Combined Pharmacotherapies for the Management of Alcoholism: Rationale and Evidence to Date

Abstract

Pharmacotherapies for alcohol use disorders (AUDs) have limited efficacy. One approach to improving treatment outcomes for AUDs is to combine pharmacotherapies that have shown some efficacy as individual agents. The rationale for combining medications rests on the following principles: a combination of medications can target more than one neurotransmitter system that is dysfunctional in AUDs, can target different drinking behaviors (i.e., positive and negative reinforcement), can treat comorbid psychiatric and medical disorders, and can minimize side effects, improving adherence to treatment by using lower doses of each drug in combination. Combined pharmacotherapy strategies may produce additive or even synergistic effects to decrease alcohol craving and consumption. Here, we reviewed the literature investigating the effect on alcohol-related outcomes of combinations of medications that have shown efficacy as single agents to reduce drinking in animal studies and clinical trials. We focused on 17 clinical studies investigating the combination of medications in AUDs, 11 of which were randomized, double-blind, and placebo-controlled. Ten of the 11 studies showed the combination to be superior to placebo, but only three showed an advantage of the combination compared with the single agent. Overall, these studies used diverse methodologies, assessments of severity, outcome measures, and adjunctive psychosocial treatments. Limitations of the current published studies and possible future directions for new combinations are discussed.

1. Introduction

Alcohol use disorders (AUDs), comprising clinical diagnoses of abuse or dependence, are among the leading causes of morbidity and mortality, increased healthcare costs, and lost work hours [1]. Not only are the prevalence and the morbidity of AUDs similar to other common psychiatric disorders (e.g., major depression and anxiety disorders), but the financial cost of AUDs is considerable and similar to the costs of other common medical problems, such as cardiovascular diseases and diabetes mellitus [2].

It is, therefore, recognized that AUDs represent chronic, relapsing medical conditions with a multifactorial etiology that includes genetic, neurobiological, psychological, and environmental components [3]. The neurobiology of AUDs is complex with involvement of and alterations in several neurotransmitter systems that have been shown to mediate alcohol’s rewarding effects and abuse liability. These include but are not limited to dopamine, opioids, γ-aminobutyric acid (GABA), N-methyl-d-aspartate (NMDA), and serotonin (for extensive reviews, see Koob [3, 4] and Koob and Volkow [3, 4]). Understanding the influence of these neurotransmitter systems on alcohol-seeking behaviors has led to investigating their role as possible neuropharmacological targets for AUDs treatment [5]. Pharmacological agents may be capable of modifying the functions of neurotransmitter systems and, hence, modify dependent drinking behavior [6]. The increased interest in using pharmacotherapies to treat alcohol dependence is further driven by the recent development of pharmaceutical agents that reduce alcohol consumption in animal models of alcohol dependence, which may be effectively applied to clinical trials [7]. For several decades, only one medication (i.e., disulfiram) was approved in the USA to treat alcohol dependence. Then, starting in 1994, three additional medications have been approved by the US FDA (naltrexone tablets, naltrexone intramuscular, and acamprosate) and several other pharmacotherapies are currently under investigation [8]. Additional medications are approved in other countries, including, for example, the recent approval of nalmefene in Europe [9] and the approval of the GABAergic medication sodium oxibate under strict medical monitoring in a few European countries [10].

There are several reasons supporting the need to develop novel effective medications for AUDs. The availability of effective medications for AUDs in the primary care setting, where most patients are seen for problem drinking, will greatly expand treatment for this disorder, whether in the primary or referral treatment setting. Further, these medications can be combined with other interventions [e.g., Alcoholics Anonymous, cognitive behavioral therapy (CBT)] and this combined approach may play a crucial role in improving addiction treatment, as it can address both the biological and psychosocial aspects of the disease [8].

Given the complex pharmacology of alcohol and neurobiology of AUDs, one approach to developing pharmacological treatments is to investigate the efficacy of combinations of medications to manipulate multiple neurotransmitter systems to reduce craving and/or block the reinforcing effects of alcohol [11]. In fact, FDA-approved medications (i.e., disulfiram, naltrexone, and acamprosate) have, as single therapeutic modalities, suboptimal efficacy [12]. Combining medications, might, therefore, improve outcomes for treatment of AUDs. The rationale for combining medications rests on the following principles: a combination of medications can target more than one neurotransmitter system that is dysfunctional in AUDs, can target different drinking behaviors, i.e., positive and negative reinforcement, can treat co-morbid psychiatric and medical disorders, and can minimize side effects, improving adherence to treatment by using lower doses of each drug in combination [13].

Animal and human studies of combination pharmacotherapies for the treatment of AUDs are reviewed here. Human studies are summarized in Table 1. The focus is on combination therapies to treat primary AUDs and does not include those combinations used to target, explicitly, AUDs with co-morbid psychiatric disorders, although those studies are cited here secondary to reporting the combination’s use to treat primary AUDs. Data were obtained for the review by searching the published medical literature until 31 October 2013. No language or date restrictions were applied. The searches were conducted through PubMed, Web of Science, and EMBASE. Additional references were identified from the reference lists of published articles.

