder (as verified by the Structured Clinical Interview for DSM-5 – Research Version), were enrolled at all 12 clinical sites that conducted screening. Participants were required to be capable of understanding and providing written informed consent and to be in stable health. In addition, participants were required to have reported at least one day of cocaine use during the 30 days prior to screening, and at least one positive on-site urine test for BE was required during screening. A total body weight of >50 kg (110 pounds) and a body mass index (BMI) of > 20 was required at screening, and participants who were capable of conceiving were required to agree to use an acceptable method of birth control. Exclusion criteria included: pregnancy or lactation; use of lorcaserin during the 30 days prior to screening; a diagnosed substance use disorder for any psychoactive substance other than cocaine, benzodiazepine, alcohol, nicotine, caffeine, or marijuana; a history of methadone or buprenorphine treatment during the year prior to screening; one or more positive on-site urine drug test results for methamphetamine, amphetamine, methadone, buprenorphine, oxycodone or other opioids during screening; alcohol use disorder judged to require medically supervised detoxification; ongoing mandatory drug testing (e.g., as a condition of probation); a Beck Depression Inventory-Second Edition (BDI-II) score > 20 at screening; acute suicidality or a history of suicidal behavior; and schizophrenia, bipolar I or II disorder, or any other diagnosed psychotic disorder. In addition, potential study participants were excluded: if they had unstable health; if they were participating in screening for another clinical trial of an investigational medication; if they participated in any clinical trial of an investigational medication during the six months prior to screening; if they refused to sign a Verified Clinical Trials Authorization (allowing their study participation to be documented in the Verified Clinical Trials subject registry, and allowing verification that they were not participating in conflicting trails covered by the registry); or if they were not expected to attend regular study visits or complete protocol assessments. Additional exclusions included co-occurring disorders that would place the subject at greater risk or prejudice evaluation of study medication safety and efficacy.

2.3. Drug Treatment

Lorcaserin 10 mg tablets (trade name Belviq^®) were purchased and over-encapsulated in size 0 light blue opaque hard gelatin capsules. Lorcaserin placebo capsules, containing a mixture of microcrystalline cellulose and magnesium stearate, as well as acetazolamide 15 mg capsules, were an identical match to the over-encapsulated lorcaserin capsules. All formulation work was conducted by Murty Pharmaceuticals Inc. (Lexington, KY).

2.4. Endpoints and data analysis

The primary efficacy outcome measure was the proportion of subjects achieving abstinence from cocaine during the last 3 weeks of treatment (Study Weeks 11–13), and comparison of this proportion in participants receiving lorcaserin vs. placebo within the efficacy population constituted the primary endpoint analysis. According to the study’s enrichment strategy, described above, participants qualified for inclusion in the efficacy population if they had a Study Day 8 plasma acetazolamide level ≥15 ng/ml and they reported 8 or more days of cocaine use during the 30 days prior to screening. A participant was regarded abstinent during the last 3 weeks of treatment if: 1) there were self-report data indicating no cocaine use on each day of the 3-week period; 2) there was at least one BE assay result for a urine sample collected within the 3-week period; and 3) BE assay results were negative (BE < 150 ng/ml) for all urine samples collected during the 3-week period. Subjects not meeting these three criteria (including dropouts) were regarded as having failed to achieve abstinence from cocaine. While the study was ongoing but prior to finalization of the statistical analysis plan (SAP), it became apparent that there would be a small amount of missing self-report data on cocaine and alcohol use for some participants who completed the study, and the SAP was written to specify a conservative imputation method that resulted in the assumption of “reported cocaine use” on each day with missing self-report data.

The secondary efficacy endpoint was the proportion of subjects successfully achieving abstinence from cocaine during the last 3 weeks of treatment in a subset of the efficacy population that were either non-drinkers (by self-report, confirmed with negative urine tests for EtG throughout screening and on Study Day 8) or attempting alcohol abstinence (as evidenced by self-report and willingness to continue the alcohol-focused, computer-guided bibliotherapy on Study Day 8, prior to randomization). The analysis was conducted as described for the primary endpoint.

