Feasibility of Double-Blind Clinical Trials with Oral Diacetylmorphine: A Randomized Controlled Phase II Study in an Inpatient Setting

Heroin dependence is a devastating, chronic psychiatric disorder. The recent increase of opium production has resulted in an increase in heroin use in many nations [1, 2]. Between 12 and 21 million people are opiate abusers [2]. Heroin is detected in 80% of acute drug-related deaths in the EU, and almost half of treatment requests for illicit drug use were for heroin dependence [1]. Crude mortality rate and standardized mortality rate among opioid-dependent users have been recently found to be, respectively, 2.09 per 100 person-years and 14.66 standardized mortality ratios, particularly during out-of-treatment periods [3].

Agonist maintenance treatment has become a treatment of choice for chronic opioid-dependent patients [4] who cannot achieve abstinence. Methadone maintenance therapy (MMT) has shown to reduce heroin use [5, 6, 7], risk of HIV transmission [8], mortality [9] and crime [10, 11]. Further, MMT is associated with global health [12], social functioning improvement [13], increased treatment retention [7, 14] and reduction in criminality [15, 16]. The use of MMT has been proven to be valid for a wide range of patients even irrespective of psychopathology [17].

However, MMTs have limited effectiveness for some patients [7, 18, 19] due to a variety of pharmacological, individual and treatment factors. This has led to the development of alternative forms of maintenance treatments. At present, buprenorphine, a partial µ receptor agonist, although more expensive than methadone [20], has proven its effectiveness [21] and is currently being progressively introduced in both the United States and the EU.

Due to the limitations of methadone and buprenorphine, along with the increases in incidence of infectious diseases associated with heroin use (i.e. HIV, hepatitis B and C, and tuberculosis), various European countries have begun to consider alternative treatments [4], namely oral slow-release morphine [22, 23, 24], and parenteral diacetylmorphine (DAM) [25, 26, 27].

In the 1990s, Swiss Health Authorities started several clinical trials studying DAM as a maintenance therapy for heroin-addicted patients who did not benefit from existing forms of opioid maintenance therapy [28]. Since then, clinical trials with intravenous or intrapulmonary DAM have been carried out in several countries (Switzerland, Netherlands, UK, Spain, Germany and Canada), showing DAM’s effectiveness as an alternative to conventional forms of maintenance treatment [29, 30, 31, 32, 33, 34]. Presently, in the UK, Switzerland, Netherlands, Germany and Denmark, DAM can be prescribed as a maintenance therapy in special cases, although a recent publication shows its effectiveness for patients with no previous experience in maintenance treatments [35]. Nevertheless, in many western countries, parenteral DAM is rejected due to a wide range of factors: higher risk of overdose associated with this route of administration, residual ‘high’ effect in some patients, higher costs of treatment than with MMT, social stigma, etc. Furthermore, because such trials conducted with intravenous DAM were not double blinded, the possibility of biased results cannot be ruled out. Finally, a high rate of adverse events related to the parenteral administration route were found, which can limit the implementation of large-scale DAM maintenance programs [33].

In 1989, the Swiss authorities commissioned an expert report in which oral DAM was described as having clear prolonged opiate effects without the intense and immediate ‘high’ effect reported with intravenous DAM [36]. As a result of this expert report, oral DAM has been tried in observational studies in Switzerland that showed higher retention compared to historical controls. Furthermore, oral DAM was well tolerated, and the rates of serious adverse events (SAEs) were low [37]. Oral DAM has been shown to provide a suitable pharmacokinetic profile for its clinical use in heroin-dependent patients [38], and clinical trials are being prepared in Andalusia (Spain) and Vancouver (Canada). Since oral DAM does not produce an intense and immediate ‘high’ effect and has a suitable pharmacokinetic profile, such oral formulation could be considered as an alternative to intravenous or intrapulmonary DAM.

Our group planned to conduct a pivotal clinical trial with oral DAM avoiding the risk related to intravenous or inhalant administration. Before a phase III study in an outpatient setting using objective outcome measures could be conducted, the Health Authorities required the conduction of a randomized, double-blind, phase II clinical trial involving a small number of participants, in order to generate initial data on DAM’s ability to decrease opioid withdrawal with a twice a day (b.i.d) schedule and to determine whether patients and staff could identify the administered substance that would prevent effective blinding. This will be done by means of measuring the capacity of oral DAM to block opiate withdrawal and clinicians’ ability to distinguish it from morphine and methadone. The present pilot study compared immediate-release oral DAM, slow-release oral morphine and methadone in a b.i.d. regime, in order to study the feasibility of running double-blind clinical trials with oral DAM, allowing future phase III clinical trials, following the standard clinical research procedures for the development of new medications.