Table 1.

Summary of human studies of combined pharmacotherapy for alcohol dependence

References	N	Medication combinationa	Tx duration	Psychosocial Txb	AUD severity	Last use pre-Tx (days)	Design	Other Psych.	Results		F/U	Alcohol-related and other outcomes
									Combination vs. PL	Combination vs. single drug
Landabaso et al. [17]	30	D vs. NTX vs. [NTX + D]	1 y (D); 6 mo (NTX)	No	Abuse or dependence; refractory to aversion therapy	None	R, open	No	N/A	[D + NTX] > D	1 and 2y	Abstinence rates, 6 mo, 1 and 2 y
Petrakis [18]	254	D vs. NTX vs. [D + NXT] vs. PL	12 wks	Q wk	AD; ADS ≈22	3–29	R, DB - NTX, PL	Axis I	D. NTX. [NTX + D] > PL	NS	No	Alcohol use, craving, Psych, symptoms, adverse events, liver function tests
Besson et al. [19]	118	[AC + D] vs. AC vs. D vs. PL	1 y	2×/mo	AD; MAST ≈29–34	≥5	R, DB - AC, PL [stratified for DS yes/ no]	No	AC > PL	[AC + D] > D or AC	1 y	Relapse rate; CAD
Maremmani et al. [20]	52	[GHB + D]	6 mo	No	AD; non-responder to GHB Tx (last 2 y); 5 units ETOH/day	None	Open	Yesc	N/A	[GHB + D] > GHB	6 mo	Treatment retention; relapse rates; CAD
Kiefer et al. [29]	160	[NTX + AC] vs. NTX vs. AC vs. PL	12 wks; start IP	Q wk	AD with 5 DSM-IV criteria	≥12	R, DB, PL	No	NTX + AC > PL	[NTX + AC] = NTX > AC	No	Time to first drink, time to relapse, CAD
Feeney et al. [30]	236	[NTX + AC] vs. NTX vs. AC vs. PL	12 wks	CBT	MAST ≈18	≥3	Open, matched, 4th cell with no medications	No	NTX + AC > 4th cell	NTX + AC > AC or 4th cell	No	CAD, time to relapse; program adherence
Anton et al. [31]	1,383	[NTX + AC] vs. NTX vs. AC vs. PL	16 wks	MM, CBI	AD; ADS ≈16–17	4–21	R, DB, PL	No	NS	[NTX + AC] = NTX > AC	1 y	% days abstinent; time to first HDD
Johnson et al. [36]	20	[NTX + OND] vs. PL	8 wks	CBT	AD 3 DSM-IV criteria; MAST ≈33	0	R, DB, PL	No	[NTX + OND] > PL	N/A	No	Drinks/day, drinks/drinking day, % days abstinent
Myrick et al. [23]	90	[NTX + OND] vs. NTX vs. OND vs. PL	1 wk	No	ADS 14–16	0	R, DB, PL	No	[NTX + OND] > PL	[NTX + OND] > OND = NTX = PL	No	Craving: ventral striatal activation (via fMRI): cues
Farren et al. [41]	113	[NTX + SSRI] vs. NTX	12 wks	Q wk	Drinks/drinking day ≈7	5–30	R, DB	No	N/A	[NTX + SSRI] = NTX	No	Time to first drink; time to relapse
O’Malley et al. [42]	99	[NTX + SSRI] vs. NTX vs. PL	16 wks	MM	AD; ADS 18–22	4–30	R, DB, PL	No	[NTX + SSRI] > PL	[NTX + SSRI] = NTX	1 y	Time to first HDD, total abstinence
Pettinati et al. [43]	170	[NTX + SSRI] vs. NTX vs. SSRI vs. PL	14 wks	CBT	AD; DSM-IV, 3 criteria; HAM-D ≥ 10	≥3	R, DB, PL	MDD	[NTX + SSRI] > PL	[NTX + SSRI] > NTX = SSRI	No	Total Abstinence rate; time relapse to HDD
Anton et al. [56]	150	[NTX + GP] vs. NTX vs. PL	16 wks	CBI	AD; ≈9–11 drinks/day	≥4	R, DB, PL	No	NTX + GP] > PL (duration of GP only)	[NTX + GP] > NTX (duration of GP only) = PL	No	Time to first HDD, % HDD
Anton et al. [57]	60	[GP + Flum] vs. PL	39 days	Q wk	AD; ADS; ≈12 lowCIWA; ≈19 high CIWA	≥3	R, DB, PL	No	[GP + Flum] > PL (high withdrawal history patients only)	N/A	8 wk	ETOH withdrawal; %days abstinent, time to first HDD
Caputo et al. [64]	55	[NTX + GHB] vs. NTX vs. GHB	3 mo	No	AD; DSM-IV TR; “severe” AD ≈80 %	≥7	R, open	No	N/A	[NTX + GHB] > NTX = GHB	No	Time abstinence
Stella et al. [65]	47	SSRI vs. [NTX + SSRI] vs. [GHB + SSRI] vs. [NTX + GHB + SSRI]	6 mo	2×/wk	AD; duration AD ≈ 12 y	All detox	Open	No	N/A	[NTX + GHB + SSRI] > SSRI = [NTX + SSRI] = [GHB + SSRI]	No	Relapse rate
Kenna et al. [83]	13	[Topiramate + aripirpazole]	35 days	No	Heavy drinkers	0	Open	No	N/A	N/A	No	Safety, alcohol consumption

