R E V IE W A R T I C L E
`
`International Journal of Neuropsychopharmacology (2005), 8, 267–280. Copyright f 2005 CINP
`doi :10.1017/S1461145704004997
`
`Naltrexone for the treatment of alcoholism : a
`meta-analysis of randomized controlled trials
`
`Manit Srisurapanont and Ngamwong Jarusuraisin
`
`Department of Psychiatry, Faculty of Medicine, Chiang Mai University, Amphur Muang, Chiang Mai 50200, Thailand
`
`Abstract
`
`Many trials of naltrexone have been carried out in alcohol-dependent patients. This paper is aimed to
`systematically review its benefits, adverse effects, and discontinuation of treatment. We assessed and
`extracted the data of double-blind, randomized controlled trials (RCTs) comparing naltrexone with pla-
`cebo or other treatment in people with alcoholism. Two primary outcomes were subjects who relapsed
`(including heavy drinking) and those who returned to drinking. Secondary outcomes were time to first
`drink, drinking days, number of standard drinks for a defined period, and craving. All outcomes were
`reported for the short, medium, and long term. Five common adverse effects and dropout rates in short-
`term treatment were also examined. A total of 2861 subjects in 24 RCTs presented in 32 papers were
`included. For short-term treatment, naltrexone significantly decreased relapses [relative risk (RR) 0.64,
`95 % confidence interval (CI) 0.51–0.82], but not return to drinking (RR 0.91, 95 % CI 0.81–1.02). Short-term
`treatment of naltrexone significantly increased nausea, dizziness, and fatigue in comparison to placebo
`[RRs (95 % CIs) 2.14 (1.61–2.83), 2.09 (1.28–3.39), and 1.35 (1.04–1.75)]. Naltrexone administration did not
`significantly diminish short-term discontinuation of treatment (RR 0.85, 95 % CI 0.70–1.01). Naltrexone
`should be accepted as a short-term treatment for alcoholism. As yet, we do not know the appropriate
`duration of treatment continuation in an alcohol-dependent patient who responds to short-term nal-
`trexone administration. To ensure that the real-world treatment is as effective as the research findings, a
`form of psychosocial therapy should be concomitantly given to all alcohol-dependent patients receiving
`naltrexone administration.
`
`Received 9 March 2004; Reviewed 25 July 2004; Revised 20 September 2004; Accepted 5 October 2004
`
`Key words: Alcohol, alcohol use disorder, meta-analysis, naltrexone, opioid antagonists.
`
`Introduction
`
`Alcoholism (alcohol dependence and abuse) is a com-
`mon health problem. Its health, social, and economic
`consequences are usually devastating. Although many
`individuals do achieve long-term sobriety with ther-
`apy, others continue to relapse and deteriorate despite
`multiple courses of treatment.
`Due to the limited success of psychosocial treatment
`programmes (Berglund et al., 2003), several pharma-
`cological agents have been studied in people with
`alcoholism. Disulfiram has only limited clinical utility
`for those with high motivation, good health, and good
`cooperation. Even in highly motivated individuals,
`disulfiram may partially improve alcohol-dependent
`patients in some respects, e.g. drinking frequency and
`
`Address for correspondence: M. Srisurapanont, M.D., Professor of
`Psychiatry, Department of Psychiatry, Faculty of Medicine, Chiang
`Mai University, Amphur Muang, Chiang Mai 50200, Thailand.
`Tel.: (66-53) 945422 Fax: (66-53) 945426 or (66-53) 217144
`E-mail : msrisura@mail.med.cmu.ac.th
`
`amount of alcohol consumption (Garbutt et al., 1999).
`While the results of some studies showed that lithium
`reduced drinking in alcohol-dependent patients with
`mood disorders (Fawcett et al., 1984; Merry et al.,
`1976), a randomized controlled trial (RCT) failed to
`demonstrate any benefit of this drug in either de-
`pressed or non-depressed patients (Dorus et al., 1989).
`The efficacy of selective serotonin reuptake inhibitors
`(SSRIs) remains to be tested in placebo-controlled,
`randomized trials with large sample sizes. Acam-
`prosate is a promising medication, but it has not been
`widely approved (Overman et al., 2003).
`The interaction of alcohol and the opioid system
`has not been fully understood because the interaction
`between ethanol and the mechanistic processes as-
`sociated with opioid production, secretion, and bind-
`ing is relatively complex (Herz, 1997). However, most
`animal studies suggest that the competitive binding
`of opioid antagonists to opioid receptors may have
`the propensity to diminish the rewarding effects by
`decreasing the dopamine released in the mesolimbic
`
`Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/267/755449
`by guest
`on 03 April 2018
`
`AMN1033
`IPR of Patent No. 7,919,499
`
`

`

`268 M. Srisurapanont and N. Jarusuraisin
`
`1997).