Forty-five heroin-dependent patients, aged 18–55, were enrolled in this pilot study between 2004 and 2006. Potential participants were selected from individuals not currently receiving treatment but seeking heroin-dependence treatment. Patients were referred from various outpatient substance abuse clinics in Catalonia, Spain. Participants had to meet criteria for heroin dependence according to DSM-IV-TR [39] and have undergone at least one episode of MMT lasting at least one month during which they received a minimum dose of 60 mg per day. Participants also had to test negative for methadone by urinalysis at the screening visit. Subjects were excluded if they were alcohol, cocaine or benzodiazepine dependent, had a major psychiatric or medical disorder, or were receiving treatment with other drugs that could interact with methadone. Women were excluded if they were pregnant, breastfeeding, or were unwilling to use effective contraception methods. Informed consent was obtained from all participants prior to their participation in the study. Participants received no payment for participation. This clinical trial was approved by the internal review boards of the participating hospitals and the Spanish Drug Regulatory Agency, and was conducted in accordance with the Declaration of Helsinki and subsequent amendments.

A randomized, double-blind, multicenter pilot study was carried out at the Drug Addiction Units of the Departments of Psychiatry of three General Hospitals in Catalonia, Spain. Forty-five participants were randomly assigned to three treatment groups (immediate-release oral DAM, slow-release oral morphine, or methadone), with 15 participants per group (see fig. 1). Participants were admitted to the Drug Addiction Units and received study medication for 10 days. Randomization was performed by a research pharmacist who had no contact with participants or study physicians, using a random numbers table, stratified by center. Thus, each center included 5 patients in each study group. Allocation was concealed by using sequentially numbered, opaque, sealed envelopes.

The main study variables were: (1) mean dose administered to block opioid withdrawal syndrome, (2) presence of subjective and objective opiate withdrawal symptoms, and (3) patient and staff beliefs of which type of opioid was administered. In addition, study retention, heroin craving, symptoms of depression and anxiety, physician and patient’s clinical global impression were determined. Opiate withdrawal symptoms were assessed using the Subjective Opiate Withdrawal Scale and the Objective Opiate Withdrawal Scale [40], craving with a 10-cm visual analogue scale [41], global improvement with the Clinical Global Impression Scale [42], depressive symptoms by the Beck Depression Inventory [43], anxiety symptoms by the State-Trait Anxiety Inventory [44], and severity of addiction with the Addiction Severity Index [45]. All variables were assessed daily. Patient and staff beliefs regarding the type of opiate administered were requested on the last day of treatment.

The study was conducted in an in-hospital setting. Study medication consisted of immediate-release oral DAM, slow-release oral morphine and methadone, for which minimum available dosage oral presentations were 50, 15 and 5 mg, respectively. Slow-release oral DAM was not used because when the study was conducted, the pharmacokinetic studies of this sustained release formulation were not available.

Patients were stabilized with an oral b.i.d. flexible dose regime using standard dose increments of 50, 15 and 5 mg of immediate-release DAM, slow-release morphine and methadone, respectively. All opioids were re-encapsulated with an identical protective coating for each dose (three different colors for each different dose). The protocol of drug administration was as follows: During the 1st day, patients were monitored, and each time they showed objective signs of opiate withdrawal as measured by the Opiate Withdrawal Scale, one standard dose was given. The total amount of medication administered during the first day was divided into two equal doses given in a 12-hour interval on the 2nd day. If the sum of the number of standard doses administered on the 1st day was an odd number, the amount of medication needed could not be divided into two equal doses. In this case, the doses were administered in such a way that the morning dose was higher than the afternoon one while maintaining the same total dosage per day. During this 2nd day, rescue standard doses could be administered if a patient showed objective signs of opiate withdrawal according to the Opiate Withdrawal Scale. The total amount of medication administered (scheduled and rescues) during the 2nd day was subjected to the same procedure to calculate the doses for the 3rd day. The same procedure was used during the remaining days of the trial. Once the trial ended, all study medication was stopped, and patients could choose between opioid detoxification and MMT. All participants received no standardized psychosocial management.