AC acamprosate, AD alcohol dependence, ADS alcohol dependence scale, AUD alcohol use disorder, CAD cumulative abstinence duration, CBI cognitive behavioral intervention, CBT cognitive behavioral therapy, CIWA Clinical Institute Withdrawal Assessment, D disulfiram, DB double-blind, DSM-IV Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, ETOH ethanol, Flum. flumazenil, fMRI functional magnetic resonance imaging, F/U follow-up, GHB γ-hydroxybutyric acid, GP gabapentin, HAM-D Hamilton Depression Rating Scale, HDD heavy drinking day, IP inpatient, MAST Michigan alcohol screening test, MDD major depressive disorder, MM medical management, mo month, N number, N/A not applicable, NS not significant, NTX naltrexone, OND ondansetron, PL placebo, Psych. psychiatric co-morbidity, Q wk weekly, R randomized, SSRI selective serotonin reuptake inhibitor, Tx treatment, wk week, y year

^aSee text for the doses used in each study

^bWhere no specific intervention is entered, frequency of outpatient clinic Tx as usual is specified

^cPsychiatric co-morbidities were allowed, but details are not reported in the manuscript

2. Combinations with Disulfuram

Disulfiram has been FDA approved for treatment of AUDs for the past 55 years. Its action yields an aversive reaction to alcohol via inhibition of alcohol dehydrogenase and concomitant accumulation of acetaldehyde that produces an aversive reaction when combined with alcohol. Despite the limited efficacy as a single agent to prevent relapse [14], with the exception of improvement in short-term abstinence [15] or supervised use [16], there has been interest in combining disulfiram with anti-craving medications such as naltrexone, acamprosate, and sodium oxybate. Decreasing craving might allow for improved adherence to disulfiram treatment with concomitant avoidance of drinking.

In a randomized clinical trial (RCT) with 30 patients who met criteria for alcohol dependence or abuse and were refractory to therapy with disulfiram alone, Landabaso et al. [17] compared treatment with daily disulfiram for 1 year combined with naltrexone 25 mg daily for 6 months versus daily disulfiram alone for 1 year. Abstinence rates at 1- and 2-year follow-up after treatment with the combination were superior to disulfiram alone. Petrakis et al. [18] investigated the same combination in 254 alcohol-dependent patients with a co-morbid Axis I disorder [36 % with post-traumatic stress disorder (PTSD)] enrolled at three Veterans Administration outpatient clinics. Treatment consisted of 12 weeks with open randomization to disulfiram or no disulfiram, and double-blind randomization to naltrexone or placebo, resulting in four treatment groups. While each active treatment group was superior to placebo in terms of consecutive days of abstinence and drinking days per week, there was no advantage of the combination over either modality alone. However, in this study, patients were not selected for treatment resistance to aversion therapy as in the previous study by Landabaso et al. [17]. In both studies [17, 18], safety and tolerability were fair in all medication conditions.

Besson et al. [19] investigated the combination of disulfiram with acamprosate in 118 alcohol-dependent patients randomized to 1 year of treatment with acamprosate or placebo and 1 year of follow-up. Disulfiram as a combined medication was allowed in those patients who wished to receive it, and the two treatment groups (acamprosate or placebo) were stratified accordingly, leading to four treatment groups, i.e., acamprosate(blind), placebo(blind), acamprosate(blind)/disulfiram(open), and placebo(blind)/disulfiram(open). Outcome measures were relapse rate and cumulative abstinence duration. This study showed a significant effect of acamprosate, compared with placebo, on alcohol-related outcomes. Furthermore, the combination of acamprosate and disulfiram was superior to either medication alone or placebo with respect to cumulative abstinence duration and 30-day relapse rates, suggesting the concomitant administration of disulfiram may improve the efficacy of acamprosate. An important limitation of this study was the high withdrawal rate, as well as the design whereby patients were allocated to disulfiram based on their wishes (although it was controlled by stratifying for this condition).