`and Zieglgansberger,
`pathway (Spanagel
`Due to these findings, many clinical
`trials have
`investigated both the harm and benefit of opioid
`antagonists
`in people with alcoholism. Among
`them, naloxone has a very short half-life and, there-
`fore, very limited clinical utility. Nalmefene has
`been examined in at least two RCTs, but is not yet
`approved (Mason et al., 1994, 1999). Naltrexone is the
`agent studied most, and it has been approved for
`the treatment of alcoholism in several countries. We,
`therefore, proposed to systematically review its ben-
`efits, adverse effects, and discontinuation of treatment
`for alcoholism.
`
`Methods
`
`Inclusion criteria
`
`This review included all double-blind RCTs that
`compared naltrexone with placebo and/or other
`treatment in people with alcohol dependence or abuse.
`Subjects with alcohol dependence or abuse had to be
`diagnosed by the use of clearly defined diagnostic
`systems, e.g. DSM-III, DSM-III-R, DSM-IV, and ICD-
`10. All data of the patients were included regardless of
`age, gender, nationality, comorbidity, and hospital-
`ization status. No language or publication restriction
`was applied.
`
`Outcomes
`
`Two primary outcomes were subjects who relapsed
`(including heavy drinking) and those who returned to
`drinking. We assigned these outcomes as being pri-
`marily beneficial because they are of concern for most
`researchers, clinicians, and patients. In addition, as
`dichotomous data,
`they can be interpreted and
`understood easily. Secondary outcomes were time to
`first drink, drinking days (in % or number), number of
`standard drinks for a defined period (e.g. week, study
`duration, and drinking day), and craving. As alcohol-
`ism is a chronic disease with a high relapse rate, all
`outcomes were reported for the short- (up to and in-
`cluding 12 wk), medium- (more than 12 wk and up to
`and including 12 months) and long-term (more than
`12 months). For any outcome assessed more than once
`in a particular term, we extracted results of the longest
`duration only.
`The analysis also included short-term adverse
`effects and discontinuation of treatment because of
`their importance in this treatment period. The five
`new-onset adverse clinical events most frequently
`found in the largest comparison, but not in a random-
`ized trial of naltrexone, were examined (Croop et al.,
`
`1997). They comprised nausea, headache, dizziness,
`fatigue, and nervousness.
`
`Locating trials
`
`To look into the comparative trials, we used four
`electronic databases (MEDLINE, EMBASE, CINHAL, and the
`Cochrane Central Register of Controlled Trails) in
`September 2003. The terms used to identify articles
`were [naltrexone or narcotic antagonist or opioid an-
`tagonist] and [alcohol or ethanol]. To identify further
`reports, we checked the references of this preliminary
`list of selected studies along with references of other
`relevant review papers. Du Pont Pharmaceutical, the
`only producer of naltrexone, was contacted for infor-
`mation about unpublished trials. Of all papers found,
`only RCTs comparing naltrexone with placebo and/or
`other treatment in people with alcohol dependence or
`abuse were included.
`
`Quality assessment of included trials and data
`extraction
`
`We assessed the methodological quality of each trial
`included by examining its randomization (Schulz et al.,
`1995). Trial characteristics and the data relevant to the
`reviewed outcomes were extracted and recorded in a
`data record form. If the data in each study were rel-
`evant to two primary, four secondary, five adverse, or
`dropout outcomes presented in figures, they would
`be extracted. Because treatment or the controlled
`group of some studies was divided into a number of
`subgroups (mostly due to the difference of concomi-
`tant treatment), a continuous outcome of these sub-
`groups could not be combined as an outcome of the
`whole group. In this case, the outcome of the subgroup
`receiving the most rigorous treatment, e.g. highest
`dose of drug treatment and most intensive psycho-
`therapy, was used to represent the group.
`
`Statistical analysis
`
`A relative risk (RR) with 95 % confidence interval (CI)
`was an effect measure used for dichotomous out-
`comes. A number-needed-to-treatment
`(NNT) and
`number-needed-to-harm (NNH) were also computed
`for outcomes, with a significantly different benefit and
`more adverse effects respectively.
`A weighted mean difference (WMD) with 95 % CI
`was used to synthesize the outcomes of time to first
`drink and drinking days. The outcomes relevant to
`standard drinks and craving were likely to be
`measured by various units (e.g. standard drinks per
`week, per month, or per drinking day) or scales (e.g. a
`
`Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/267/755449
`by guest
`on 03 April 2018
`
`AMN1033
`IPR of Patent No. 7,919,499
`
`

`

`visual analogue scale or specially designed rating
`scales for craving) and, therefore, combined by using
`standardized mean differences (SMDs) with 95 % CIs.