Statistical analysis was performed by intention to treat. The study examined differences within each group of treatment comparing the changes in scores at baseline and at 10 days. The difference between baseline and final values was used to carry out the analysis of treatment efficacy, and Cohen’s d of these differences were calculated with 95% confidence intervals. Last observation carried forward methodology was used for imputing missing values. Sociodemographic data were analyzed using χ² or mean comparisons according the nature of variables. Analysis of variance (ANOVA) was used to test differences between the three treatment groups on differences between baseline and last moment. Hypotheses were two tailed and 95% confidence level was used.

As shown in table 1, the sample was composed of 84.4% men, and mean age was 34.7 (SD = 5.1). 93.3% had at least primary studies and 44.7% were employed. The most common route of heroin use was intravenous (71.4%), followed by intrapulmonary (21.4%) and intranasal (7.1%). Most patients (71.1%) had failed previously one MMT episode, and the remaining 28.9% had failed two or more. Nicotine dependence was present in 85.4% of the sample, HIV infection in 15.6% and HCV infection in 75.6%. No statistically significant differences were found between treatment groups in baseline characteristics. No statistically significant differences in Addiction Severity Index scores were observed between the study groups at baseline.

Patients were stabilized with a mean of 350 mg (SD = 193) of oral immediate-release heroin, 108 mg (SD = 46.2) of oral slow-release morphine and 40 mg (SD = 17.9) of methadone. Table 2 shows the dose of study medication that was administered every study day. The majority of patients (82.2%) completed the study. No group effect was found regarding retention in treatment. However, as it was the principal study outcome, it was further explored. Accordingly, in the context of post-hoc comparisons, morphine treatment was found to be statistically lower than heroin and methadone (z = 2.12, p = 0.034). Nevertheless, this effect disappeared after Bonferroni adjustment for multiple comparisons (0.05/3 = 0.017). Treatment improved the scores in all areas (withdrawal symptoms, craving, mood, anxiety and clinical impression) assessed for each group separately from baseline to study end (table 3). Nevertheless, no statistical differences were found in any of these measures between the three different treatment groups.

Blinding was good as proved by the fact that the sum of correct judgments about the administered opioid was 49% for physicians and 29% for patients. No statistically significant differences were found between proportions of concordance in the two groups. Neither clinicians nor patients reported a statistically different rate of correct judgments than expected from chance.

Regarding safety, 20 different types of SAEs were reported during the 10-day study period (table 4). Only one adverse event (comitial crisis in a patient with history of seizures) was considered serious, and entailed the withdrawal of the patient from the study. Finally, no significant differences were found in the number of adverse events between study groups.

This first phase II study requested by the Health Authorities constitutes the first randomized clinical trial comparing b.i.d. immediate-release oral DAM, slow-release oral morphine and methadone supporting the feasibility of running double-blind clinical trials with oral DAM in opioid-dependent patients. This is supported by first, the finding that the administration of immediate-release oral DAM could effectively reduce the opioid withdrawal syndrome, decrease heroin craving and improve depressive and anxiety symptoms similar to slow-release oral morphine and methadone; second, because neither patients nor clinicians seem to be able to identify which was the administered drug.

The dose needed to block opiate withdrawal with immediate-release oral DAM, slow-release oral morphine and methadone was 350, 108 and 40 mg/day, respectively. The oral DAM to methadone ratio in our study was approximately 9, which is quite similar to the bioequivalence found in the Swiss heroin-assisted treatment studies [37, 46]. Regarding morphine stabilization dose, it was expected to be higher because morphine bioavailability appears to be higher with oral DAM than with oral morphine [47]. Other studies have found that the slow-release morphine to methadone ratio was 4.5 [48], while in our study it was of 2.7. It is likely that the slow-release morphine dose in our study was too low, thereby explaining the lower retention in treatment amongst patients randomized to slow-release morphine compared to methadone or DAM. Given this limitation, the bioequivalence between oral methadone and oral DAM must be seen as more convincing than those between methadone and morphine.

It is notable that patients were stabilized with a b.i.d. schedule of the three study medications. This finding, striking for oral DAM, is consistent with that of recent studies showing that heroin-dependent patients are able to be maintained with an average of 2 intrapulmonary or intravenous DAM administrations [30, 33]. It can be explained because, as shown in pharmacokinetic studies [38, 49], DAM metabolization yields the production of active metabolites including 6-mono-acetyl-morphine, morphine, and morphine-6-glucoronide with long half-life. The existence of these metabolites could explain why oral and parenteral DAM is able to block opiate withdrawal syndrome with a b.i.d. administration, which is a suitable regime for a maintenance treatment. As slow-release oral DAM is being developed at present [50], it may be possible in a near future to reliably stabilize heroin addicts with a once-a-day regime of slow-release oral DAM.