Six months’ treatment with a combination of disulfiram and sodium oxybate [also known as γ-hydroxybutyric acid (GHB); more information about this drug is provided in Sect. 3.6] in alcohol-dependent patients who did not respond to sodium oxybate treatment alone yielded reduced relapse rates, and increased days of abstinence and treatment adherence [20]. In this open-label retrospective study, 53 alcohol-dependent patients, the majority of whom had psychiatric co-morbidity, were treated with 1 week of sodium oxybate (50 mg/kg/day), then disulfiram 400 mg/day was added and sodium oxybate was increased to a maximum of 100 mg/kg/day according to individual response.

3. Combinations with Naltrexone

3.1. Naltrexone and Acamprosate

The use of naltrexone and acamprosate, the most common combination investigated in the alcoholism pharmacotherapy research, is based on clinical evidence of each drug’s efficacy with respect to treatment outcomes for alcohol dependence (reviewed in Litten et al. [21]). In addition, preclinical studies suggest a potential additive effect as each medication targets different neurobiological pathways: naltrexone has been shown to reduce the positive reinforcing properties of alcohol [22, 23], and acamprosate the negative reinforcing properties [24]. Kim et al. [25] showed an advantage for the combination over naltrexone alone with respect to alcohol administration in C57BL/6 mice, and the combination has been shown to be advantageous over low-dose naltrexone to reduce post-deprivation alcohol administration, an animal model of relapse [26].

The combination of naltrexone and acamprosate has been shown to be safe in healthy volunteers [27] and alcohol-dependent patients [28]. Efficacy in the treatment of alcohol dependence in recent abstinence has been studied in two single-site studies. In Germany, Kiefer et al. [29] randomized 160 alcohol-dependent patients in a four-cell, double-blind, placebo-controlled RCT to naltrexone (50 mg/day), acamprosate (1,998 mg/day), their combination, or placebo for 12 weeks’ treatment. In Australia, Feeney et al. [30] examined 12 weeks’ treatment with naltrexone (50 mg/day), acamprosate (1,998 mg/day), and their combination in four matched groups of 59 subjects each, with no medication as the comparator group (i.e., a group of patients who declined receiving medications). This was an open study where at different periods of time, patients were treated with either naltrexone, acamprosate, or their combination, and then patients were selected based on matching criteria (i.e., sex, age, prior medically supervised detoxification, alcohol problem severity) in order to analyze alcohol-related outcomes. All four groups also received CBT.

The largest single multisite study to date investigating a pharmacotherapeutic combination in alcoholic patients is the COMBINE (Combined Pharmacotherapies and Behavioral Interventions) study [31]. In this study, 1,383 alcohol-dependent patients in early abstinence were randomized to 16 weeks’ treatment with the following: naltrexone (100 mg/day) or acamporsate (3 g/day), or both or both placebos. Each of these four groups was further divided into receiving or not combined behavioral intervention (CBI). All eight groups received medication management. A ninth group received CBI only (no medications). Results of this study indicated that patients receiving medical management with naltrexone, CBI, or both had better drinking outcomes.

These studies [29–31] confirmed an effect of naltrexone in alcoholism and, in some cases, suggested an effect of the combination of naltrexone and acamprosate, as compared with placebo or acamprosate alone. By contrast, none of these studies demonstrated that the addition of acamprosate to naltrexone was superior to naltrexone or placebo alone with respect to relapse rates or cumulative time of abstinence. The PREDICT study [32], conducted at multiple sites in Germany in 426 alcohol-dependent patients, was based on a design similar to the US COMBINE study except for the absence of the combination arm and the CBI intervention in the primary study. Results showed no separation of groups in the time to first occurrence of heavy drinking, in contrast to the COMBINE results which favored the naltrexone arm. A baseline difference in patients’ severity of alcohol dependence was posited as a reason for the difference in results as patients in PREDICT consumed significantly more alcohol before admission and reported significantly more Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria for alcohol dependence [33]. However, it should be noted that PREDICT found neither acamprosate nor naltrexone significantly beneficial compared with placebo [33].

3.2. Naltrexone and the Serotonin 5-HT₃ Antagonist Ondansetron

Preclinical evidence exists for the synergistic interaction between the serotonin 5-HT₃ receptor and endogenous opioids in the mesocorticolimbic dopamine reward system [34] where 5-HT₃ receptors may mediate alcohol reward processing via activation of the endogenous opioid system. In rodents, alcohol intake was not affected when naltrexone and ondansetron were administered individually and in low dosages; however, when naltrexone and ondansetron were administered together there was a marked reduction in alcohol intake with no change in water intake (reviewed in Le et al. [35]).

