Substance use disorder is associated with alcohol-associated liver disease in patients with alcohol use disorder

Background and Aims: Substance use disorder (SUD) commonly associates with alcohol use disorder (AUD), and certain substances have independently been shown to drive liver injury. In this work, we sought to determine if co-existing SUD in patients with AUD associated with the presence of alcohol-associated liver disease (ALD). Methods: We performed a cross-sectional analysis using the Mass General Brigham Biobank to identify patients based on ICD-10 codes. We performed multivariate analyses accounting for a wide range of demographic and clinical variables to evaluate the association between SUD and ALD. We subsequently used the same method to evaluate the association between SUD and hepatic decompensation. Results: We identified 2848 patients with a diagnosis of AUD, 9.0% of which had ALD. 25.2% had a history of SUD. In multivariate analyses, patients with SUD were more frequently diagnosed with ALD compared to those without SUD (OR = 1.95, P = 0.001). Furthermore, the number of concurrent SUDs was positively associated with the diagnosis of ALD (OR: 1.33, P < 0.001). Independent of the presence of other SUDs, opioid use disorder in patients with AUD was associated with ALD (OR = 1.902, P = 0.02). In subsequent analyses, we found that sedative use disorder was associated with hepatic decompensation (OR: 2.068, P = 0.03). Conclusions: In patients with AUD, SUD, and in particular opioid use disorder, was independently associated with the diagnosis of ALD.


INTRODUCTION
Excessive alcohol use is highly prevalent and contributes to significant morbidity and mortality in society (1,2). It is well-established that excessive alcohol intake increases the risk for developing Alcohol-associated Liver Disease (ALD) (4)(5), a leading cause of morbidity among alcohol users (5). While most patients with excessive alcohol use do not experience severe liver injury, a minority can develop alcohol-related hepatitis or cirrhosis. Severe alcohol-related hepatitis, the most aggressive form of ALD, carries a dismal prognosis (6)(7)(8). Similarly, patients with cirrhosis, and in particular those who have experienced a liver-related decompensating event, face high two-year mortality rates (9). Accordingly, identifying modifiable risk factors for advanced liver disease among patients with alcohol use disorder (AUD) is critical.
In recent years, the use of non-alcohol substances, such as marijuana, opioids, benzodiazepines and cocaine, has increased and exacted significant burden on patients and the medical system (10,15). Due to common epidemiologic, socioeconomic and psychiatric drivers, there is frequently a comorbidity between AUD and other substance use disorders (SUD) (11)(12)(13)(14). In addition, it has been demonstrated that hospitalized patients with ALD have a significantly higher prevalence of non-alcohol SUDs when compared to hospitalized patients without underlying liver diseases (16). Notably, for each of these substances, there exists mechanistic data demonstrating their ability to independently drive hepatotoxicity, thereby potentially accentuating the negative effect of alcohol on the liver in patients with AUD (17)(18)(19)(20).
Because of the increased frequency of non-alcohol substance use in patients with AUD and the potential hepatotoxicity of these substances, we hypothesize that patients with AUD and other substance use history will be more frequently diagnosed with ALD compared to those patients with AUD and no history of other substance use. In this work, we leverage a large and well-characterized cohort of patients with AUD who exhibit varying degrees of liver injury and substance use patterns to examine the association between substance use and the presence and severity of ALD.

Patient cohort
For this cross-sectional associative study, we identified eligible patients in the Massachusetts General Brigham (MGB) Biobank (21), which is a large integrated database comprised of over 125,000 consented subjects within the MGB healthcare system. Robust demographic, clinical, social, and demographic data are available for each patient and accessible through an online platform. Data are derived from the electronic health record and health surveys. All patients included in the database provided informed consent. To identify patients with alcohol use disorder, we queried the MGB Biobank for patients with the following ICD-10 diagnoses: alcohol abuse (F10.1) or alcohol dependence (F10.2). We included all patients with one of these diagnoses that also completed an alcohol questionnaire to quantify their alcohol use.