`The data synthesis was done on an intention-to-
`treat basis. Because the results obtained from a ran-
`dom effect model of data synthesis might be more
`generalized, this model was used throughout the
`review for calculating RRs, WMDs, and SMDs
`(DerSimonian and Laird, 1986). The inconsistency of
`data was examined by looking at the graphical display
`of the results and also by using an I-square (I2)
`(Higgins et al., 2003). As recommended, an I2 of 75 % or
`more indicates high inconsistency of data. To test the
`robustness of the results, relative to features of the
`primary studies (those carried out only in individuals
`with alcoholism), a sensitivity analysis was performed
`to examine the results including and excluding the
`studies conducted in alcohol-dependent patients with
`comorbidity. The statistical analysis was performed by
`the use of Review Manager 4.2 (Cochrane Collabor-
`ation, Oxford, England).
`
`Results
`
`Study inclusion and characteristics
`
`Our searches found 28 RCTs of naltrexone in people
`with alcohol dependence or abuse. We excluded four
`studies not using the double-blindness design (Lee
`et al., 2001 ; Rubio et al., 2001, 2002) and a clearly
`defined diagnostic system (Huang et al., 2002).
`Twenty-four papers were identified as original articles
`presenting main findings of 24 RCTs (see Table 1).
`Eight papers were classified as duplicated reports
`presenting only additional data or representing the
`data (Anton et al., 2001; Jaffe et al., 1996; Modesto-
`Lowe et al., 1997; O’Malley et al., 1996a,b; Oslin et al.,
`1997a; Rohsenow et al., 2000; Volpicelli et al., 1995) of
`six original papers (Anton et al., 1999; Hersh et al.,
`1998; Monti et al., 2001; O’Malley et al., 1992; Oslin
`et al., 1997b; Volpicelli et al., 1992). Although some data
`presented in duplicated papers were also included in
`this review, the following parts referenced only the
`original articles to cause less confusion. It was noted
`that a nested sequence of three trials presented in a
`paper was considered as a trial, since the subjects in all
`three trials were the same (O’Malley et al., 2003). In this
`study, the investigators started by conducting a 10-wk
`RCT comparison of naltrexone+primary care manage-
`ment (PCM) and naltrexone+cognitive–behavioural
`therapy (CBT) and placebo, followed by two 24-wk
`RCTs in those responding to first trial treatment.
`The total number of subjects included in this review
`was 2861. Of those, 1709 were assigned to receive
`
`Naltrexone for alcoholism 269
`
`naltrexone treatment. All trials diagnosed the subjects
`by using DSM-III-R or DSM-IV. Apart from 82 patients
`with dual alcohol and cocaine dependence or abuse in
`two trials (Carroll et al., 1993; Hersh et al., 1998) and
`six individuals with alcohol abuse in a study (Chick et
`al., 2000), all other subjects were alcohol-dependent
`patients. All were aged 18 yr or more. The sample
`sizes of most trials were between 10–99 in each arm.
`While one study had only nine subjects in each arm
`(Carroll et al., 1993), two trials had 100–200 subjects in
`each arm (Guardia et al., 2002; Krystal et al., 2001).
`Only a few studies stated clearly the locations or
`countries in which the studies were carried out.
`According to the investigators’ affiliations,
`it was
`understood that 14 studies were conducted in North
`American (including Puerto Rico), seven in Europe,
`one in Asia, and two in Australia.
`Of 24 RCTs, only five provided the details of tech-
`niques used for randomization (Balldin et al., 2003;
`Kiefer et al., 2003; Latt et al., 2002; O’Malley et al.,
`2003; Volpicelli et al., 1997). With regard to study
`duration, only eight trials were carried out for longer
`than 12 wk (Anton et al., 1999; Balldin et al., 2003;
`Heinala et al., 2001; Knox and Donovan, 1999;
`Landabaso et al., 1999; Monti et al., 2001; O’Malley
`et al., 1992, 2003).
`Of the 24 RCTs included in this review, only two
`did not have a placebo arm (Carroll et al., 1993;
`Landabaso et al., 1999). Apart
`from three trials
`(Galarza et al., 1997; Landabaso et al., 1999; Oslin et al.,
`1997b), naltrexone in all studies was administered
`daily at a dose of 50 mg/d. It was noted that one trial
`gave naltrexone continuously in the first 3 months and
`as targeted medication (only when alcohol consump-
`tion was likely) in the last 3 months (Heinala et al.,
`2001). All trials clearly defined the psychosocial treat-
`ment concomitantly given with naltrexone. While a
`trial gave a simple psychosocial
`treatment called
`medical advice (Latt et al., 2002), each of the remainder
`gave at least one form of intensive psychosocial treat-
`ment, e.g. coping skills and CBT.
`Regarding the outcome measures, those related to
`drinking behaviour were reported in figures in most
`trials. Seven (Ahmadi and Ahmadi, 2002 ; Chick et al.,
`2000; Kiefer et al., 2003; Kranzler et al., 2000; Monti
`et al., 2001; Morris et al., 2001 ; O’Malley et al., 1992)
`and three trials (Hersh et al., 1998; Kiefer et al., 2003;
`Morris et al., 2001) presented the outcomes of subjects
`with relapses and return to drinking in graphs and p
`values respectively. Therefore, these data could not be
`included in the analysis. Functional outcomes were
`presented in only one trial (Knox and Donovan, 1999).