Prespecified exploratory endpoints included baseline (pre-randomization) to Study Week 13 changes in scores obtained with the Addiction Severity Index-Lite assessment tool (ASI-Lite; Cacciola et al., 2007) and the Brief Substance Craving Scale (BSCS; Mezinskis et al., 2001). In addition, the proportion of subjects successfully achieving abstinence from alcohol during the last 3 weeks of treatment was determined for the subset of study participants who were diagnosed with AUD during screening and were attempting alcohol abstinence during treatment. In the related analysis, a participant was regarded abstinent from alcohol during the last 3 weeks of treatment if: 1) there were self-report data indicating no alcohol use on each day of the 3-week period; 2) there was at least one EtG assay result for a urine sample collected within the 3-week period; and 3) EtG assay results were negative (EtG < 500 ng/ml) for all urine samples collected during the 3-week period. Subjects not meeting these three criteria (including dropouts) were regarded as having failed to achieve abstinence from alcohol. To address missing self-report data for alcohol use, each day of missing data was imputed as “reported alcohol use.” Safety outcomes included body weight and vital sign measurements, physical examination data, depression severity (BDI-II), suicidality (Columbia Suicide Severity Rating Scale) and reported adverse events.

Following the methods of Santos et al. (2021), two post-hoc analyses were conducted on data from the efficacy population. One analysis focused on changes in self-reported cocaine use alone, without consideration of urine testing results, and the other focused on changes in rates of urine BE positivity (BE ≥ 150 ng/ml) alone, without consideration of cocaine use self-report data. Data from Study Week 1 (the placebo run-in period) were evaluated as the baseline for these analyses and data from Study Weeks 2–13 were subjected to GEE analysis. In addition, post-hoc analyses focused on heavy drinking days, defined as 4 or more drinks per day for women and 5 or more drinks per day for men (Allen, 2003; Falk et al., 2010), were conducted for the subset of study participants with AUD who were attempting alcohol abstinence during treatment. Within this subgroup, the proportion of participants reporting no heavy drinking days during the last 3 weeks of treatment was compared for study completers receiving lorcaserin vs. placebo. Pre-randomization (Study Week 1) drinking levels were similarly compared.

To evaluate statistical significance, unless otherwise noted, chi-square tests or Fisher’s exact tests (if chi-square was not appropriate due to small cell size) were used for categorical variables, and t-tests were used for continuous variables. Two-tailed tests were performed for all analyses, with a significance level of α = 0.05. The sample size estimate was based on potential cocaine abstinence success rates of 6% and 20% for participants randomized to placebo and lorcaserin, respectively. Assuming these success rates, it was calculated that 90 subjects per treatment in the efficacy population would yield 80% power (Pearson’s Chi-square test for two proportions; two-sided comparison, α = 0.05). Based on estimated rates for dropout prior to randomization and failure to qualify for the efficacy population, it was calculated that enrollment of 272 study participants would yield an efficacy population of 180 (approximately 90 for each treatment).

3. Results

3.1. Participant Disposition and Demographics

From June 2017 through June 2018, a total of 593 individuals were screened for study eligibility at 12 sites within the United States. Fig. 1 shows the disposition of all individuals who were screened. A total of 318 did not progress to enrollment because they did not meet inclusion/exclusion criteria, three individuals who were eligible for enrollment did not return to the clinic after screening, and 272 were enrolled in the 1-week, single-blind run-in period with twice daily 15 mg acetazolamide capsules. Thirty enrolled participants were not randomized on Study Day 8, and the majority of them (24) simply failed to return to the clinic that day. Overall retention in treatment after randomization was 75%, and similar retention was seen in those randomized to placebo (73%) and lorcaserin (77%).

Fig. 1.

