Naltrexone Is Superior to Placebo for Abstinence and Craving Reduction in Alcohol‐Associated Cirrhosis: NAL‐CI Trial

ABSTRACT

Background and Aims

Alcohol use disorder (AUD) coexisting with cirrhosis carries high morbidity and mortality, with no approved pharmacotherapy for AUD. We evaluated the safety and efficacy of naltrexone, an opioid receptor antagonist, in patients with compensated alcohol‐associated cirrhosis (AaC) and AUD.

Methods

One hundred patients with compensated AaC and DSM‐5 AUD were randomised 1:1 to naltrexone (50 mg/day) or placebo for 12 weeks. The primary endpoint was point‐prevalence abstinence at 12 weeks, defined as no alcohol use in the four preceding weeks. Secondary endpoints included craving (Obsessive Compulsive Drinking Scale [OCDS]—Obsessive and Compulsive subscales), lapses, relapses, and hepatic safety. Standardised psychosocial support was provided to both arms.

Results

Baseline characteristics were well matched between groups (mean MELD 12.6 vs. 12.7; CTP score 5.9 vs. 6.2; age 42.9 vs. 44.3 years). AUDIT and OCDS scores were comparable between groups. Abstinence at 12 weeks was significantly higher with naltrexone: 64% (32/50) versus 22% (11/50), p < 0.001; OR 10.86 (95% CI: 1.89–62.2). Naltrexone significantly reduced lapses at 3 months (28% vs. 54%, p = 0.008) and showed a trend toward fewer heavy‐drinking relapses (12% vs. 28%, p = 0.07). Maintenance of abstinence at 6 months favoured naltrexone (22% vs. 8%, p = 0.09). No patient developed hepatic decompensation attributable to study medication, and no AST/ALT elevation exceeding 5× ULN was observed in either group. Mean craving scores were lower with naltrexone by week 12 than with placebo: OCDS‐O score (6.63 ± 1.16 vs. 9.29 ± 1.78, p < 0.01) and OCDS‐C score (6.35 ± 1.23 vs. 9.02 ± 1.86, p < 0.01). Adverse events were comparable between the groups.

Conclusion

Naltrexone is safe and effective in patients with compensated alcohol‐associated cirrhosis, achieving a threefold higher abstinence rate and significantly reducing craving compared with placebo. These findings support the use of naltrexone as a pharmacological option in patients with compensated AaC and AUD.

Trial Registration

NCT04391764

Key Points

• Alcohol use disorder is common in patients with compensated alcohol‐associated cirrhosis, yet no pharmacotherapy is approved for this group owing to concerns about hepatic safety.

• In this double‐blind, placebo‐controlled randomised trial of 100 patients, naltrexone 50 mg/day markedly increased 12‐week abstinence compared with placebo (64% vs. 22%; OR 10.86).

• Naltrexone also significantly reduced alcohol craving and lapses and was not associated with any hepatic decompensation or clinically significant hepatotoxicity.

Summary

Patients with stable alcohol‐related cirrhosis who keep drinking have poorer survival, yet no medicine is approved to help them stop because of concerns about liver safety. In this randomised trial of 100 such patients, the anti‐craving drug naltrexone was safe and helped far more patients stop drinking than a dummy tablet (64% vs. 22%). Naltrexone could therefore offer a safe and effective treatment option for a group that currently has none.

Abbreviations

AaC

alcohol‐associated cirrhosis

ALT

alanine aminotransferase

AST

aspartate aminotransferase

AUD

alcohol use disorder

AUDIT

Alcohol Use Disorders Identification Test

CTP

Child–Turcotte–Pugh

DBRCT

double‐blind, placebo‐controlled, randomised trial

DM

diabetes mellitus

DSM‐5

Diagnostic and Statistical Manual of Mental Disorders, 5th Edition

EtG

ethyl glucuronide

FDA

Food and Drug Administration

GGT

gamma‐glutamyl transferase

HE

hepatic encephalopathy

HRAR

high‐risk alcohol relapse

INR

international normalised ratio

MELD

Model for End‐Stage Liver Disease

OCDS

Obsessive Compulsive Drinking Scale

OCDS‐C

Obsessive Compulsive Drinking Scale‐Compulsive

OCDS‐O

Obsessive Compulsive Drinking Scale‐Obsessive

SNOSE

sequentially numbered opaque sealed envelope

VAS

visual analogue scale

1. Introduction

Alcohol‐associated cirrhosis (AaC) is a leading cause of cirrhosis‐related mortality worldwide. Continued alcohol use drives accelerated disease progression, hepatic decompensation and death, with 5‐year survival rates of approximately 30%–40% in those who persist in drinking compared with 60%–80% in those who achieve sustained abstinence [1, 2]. Sustained abstinence is therefore the single most powerful therapeutic intervention in AaC—improving hepatic synthetic function, promoting fibrosis regression, reducing portal hypertensive complications and lowering the risk of hepatocellular carcinoma [1, 2, 3, 4]. Despite this imperative, pharmacological management of alcohol use disorder (AUD) in patients with cirrhosis remains profoundly restricted. None of the three agents with regulatory approval for AUD—acamprosate, disulfiram and naltrexone—is currently approved in cirrhosis, owing to concerns regarding hepatotoxicity and the systematic exclusion of cirrhotic patients from pharmacotherapy trials [3, 4, 5]. The persistence of drinking despite severe medical consequences such as cirrhosis is reinforced by the neuropsychological profile of severe AUD, which is characterised by impaired decision‐making and reduced empathic capacity that undermine sustained behavioural change [6]; this provides a strong rationale for pharmacological anti‐craving support alongside psychosocial intervention.