In a pilot double-blind placebo-controlled trial of 20 early-onset alcoholics (EOAs, i.e., individuals with an age of onset of alcohol problems <25 years; for an extensive review on EOAs and other typologies, see Leggio et al. [22]), Johnson et al. [36] reported that the combination of ondansetron (4 μg/kg twice daily) and naltrexone (25 mg twice daily) reduced drinking compared with placebo and that this reduction was related to a reduced serum carbohydrate-deficient transferrin (CDT) level, a sensitive and validated marker of transient alcohol consumption [37]. In a human laboratory and functional magnetic resonance imaging (fMRI) study, Myrick et al. [23] found that the combination of ondansetron and naltrexone reduced alcohol cue-induced craving in 90 non-treatment-seeking alcohol-dependent patients compared with placebo, each drug alone, and a comparator group consisting of 17 social drinkers. Subjects were randomly assigned to 7 days of daily dosing with naltrexone (50 mg; n = 23), ondansetron (0.25 mg twice daily; n = 23), the combination (same doses; n = 20), or matching placebo (n = 14). Further, the combination, naltrexone, and comparator groups showed significantly reduced activation to alcohol cues in the ventral striatum compared with placebo, suggesting a synergistic action of the two medications. Interestingly, there was no difference in alcohol-elicited craving during the fMRI session between the placebo and naltrexone alone, or placebo and ondansetron alone groups. To date, there have been no treatment studies comparing the combination with either medication alone.

3.3. Naltrexone and Serotonin Reuptake Inhibitors

Selective serotonin reuptake inhibitors (SSRIs) have been suggested as possible pharmacological treatments for AUDs, especially for specific subtypes of patients (for review, see Kenna [24]). Additionally, some lines of research suggest that SSRIs may be useful for augmenting naltrexone response in AUDs. Serotonergic neurotransmission has been found to be abnormal in preclinical and clinical studies of alcoholism [38]. Preclinical models have also shown that the combination of an SSRI and naltrexone results in greater suppression of drinking than either drug alone [39]. Potential additive effects are also suggested by additional preclinical work showing that fluoxetine reduces stress-induced reinstatement of drinking and naltrexone reduces alcohol reinstatement [40].

In a double-blind, placebo-controlled trial in alcohol-dependent patients (n = 113) by Farren et al. [41], subjects were randomized to 12 weeks’ treatment with either naltrexone and placebo, or naltrexone and sertraline. Naltrexone was administered at the dose of 12.5 mg/day for 3 days, then 25 mg/day for 4 days, and finally 50 mg/day for the next 11 weeks; and sertraline was administered at the dose of 50 mg/day for 2 weeks, increasing to 100 mg/day for the remaining 10 weeks. The combination was equivalent to naltrexone alone with respect to measures of abstinence; hence, not supporting the hypothesis that sertraline would augment the efficacy of naltrexone in patients with alcohol dependence. O’Malley et al. [42] also reported a negative double-blind placebo-controlled RCT in 101 Alaskans with alcohol dependence residing in rural settings, 68 of whom were American Indians/Alaska Natives. Patients were randomized to 16 weeks’ treatment with either placebo–naltrexone and placebo–sertraline, naltrexone 50 mg/day and placebo–sertraline, or naltrexone 50 mg/day and sertraline 100 mg/day (both naltrexone and sertraline were titrated up at the beginning of the trial). In this specific setting, where access to treatment is limited, naltrexone alone resulted in a significantly higher percentage of total abstinence and percentage days abstinent and reduced drinking-related consequences, while the combination did not differ from naltrexone alone. Thus, this trial also failed to demonstrate an additive effect of sertraline to naltrexone in enhancing abstinence rates. However, another double-blind, placebo-controlled randomized study (n = 170) found that during a 14-week treatment period, the combination resulted in a higher abstinence rate and longer time to relapse to heavy drinking than either treatment alone or placebo [43]. Notably, the latter study enrolled alcohol-dependent patients with comorbid major depressive disorder (MDD), suggesting that alcoholic patients with this psychiatric co-morbidity may respond better to the combination of naltrexone and sertraline. Furthermore, compared with the two previous studies [41, 42], Pettinati et al. [41, 42] used higher doses of both medications (after titration)—sertraline 200 mg/day and naltrexone 100 mg/day—thus suggesting that perhaps this study was more likely to allow for maximum therapeutic benefits of sertraline and/or naltrexone (notably, the combination of naltrexone and sertraline was always superior to the other three medication conditions, thus not allowing one to speculate possible maximum therapeutic benefits specifically for one of the two tested medications).

3.4. Naltrexone and Baclofen

The GABA_B receptor has been shown to be involved in the neural circuitry mediating the stimulating effect of morphine on alcohol consumption in Sardinian alcohol-preferring (sP) rats [44]. Of note, the ability of the GABA_B receptor agonist baclofen to reduce measures of anxiety [45] suggests that it may primarily lessen relief-type craving, which would lead to the potential for additive effects with a treatment such as naltrexone that targets reward-type craving [40]. Further to this, preclinical studies also report that the combination of naltrexone and baclofen result in a potentiation of the effect of each drug to reduce alcohol intake in alcohol-preferring sP rats without affecting food intake [46] and in alcohol-experienced Wistar rats [47]. Despite clinical evidence that baclofen is efficacious in treating withdrawal and maintaining abstinence in alcohol-dependent patients (reviewed in Leggio et al. [45], but see Garbutt et al. [48] for conflicting results), no clinical studies have been published investigating the combination of baclofen with naltrexone.