Flow diagram of participants through the study of lorcaserin in cocaine use disorder

Among the 242 randomized participants, 183 (75.6%) qualified for inclusion in the efficacy population, with 92 randomized to placebo and 91 randomized to lorcaserin. Among the 59 randomized participants excluded from the efficacy population, 40 did not qualify only because they reported < 8/30 days baseline cocaine use, 11 were found to have pre-randomization acetazolamide plasma levels < 15 ng/ml, and eight had no acetazolamide assay result for a pre-randomization plasma sample. Within the efficacy population, retention in treatment was 75% (69/92) for those receiving placebo and 77% (70/91) for those receiving lorcaserin.

During data analysis, when categorizing the severity of CUD for all study participants, it was discovered that one enrolled and randomized study participant did not meet criteria for CUD (at any severity level) and, therefore, should not have been enrolled. This individual did not qualify for inclusion in the efficacy population.

As shown in Table 1, participants receiving placebo and lorcaserin had similar demographic and baseline characteristics. Among all randomized participants, the average age was 50.2 years, 72% were male, 28% were female, 66% were Black or African American, 27% were white, and 7.4 % were Hispanic or Latino. Smoking was the usual route of cocaine administration for 76% of randomized participants, while nasal administration was the usual route for 23%. Based on the DSM-5 criteria for severity of the CUD diagnosis, 81 % of randomized participants had severe CUD and 14% demonstrated moderate severity. As previously noted, the last three items in Table 1 served as stratification factors during randomization.

Table 1.

Characteristics of all randomized participants by treatment

Demographic/Baseline Characteristic	Placebo (n = 124)	Lorcaserin (n = 118)
Gender
Female: n (%)	40 (32.3%)	28 (23.7%)
Male: n (%)	84 (67.7%)	90 (76.3%)
Age in years: mean (SD)	50.3 (9.99)	50.0 (9.88)
Body weight, lb.: mean (SD)	186.7 (41.8)	185.2 (41.7)
BMI, kg/m2: mean (SD)	27.8 (5.24)	27.6 (5.58)
Race
American Indian or Alaska Native: n (%)	0 (0.0%)	1 (0.8%)
Asian: n (%)	1 (0.8%)	0 (0.0%)
Black or African American: n (%)	83 (66.9%)	77 (65.3%)
White: n (%)	30 (24.2%)	35 (29.7%)
Other: n (%)	9 (7.3%)	4 (3.4%)
Ethnicity
Hispanic or Latino: n (%)	10 (8.1%)	8 (6.8%)
Not Hispanic or Latino: n (%)	113 (91.1%)	109 (92.4%)
Marital status
Legally married: n (%)	12 (9.7%)	16 (13.6%)
Living with partner/cohabitating: n (%)	3 (2.4%)	6 (5.1%)
Widowed: n (%)	4 (3.2%)	4 (3.4%)
Separated: n (%)	6 (4.8%)	8 (6.8%)
Divorced: n (%)	35 (28.2%)	28 (23.7%)
Never married: n (%)	64 (51.6%)	56 (47.5%)
Years of formal education: mean (SD)	13.2 (2.05)	12.8 (2.10)
Employment pattern
Full time (35+ hours/week): n (%)	30 (24.2%)	28 (23.7%)
Unemployed: n (%)	54 (43.5%)	47 (39.8%)
Usual route of cocaine administration
Nasal: n (%)	30 (24.6%)	24 (20.9%)
Smoking: n (%)	91 (74.6%)	89 (77.4%)
IV: n (%)	1 (0.8%)	0 (0%)
Oral: n (%)	0 (0%)	2 (1.7%)
Severity of cocaine use disorder (DSM-5 diagnosis at screening)
Severe: n (%)	103 (83.1%)	93 (78.8%)
Moderate: n (%)	16 (12.9%)	18 (15.3%)
Mild: n (%)	5 (4.0%)	6 (5.1%)
Participant-reported cocaine use during the 30 days prior to screening
< 8 days use: n (%)	24 (19.4%)	21 (17.8%)
8 or more days use: n (%)	100 (80.6%)	97 (82.2%)
Optional viewing of alcohol abstinence-focused bibliotherapy module on Study Day 8
Yes: n (%)	47 (37.9%)	47 (39.8%)
No: n (%)	77 (62.1%)	71 (60.2%)
Alcohol and/or benzodiazepine dependence
Yes: n (%)	23 (18.5%)	23 (19.5%)
No: n (%)	101 (81.5%)	95 (80.5%)