Of agents evaluated in this population, baclofen—a GABA‐B receptor agonist—is the only agent tested in a randomised controlled trial in cirrhotic patients with AUD, achieving abstinence in 71% of patients compared with placebo [7]. However, meta‐analytic evidence demonstrates no significant craving reduction at the hepatically safe dose of 30 mg/day [8], and dose‐limiting somnolence and fatigue preclude effective higher dosing [9, 10]. Beyond baclofen, other agents including topiramate and pregabalin have shown anti‐craving efficacy in general‐population alcohol use disorder: pregabalin was comparable to naltrexone in a randomised head‐to‐head trial [11], and topiramate has demonstrated benefit across several trials [12], while pregabalin and related agents have also been evaluated in the management of alcohol withdrawal [13]. However, none of these agents has been evaluated in a randomised trial in patients with cirrhosis, and their hepatic safety in this setting remains undefined. Naltrexone, a competitive μ‐opioid receptor antagonist that attenuates the mesolimbic dopaminergic reward pathway activated by alcohol, is highly effective for AUD in non‐cirrhotic patients: multiple randomised trials and meta‐analyses demonstrate significant reductions in heavy drinking, monthly alcohol consumption and craving, with improved abstinence rates compared with placebo [14, 15, 16, 17, 18]. Its use in liver disease has been restricted by a product‐label hepatotoxicity warning derived from supratherapeutic doses of ≥ 300 mg/day—more than five times the standard AUD therapeutic dose. Accumulating real‐world evidence, including retrospective data from multiple centres and a 4‐week prospective safety observation from our own group, has found no clinically meaningful hepatotoxicity at 50 mg/day in patients with liver disease [19, 20, 21]. Nevertheless, prospective randomised evidence evaluating naltrexone specifically in patients with compensated AaC has remained entirely absent. In a national cohort of 9131 patients with alcohol‐associated liver disease, only 5.7% had received naltrexone, despite its association with improved survival [22]—underscoring the magnitude of this evidence gap.

We therefore designed the NAL‐CI Trial—a double‐blind, placebo‐controlled randomised controlled trial of naltrexone 50 mg/day for 12 weeks in patients with compensated AaC and DSM‐5 AUD—to provide the first prospective randomised evidence on the safety and efficacy of naltrexone in this population. The primary endpoint was point‐prevalence abstinence at 12 weeks; secondary endpoints included alcohol craving (Obsessive Compulsive Drinking Scale, OCDS), lapses, relapses and hepatic safety. Standardised brief psychosocial intervention based on the WHO brief intervention module was provided to both arms. The trial was registered prospectively at ClinicalTrials.gov (NCT04391764) and approved by the ILBS Institutional Review Board (IEC/2020/76/NA04).

1.1. Trial Design

This is a single‐centre, double‐blind, placebo‐controlled, randomised trial of naltrexone versus placebo in patients with compensated AaC and DSM‐5 AUD.

2. Patients and Methods

Consecutive patients with alcohol‐associated cirrhosis (AaC) and continued alcohol consumption, identified from the combined Hepatology and Psychiatry outpatient and inpatient services at the Institute of Liver and Biliary Sciences (ILBS), a tertiary liver referral centre, aged 18–60 years, fulfilling DSM‐5 criteria for alcohol use disorder, were screened and assessed for enrolment into the study conducted between April 2020 and September 2022.

The study was approved by the ILBS Institutional Review Board (IEC/2020/76/NA04). The study was registered with ClinicalTrials.gov (identifier number: NCT04391764). Informed consent was obtained from the participants, and the work was done in accordance with the Declaration of Helsinki. Compensated cirrhosis was defined as Child–Turcotte–Pugh (CTP) class A or stable class B (score ≤ 8). Exclusion criteria: patients with ongoing hepatic encephalopathy, total bilirubin > 3 mg/dL, Child–Turcotte–Pugh class C or score ≥ 9, recent GI bleed, treatment with prednisolone for alcohol‐associated hepatitis within the previous 60 days, unwillingness to participate, dependence on any other substance (except nicotine), psychotic disorder requiring treatment, and concomitant use of any other agent with a plausible effect on alcohol consumption (including baclofen, acamprosate, disulfiram, topiramate or GLP‐1 receptor agonists).

We diagnosed liver cirrhosis based on physical examination, laboratory tests, imaging studies, non‐invasive tests and endoscopic evidence of varices. Alcohol use disorder was diagnosed by a psychiatrist using DSM‐5 criteria based on a structured clinical interview. We graded the clinical condition of patients at the time of enrolment according to the Child‐Pugh classification.