3.5. Naltrexone and Gabapentin

Another GABA modulator, gabapentin, has been studied in alcohol-dependent patients and has been shown to lessen withdrawal symptoms [49–51] and reduce risk for relapse, with inconsistent effects on alcohol craving [52–54]. A recent RCT confirmed the efficacy of gabapentin in treating alcohol dependence and relapse-related symptoms of insomnia, dysphoria, and craving [55]. Anton et al. [56] studied the combination of gabapentin with naltrexone, with the rationale that the two drugs will mediate negative and positive reinforcing effects of alcohol, respectively. In this study, 150 alcohol-dependent patients with at least 4 days’ abstinence were randomized to three groups of 50 patient each receiving 16 weeks’ treatment with naltrexone alone (50 mg/day), naltrexone (50 mg/day) and gabapentin (up to 1,200 mg /day) added for the first 6 weeks or double placebo. The combination resulted in longer time to relapse to heavy drinking, fewer heavy drinking days, fewer drinks per day than naltrexone alone, while naltrexone alone did not result in better outcomes than placebo. However, the significant beneficial effects in the combined cell only lasted for the 6-week duration of gabapentin/naltrexone treatment only, while no effects were seen after gabapentin was discontinued (i.e., for the remaining 10 weeks of the study).

Although it does not include naltrexone, we also would like to briefly mention here the possible combination of gabapentin with flumazenil. A few studies have investigated the effect of the combination of gabapentin and flumazenil, a benzodiazepine antagonist, on alcohol withdrawal as a way to modulate GABA, which is known to mediate alcohol withdrawal. Anton et al. [57] enrolled 60 alcoholic patients who were pre-treated with hydroxyne (50 mg orally) and divided in two groups, i.e., patients with low pre-treatment alcohol withdrawal (n = 44) and patients with high alcohol withdrawal (n = 16). Patients were then randomized to receive either gabapentin and flumazenil combined or their inactive placebos. The main results of this study indicated that the combination of intravenous flumazenil (2 mg of incremental bolus for 20 min) for 2 consecutive days and gabapentin (up to 1,200 mg daily for 39 days) compared with their placebos differentially affected the subgroup of patients with pre-treatment histories of more severe alcohol withdrawal symptoms. Specifically, this group had greater reduction in withdrawal symptoms than placebo and had greater percentage days abstinent and time to first heavy drinking than when treated with placebo. The group with low pretreatment withdrawal symptoms showed the opposite treatment effect with respect to these measures. Neurocognitive measures at baseline and at 2 weeks into treatment showed that the group with elevated withdrawal symptom histories also showed improvement in response inhibition over pretreatment levels on the combination treatment compared with the low withdrawal group or the placebo groups [58]. Further, the high withdrawal symptom group treated with the combination showed greater dorsal anterior cingulate cortex activation to alcohol cues, similar to the lower symptom withdrawal group treated with placebo. This was related to task deactivation and cognitive control over alcohol-related thoughts [59].

3.6. Naltrexone and Sodium Oxybate (γ-Hydroxybutyric Acid)

Sodium oxybate (also called GHB), a short-chain fatty acid, structurally similar to GABA, is used as a treatment for alcohol dependence, both withdrawal [60] and maintenance of abstinence [61, 62], in a few European countries [63]. Investigating the combination of sodium oxybate and naltrexone with the rationale of promoting abstinence with sodium oxybate and decreasing relapse rates, especially to heavy drinking, with naltrexone, Caputo et al. [64] reported a randomized open-label clinical trial of 55 alcohol-dependent subjects, the majority (77–81%) with severe dependence and post-alcohol detoxification, who were treated for 3 months with either sodium oxybate alone (50 mg/kg/day), naltrexone alone (50 mg/day), or sodium oxybate (50 mg/kg/day) and naltrexone 50 mg/day combined. The combination resulted in significantly longer periods of abstinence than either treatment alone with no significant between-group differences in relapse drinking. Daily intake after relapse was significantly lower in the combination group than in either drug alone. These results were supported by a second small, open-label study in 47 alcohol-dependent patients assigned to either escitalopram (20 mg/day) alone, or naltrexone (50 mg/day) and escitalopram (20 mg/day) combined, or sodium oxybate (75 mg/kg/day) and escitalopram (20 mg/day) combined, or naltrexone (50 mg/day) and sodium oxybate (75 mg/kg/day) and escitalopram (20 mg/day) combined altogether. The combination of the three drugs altogether was superior to either medication with escitalopram or escitalopram alone [65].

It is important to keep in mind that while sodium oxybate is used in some European countries for alcoholism, in the USA, through a limited distribution program, the FDA approved it as a Schedule III Controlled Substance to treat a small subset of patients with narcolepsy who have episodes of weak or paralyzed muscles (i.e., cataplexy). However, as a treatment for alcoholism, sodium oxybate remains a drug difficult to manage considering its short half-life (thus, the need for up to six administrations per day) and the concern for abuse liability properties. Nonetheless, these small studies mentioned here and above (Sect. 2) with sodium oxybate, despite limitations (e.g., small samples, open design), might represent potentially interesting proof-of-concepts around the idea of combining naltrexone with medication with GABAergic actions and/or serotonergic actions.