3.2. Evaluation of Lorcaserin Efficacy vs. CUD

As shown in Table 2, lorcaserin failed to facilitate achievement of cocaine abstinence in the efficacy population (designated for the primary efficacy endpoint), in the subset of the efficacy population who were either nondrinkers or participating in optional bibliotherapy sessions to promote abstinence from alcohol (the population designated for the secondary efficacy endpoint), and in all randomized study participants. In fact, within all three populations, the nonsignificant trend was in the direction of greater success achieving cocaine abstinence with placebo treatment.

Table 2.

Evaluation of lorcaserin efficacy in cocaine use disorder

Population	Outcome (last 3 weeks of treatment)	Placebo n (%)	Lorcaserin n (%)	p-value
Efficacy Populationa	Cocaine-Abstinent	4/92 (4.3%)	1/91 (1.1%)	0.368
	Not Abstinent	88/92 (95.7%)	90/91 (98.9%)
Subset of Efficacy Population Attempting Alcohol Abstinence or Nondrinkersb	Cocaine-Abstinent	3/45 (6.7%)	0/49 (0.0%)	0.106
	Not Abstinent	42/45 (93.3%)	49/49 (100%)
All Randomized Participants	Cocaine-Abstinent	7/124 (5.6%)	3/118 (2.5%)	0.225
	Not Abstinent	117/124 (94.4%)	115/118 (97.5%)

^aEvaluation in this population constitutes the trial’s primary efficacy endpoint.

^bEvaluation in this population constitutes the secondary efficacy endpoint.

While the prespecified imputation method for missing self-report data on cocaine use was conservative, resulting in the assumption of “used cocaine” for all missing days of self-report, the amount of missing data was minimal, and the conservative imputation method did not mask lorcaserin efficacy. Within the efficacy population, only 5 subjects (1 receiving placebo and 4 receiving lorcaserin) who completed treatment were categorized as “not abstinent” due to missing self-report data. Using the opposite, most liberal imputation method (assuming “did not use cocaine” for all missing self-report data), the outcome in the efficacy population would have been 5.4% (5/92) success with placebo treatment and 5.5% (5/91) success with lorcaserin treatment (p = 0.986). Using this same liberal imputation method in all randomized subjects, the outcome would have been 6.5% (8/124) success with placebo treatment and 8.5% (10/118) success with lorcaserin treatment (p = 0.549).

Within the efficacy population, prespecified exploratory evaluations of baseline (pre-randomization) to Study Week 13 changes in scores for the BSCS and the ASI-Lite also revealed no indication of lorcaserin efficacy in the treatment of CUD. The mean change in craving severity (BSCS) score from baseline to Study Week 13 was –3.5 (SD 3.23) for placebo and –3.4 (SD 3.05) for lorcaserin (p = 0.912). Questions related to problems with drug and alcohol use were analyzed as separate domains within the ASI-Lite. The scores in both domains decreased from baseline to Study Week 13; however, placebo treatment was associated with significantly greater decreases in scores for reported problems with drug and alcohol use. In other words, greater improvements were seen in those receiving placebo. The mean change in score for the “drugs” ASI domain from baseline to Study Week 13 was −0.091 (SD 0.106) for placebo and – 0.053 (SD 0.088) for lorcaserin (p = 0.033). The mean change in score for the “alcohol” ASI domain from baseline to Study Week 13 was −0.114 (SD 0.190) for placebo and – 0.015 (SD 0.160) for lorcaserin (p = 0.003). No significant baseline to Study Week 13 changes were seen for the other 5 domains of the ASI-Lite.