`No trial reported the outcome of patient satisfaction,
`
`Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/267/755449
`by guest
`on 03 April 2018
`
`AMN1033
`IPR of Patent No. 7,919,499
`
`

`

`270 M. Srisurapanont and N. Jarusuraisin
`
`Table 1. Characteristics of RCTs comparing naltrexone with placebo or other treatment in people with alcoholism
`
`Authors
`
`Methods
`
`Subjects
`
`Interventions
`
`O’Malley
`et al. (1992)
`
`Double-blind, placebo-controlled,
`12-wk study with 6-month
`follow-up after the completion
`of 12-wk treatment in the USA
`
`Outpatients with alcohol
`dependence (DSM-III-R),
`18–68 yr old
`
`Volpicelli
`et al. (1992)
`
`Double-blind, placebo-controlled,
`12-wk study in the USA
`
`Carroll et al.
`(1993)
`
`Double-blind, 12-wk study in the
`USA
`
`Galarza
`et al. (1997)
`
`Double-blind, placebo-controlled,
`4-wk study in Puerto Rico
`
`Outpatients with alcohol
`dependence (DSM-III-R);
`21–65 yr old
`Outpatients with dual alcohol and
`cocaine dependence or abuse
`(DSM-III-R), no age specified
`
`Outpatients with alcohol
`dependence (DSM-IV); 21–75 yr
`old; male only
`
`Oslin et al.
`(1997b)
`
`Double-blind, placebo-controlled,
`12-wk study in the USA
`
`Patients with alcohol dependence
`(DSM-III-R); 50–70 yr old
`
`Volpicelli
`et al. (1997)
`
`Double-blind, placebo-controlled,
`12-wk study in the USA
`
`Hersh et al.
`(1998)
`
`Double-blind, placebo-controlled,
`8-wk study in the USA
`
`Outpatients with alcohol
`dependence (DSM-III-R);
`21–65 yr old
`
`Patients with dual alcohol and
`cocaine dependence or abuse
`(DSM-III-R); 18–45 yr old
`
`Anton et al.
`(1999)
`
`Knox and
`Donovan
`(1999)
`
`Double-blind, placebo-controlled,
`12-wk study with a 14-wk
`follow-up after the completion
`of 12 wk treatment in the USA
`
`Outpatients with alcohol
`dependence (DSM-III-R);
`21–65 yr old
`
`Double-blind, placebo-controlled,
`6-month study in the USA
`
`Patients with alcohol dependence
`(DSM-IV); 18–65 yr old
`
`Landabaso
`et al. (1999)
`
`Double-blind, 24-month study in
`Spain
`
`Patients with alcohol dependence
`(DSM-IV); mean age=30.6 yr
`old
`
`Chick et al.
`(2000)
`
`Double-blind, placebo-controlled,
`multicentre, 12-wk study in the
`UK
`
`Outpatients with alcohol
`dependence or abuse (DSM-III-
`R); 18–65 yr old
`
`50 mg/d naltrexone+coping
`skills (n=29) vs. 50 mg/d
`naltrexone+supportive therapy
`(n=23) vs. placebo+coping
`skills (n=25) vs. placebo+
`supportive therapy (n=27);
`no intervention given during
`follow-up period
`50 mg/d naltrexone (n=35) vs.
`placebo (n=35); all received
`rehabilitation treatment
`50 mg/d naltrexone (n=9) vs.
`disulfiram (n=9); all received
`weekly individual
`psychotherapy
`Naltrexone (undefined dose)
`(n=10) vs. placebo (n=10); all
`received regular psychosocial
`treatment
`100 mg/d naltrexone on
`Monday/Wednesday and
`150 mg naltrexone on Friday
`(n=21) vs. placebo (n=23); all
`received weekly group therapy
`and bi-weekly case management
`50 mg/d naltrexone (n=48) vs.
`placebo (n=49); all received
`individual psychotherapy and
`counselling
`50 mg/d naltrexone (n=31) vs.
`placebo (n=33); all received
`individual relapse prevention
`psychotherapy
`50 mg/d naltrexone (n=68)
`vs. placebo (n=63); all
`received weekly CBT; no
`intervention given during
`follow-up period
`50 mg/d naltrexone (n=31) vs.
`placebo (n=32); all received
`21 d in-patient chemical
`dependency treatment followed
`by a 6-month outpatient
`programme
`25 mg/d naltrexone+an aversive
`agent (n=15) vs. an aversive
`agent alone (n=15); 6-month
`naltrexone treatment; 12-month
`aversive therapy
`50 mg/d naltrexone (n=90) vs.
