Alcohol use disorder with cirrhosis has no approved pharmacotherapy, and the reason is circular: naltrexone carries a historical hepatotoxicity warning, so patients with liver disease were excluded from the trials that would have told us whether it is safe in liver disease. NAL-CI broke that loop.
This randomised trial enrolled 100 patients with compensated alcohol-associated cirrhosis and DSM-5 alcohol use disorder, assigning them 1:1 to naltrexone 50 mg daily or placebo for 12 weeks. Both arms received standardised psychosocial support. Groups were well matched — mean MELD 12.6 versus 12.7, Child-Turcotte-Pugh 5.9 versus 6.2, mean age about 43.
Point-prevalence abstinence at 12 weeks, defined as no alcohol in the preceding four weeks, was 64% with naltrexone and 22% with placebo (odds ratio 10.86, 95% CI 1.89 to 62.2, p<0.001). Lapses at three months fell from 54% to 28% (p=0.008). Heavy-drinking relapse showed a trend, 12% versus 28% (p=0.07), and abstinence maintained at six months was 22% versus 8% (p=0.09). Craving scores on both Obsessive Compulsive Drinking Scale subscales were significantly lower.
The safety result is what unlocks practice. No patient developed hepatic decompensation attributable to the drug, and no transaminase exceeded five times the upper limit of normal in either arm.
This is an Indian trial in Indian patients, published from a centre where alcohol-associated liver disease is a dominant workload, and the mean age of 43 tells its own story about who is presenting with cirrhosis. Where addiction services are scarce, a 50 mg tablet that a hepatologist can prescribe in the liver clinic is a different proposition from a referral that will not happen.
- Consider naltrexone 50 mg daily in compensated alcohol-associated cirrhosis with alcohol use disorder, alongside psychosocial support
- Do not withhold it on hepatotoxicity grounds in compensated disease — no decompensation attributable to the drug and no transaminase above five times normal
- This applies to compensated cirrhosis only; decompensated patients were not studied
- Both arms received structured psychosocial support, so prescribe the drug as an addition to counselling, not a replacement
- Check baseline transaminases and repeat them, and confirm the patient is not on opioids before starting
The statistics, in plain English
An odds ratio of 10.86 with a confidence interval running from 1.89 to 62.2 is an extreme width, and it comes from a small trial with 100 patients. The interval excludes no effect, so a benefit exists, but its size could be anything from modest to enormous — quote the absolute figures instead, 64% versus 22%, which are far more informative. Two secondary results, heavy-drinking relapse at p=0.07 and six-month abstinence at p=0.09, did not reach significance and should not be quoted as findings. Note the endpoint is point-prevalence abstinence, meaning no drinking in the four weeks before the 12-week assessment, not continuous abstinence throughout.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for gastroenterology & hepatology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free