Post-hoc analyses focused on urine BE positivity alone (without consideration of cocaine use self-report data) and focused on self-reported cocaine use alone (without consideration of urine testing results) revealed comparable baseline values in the efficacy population for participants randomized to lorcaserin and placebo. At baseline (during Study Week 1, the placebo run-in period), BE-positive urines were obtained from 87% (78/90) and 88% (79/90) of participants randomized to lorcaserin and placebo, respectively. During the same baseline period, cocaine use was reported by 95% (86/91) and 92% (85/92) of participants randomized to lorcaserin and placebo, respectively. While there was an overall decrease in urine BE positivity over time during Study Weeks 2–13, GEE analysis revealed no significant difference between groups (incidence risk ratio 1.0648; 95% CI 0.9360 – 1.2113; p = 0.3402). During Study Week 13, BE-positive urines were obtained from 73% (45/62) and 72% (49/68) of those randomized to lorcaserin or placebo, respectively. There was also an overall decrease in self-reported cocaine use over time during Study Weeks 2–13, but GEE analysis revealed a significantly greater decrease among participants receiving placebo (IRR 1.3233; 95% CI 1.048 – 1.6694; p = 0.0181). During Study Week 13, cocaine use was reported by 78% (52/67) and 60% (39/65) of those randomized to lorcaserin or placebo, respectively. The finding of less self-reported cocaine use among participants receiving placebo is contrary to the findings of Santos et al. (2021) but is congruent with the placebo group’s greater decrease in ASI-Lite subscale scores for reported problems with drug use, noted above.

3.3. Evaluation of Lorcaserin Efficacy vs. AUD

In total, 30 study participants who met DSM-5 criteria for AUD during screening attempted to achieve abstinence from alcohol during the trial (as evidence by endorsement of the goal and continued viewing of the optional bibliotherapy to promote alcohol abstinence on Study Day 8). Among these participants, 7.1% (1/14) receiving placebo and 6.3% (1/16) receiving lorcaserin were successful in achieving abstinence from alcohol during the last 3 weeks of treatment (p > 0.99). Similarly, a nonsignificant trend toward greater success in achieving alcohol abstinence with placebo was seen in all randomized participants with AUD (2/22 or 9.1% success for placebo; 1/23 or 4.3% success for lorcaserin) and in all randomized participants (14/124 or 11.3% success for placebo; 11/118 or 9.3% success for lorcaserin).

A post-hoc analysis focused on study completers within the subpopulation of study participants with AUD who were attempting alcohol abstinence revealed that 82% (9/11) receiving placebo and 50% (7/14) receiving lorcaserin reported no heavy drinking days during the last 3 weeks of treatment (p = 0.21). A similar trend was seen when pre-randomization drinking levels were compared in these same individuals. During Study Week 1, 82% (9/11) who were subsequently randomized to placebo and 57% (8/14) who were subsequently randomized to lorcaserin reported no heavy drinking days (p = 0.23). It follows that the current results suggest neither a favorable nor unfavorable effect of lorcaserin on drinking outcomes.

3.4. Medication Adherence and Lorcaserin Plasma Levels

Based on data transmitted with AiCure software during dosing events, study medication adherence averaged 77.3% from randomization (Study Day 8) through the end of treatment. It is likely that some dosing events were not captured by the AiCure software due to study participant failure to keep cell phones charged or due to misspaced cell phones. Because self-report data on medication adherence may not be accurate or honest, only software-captured dosing events were utilized in the calculation of medication adherence; claims of medication adherence without device use were not captured in the database. This approach may have led to an underestimation of adherence for some participants. On the other hand, it must be noted that the cellphone-generated data resulted in an overestimation of adherence for participants who exhibited intentional nonadherence (participants who pretended to take their study medication in front of their cell phones but did not swallow their capsules). In a review of data from the first week of cell phone-captured dosing events (beginning on Study Day 8), AiCure staff identified at least one occurrence of intentional nonadherence in 16% (39/242) of randomized participants.