`placebo (n=85); all received the
`usual psychosocial treatment
`programme
`
`Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/267/755449
`by guest
`on 03 April 2018
`
`AMN1033
`IPR of Patent No. 7,919,499
`
`

`

`Table 1 (cont.)
`
`Authors
`
`Methods
`
`Subjects
`
`Interventions
`
`Naltrexone for alcoholism 271
`
`Johnson
`et al. (2000)
`
`Double-blind, placebo-controlled,
`8-wk study in the USA
`
`Kranzler
`et al. (2000)
`
`Double-blind, placebo-controlled,
`11-wk study in the USA
`
`Outpatients with alcohol
`dependence (DSM-IV) with age
`of alcoholic onset <25 yr old,
`age 25–65 yr old
`Patients with alcohol dependence
`(DSM-III-R); 18–60 yr old
`
`Heinala et al.
`(2001)
`
`Double-blind, placebo-controlled,
`12-wk regular treatment study
`with 20-wk targeted medication
`treatment in Finland
`
`Outpatients with alcohol
`dependence (DSM-IV); 21–65 yr
`old
`
`Krystal et al.
`(2001)
`
`Double-blind, placebo-controlled,
`12-wk study in the USA
`
`Outpatients with alcohol
`dependence (DSM-IV), >18 yr
`old
`
`Monti et al.
`(2001)
`
`Double-blind, placebo-controlled,
`12-month study in the USA
`
`Outpatients with alcohol
`dependence (DSM-IV), mean
`age of 39.2 yr old
`
`Morris et al.
`(2001)
`
`Double-blind, placebo-controlled,
`12-wk study in Australia
`
`Outpatients alcohol dependence
`(DSM-III-R), 18–65 yr old
`
`Ahmadi and
`Ahmadi (2002)
`
`Double-blind, placebo-controlled,
`12-wk study in Iran
`
`Outpatients with alcohol
`dependence (DSM-IV), 23–56 yr
`old; male only
`
`Gastpar et al.
`(2002)
`
`Double-blind, placebo-controlled,
`multicentre, 12-wk study in
`Germany
`
`Out- and in-patients with alcohol
`dependence (DSM-III-R); mean
`age (SD)=42.7 (9.7) yr
`
`Guardia et al.
`(2002)
`
`Latt et al.
`(2002)
`
`Balldin et al.
`(2003)
`
`Double-blind, placebo-controlled,
`multicentre, 12-wk study in
`Spain
`Double-blind, placebo-controlled,
`multicentre, 12-wk study in
`Australia
`Double-blind, placebo-controlled,
`multicentre, 6-month study in
`Sweden
`
`Outpatients with alcohol
`dependence (DSM-IV); 18–60 yr
`old
`Patients with alcohol dependence
`(DSM-IV); 18–70 yr old
`
`Outpatients with alcohol
`dependence (DSM-IV); 18–65 yr
`old
`
`50 mg/dnaltrexone+ondansetron
`4 mg/kg (n=10) vs. placebo
`(n=10); all participants received
`group CBT
`50 mg/d naltrexone (n=61) vs.
`400–600 mg/d nefazodone
`(n=59) vs. placebo (n=63); all
`received coping skill training
`First 12 wk, 50 mg/d naltrexone
`(n=63) vs. placebo (n=58)
`+either cognitive coping skill
`(n=67) or supportive
`psychotherapy (n=54); for
`20 wk duration, naltrexone
`(undefined dose) given only
`when alcohol drinking was
`likely (targeted medication)
`50 mg/d naltrexone for 12
`months (n=209) vs. 50 mg/d
`naltrexone for 3 months
`followed by a placebo for 9
`months (n=209); all received
`12-step facilitation counselling
`50 mg/d naltrexone (n=64) vs.
`placebo (n=64) for 12 wk; all
`received 1–2 wk of CET+CST or
`ERC
`50 mg/d naltrexone (n=55) vs.
`placebo (n=56); all received
`group psychoeducation and
`social support
`50 mg/d naltrexone (n=58) vs.
`placebo (n=58); all received
`individual counselling and
`relapse prevention programme
`50 mg/d naltrexone (n=87) vs.
`placebo (n=84); all received
`psychosocial alcoholic treatment
`programme
`50 mg/d naltrexone (n=101) vs.
`placebo (n=101); all received
`rehabilitation treatment
`50 mg/d naltrexone (n=56) vs.
`placebo (n=51); all received
`medical advice
`50 mg/d naltrexone+CBT
`(n=25) vs. 50 mg/d
`naltrexone+supportive therapy
`(n=31) vs. placebo+CBT
`(n=30) vs. placebo+supportive
`therapy (n=32)
`
`[continued overleaf
`
`Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/267/755449
`by guest
`on 03 April 2018
`
`AMN1033
`IPR of Patent No. 7,919,499
`
`

`

`272 M. Srisurapanont and N. Jarusuraisin
`
`Table 1 (cont.)