2.1. Interventions

The included patients were given naltrexone 50 mg or matching placebo daily for 12 weeks by an independent trial coordinator.

At every visit, patients were assessed for review of consumption and assessment of abstinence from alcohol, overall functioning, difficulties with treatment adherence (pill counts), adverse effects—drug/withdrawal‐related, psychological support and relapse‐prevention counselling at every visit by a psychiatrist. Naltrexone/placebo tablets were entrusted to a family member to administer every dose and monitor for side effects. Patients were asked to bring the empty blisters on every follow‐up visit to assess compliance, and they were also dispensed the pills until the next visit. Brief psychosocial intervention based on the WHO standard brief intervention module was provided at each visit by trained psychiatrists, with standardised supervision to ensure consistency across study arms.

2.2. Assessment and Follow‐Up of Patients

Assessment of alcohol use–related parameters was done using self‐reported history (corroborated by a family member), DSM‐5 clinical diagnosis of AUD, and structured evaluation of abstinence, lapses, relapses, and craving scores at baseline and weeks 4, 8 and 12, with a follow‐up visit at 6 months.

The primary endpoint (abstinence at 12 weeks) was assessed at the week‐12 visit. Lapse and relapse outcomes were recorded at weeks 12 and 24 (6 months), based on alcohol intake in the preceding month. Study medication (naltrexone or placebo) was administered only during the 12‐week treatment period. Between weeks 12 and 24, no participant received open‐label naltrexone or any other pharmacotherapy for alcohol use disorder; this window constituted a medication‐free observational follow‐up, during which all participants continued to receive standard psychosocial support. Alcohol Use Disorders Identification Test (AUDIT) scale was measured at baseline. Craving measurement was done using OCDS [23], visual analogue scale (VAS) [24] and High‐Risk Alcohol Relapse scale (HRAR) (at baseline).

Serum ethyl glucuronide (EtG) was measured selectively in cases as an objective biomarker where there was a discrepancy between patient self‐reported use and the family member interview. EtG was measured using an enzyme‐linked immunosorbent assay (ELISA), according to the manufacturer's protocol. A serum EtG value < 100 ng/mL was considered negative, while values ≥ 100 ng/mL were considered positive for recent alcohol exposure. Serum EtG was interpreted in the context of its short detection window, typically reflecting alcohol intake within the preceding 48–72 h and, in some cases, up to 5 days depending on the timing and quantity of alcohol consumption. EtG was therefore used as an adjunctive tool to support abstinence assessment in discordant cases and was not used as a routine screening test or as a stand‐alone marker of long‐term abstinence.

Baseline investigations included complete blood count, liver and kidney function tests, prothrombin time (international normalised ratio), transient elastography for liver stiffness measurement and contrast‐enhanced CT scan of the upper abdomen. The blood tests were repeated at every follow‐up, and patients were asked to have their tests done in the morning with overnight fasting (8 h). Clinical follow‐up visits occurred weekly during the first month and every 2 weeks thereafter; laboratory assessments followed the same schedule.

2.3. Study Outcomes

The primary outcome of the study was to assess the proportion of patients achieving point‐prevalence abstinence at 12 weeks, defined as non‐consumption of alcohol during the 4 weeks preceding the week‐12 visit. Secondary outcomes of the study included alcohol craving at 12 weeks, lapse and relapse during active treatment (12 weeks), abstinence at 24 weeks and liver‐related adverse events. All secondary outcomes were considered exploratory.

2.4. Study Definitions

Study definitions were prespecified in the 2020 study protocol and approved by the Institutional Ethics Committee. Abstinence was defined as non‐consumption of alcohol for four consecutive weeks preceding the assessment visit (point‐prevalence abstinence). A lapse was defined as consumption of fewer than four standard drinks in a month, and relapse as consumption of four or more standard drinks in a month.

These definitions were selected to reflect clinically meaningful alcohol exposure in patients with compensated cirrhosis, in whom even low‐level intake may adversely affect outcomes. These thresholds correspond to point‐prevalence rather than continuous abstinence and should be interpreted accordingly. Similar pragmatic definitions have been used in earlier alcohol‐associated liver disease trials conducted prior to recent consensus recommendations [25].

2.5. Stopping Rules

Study medication was discontinued in cases of clinical deterioration defined as progression to Child–Turcotte–Pugh class C (score ≥ 10), development of hepatic decompensation, AST or ALT > 5× upper limit of normal, or withdrawal of consent.

2.6. Sample Size

The primary outcome was binary; therefore, sample size estimation was based on a two‐sample comparison of proportions. Assuming a clinically meaningful abstinence rate of 50% in the naltrexone group and 25% in the placebo group, with a two‐sided alpha of 0.05% and 80% power, approximately 45 participants per group were required.

Prior meta‐analyses reporting standardised effect sizes [14] were used to inform clinical plausibility but were not directly applied for sample size estimation, as they derive from continuous outcomes in heterogeneous, predominantly non‐cirrhotic populations.