Lorcaserin plasma levels were captured at two times. For samples drawn during Study Week 3 visits, following one week of lorcaserin dosing, the median lorcaserin plasma concentration was 40.3 ng/ml, ranging from a maximum of 118 ng/ml to undetectable (< 0.5 ng/ml). There were 108 plasma assay results from Study Week 3, and lorcaserin was undetectable in 6 samples (5.6%). For samples drawn during Study Week 7 visits, following five weeks of lorcaserin dosing, the median lorcaserin plasma concentration was 39.7 ng/ml, ranging from 167 ng/ml to undetectable. There were 93 plasma assay results from Study Week 7, and lorcaserin was undetectable in 5 samples (5.4%). Substantial within-participant variation was observed. For example, the participant with the highest lorcaserin plasma concentration (167 ng/ml) in Week 7 had no detectable lorcaserin in Week 3. Also, while a total of 15 plasma samples had a lorcaserin concentration of at least 80 ng/ml, only two participants met this threshold at both sampling times; notably, neither achieved end-of-study abstinence, as defined for the primary endpoint.

3.5. Safety

Lorcaserin was generally well tolerated, and the established efficacy of lorcaserin in promoting weight loss was apparent from safety data. After randomization, there were 391 adverse events reported from 135 individuals (55.8% of randomized participants), with 194 events reported from 66 participants (55.9%) receiving lorcaserin and 197 events reported from 69 participants (55.6%) receiving placebo. The most frequent types of AEs were “investigations” (56), “gastrointestinal disorders” (50), “infections and infestations” (45), and “nervous system disorders” (45). Among all reported AEs, the only statistically significant difference between treatments was found for the condition “weight decreased,” which was an AE for 5.9% of participants randomized to lorcaserin and 0% of those randomized to placebo. When changes in vital signs and weights were evaluated over the course of treatment for all randomized participants, lorcaserin was found to significantly reduce both body weight and BMI. For body weight, the Study Day 8 to Study Week 13 change was a mean increase of 2.3 pounds (SD 6.02) for placebo and a mean decrease of 1.9 pounds (SD 7.01) for lorcaserin treatment (p < 0.001). For BMI, the Study Day 8 to Study Week 13 change was a mean increase of 0.3 kg/m² (SD 0.90) for placebo and a mean decrease of 0.3 kg/m² (SD 1.09) for lorcaserin treatment (p < 0.001).

Of 13 total serious adverse events (SAEs) reported after randomization, eight occurred in participants receiving placebo and five occurred in those receiving lorcaserin. There were no deaths and no SAEs categorized as “life-threatening (without hospitalization),” “persistent or significant disability/incapacity,” or “congenital abnormality/birth defect.” For placebo, seven SAEs were categorized as “hospitalization” and one as “other serious (important medical event).” For lorcaserin, three SAEs were categorized as “hospitalization” and two as “life-threatening and hospitalization.” All SAEs were considered not related to study medication or procedures. Comparing physical exam data from screening with data from Study Week 16 or early termination, there were no statistically significant differences between participants randomized to placebo and lorcaserin. There were no BDI-II scores that required early study termination (the protocol specified a score > 28 as the related threshold), and there were no reports of acute suicidality at any rating period for study participants receiving either treatment.

4. Discussion

From the current results, it may be concluded that: 1) Medication adherence and sample size were sufficient to detect a significant pharmacological effect of lorcaserin, as evidenced by detection of significant weight loss in study participants receiving lorcaserin vs. placebo; 2) Lorcaserin failed to facilitate abstinence from cocaine in individuals with CUD; and 3) Lorcaserin failed to facilitate abstinence from alcohol in a subgroup of study participants with comorbid AUD/CUD who were also trying to stop drinking. Regarding attempted abstinence from cocaine, it is noteworthy that the non-significant trend was in the wrong direction (a slightly higher percentage of subjects in the placebo group achieved abstinence from cocaine during the last three weeks of treatment). Considering the lack of efficacy in CUD, data from relevant human laboratory and preclinical studies are discussed below. Finally, the current study is discussed in the context of addressing medication nonadherence in clinical trials.