`
`Authors
`
`Methods
`
`Subjects
`
`Interventions
`
`Keifer et al.
`(2003)
`
`Double-blind, placebo-controlled,
`multicentre, 12-wk study in
`Germany
`
`Patients with alcohol dependence
`(DSM-IV); 18–65 yr old
`
`O’Malley
`et al. (2003)
`
`Open-label, 10-wk study followed
`by 24-wk, randomized, double-
`blind, placebo-controlled study
`in treatment responders in the
`USA
`
`Outpatients with alcohol
`dependence (DSM-III-R);
`18–65 yr old
`
`50 mg/d naltrexone (n=40) vs.
`1998 mg/d acamprosate (n=40)
`vs. naltrexone+acamprosate
`(n=40) vs. placebo (n=40); all
`received group CBT
`First 10 wk, 50 mg/d
`naltrexone+PCM (n=93) vs.
`50 mg/d naltrexone+CBT
`(n=97); 24 wk follow-up (in
`treatment responders only),
`naltrexone+PCM (n=26) vs.
`placebo+PCM (n=27) and
`naltrexone+CBT (n=30) vs.
`placebo+CBT (n=30)
`
`CBT, cognitive–behavioural therapy; CDP, controlled drinking programme ; CET, cue-exposure treatment; CST, communication
`skills training; ERC, education and relaxation control; PCM, primary care management.
`
`Study or subcategory
`(first-named author)
`
`Naltrexone
`n/N
`
`01 Number of subjects with relapses
`Volpicelli (1992)
`Oslin (1997b)
`Volpicelli (1997)
`Anton (1999)
`Gastpar (2002)
`Guardia (2002)
`Latt (2002)
`
`8/35
`3/23
`17/48
`26/68
`34/84
`8/101
`19/56
`
`Subtotal (95% CI)
`415
`Total events: 115 (naltrexone), 173 (placebo)
`Test for heterogeneity: χ2 = 8.86, df = 6 (p = 0.18), I2 = 32.3%
`Test for overall effect: Z = 3.65 (p = 0.0003)
`
`02 Number of subjects with return to drinking
`30/52
`O'Malley (1992)
`16/35
`Volpicelli (1992)
`Oslin (1997b)
`6/21
`27/48
`Volpicelli (1997)
`36/68
`Anton (1999)
`74/90
`Chick (2000)
`43/61
`Kranzler (2000)
`41/84
`Gastpar (2002)
`
`459
`Subtotal (95% CI)
`Total events: 273 (naltrexone), 299 (placebo)
`Test for heterogeneity: χ2 = 8.91, df = 7 (p = 0.26), I2 = 21.4%
`Test for overall effect: Z = 1.64 (p = 0.10)
`
`Placebo
`n/N
`
`19/35
`8/21
`26/49
`38/63
`36/87
`19/101
`27/51
`
`407
`
`42/52
`20/35
`8/23
`32/49
`42/63
`69/85
`41/63
`45/87
`
`457
`
`RR (random)
`95% CI
`
`Weight
`%
`
`9.72
`3.73
`16.74
`22.18
`22.31
`7.82
`17.50
`
`100.00
`
`14.35
`8.40
`3.18
`12.36
`13.74
`19.44
`15.33
`13.20
`
`100.00
`
`RR (random)
`95% CI
`
`0.42
`0.34
`0.67
`0.63
`0.98
`0.42
`0.64
`
`0.64
`
`0.71
`0.80
`0.82
`0.86
`0.79
`1.01
`1.08
`0.94
`
`0.91
`
`(0.21–0.83)
`(0.10–1.12)
`(0.42–1.06)
`(0.44–0.91)
`(0.68–1.40)
`(0.19–0.92)
`(0.41–1.00)
`
`(0.51–0.82)
`
`(0.55–0.93)
`(0.50–1.27)
`(0.34–1.98)
`(0.62–1.19)
`(0.60–1.06)
`(0.88–1.17)
`(0.85–1.38)
`(0.70–1.27)
`
`(0.81–1.02)
`
`Figure 1. Short-term comparisons between naltrexone and placebo in respect of subjects with relapses and those who return to
`drinking.
`
`0.1
`0.2
`0.5
`Favours naltrexone
`
`1
`
`2
`5
`Favours placebo
`
`10
`
`health-related quality of life, cost, or mortality rates.
`Apart from three studies (Johnson et al., 2000; Heinala
`et al., 2001; Monti et al., 2001), all placebo-controlled
`trials with a short-term study duration reported the
`dropout rates. Eleven placebo-controlled trials, with a
`short-term study duration, reported at least one of five
`interestingly adverse effects (Ahmadi and Ahmadi,
`2002; Chick et al., 2000; Gastpar et al., 2002; Guardia
`et al., 2002; Hersh et al., 1998; Kranzler et al., 2000;
`
`Krystal et al., 2001; Latt et al., 2002; Monti et al., 2001 ;
`O’Malley et al., 1992 ; Oslin et al., 1997b).
