Predicting steroid response in acute alcohol-associated hepatitis: Beyond biomarkers of alcohol consumption

Abstract

Acute alcohol-associated hepatitis (AAH) is a life-threatening condition with high mortality, and steroid therapy remains the mainstay of treatment despite variable efficacy. The study by Sabatose et al explores patient factors distinguishing responders and non-responders to steroid therapy for AAH, focusing on phosphatidylethanol (PEth)-a biomarker of alcohol consumption-and other clinical variables. Their findings indicate that PEth, abstinence duration, and pre-treatment alcohol intake do not predict steroid response, while older age, lower pre-steroid albumin, and higher pre-steroid bilirubin are associated with non-response. Non-responders exhibit higher mortality and healthcare costs, underscoring the need for early identification to guide liver transplantation referrals. This commentary evaluates the implications of these findings-specifically, how prioritizing pre-steroid albumin, bilirubin, and age over alcohol biomarkers can improve clinical decision-making by reducing unnecessary steroid exposure and expediting transplantation referrals for high-risk non-responders-contextualizes them within existing literature, and highlights directions for future research to optimize AAH management.

Key Words: Acute alcohol-associated hepatitis; Steroid therapy; Treatment response; Biomarkers; Liver transplantation; Phosphatidylethanol

Core Tip: This manuscript offers a thorough analysis of Sabatose et al's study exploring predictors of steroid response in acute alcohol-associated hepatitis (AAH). The study examines the role of alcohol consumption biomarkers (such as phosphatidylethanol) and clinical factors, revealing that age, pre-steroid albumin, and bilirubin levels-rather than alcohol-related markers-are key predictors of non-response. The identification of poorer outcomes in non-responders, coupled with insights into limitations and future research directions, provides valuable guidance for clinical decision-making and the development of improved strategies for AAH management.

TO THE EDITOR

Acute alcohol-associated hepatitis (AAH) is a severe manifestation of alcohol-associated liver disease, characterized by acute-on-chronic liver failure and a 30%-60% mortality rate without intervention[1]. Steroid therapy (e.g., prednisolone) is the standard treatment, yet only approximately 50% of eligible patients respond, and no reliable pre-treatment predictors of response exist[2,3]. The Lille score, which incorporates clinical and biochemical parameters measured 4-7 days after treatment initiation, is used to assess response but cannot guide therapy before steroids are administered[2-5]. Identifying pre-treatment factors associated with non-response is critical to avoid unnecessary steroid exposure (with risks like hyperglycemia and infection) and to expedite alternative interventions, such as liver transplantation[2,6]. Sabatose et al’s[7] study addresses this gap by examining phosphatidylethanol (PEth)-a validated marker of recent alcohol intake[8-12]-and other patient factors as potential predictors, making it a valuable contribution to AAH management research.

METHODS

Sabatose et al[7] conducted a retrospective case-control study of adult patients with AAH (defined by Maddrey’s Discriminant Factor ≥ 32) who received ≥ 4 days of steroid therapy at a single institution between July 2019 and June 2022. A total of 2087 patients were screened, with 64 meeting inclusion criteria. Patients were classified as responders (Lille score < 0.45) or non-responders (Lille score ≥ 0.45) based on day 4 or 7 assessments[4,5].

Data collected included demographics (age, gender, ethnicity), alcohol-related metrics (PEth levels, abstinence duration, weekly standard drinks), and pre-steroid biochemical parameters (albumin, bilirubin, creatinine, prothrombin time). Outcomes (6-month disposition: Death, transplantation, relapse, etc.) were also recorded. Statistical analyses used independent sample t-tests (continuous variables) and χ² tests (categorical variables), with significance set at P < 0.05[13].

KEY FINDINGS

The study's key findings reveal several important insights into the factors associated with steroid response and outcomes in the studied population. First, alcohol consumption biomarkers were found to have no predictive value for steroid response; specifically, PEth levels (whether considered continuously or dichotomized at > 20 ng/mL), the duration of abstinence from alcohol prior to steroid administration, and weekly alcohol intake all showed no significant association with response, with P values of 0.210, 0.456, 0.123, and 0.136, respectively. Second, certain biochemical and demographic factors emerged as predictors of non-response: (1) Non-responders were significantly older, with a mean age of 49 years compared to 42 years among responders (P = 0.024); (2) Had lower pre-steroid albumin levels (2.1 g/dL vs 2.6 g/dL, P = 0.003); and (3) Higher pre-steroid bilirubin levels (24.5 mg/dL vs 16.9 mg/dL, P = 0.010). Third, non-responders experienced notably poorer outcomes within six months, including higher rates of death (9% vs 0%, P = 0.001), liver transplantation (17% vs 2%), and admission to rehab or intensive care units (17% vs 0%), while also having fewer discharges (26% vs 54%) compared to responders. Finally, ethnic background showed no correlation with steroid response, as indicated by a P value of 0.254.