2.7. Randomization and Allocation Concealment

In this double‐blind, randomised controlled trial, cases were randomly allocated in a 1:1 ratio using a block randomization method with block sizes of 10. Allocation concealment was done using the sequentially numbered opaque sealed envelopes (SNOSEs) method. Concealed allocation was done by the trial coordinator. The investigators, participants and research team were blinded to the treatment allocation. Naltrexone (ADDTREX) 50 mg and identical placebos were provided by Rusan Pharma, Mumbai, India, free of cost. The code was opened at the end of the study by the trial coordinator. The patients and the researchers were not aware of the group to which the cases were being allocated until completion of enrolment.

2.8. Statistical Methods

Statistical analysis was done using SPSS (IBM CORP Ltd., Armonk, New York, USA), version 22.0. Descriptive statistics were presented as proportions, means ± standard deviations and medians with interquartile range. Comparative analysis was performed using Student's t‐test/Mann–Whitney U‐test for continuous variables and chi‐square test/Fisher's exact test for qualitative variables. Univariate and multivariate logistic regression were also used to find the predictors of treatment response. For multivariate analysis, variables with p < 0.05 on univariate logistic regression were entered as candidates; variables with a variance inflation factor (VIF) > 10 were excluded owing to high multicollinearity. Kaplan–Meier analysis was used to calculate cumulative probabilities, and different factors were compared with the log‐rank test. Cox regression analysis was used to depict the predictors of survival. All statistical tests were two‐tailed, and a p < 0.05 was considered significant. Secondary outcomes were exploratory, and no formal adjustment was made for multiple comparisons.

3. Results

Of 127 patients screened, 27 were excluded for not meeting eligibility criteria (hepatic encephalopathy n = 5; bilirubin > 3 mg/dL n = 12; recent GI bleed n = 4; other substance dependence n = 3; unwilling n = 3). One hundred patients met eligibility criteria and were randomised, 50 per arm. One patient in each group was lost to follow‐up, and non‐compliance was recorded in both arms (Figure 1). Intention to treat (ITT) analysis was performed to assess the objectives. Baseline characteristics are presented in Table 1. The patient disposition is shown in the CONSORT flow diagram (Figure 1). Although TLC and INR showed statistically significant differences between groups (p = 0.02 for both), neither difference was considered clinically meaningful, and both parameters remained within clinically acceptable ranges for compensated cirrhosis.

FIGURE 1.

CONSORT flow diagram for the NAL‐CI Trial.

TABLE 1.

Baseline characteristics of the study groups.

Variable	Naltrexone mean ± SD (%)	Placebo mean ± SD (%)	p
Male	50 (100)	50 (100)
Age (years)	42.86 ± 7.8	44.32 ± 8.72	0.21
Comorbidity (n, %)
None	16 (55.2)	13 (48.1)
DM	3 (10.5)	4 (14.8)
Hypertension	1 (3.4)	6 (22.2)
Hypothyroidism	4 (13.8)	2 (7.4)
Smoking (n, %)	15 (30)	20 (40)	0.29
Smokeless tobacco	6 (12)	8 (16)	0.56
Serum creatinine (mg/dL)	0.7 ± 0.8	0.54 ± 0.46	0.89
Haemoglobin (g/dL)	11.2 ± 2.6	11.78 ± 3.40	0.39
TLC (×103/cumm)	5.86 ± 1.30	6.60 ± 1.8	0.02
Platelets (×103 cells/cumm)	154.26 ± 43.47	160.62 ± 77.33	0.61
Total bilirubin (mg/dL)	0.78 ± 0.54	1.42 ± 0.92	0.34
ALT (IU/L)	51.98 ± 27.7	57.38 ± 33.5	0.38
AST (IU/L)	80.06 ± 39.59	81.81 ± 39.1	0.82
GGT (IU/L)	478.5 ± 319.2	412.3 ± 298.9	0.28
Serum albumin (g/dL)	3.56 ± 1.78	4.10 ± 0.81	0.05
INR	1.48 ± 0.37	1.31 ± 0.32	0.02
Baseline CTP	5.92 ± 1.17	6.30 ± 0.78	0.06
Baseline MELD	12.56 ± 6.13	12.70 ± 4.06	0.89
Serum EtG (ng/ml)	75 ± 23.4	66 ± 19.9	0.43
HRAR‐1	3 (10.30)	0	0.12
2	16 (55.2)	15 (55.6)
3	8 (27.60)	12 (44.40)
AUDIT	22.26 ± 2.87	21.86 ± 6.67	0.69
OCDS
–O (obsessive)	14.78 ± 5.54	15.02 ± 5.11	0.82
–C (compulsive)	9.96 ± 2.65	8.60 ± 4.12	0.06
VAS (range 0–10)	7.10 ± 1.61	6.40 ± 3.11	0.19
Number of drinks per month	17.42 ± 5.93	18.8 ± 6.64	0.27
Number of drinks/drinking day	9.70 ± 1.07	9.46 ± 1.43	0.28
Longest abstinence duration in past (weeks)	7.22 ± 2.71	6.68 ± 2.55	0.31

Abbreviations: AUDIT, Alcohol Use Disorders Identification Test; DM, diabetes mellitus; EtG, ethyl glucuronide; HRAR, high risk alcohol relapse scale; OCDS, Obsessive Compulsive Drinking Scale; TLC, total leucocyte count; VAS, visual analogue scale for craving.