`Two RCTs were carried out in patients with dual
`cocaine and alcohol dependence or abuse (Carroll
`et al., 1993; Hersh et al., 1998). Both of these trials
`reported only secondary outcomes. However, accord-
`ing to the treatment comparison, only the results
`of short-term, placebo-controlled RCTs could be com-
`bined with others (Hersh et al., 1998). We, therefore,
`
`Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/267/755449
`by guest
`on 03 April 2018
`
`AMN1033
`IPR of Patent No. 7,919,499
`
`

`

`performed sensitivity analyses of the short-term out-
`comes that compared naltrexone with placebo with
`regard to time to first drink, drinking days, standard
`drinks, headache, dizziness, fatigue, and discontinu-
`ation of treatment. It was found that the inclusion
`or exclusion of the outcomes obtained from Hersh’s
`study were not much different in respect of data in-
`consistency or significant/non-significant differences
`of outcome measures. The data of that study were,
`therefore, included in the following presentation.
`
`Naltrexone vs. placebo
`
`Figure 1 shows short-term comparisons between
`naltrexone and placebo in respect of subjects who
`relapsed and those who returned to drinking. In the
`first 12 wk of
`treatment, naltrexone significantly
`diminished the risk of relapse (RR 0.64, 95 % CI
`0.51–0.82), but not the risk of return to drinking (RR
`0.91, 95 % CI 0.81–1.02). No significant difference be-
`tween groups and no significant heterogeneity of data
`were found in any secondary outcomes of short-term
`treatment (see Table 2).
`Of five adverse effects of interest, only nausea,
`dizziness, and fatigue were significantly more fre-
`quent
`in the naltrexone-treated group, with RRs
`(95 % CIs) of 2.14 (1.61–2.83), 2.09 (1.28–3.39), and
`1.35 (1.04–1.75) respectively (see Figure 2). Figure 3
`shows short-term comparisons between naltrexone
`and placebo regarding discontinuation of treatment.
`Discontinuation of
`treatment rates were not sig-
`nificantly different between groups (RR 0.85, 95 % CI
`0.72–1.01).
`For medium-term treatment, no significant differ-
`ence between groups was found in respect of relapse.
`Two secondary outcomes were significantly different,
`in which naltrexone was superior in increased time
`(days)
`to first drink and decreased craving (see
`Table 2).
`
`Naltrexone (°other treatment) vs. placebo/other
`treatment (see Table 2)
`
`Naltrexone+ondansetron vs. placebo
`
`Only a secondary outcome of standard drinks was
`available, with a significant preference towards
`naltrexone+ondansetron treatment.
`
`Naltrexone vs. acamprosate vs. naltrexone+acamprosate
`vs. placebo
`
`Only a short-term, placebo-controlled trial was carried
`out to compare naltrexone vs. acamprosate vs. nal-
`trexone+acamprosate (Kiefer et al., 2003). Because the
`
`Naltrexone for alcoholism 273
`
`study findings were presented only in graphs, no RR,
`WMD, or SMD could be computed.
`
`Naltrexone vs. nefazodone
`
`No significant difference between groups was found
`in respect of return to drinking, time to first drink,
`drinking days, standard drinks, and craving.
`
`Naltrexone vs. intensive psychosocial treatment alone
`
`Because disulfiram mainly works through its behav-
`ioural but not pharmadynamic effects, we classified
`disulfiram administration as an intensive psychosocial
`treatment. In this small-sample-size study (n=9 in
`each arm), naltrexone was significantly inferior to
`disulfiram regarding time to first drink and drinking
`days.
`
`Naltrexone+intensive psychosocial treatment vs.
`intensive psychosocial treatment alone
`
`intensive
`trials applied a form of
`Because most
`psychosocial treatment and we have already included
`those study findings in the section under naltrexone
`vs. placebo, only the data comparison between nal-
`trexone+intensive psychosocial treatment vs. inten-
`sive psychosocial treatment alone (without a placebo)
`is included in this section.
`Only medium- and long-term comparisons were
`reported in one trial (Landabaso et al., 1999). The in-
`tensive psychosocial treatment used in that trial was
`an aversive agent (not specified). Medium- and long-
`term adjuncts of naltrexone significantly decreased the
`risks of relapse and return to drinking.
`
`Naltrexone+intensive psychosocial treatment vs.
`naltrexone+simple psychosocial treatment
`
`and medium-term outcomes were
`Both short-
`available in three placebo-controlled trials (Balldin
`et al., 2003; O’Malley et al., 1992, 2003). While the in-
`tensive psychosocial treatment was CBT, the simple
`psychosocial treatment was supportive psychotherapy
`(Balldin et al., 2003; O’Malley et al., 1992) and PCM
`(O’Malley et al., 2003).