3.7. Naltrexone and Topiramate

The anticonvulsant topiramate has several pharmacological mechanisms of action, i.e., it increases GABA_A-facilitated neuronal activity, antagonizes AMPA and kainite glutamate receptors with a consequent reduction of dopamine release in the nucleus accumbens (NAc), and modulates ionotropic channels, inhibiting L-type calcium channels, limiting the activity of voltage-dependent sodium channels and facilitating potassium conductance. Additionally, topiramate is weak inhibition of the carbonic anhydrase isoenzymes, CA-II and CA-IV, in the brain and in the kidney [66, 67].

Topiramate has been shown in clinical studies to reduce alcohol consumption and relapse (for reviews, see previous publications [48, 68, 69]). A preclinical study [70] explored the combination of naltrexone and topiramate in alcohol-preferring, C27BL/6 mice on the motivation to drink and on ethanol consumption. The combination of naltrexone and topiramate resulted in significantly reduced ethanol self-administration and motivation to drink, measured by progressive ratio paradigm, compared with either medication alone. In humans, when compared with naltrexone, topiramate was superior to placebo, with trend significance superiority to naltrexone with respect to measures of abstinence and relapse [71] and with respect to alcohol craving and social functioning [72]. Of note, the study by Baltieri et al. [71] included a placebo condition, which was not present in the study by Florez et al. [72]; neither of the two studies, however, tested the combination of topiramate and naltrexone together.

3.8. Naltrexone and Pioglitazone

The gamma isoform of the peroxisome proliferator-activated receptor (PPAR) is highly expressed in the lateral hypothalamus and in the ventral tegmental area (VTA) on dopaminergic cells [73]. Activation of PPAR-γ receptors by pioglitazone, a prototypical thiazolidinedione approved for the treatment of insulin resistance and type 2 diabetes, reduced alcohol drinking and stress, not cue-induced relapse to alcohol seeking in genetically selected alcohol-preferring Marchigian Sardinian (msP) rats [74]. Following up this first evidence on pioglitazone, the same research team reported another preclinical study [75] in msP rats on the effect of the combination of pioglitazone and naltrexone. Co-administration of the two drugs resulted in an additive effect in reducing ethanol intake with a significant reduction in drinking occurring following administration of naltrexone 0.25 mg/kg and pioglitazone 10 mg/kg. The drug combination resulted in a marked inhibition of reinstatement elicited by cues and also by stress. These findings suggest that adding pioglitazone to naltrexone treatment may increase the anti-relapse potential of this opioid antagonist by expanding its efficacy toward stress-induced alcohol seeking.

3.9. Naltrexone and Prazosin

Froehlich et al. [76], in a preclinical study, examined the combination of naltrexone and prazosin in alcohol-preferring ‘P’ rats. Prazosin reduces central noradrenergic signaling via alpha blockade. It has been used to treat PTSD as well as PTSD/AUDs co-morbidity [77]. In alcohol-preferring ‘P’ rats, the combination of low-dose naltrexone (10 mg/kg) and prazosin (2 mg/kg) was effective to decrease alcohol drinking in doses that were each ineffective when administered alone [76]. The effect was rapid acting, decreasing drinking in the first week of administration, and was found to be additive. No human studies of this combination have been published to date.

4. Combinations with Topiramate

The combination of topiramate and naltrexone has already been discussed above (Sect. 3.8). Preliminary studies have investigated the combination of topiramate with either ondansetron or sertraline, as detailed in the following section.

4.1. Topiramate and Ondansetron

Ondansetron acts by antagonizing 5-HT₃ receptors that are located in mesolimbic brain regions leading to an overall suppression of dopamine signaling in these areas [78], and suppresses alcohol-induced dopamine concentration in the NAc [79]. Topiramate and ondansetron each decreases alcohol consumption in rat models [80, 81], and each reduces dopamine levels in the cortico-mesolimbic pathway, albeit via different neural mechanisms. Lynch et al. [82] examined ondansetron and topiramate alone and their combination on ethanol consumption in a free choice paradigm as well as on alcohol consumption post deprivation in alcohol-preferring (P) rats and less-preferring rats (Wistar). The combination of ondansetron and topiramate decreased ethanol consumption in heavy not lighter drinking rats (i.e., P rats and Wistars, respectively) compared with either drug alone. Both topiramate alone and the combination attenuated the alcohol deprivation effect in both types of rats with a trend favoring the combination. No clinical studies have been published on the possible benefits of this combination of pharmacotherapies to treat AUDs.