3.1. Primary Objective

Point‐prevalence abstinence at 12 weeks was achieved by 32/50 (64%) patients in the naltrexone group, compared with 11/50 (22%) in the placebo group (p < 0.001). On further analysis, of the 32 patients who achieved abstinence at 3 months: 10 patients (20%) did so within the first month and 18 (36%) within the second month.

3.2. Secondary Objectives

Maintenance of abstinence was seen in 11/50 (22%) in the naltrexone group compared with 4/50 (8%) in the placebo group at 6 months, showing a favourable trend (p = 0.09). Missed visits and non‐compliance were infrequent and comparable between groups; intention‐to‐treat analysis was used for all efficacy outcomes.

Alcohol lapses were seen in 28% (14/50) at 3 months and 58% (29/50) at 6 months in the naltrexone group compared with 54% (27/50) and 78% (38/50) at 3 months (p = 0.008) and 6 months (p = 0.03) respectively, in the placebo group.

Similarly, alcohol relapses were less frequent in the naltrexone group: 12% (6/50) at both 3 and 6 months, compared with 28% (14/50) and 22% (11/50) respectively in the placebo arm.

Mean craving scores were lower with naltrexone by week 12: OCDS‐O score (6.63 ± 1.16 vs. 9.29 ± 1.78, p < 0.01) and OCDS‐C score (6.35 ± 1.23 vs. 9.02 ± 1.86, p < 0.01), visual analogue scale for craving (4.27 ± 1.301 vs. 6.51 ± 1.27, p < 0.01). Additionally, higher delta change in mean craving measures were noted in the naltrexone group at 3 months compared with the placebo group (Table 2). Figure 2 shows the cumulative probability of abstinence at 3 and 6 months. Figure 3 shows probability of lapse and relapse during the study period.

TABLE 2.

Delta change between Naltrexone and placebo in craving scores and drinks at the end of 12 weeks.

Variable	Δ values naltrexone (n = 50) Mean ± SD/%	Δ values placebo (n = 50) Mean ± SD/%	p
OCDS‐O	−4.6 ± 1.5	−2.11 ± 0.39	< 0.01
OCDS‐C	−7.65 ± 1.31	−3.3 ± 1.03	< 0.01
Number of drinks per month	−12.24 ± 4.22	−2.32 ± 4.67	< 0.01
Number of drinks per drinking day	−3.41 ± 0.69	−1.02 ± 0.21	< 0.01

Abbreviations: C, Compulsive; O, Obsessive; OCDS, Obsessive Compulsive Drinking Scale; VAS, visual analogue scale.

FIGURE 2.

Cumulative probability of abstinence over time in the naltrexone and placebo groups. The solid line represents the on‐treatment period (0–12 weeks); the dashed line represents the post‐treatment follow‐up period (12–24 weeks). Shaded bands indicate approximate 95% confidence intervals. ITT analysis; n = 100.

FIGURE 3.

Cumulative probability of lapse (Panel A) and heavy‐drinking relapse (Panel B) during the study period. A lapse was defined as fewer than four standard drinks in a month; a relapse as four or more. The dotted vertical line denotes the end of treatment at 12 weeks. ITT analysis; n = 100.

The reduction in the number of drinks per day and the number of drinks per month was greater in the naltrexone group compared with the placebo group. The patients on naltrexone also showed a significant decrease in craving scores (OCDS and VAS) at the end of 12 weeks.

None of the patients developed an upper GI bleed or encephalopathy during our study. None of the patients (on naltrexone or placebo) showed elevated liver enzymes of more than 5 times the upper limit of normal. Non‐compliance with drugs was noted in two patients in the naltrexone group and three patients in the placebo group.

3.3. Predictors of Abstinence

On univariate analysis, concurrent smoking showed a non‐significant trend towards lower abstinence (OR: 0.22, 95% CI: 0.04–1.05, p = 0.06). A higher frequency of binge drinking at baseline significantly reduced the probability of abstinence on both univariate and multivariate analyses (multivariate OR: 0.42, 95% CI: 0.25–0.70, p < 0.05). Higher baseline VAS craving scores were associated with a greater likelihood of abstinence on univariate analysis (OR: 2.47, 95% CI: 1.54–3.96, p < 0.01), potentially reflecting heightened motivation for treatment in more severely craving patients; however, this association did not remain significant in multivariate analysis (OR: 0.83, 95% CI: 0.33–1.34, p = 0.45). Higher AUDIT scores and lower baseline bilirubin, GGT and albumin levels showed significant associations with abstinence on univariate analysis but were not retained in the multivariate model. Notably, allocation to the naltrexone group was the strongest independent predictor of abstinence (OR: 10.86, 95% CI: 1.89–62.2, p < 0.05), as shown in Table 3.

TABLE 3.

Predictors of abstinence at 3 months in AaC patients.