`A short-term adjunct of intensive psychotherapy
`was not superior in any respect of primary or second-
`ary outcomes. For a medium-term adjunct of in-
`tensive psychosocial treatment, no superiority was
`found on the risks of relapse and return to drinking.
`However, the intensive psychosocial adjunct signifi-
`cantly increased time (days) to first drink and de-
`creased craving.
`
`Downloaded from https://academic.oup.com/ijnp/article-abstract/8/2/267/755449
`by guest
`on 03 April 2018
`
`AMN1033
`IPR of Patent No. 7,919,499
`
`

`

`274 M. Srisurapanont and N. Jarusuraisin
`
`Table 2. Effects of naltrexone treatment, apart from short-term, primary outcome comparisons with placebo
`
`Outcome comparison (studies)
`
`Total
`number of
`patients
`
`Test for
`heterogeneity
`
`Mean effect size (95 % CI)
`
`Naltrexone vs. placebo (short-term secondary outcomes only)
`Time to first drink (Anton et al., 1999;
`511
`Guardia et al., 2002; Hersh et al., 1998;
`Kranzler et al., 2000)
`Drinking days (Hersh et al., 1998;
`Kranzler et al., 2000; Latt et al., 2002;
`O’Malley et al., 1992; Volpicelli et al.,
`1997)
`Standard drinks (Anton et al., 1999;
`Chick et al., 2000; Guardia et al., 2002;
`Hersh et al., 1998; Kranzler et al., 2000;
`O’Malley et al., 1992)
`Craving (Anton et al., 1999; Kranzler
`et al., 2000; O’Malley et al., 1992;
`Volpicelli et al., 1997)
`
`489
`
`772
`
`400
`
`I2=0 %
`
`WMD x0.06 (x1.04 to 0.93)
`
`I2=61.5 %
`
`WMD x1.96 (x5.47 to 1.56)
`
`I2=65.3 %
`
`SMD x0.21 (x0.46 to 0.04)
`
`I2=34.1 %
`
`SMD x0.10 (x0.35, 0.15)
`
`Naltrexone vs. placebo (medium-term)
`Relapses (Balldin et al., 2003)
`Time to first drink (Balldin et al., 2003)
`Drinking days (Balldin et al., 2003;
`Krystal et al., 2001)
`Standard drinks (Krystal et al., 2001)
`385
`na
`Craving (Balldin et al., 2003)
`55
`na
`Naltrexone+ondansetron vs. placebo (short-term) (Johnson et al., 2000)
`Standard drinks
`20
`na
`
`118
`55
`440
`
`na
`na
`I2=58.3 %
`
`Naltrexone vs. nefazodone (short-term) (Kranzler et al., 2000)
`Return to drinking
`120
`Time to first drink
`120
`Drinking days
`120
`Standard drinks
`120
`Craving
`120
`
`na
`na
`na
`na
`na
`
`RR 1.01 (0.92 to 1.11)
`WMD 35.00 (29.83 to 40.17)*
`WMD x6.92 (x18.09 to 4.25)
`
`SMD 0.03 (x0.17 to 0.23)
`SMD x0.88 (x1.44 to x0.33)*
`
`SMD x4.19 (x5.88 to x2.50)*
`
`RR 1.04 (0.82 to 1.32)
`WMD 0.70 (x0.96 to 2.36)
`WMD 4.40 (x4.95 to 13.75)
`SMD 0.09 (x0.27 to 0.45)
`SMD 0.28 (x0.08 to 0.64)
`
`Naltrexone vs. intensive PST alone (short-term) (Carroll et al., 1993)
`WMD x5.60 (x7.94 to x3.26)*
`Time to first drink
`18
`na
`Drinking days
`18
`na
`WMD 22.30 (22.18 to 22.42)*
`SMD 0.90 (x0.08 to 1.88)
`Standard drinks
`18
`na
`Naltrexone+intensive PST vs. intensive PST alone (medium-term) (Landabaso et al., 1999)
`Relapses
`30
`na
`RR 0.33 (0.14 to 0.80)*
`Return to drinking
`30
`na
`RR 0.33 (0.14 to 0.80)*
`Naltrexone+intensive PST vs. intensive PST alone (long-term) (Landabaso et al., 1999)
`Relapses
`30
`na
`Return to drinking
`30
`na
`Naltrexone+intensive PST vs. naltrexone+simple PST (short-term)
`Relapses (O’Malley et al., 2003)
`190
`na
`I2=85.4 %
`Return to drinking (O’Malley et al.,
`242
`1992, 2003)
`Drinking days (O’Malley et al., 1992)
`Standard drinks (O’Malley et al., 1992,
`2003)
`Craving (O’Malley et al., 1992, 2003)
`
`RR 0.60 (0.40 to 0.91)*
`RR 0.60 (0.40 to 0.91)*
`
`RR 0.87 (0.61 to 1.22)
`RR 1.19 (0.54 to 2.61)