4.1. Current Results vs. Human Laboratory Findings

In parallel with the current outpatient efficacy trial, two human laboratory studies were conducted to characterize the effects of lorcaserin in cocaine users. In the first (Pirtle et al., 2019), subjects received a single 10 mg dose of lorcaserin or placebo, while in the second (Johns et al., 2021), repeated lorcaserin doses were administered over 11 days, with dose escalation to 10 mg lorcaserin or placebo twice daily. Due to the dosing regimen, the latter study is most relevant to the current results. Consonant with results of the current trial, no suggestion of lorcaserin efficacy was apparent from cocaine vs. money choice procedures; lorcaserin either had no effect on choice (Pirtle et al., 2019) or increased cocaine choice (Johns et al., 2021). Both studies detected significant reductions in reported craving for cocaine following lorcaserin treatment. In the study by Johns et al., lorcaserin treatment was also associated with significant reductions in ratings of “feel irritable.” While reductions in cocaine craving were not predictive of results in the choice procedure or the negative outcome in the current efficacy trial, this does not necessarily argue against the potential usefulness of craving assessments in future studies. A medication with a more selective effect to reduce cocaine craving (i.e., one that does not also increase cocaine self-administration in a choice procedure) may show efficacy in the treatment of CUD. Although lorcaserin’s lack of effect on craving in the current study may appear at odds with the human laboratory findings, the difference could stem from inherent methodological differences in a highly controlled human laboratory study vs. outpatient treatment studies; craving assessed in outpatient studies, which allow access to cocaine outside of clinic visits, may be too variable for the detection of effects observed in the human laboratory setting. In addition, the fact that different craving assessment tools were used in the human laboratory studies and the current efficacy trial may have contributed to the disparate findings.

4.2. Current Results vs. Preclinical Findings

In medication discovery efforts focused on CUD treatment, the potential predictive value of preclinical choice procedures is supported by the results of Banks and Negus (2017); under a choice schedule with concurrent cocaine and food availability, rhesus monkeys showed a strong trend toward increased cocaine choice following lorcaserin treatment, consonant with the human laboratory findings of Johns et al. (2021) and the negative efficacy trial results. Although the shift toward increased cocaine choice in monkeys was not significant, this may have stemmed from the extremely small number of monkeys (2 to 4 per group, depending on dosage comparisons). Because small group size often challenges the interpretation of results from primate studies, recent advances in the utilization of rats for similar choice procedures (e.g., Townsend et al., 2021) are encouraging, and future drug discovery efforts may benefit from more robust choice data.

Despite the negative results in the current study, findings from animal models suggestive of lorcaserin efficacy vs. CUD should not automatically be regarded as “false positives;” it is important to consider if comparable doses were evaluated in preclinical and clinical testing. Grasing et al. (2022) have noted that doses of lorcaserin reported to produce positive findings vs. cocaine in preclinical test procedures are substantially higher, on a mg/kg basis, than the lorcaserin dose approved for clinical use. However, metabolic rates are higher in small vs. large animals, and it is well-established that higher doses of drugs, on a mg/kg basis, are required in laboratory animals vs. humans to achieve similar plasma levels (for a recent review on species comparisons, see Nair and Jacob, 2016). Because of species differences in metabolism, the comparability of preclinical and clinical lorcaserin dosing is best evaluated through a comparison of drug plasma levels. Unfortunately, lorcaserin plasma levels were not measured in most preclinical studies suggestive of efficacy in CUD. The one exception is a rhesus monkey study (Collins et al., 2016) in which an average maximum lorcaserin plasma level of 80 ng/ml was observed in association with decreased cocaine self-administration. In comparison, after 7 days of twice daily 10 mg oral lorcaserin (the same dosing regimen used in the current study), an average minimum plasma concentration of 27.4 ng/ml and an average maximum plasma concentration of 63.1 ng/ml were observed during the clinical development of lorcaserin (FDA Clinical Pharmacology and Biopharmaceutics Review; Application number 022529Orig1s000). This is consonant with median lorcaserin plasma levels (approximately 40 ng/ml) measured at two time points in the current study. Although two study participants, both of whom failed to achieve abstinence from cocaine, had lorcaserin plasma levels exceeding 80 ng/ml at both sampling times (during Study Weeks 3 and 7), such levels were not necessarily maintained around the clock or throughout the study, and it is likely that most participants did not achieve lorcaserin plasma levels comparable to levels associated with decreased cocaine self-administration in rhesus monkeys. It follows that the findings of Collins et al. (2016) should not necessarily be regarded as a “false positive;” the possibility that higher clinical doses of lorcaserin may have shown efficacy in CUD cannot be ruled out based on the present data. Possible lorcaserin efficacy at higher clinical doses is speculative, and higher doses would be expected to pose a greater risk of side effects. In addition, trends in the current study are in the direction of greater abstinence (p = 0.368) and less self-reported cocaine use (p = 0.018) with placebo treatment. However, as noted by Grasing et al. (2022), medications can have bi-phasic effects, and findings at low doses do not necessarily portend findings at high doses.