Variables	Univariate OR (95% CI)	p	Multivariate OR (95% CI)	p
Age (years)	0.98 (0.90–1.07)	0.65
Drinks per month at baseline	1.02 (0.94–1.11)	0.62
Smoking	0.22 (0.04–1.05)	0.06	0.32 (0.04–1.12)	0.08
Tobacco	0.17 (0.02–1.51)	0.11
Binge episodes/month	0.39 (0.23–0.66)	< 0.01	0.42 (0.25–0.70)	< 0.05
TLC (×103/cumm)	0.90 (0.59–1.37)	0.61
Hb (g/dL)	1.16 (0.91–1.47)	0.23
Platelet (×103 cells/cumm)	0.99 (0.98–1.00)	0.23
GGT (IU/L)	0.99 (0.99–1.00)	0.05
Bilirubin (mg/dL)	0.32 (0.13–0.78)	< 0.05
Albumin (g/dL)	0.61 (0.40–0.94)	< 0.05
HRAR	1.85 (0.93–3.71)	0.08
AUDIT	0.81 (0.67–0.96)	< 0.05
VAS	2.47 (1.54–3.96)	< 0.01	0.83 (0.33–1.34)	0.45
OCDS	1.15 (0.99–1.33)	0.08
CTP	1.09 (0.59–1.99)	0.79
MELD	1.01 (0.90–1.12)	0.881
Allocation to naltrexone group	6.30 (2.60–15.25)	< 0.01	10.86 (1.89–62.2)	< 0.05

Note: Following variables not reported in multivariate analysis due to high collinearity (VIF > 10): Age (VIF = 25.83); HRAR (VIF = 22.30); Hb (VIF = 19.27); AUDIT (VIF = 18.78); TLC (VIF = 15.39); OCDS (VIF = 10.66); CTP Baseline (VIF = 37.46).

3.4. Adverse Events

These were carefully recorded by patients and relatives. The frequency of liver‐related adverse events was comparable between groups. No participant developed hepatic decompensation attributable to study medication. Overall, adverse events were infrequent in both groups, with no adverse event reported in 36/50 (72%) patients in the naltrexone group and 41/50 (82%) patients in the placebo group. The overall frequency of adverse events was not significantly different between groups. Gastritis was the most reported adverse event, occurring in 5/50 (10%) patients receiving naltrexone and 3/50 (6%) receiving placebo. Other adverse events, including nausea, vomiting and sedation, were uncommon and comparable between groups, as shown in Table 4. 2/50 (4%) patients in the naltrexone group developed jaundice with bilirubin values > 3 mg/dL; both episodes occurred in the context of clinical and biochemical features consistent with ongoing alcohol use; however, drug‐induced liver injury could not be definitively excluded. These patients were given one additional session of brief intervention counselling by the treating psychiatrist, and naltrexone was continued at the same dose.

TABLE 4.

Adverse events between the groups.

Adverse events	Naltrexone (N = 50) (%)	Placebo (N = 50) (%)	p
None	36 (72)	41 (82)	0.34
Nausea	2 (4)	3 (21)	1.00
Vomiting	3 (21)	2 (4)	1.00
Sedation	2 (4)	1 (2)	1.00
Gastritis	5 (22)	3 (21)	0.71
Jaundice	2 (4)	0	0.49

4. Discussion

Sustained alcohol abstinence is the single most important modifiable determinant of long‐term survival in patients with alcohol‐associated cirrhosis (AaC). Several prospective studies have demonstrated that continued alcohol use independently predicts disease progression, hepatic decompensation, and mortality [1, 2]. Despite this, no pharmacotherapy for AUD is currently approved for use in patients with cirrhosis. Acamprosate, disulfiram and naltrexone are each FDA‐approved for AUD in non‐cirrhotic patients but have been routinely withheld from the cirrhotic population, primarily due to concerns regarding hepatotoxicity [3, 4, 5]. This therapeutic gap leaves clinicians managing patients with compensated AaC without a pharmacological tool to complement behavioural intervention.

This double‐blind, placebo‐controlled RCT is, to our knowledge, the first prospective randomised trial to evaluate naltrexone in patients with compensated AaC and DSM‐5 AUD. Patients randomised to naltrexone were 10.86 times more likely to achieve point‐prevalence abstinence at 12 weeks compared with placebo (64% vs. 22%, OR: 10.86, 95% CI: 1.89–62.2, p < 0.001). These findings are consistent with the established efficacy of naltrexone in non‐cirrhotic AUD populations and substantially extend the evidence base to a group that has been systematically excluded from pharmacological trials. The between‐group difference in abstinence rates at 12 weeks is clinically large and compares favourably with active comparators previously evaluated in non‐cirrhotic cohorts [14, 17, 18]. The magnitude of this effect is larger than that reported in general‐population meta‐analyses, in which oral naltrexone yields an absolute reduction in return to drinking of approximately 5% and a number needed to treat of 12–20 [26]. Acamprosate, the other agent with the strongest evidence base in non‐cirrhotic alcohol use disorder, shows a comparable number needed to treat of around 12 [26]. We attribute our larger effect to three features of the present trial: a treatment‐seeking, severely dependent cohort (mean AUDIT≈22) in whom an established diagnosis of cirrhosis provides strong medical motivation for abstinence; directly observed, family‐supervised medication administration that minimised the nonadherence that attenuates naltrexone efficacy in routine practice; and a point‐prevalence abstinence endpoint assessed at a fixed 12‐week time point under intensive follow‐up with standardised psychosocial support in both arms. Prior meta‐analyses have consistently shown that naltrexone effect sizes are larger in men, in more severely dependent patients, and in those with lead‐in abstinence—all features of our cohort [26]. Our number‐needed‐to‐treat is therefore not directly comparable with that of general‐population maintenance trials and should be interpreted in the context of these design and population characteristics.