Demographic and Clinical Factors
We collected the following demographic and clinical data on our cohort: sex, age, race, ethnicity, BMI, viral hepatitis status, drinking history, history of nicotine dependence, substance use history, psychiatric disorder history, and the receipt of medical addiction therapy or psychotherapy (Supplementary Table 1). For lab testing, the most recent values were used in our analyses. Alcohol consumption was quantified from a patient-completed questionnaire which asked the following question: "During the past year, how many alcoholic drinks (glass/bottle/can of beer; 4oz glass of wine; drink or shot of liquor) did you usually drink in a typical week?". Participants were given the following options: 1) None, or less than one per month, 2) 1-3 per month, 3) 2-4 per week, 4) 5-6 per week, 5) 1-2 per day, 6) 3-4 per day, 7) 5-6 per day, or 9) more than 6 per day. We converted these categorical responses into a linear score (0-8) for our multivariate analyses. Viral hepatitis was assessed through analysis of lab testing and diagnosis history. Patients were classified as having hepatitis B if they had a positive hepatitis B antigen test or a diagnosis of chronic hepatitis B in their medical chart. Patients were classified as having hepatitis C if they had a positive hepatitis C antibody test or a diagnosis of hepatitis C in their medical chart. Patients that were missing either a hepatitis C antibody test or a hepatitis B antigen test and were not diagnosed with viral hepatitis were considered untested for the purpose of our multivariate analyses.

Definition of substance use disorder
We identified the subset of patients with AUD who also had a history of a substance use disorder. We determined that patients had a substance use disorder if they had an ICD-10 code in their chart that supported a diagnosis of abuse of or dependence of the following 2) and sedatives, hypnotic or anxiolytic (F13.1 & F13.2). Stimulants include substances such as amphetamines, while sedative, hypnotic or anxiolytics refer to benzodiazepines and barbiturates.

Statistical Analyses
Continuous variables were summarized using means compared using the t test with Welch correction; categorical variables were expressed as numbers and percentages and compared using the Fisher exact test. We also performed univariate logistic regressions to determine the odds ratio (OR) and 95% confidence interval (CI) of having alcohol-associated liver disease for each risk factor, demographic characteristic, and substance use disorder. Multivariate logistic regressions adjusted for the following variables: demographics, risk factors for AUD (homelessness, psychiatric disorders, non-alcohol substance use disorders), specific AUD diagnosis (abuse, dependence or both), the receipt of medical addiction therapy, psychotherapy, alcohol use history (as assessed by questionnaire) and concurrent liver diseases that may contribute to the development and progression of ALD (HBV and HCV positivity, NASH, biliary cholangitis, autoimmune hepatitis, chronic passive liver congestion, alpha 1 antitrypsin deficiency, and hemochromatosis). All analyses were conducted using GraphPad Prism (San Diego, CA).

Patient Demographics
Our analysis consisted of 2848 patients with AUD who had completed an alcohol use questionnaire ( Figure 1). Demographic data are provided in Table 1. We found no difference in he severity of AUD, based on diagnostic coding of alcohol abuse, in patients with AUD with or without SUD (81% versus 78%, respectively, p > 0.05). Many patients had multiple substance use disorders (Supplementary Figure 1)

Substance use disorder is associated with the presence of ALD in patients with AUD
Given the potential biological contribution of substance use to liver disease (17)(18)(19)(20), we first sought to determine if there was an association between the presence of SUD in patients with AUD and a diagnosis of ALD. We found that patients with a history of any drug disorder were more often diagnosed with ALD compared to those without a history of a drug disorder (OR: 1.95, p = 0.001) ( Table 2). In subgroup analyses, AUD patients with concurrent opioid disorder (OR: 1.90, p = 0.01) were more frequently diagnosed with ALD. Notably, we found no association between ALD and sedative, stimulant, cocaine, or cannabis disorder (Table 2). Furthermore, we found a positive association between the number of SUDs and the diagnosis of ALD (OR: 1.33, p < 0.001) Taken together, we found a positive association between the presence of SUD and ALD in patients with AUD, especially in those patients with a history of opioid disorder.

Sedative use disorder, but not other substance use diagnoses, associates with hepatic decompensation in patient with AUD
Given the wide spectrum of ALD, we next sought to determine if there was an association between SUD and hepatic decompensation in patients with AUD. While we did not find an association between SUD and hepatic decompensation (OR: 1.08, p = 0.68), there was a positive association between sedative use disorder and hepatic decompensation (OR: 2.07, p = 0.03) ( Table 3). We did not find an association between ALD and the number of SUDs or the presence of cannabis, cocaine, inhalant, or sedative use disorder. Taken together, we conclude that a diagnosis of sedative use disorder was associated with hepatic decompensation in patients with AUD.