4.3. Addressing Medication Nonadherence in Clinical Trials

Medication nonadherence can prevent the detection of efficacy in clinical trials, and the enrichment strategy utilized in the current study, as well as post-randomization efforts to promote medication adherence, have been advocated to address the problem (McCann et al., 2015). While the significant impact of lorcaserin on body weight in the current study is evidence of adequate medication adherence, this success cannot be attributed to the enrichment strategy that was utilized. The lorcaserin effect on body weight was detected during the evaluation of safety data from all randomized subjects (including those who did not achieve the prespecified Study Day 8 acetazolamide plasma level for inclusion in the efficacy population). It appears that the post-randomization use of cell phones and AiCure software to remotely monitor dosing events and promote medication adherence among otherwise forgetful study participants may have been sufficient, and the usefulness of a run-in period and related enrichment strategy for future studies could be questioned. On the other hand, the fact that intentional nonadherence was detected in 16% of randomized participants during their first week of cell phone/AiCure software use indicates that a subset of enrolled participants was not being honest with study investigators, and strategies minimize the impact of such individuals in future trials appear warranted. What motivates individuals to pretend to take study drug without actually swallowing the capsules is unclear. In some cases, they may be “professional subjects” who enroll in clinical studies only for financial gain. Regardless of their motivation, excluding such individuals from efficacy analyses should result in lower p-values and greater effect sizes for effective medications. Instead of the current study’s use of acetazolamide as an adherence marker, investigators designing future studies may consider a placebo run-in period with cell phone/AiCure monitoring to identify and exclude those who exhibit intentional nonadherence from inclusion in efficacy analyses.

4.4. Conclusions and Limitations

While the current results demonstrate a clear lack of efficacy for lorcaserin in the treatment of CUD, only one dosing regimen (10 mg twice daily) was evaluated. Although speculative and not supported by the current results, the possibility that higher doses of lorcaserin may have shown efficacy in CUD cannot be ruled out. In addition, the possibility that off-target effects of lorcaserin (effects other than 5-HT_2C receptor activation) may have contributed to the negative findings cannot be ruled out. Further evaluation of lorcaserin in CUD appears unlikely due to its withdrawal from the market (Sharretts et al., 2020). Given the current results, the finding of increased cocaine choice after lorcaserin treatment in a human laboratory study (Johns et al., 2021), and the high cost of IND-enabling development, it appears unlikely that a novel/alternative 5-HT2C agonist will be advanced to clinical development specifically for the CUD indication; however, such a compound may advance for weight loss or other indications. In that case, CUD could be considered as a secondary indication, and limited profiling in preclinical and human laboratory testing procedures could serve as a logical gate keeper. An improved profile (e.g., evidence of decreased cocaine choice or evidence of decreased cocaine craving without increased cocaine choice) could justify advancement to efficacy trials in CUD.

Highlights.

• Lorcaserin (10 mg twice daily) lacked efficacy in cocaine use disorder treatment

• Lorcaserin lacked efficacy in alcohol use disorder treatment in cocaine users

• Significant lorcaserin-induced weight loss indicated adequate medication adherence