The theoretical rationale for withholding naltrexone in liver disease rests on dose‐dependent hepatotoxicity observed at supratherapeutic doses in early studies, which led to a cautionary label for patients with active hepatitis or hepatic insufficiency. However, accumulating real‐world evidence challenges the clinical relevance of this concern at standard therapeutic doses. In a retrospective analysis of 100 patients with liver disease treated with naltrexone, Ayyala et al. [19] demonstrated reductions in AST and ALT levels; two patients developed hyperbilirubinaemia attributed to ongoing alcohol use rather than drug toxicity, mirroring our experience. Thompson et al. [20] similarly found no significant hepatotoxic signal in patients with cirrhosis receiving naltrexone at 50 mg/day. At the health‐system level, Rabiee et al. [22] examined 9131 patients with alcohol‐associated liver disease and observed that AUD pharmacotherapy was associated with improved survival, yet only 520 patients (5.7%) had received naltrexone, underscoring the extent of prescriber hesitancy. Our group's prior 4‐week prospective safety observation in a comparable Indian cohort provided the first published evidence of its feasibility in this population [21].

Prior to this trial, baclofen was the only pharmacological agent with randomised controlled evidence for AUD in patients with cirrhosis. In the landmark double‐blind trial by Addolorato et al. [7], baclofen achieved abstinence in 71% of cirrhotic patients compared with placebo. However, a subsequent Cochrane meta‐analysis found no significant reduction in craving at the recommended hepatically safe dose of 30 mg/day [8], and dose‐limiting adverse effects—particularly excessive fatigue and somnolence—were common at higher doses that appeared necessary for clinical effect [9]. Baclofen is consequently not currently recommended by the American Psychiatric Association for the management of AUD [10]. Naltrexone, with its mechanistically distinct opioid‐receptor antagonist profile, a favourable tolerability profile, and a 12‐week abstinence rate of 64% in our cirrhotic cohort, represents a more efficacious and better‐tolerated pharmacological option in this setting.

Naltrexone 50 mg/day was well tolerated over 12 weeks in patients with compensated cirrhosis in our study. Adverse events—nausea (4% vs. 6%), vomiting (6% vs. 4%), sedation (4% vs. 2%) and gastritis (10% vs. 6%)—were comparable between groups (p > 0.05 for all). No participant developed AST or ALT elevation exceeding 5× the upper limit of normal, and no episode of hepatic decompensation was attributable to study medication. Two patients developed jaundice during the study period; on clinical and biochemical review, both episodes occurred in the context of ongoing alcohol consumption and were consistent with alcoholic hepatitis rather than drug‐induced liver injury. These findings are consistent with prior observational reports [19, 20] and with clinical trial data obtained predominantly in non‐cirrhotic patients, which have consistently demonstrated a favourable hepatic safety profile for naltrexone at therapeutic doses [27].

The 64% abstinence rate observed in the naltrexone arm at 12 weeks is broadly consistent with prior trials conducted in non‐cirrhotic populations. Feeney et al. [17] reported abstinence rates of approximately 75% in non‐cirrhotic patients receiving naltrexone combined with cognitive behavioural therapy at 12 weeks, while Chick et al. [18] demonstrated significant reductions in heavy drinking and improvements in abstinence outcomes in a large multicentre randomised trial. Meta‐analytic data confirm that naltrexone reduces the frequency and severity of drinking episodes, lowers monthly drink counts, and significantly improves craving scores compared with placebo [14, 15, 16]. Notably, 22% of patients in the naltrexone arm maintained abstinence at 24 weeks, assessed 12 weeks after drug discontinuation. This sustained effect, while not statistically significant versus placebo at this time point, suggests a degree of behavioural carry‐over consistent with what has been described in the non‐cirrhotic naltrexone literature [17].

Craving reduction is a central mechanism through which naltrexone achieves its clinical benefit. In our cohort, both OCDS subscales (Obsessive and Compulsive) and the VAS craving score showed statistically significant improvements in the naltrexone arm at 12 weeks (OCDS‐O: −4.6 vs. −2.1, OCDS‐C: −7.65 vs. −3.3, p < 0.01 for both). These delta changes are consistent with craving reductions reported in randomised trials of naltrexone in non‐cirrhotic populations [18, 28]. Sustained craving reduction was in turn associated with lower rates of hazardous drinking: alcohol lapses were significantly less common in the naltrexone group at 3 months (28% vs. 54%, p = 0.008) and 6 months (58% vs. 78%, p = 0.03), and there was a clinically meaningful trend toward fewer heavy‐drinking relapses at 3 months (12% vs. 28%, p = 0.07). These antirelapse effects are consistent with prior randomised evidence in non‐cirrhotic populations [15, 16].