CRITICAL APPRAISAL

Sabatose et al’s[7] study fills a critical methodological gap in AAH research by integrating both validated alcohol-specific biomarkers (PEth) and liver function markers-a combination rarely assessed in prior studies of steroid response. This dual focus is a major strength, as it directly tests whether alcohol consumption history (a modifiable risk factor for AAH progression) correlates with treatment efficacy, rather than relying solely on liver biochemistry. Conceptually, this design helps disentangle two overlapping hypotheses: (1) That recent alcohol intake exacerbates steroid resistance; and (2) That baseline liver damage (not alcohol use itself) drives non-response. The study’s confirmation that liver-specific markers (albumin, bilirubin) outperform alcohol biomarkers aligns with Sarin and Sharma[13] who emphasized that AAH severity-rather than alcohol intake-is the primary determinant of therapeutic outcomes, thereby reinforcing a clinically actionable framework for predicting response.

In terms of methodological rigor, the use of Maddrey’s Discriminant Factor (≥ 32) for AAH diagnosis and the Lille score (day 4/7) for response classification adheres to international clinical guidelines[4], ensuring consistency with prior trials like STOPAH[14]. Reporting of exact P values (e.g., P = 0.024 for age, P = 0.003 for albumin) further enhances transparency, allowing readers to assess the robustness of associations beyond statistical significance.

Notwithstanding these strengths, several limitations merit deeper consideration to contextualize the findings. First, the single-center retrospective design introduces selection bias: The study population (from a single institution) may overrepresent patients with severe disease or specific demographic characteristics (e.g., ethnic homogeneity, as noted by P = 0.254 for ethnic background) that differ from real-world multi-center cohorts[14]. For example, the STOPAH trial (n = 1103) included patients across 33 centers and reported similar response rates (approximately 50%) but with greater diversity in comorbidities (e.g., diabetes, hypertension)-factors unmeasured here that could modify steroid response[14-16]. Second, the small sample size (n = 64) limits statistical power to detect subtle associations, such as potential interactions between PEth and liver markers (e.g., whether high PEth + low albumin correlates with worse outcomes). Third, unmeasured confounders-including adherence to steroid therapy (critical, as non-adherence mimics non-response[6]) and socioeconomic access to post-treatment care (which impacts transplantation referral timing[15,16])-could skew outcome analyses (e.g., higher mortality in non-responders). Finally, the study’s focus on a single PEth threshold (> 20 ng/mL) overlooks the potential for dose-dependent effects: Recent work by Perilli et al[12] suggests PEth levels > 100 ng/mL correlate with advanced liver fibrosis, which may independently predict steroid resistance-an untested hypothesis here that future studies should address.

FUTURE PERSPECTIVES

To build on Sabatose et al’s[7] findings, future research should prioritize methodological and conceptual innovations that address unmet needs in AAH management. First, prospective multi-center studies with larger cohorts (n > 500) are essential to: (1) Validate age, albumin, and bilirubin as independent predictors (adjusting for confounders like comorbidities and adherence[6,15]); (2) Explore multi-marker models that combine these factors with inflammatory markers [e.g., C-reactive protein[17], interleukin (IL)-6[18]] or imaging-based liver stiffness (e.g., transient elastography[6]); and (3) Test whether PEth thresholds > 100 ng/mL (linked to advanced fibrosis[12]) correlate with steroid resistance-a hypothesis unaddressed in the current study. Such models could move beyond binary “responder/non-responder” classifications to predict graded outcomes (e.g., 30-day vs 6-month mortality), enhancing clinical utility.

Second, the study’s finding of high mortality (9%) in non-responders demands a revaluation of transplantation criteria for AAH, which currently require 6-month abstinence[6]. Future work should test two critical questions to stimulate debate: (1) Whether non-responders with early fibrosis (vs cirrhosis) benefit more from urgent transplantation; and (2) Whether combining pre-steroid albumin/bilirubin with post-transplant abstinence predictors (e.g., social support[16], mental health services) improves long-term graft survival. This could challenge the “one-size-fits-all” abstinence rule and prioritize patients most likely to benefit from transplantation.

Third, for non-responders, trials of combination therapies (not just single agents) are needed. For example, pentoxifylline (an anti-inflammatory)[14] plus anti-tumor necrosis factor agents (e.g., infliximab[18]) may target the dual drivers of AAH-hepatic inflammation and cell death-more effectively than steroids alone. Additionally, research should explore whether “steroid-sparing” regimens (e.g., low-dose steroids + IL-6 inhibitors) reduce adverse effects (hyperglycemia, infection[2]) in non-responders, a gap in current therapeutic strategies.

Finally, cost-effectiveness analyses should quantify both direct and indirect savings of early non-responder identification: Direct savings from reduced intensive care unit stays[15] and avoided steroid-related complications; and indirect savings from shorter transplant wait times (by prioritizing high-risk non-responders). Such data would be critical to advocate for policy changes-e.g., reimbursement for multimarker predictive tests-in resource-limited settings.

CONCLUSION

Sabatose et al’s[7] study highlights that liver-specific biochemistry (albumin, bilirubin) and age, rather than alcohol consumption markers like PEth, predict steroid response in AAH. These findings empower clinicians to identify non-responders early, guiding timely liver transplantation referrals and reducing unnecessary costs. Future research should focus on validating these predictors in larger cohorts and integrating them into clinical decision-making to improve AAH outcomes.