Patients were stabilized with 40 mg of methadone. It must be noted that patients were admitted to a detoxification unit, and in this setting it is likely that the exposure to cues is lower thereby decreasing craving, anxiety and withdrawal symptoms and the need for methadone. Besides, it must also be stressed that this dose is lower than the one used in methadone maintenance programs for which doses above 60 mg/day are recommended. Higher doses are needed in methadone maintenance programs because their aim is not only to prevent heroin withdrawal but also to cause a narcotic blockade [51]. Given that our study has shown that the equivalent dose to treat the opioid withdrawal syndrome with oral DAM and methadone was 350 and 40 mg/day, respectively, and considering that in methadone maintenance programs patients are stabilized with doses ranging between 60 and 125 mg/day, we reckon that the dose of oral DAM needed in maintenance programs will be between 540 and 1,100 mg/day.

This study also shows that oral DAM has a good safety profile. No serious physical complaints were reported, and although high tolerance has been documented for DAM, this study shows that when it is administered under controlled clinical conditions, even in a flexible regime, it has low risk of physical tolerance. Finally, no overdoses were registered. In intravenous DAM studies, overdose was the most hazardous SAE; nevertheless, it mostly occurred in heroin administration clinics and were treated and resolved without sequelae or hospital admission [33].

Some study limitations must be stressed. The sample size was small and, as a consequence, the power of statistical results was also low. In this study, inferential statistics have only an exploratory purpose. However, despite the large confidence intervals observed, also a consequence of the sample size, some size effects obtained comparing basal and final data can be interpreted as large [52]. Under our point of view, it is important to remark that some percentual differences between groups should be perceived as clinically relevant (for example different retention percentages). A placebo arm was not included in order to provide an active treatment to all study participants according to the European regulation. Patients were treated in an inpatient regime in the context of a short follow-up, with reduced exposition to environmental stimulus that may lead to drug craving. This artificial setting was needed to monitor patient safety and warrant double-blind conditions. Another issue that limits the external validity of our study is that we excluded patients with comorbid benzodiazepine, cocaine and alcohol dependence, which are rather prevalent amongst heroin-dependent patients. Besides, our study focused on clinical outcome measures, but we did not fully investigate the behavioral effects of oral DAM or its impact on physiological variables, which should be studied in the future to characterize the clinical pharmacology of this drug. Although administering methadone twice a day limits the external validity of our study, dividing methadone dose is usually recommended in patients that can be very fast methadone metabolizers or in patients for whom once-a-day methadone administration cannot prevent heroin craving for 24 h [53, 54]. Furthermore, using once-a-day oral DAM was not possible given that, by the time this study was designed and approved, the slow-release formulation had not completed the phase I studies [50]. Therefore, a twice-a-day administration of all study medication was needed to allow double-blind conditions. Given that slow-release DAM formulation is available nowadays, caution is recommended when extrapolating our findings to this formulation. The present study was not aimed at investigating oral DAM efficacy with respect to morphine and methadone, or proposing it as a maintenance treatment but at investigating the feasibility of double-blind clinical trials with oral DAM. Our findings suggest that the oral formulation of DAM could be effective and safe, allowing double-blind clinical trials in future research with DAM as a maintenance treatment in outpatient resources.

Further research is warranted using double-blind, randomized clinical trials with oral DAM in the process of introducing oral DAM as a medication in heroin addiction.

The Catalan Oral Heroin Study Group is composed of the following researchers (in alphabetical order): Alvarós Costa J., Arroyo Zamora C., Batlle F., Blasi López A., Bosch Munsó R., Bruguera Cortada E., Casas Brugué M., Castells Cervelló X., Collazos Sánchez F., Colom Farran J., del Río Meyer M., Egido Polo A., Eiroá Orosa F.J., Gonzalvo Cirac B., Majó Roca X., Mir Duñach J., Pérez de los Cobos Peris J.C., Pinet Ogué C., Ramírez Vila M., Ramos Quiroga J.A., Ribalta Corona E., Roncero Alonso C., Sanz Ruíz R., Tascón López Y., Tejero Pociello A., Trujols Albet J., Valero Ventura S.

We would like to thank the patients and staff who participated in the study, as well as the Department of Health of the Government of Catalonia (Generalitat de Catalunya) for funding the study.

This trial was commissioned and funded by a working group of the Government of Catalonia, Spain. The first author is a member of this institution and had a further role in the study development. The remaining authors have no conflicts of interest.

Names of members of the Catalan Oral Heroin Study Group are listed in the appendix.