The baseline drinking characteristics of this cohort warrant brief contextualisation. Although the mean number of drinks per month appeared relatively modest (17.4 ± 5.9 vs. 18.8 ± 6.6), this is unlikely to reflect true consumption. Mean AUDIT scores of approximately 22 in both arms are consistent with severe alcohol dependence by established thresholds. The mean number of drinks per drinking day was 9–10, indicating concentrated, episodic binge‐pattern consumption—a drinking phenotype well‐recognised in South Asian men with AaC, in whom alcohol use is typically concentrated over fewer days per week but with high‐quantity episodes [25]. High‐Risk Alcohol Relapse (HRAR) scores were elevated in the majority of participants. Serum ethyl glucuronide (EtG) testing was used to corroborate self‐reported abstinence in cases of discrepancy between patient and family reports, in line with current recommendations for alcohol‐associated liver disease trials [25]. Despite these measures, residual underreporting of alcohol intake remains a recognised limitation of self‐report methodology in AUD research.

This study has several limitations that merit consideration. First, it was conducted at a single tertiary referral centre and enrolled an exclusively male cohort, which may limit generalisability to women and to settings with different clinical or social contexts. Second, the 6‐month observational follow‐up may be insufficient to assess the long‐term durability of abstinence, the safety of naltrexone beyond the treatment period, or late relapse rates. Third, although self‐reported alcohol consumption was corroborated by family member interview and supplemented with serum EtG testing in discordant cases, phosphatidylethanol (PEth)—a more sensitive and specific biomarker of recent alcohol intake—was not available at our centre and was therefore not used; this represents a limitation in biomarker‐based abstinence verification [25]. Fourth, patients with decompensated cirrhosis, bilirubin > 3 mg/dL, or active hepatic complications were excluded; findings should therefore not be extrapolated to more advanced liver disease. Fifth, psychosocial support, although standardised to the WHO brief intervention module, was delivered by different members of the psychiatry team, which may have introduced variability in its application. Finally, the trial was powered for the primary 12‐week abstinence endpoint; all secondary and 6‐month outcomes are exploratory and should be interpreted accordingly. Larger, multicentre randomised trials with longer follow‐up, inclusion of female patients and more advanced disease stages, and rigorous biomarker‐based alcohol verification are needed to confirm and extend these findings.

In conclusion, this double‐blind, placebo‐controlled RCT demonstrates that naltrexone 50 mg/day is safe and effective in patients with compensated alcohol‐associated cirrhosis and AUD, producing a threefold improvement in 12‐week abstinence rates and significant reductions in craving and lapses compared with placebo. No clinically significant hepatotoxicity attributable to study medication was observed. Naltrexone has been in clinical use for alcohol use disorder for more than two decades and is already prescribed off‐label to patients with cirrhosis in routine practice, despite the cautionary product label; our trial provides the first prospective randomised confirmation that this real‐world practice is both safe and effective in compensated disease. These findings provide the first prospective randomised evidence supporting naltrexone as a pharmacological option in this population and directly challenge the practice of routinely withholding it from patients with compensated cirrhosis. Larger multicentre trials with longer follow‐up, inclusion of female patients and more advanced disease stages, and biomarker‐based alcohol verification are warranted to establish the generalisability and long‐term durability of these findings.

Author Contributions

S. M Shasthry, Mohit Varshney and Shiv Kumar Sarin conceptualised the study; Manasa Alla monitored the study; Manasa Alla, S. M Shasthry, Mohit Varshney and Shiv Kumar Sarin enrolled the patients, acquisition of data, analysis and interpretation; Ankit Bhardwaj, Guresh Kumar Analysis and data interpretation; Manasa Alla, Mohit Varshney, S. M Shasthry drafted the manuscript and revisions; S. M Shasthry, Shiv Kumar Sarin provided critical intellectual inputs.

Funding

No external funding was received. Naltrexone and matching placebo tablets were provided free of cost by Rusan Pharma, Mumbai, India. Rusan Pharma had no role in study design, conduct, analysis, interpretation, or manuscript preparation.

Ethics Statement

Approved by the ILBS Institutional Review Board (IEC/2020/76/NA04).

Consent

Informed written consent was obtained from all participants. The study was conducted in accordance with the Declaration of Helsinki.

Conflicts of Interest

The authors declare no conflicts of interest.

Alla M., Varshney M., Bhardwaj A., Kumar G., Shasthry S. M., and Sarin S. K., “Naltrexone Is Superior to Placebo for Abstinence and Craving Reduction in Alcohol‐Associated Cirrhosis: NAL‐CI Trial,” Liver International 46, no. 9 (2026): e70845, 10.1111/liv.70845.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.