DISCUSSION
In this cross-sectional study, we found a positive association between the presence of ALD and a history of SUD in large cohort of patients with AUD. Notably, opioid use disorder significantly associated with the presence of ALD. Moreover, we found that an increasing number of SUD diagnoses in a patient with AUD also associated with ALD. Finally, we also found that sedative use disorder, but not other SUDs, associated with hepatic decompensation in patients with AUD. These associations were identified in multivariate analyses, suggesting drug use may independently contribute to liver injury in patients with AUD.
It is well established that there is an association between alcohol use and liver disease, ranging from reversible hepatic steatosis, to steatohepatitis, to chronic hepatic fibrosis. The wide spectrum of liver disease observed in patients with alcohol use disorder highlights the numerous factors independent of quantity of alcohol consumed that may contribute to liver injury. While many of these factors are nonmodifiable, there is an urgent and unmet need to identify modifiable risk factors in these patients. We hypothesized that concurrent drug use would represent a key variable in driving liver disease in patients with AUD. This hypothesis was based on established data which highlight the ability of certain substances to mechanistically drive liver injury. In particular, opioids are metabolized in the liver and may accentuate liver injury through lipid peroxidation and mitochondrial injury (23). Furthermore, hepatocytes have numerous opioid receptors that drive pathways which may contribute to liver injury (24). Opioids have been implicated in altering bile acid metabolism as well, which could potentially effect liver homeostasis (25). Accordingly, there is biological plausibility to support our observed findings that substance use associates with liver disease in patients with AUD.
Previous work has highlighted an increase in SUD prevalence among patients with ALD. An analysis of the National Inpatient Sample data from 2011 found an increased likelihood for hospitalized patients with ALD to have SUD compared to hospitalized patients without ALD (16). A more recent analysis of the National Inpatient Sample data found that compared with the patients with chronic liver disease, patients with comorbid chronic liver disease and SUD were more likely to have ALD (26). Our study builds on these data by in multiple ways. First, we study the effect of SUD on liver disease in only those patients with AUD. Second, we demonstrate the association of individual SUDs with ALD. Finally, we show the effect of SUDs on hepatic decompensation in patients with AUD.
Overall, there is an enrichment of substance use in patients with AUD. A recent crosssectional study that demonstrated that 8% of those who use alcohol engage in other drugs; and among those who have AUD 15% have another substance use disorder including hard drugs and marijuana (27). Those with concurrent substance use disorders tend to be male, have lower annual incomes, have more severe alcohol use disorder and tend to have more comorbid psychiatric disorders (27). In our analysis, most patients with AUD did not have a concurrent substance use disorder and controlling for alcohol use did not eliminate the association between hard drug use and liver disease.
It is impossible to examine the effect of drug use simply from a biologic perspective without recognizing the societal context that these patients live in and the stigma they experience. Many people who inject drugs feel judged by the healthcare system and even prior to engaging in it feel like their complaints will be dismissed or they will be turned away (28)(29). Studies utilizing qualitative interviews of people with substance use disorder, revealed that they use the coping strategies of delaying care or seeking care with alternative medicine (29). These feelings are not simply imagined; in reality, many health care providers do harbor negative feelings towards people who use drugs (30). This results in reduced collaboration between professionals and patients, short and more task-oriented visits. Patients under these circumstances are less likely to complete treatment. Thus, even if drug use does not biologically contribute to liver disease, it's stigmatizing effects can create a schism in healthcare equity (29).
Our study has several limitations. First, the observational and cross-sectional nature of our study limits the findings to associations and may have potential co-founders not accounted for on statistical analysis. This is an inherent limitation of an associative study; by adjusting for many factors in our multivariate analyses, we limit this effect as much as possible. Necessarily, however, we cannot be sure that we have completely avoided residual confounding. Second, the determination of AUD and SUD was made by ICD-10 codes, and each patient may not have been fully evaluated by a provider concentrating in addiction medicine. However, while detailed SUD histories taken by addiction specialists would have reduced the noise in our cohort, this should not sum to produce the associations that we demonstrated in this study. Rather, greater noise introduced by imperfect coding increases statistical uncertainty and makes such associations more difficult to demonstrate Despite these limitations, we conclude that in our cohort of patients with AUD, SUD was independently associated with the presence of ALD. Prospective cohort studies are warranted to conclusively establish the altered risk of patients with concurrent SUD and AUD to develop ALD. The flow diagram details the method for patient selection in our analyses. AUD: alcohol use disorder; ALD: alcohol-associated liver disease; SUD: substance use disorder; ICD: international classification of diseases.