4.2. Topiramate and Aripiprazole

A small open-label human laboratory trial in 13 heavy-drinking subjects titrated over 35 days to aripiprazole (30 mg/day) and topiramate (300 mg/day) showed that the combination did not yield an additive side effect profile, even when combined with a fixed dose of alcohol administered in a human laboratory setting. However, there was a trend to reduced alcohol use during the naturalistic drinking phase of the study [83].

5. Future Directions—Possible New Combinations

In addition to the combinations mentioned above that are supported by preclinical evidence but currently remain unstudied in humans (i.e., combinations of naltrexone with ondansetron, topiramate, baclofen, pioglitazone, prazosin as well as the combination of ondansetron and topiramate), the recent report of efficacy of varenicline, a partial α4β2 nicotinic agonist, in alcohol-dependent patients to reduce drinking raises the possibility of new potentially interesting combinations. In a multisite, randomized, placebo-controlled, 13-week trial of varenicline in alcohol-dependent patients, varenicline reduced drinking and alcohol craving, regardless of patients’ smoking status [84].

Activation of nicotinic acetylcholine receptors (nAChR) enhances alcohol-induced mesolimbic dopaminergic signaling [27]. Alcohol acts, not only on nAChRs, but also on several related cysteine-loop ligand-gated ion channels including glycine receptors, GABA_A and 5-HT₃ receptors [85] to modulate dopamine in the VTA and NAc. Combining pharmacotherapies that target these receptors, therefore, might yield additive or even synergistic effects to reduce alcohol’s reinforcing effects, such as the combination of varenicline with either topiramate (GABA_A antagonist) or ondansetron (5-HT₃ antagonist). Clinically, the combination of varenicline or topiramate with baclofen might offer a promising combination given their individual efficacy to decrease alcohol consumption, and their renal metabolism might be advantageous in patients with liver disease. Lastly, with respect to alcohol withdrawal, the peptide oxytocin delivered intranasally has recently been shown, in a small study, to reduce physiologic symptoms, alcohol craving, and the benzodiazepine dose required during alcohol withdrawal [86]. Since oxytocin has a short half-life, it could be used as an ‘as needed’ adjunct to a longer-acting medication to target craving and withdrawal symptoms. Therefore, combining intranasal oxytocin as needed with, for example, baclofen which also reduces withdrawal craving and anxiety warrants study, particularly as the two medications are known to modulate GABA.

6. Conclusions

This article provides a brief review of the animal and human studies performed on combining pharmacotherapies for alcohol dependence. Of 17 clinical studies investigating the combination of psychopharmacologic medications in AUDs, 11 were randomized, double-blind, placebo-controlled RCTs. Ten of the 11 studies showed the combination superior to placebo, but only three showed an advantage of a combination (each of the three showing superiority of a unique combination: acamprosate + disulfiram; naltrexone + SSRI; gabapentin + naltrexone) compared with a single agent. Overall, these studies used diverse methodologies, assessments of severity, outcome measures, and adjunctive psychosocial treatments. In particular, the length of abstinence before randomization was highly variable between and within these studies. This parameter has been shown to impact outcomes, such as placebo response rates, thereby producing conflicting results when attempting to replicate clinical trials [33]. Further, this variable is not routinely used as a covariate in analyses of outcomes. Lastly, combining two medications ideally requires a four-cell design; in this case, there are practical barriers to consider, most of all budget issues that may limit the ability of many scientists to run fully powered studies with large samples.

To date, the rationale of using combinations based on clinical observation, i.e., combining drugs that individually have shown an effect to reduce drinking or based on the theoretical rationale of impacting both positive and negative reinforcement processes in AUDs, has not yielded promising results. Of course, it is important to consider study population characteristics such as severity of AUDs as well as genetics to identify subpopulations that would respond differentially to a given approach. Indeed, with the emergence of the DSM-V criteria for AUDs, subjects and patients will be characterized with respect to craving and sub-grouped according to severity. This may help to design studies using more homogenous, well-characterized populations, reducing the variance in outcome inherent in heterogeneous study populations.

There are also important questions that will need to be addressed in the future, the first being whether combined pharmacotherapies should be used as a primary pharmacologic treatment or reserved only for those who fail monotherapy. Additionally, primary care physicians, unlike addiction specialists, may be loathe to prescribe combined pharmacotherapies for patients with AUDs. This highlights the need to educate physicians about addiction, including its diagnosis and treatment [87].

Apart from study methodology and subject characterization, identification of novel molecular targets involved in the pathogenesis of AUDs is essential to expand what is to date a limited number of pharmacologic options for treatment of alcoholism. Combining effective medications to treat chronic health disorders is routinely done in clinical practice for both medical and mental health problems (e.g., diabetes and depression, just to cite two examples). In the addiction field, combining medications to increase the likelihood of better outcomes for our patients is still in its infancy. Above all, with only three medications approved by the FDA, the pharmacological treatment of AUDs is still in an early stage; therefore, there is a need to identify more effective medications in order to test them, and in order to investigate whether combinations of different medications may be beneficial for patients with AUDs.