Highlight
In adults with steatotic liver disease, the MetALD-ALD Prediction Index (MAPI) used five routinely available variables: sex, mean corpuscular volume, gamma-glutamyltransferase, high-density lipoprotein cholesterol, and hemoglobin A1c, to estimate the likelihood of excessive alcohol use as defined with phosphatidylethanol (PEth).
MAPI showed good and reproducible discrimination in both derivation and external validation cohorts, with AUROCs of 0.76 and 0.75, respectively.
Among commonly used indirect alcohol biomarker models, MAPI was the top performer, supporting its potential role as a pragmatic screening tool when direct PEth testing is unavailable or needs to be selectively deployed.
The study is clinically relevant because distinguishing metabolic dysfunction–associated steatotic liver disease with excess alcohol use from purely metabolic disease remains difficult in routine practice, yet the distinction affects counseling, prognosis, research classification, and potentially therapeutic strategy.
Background and Clinical Context
Accurate assessment of alcohol exposure in patients with steatotic liver disease has become increasingly important as liver disease nomenclature and classification evolve. The newer framework recognizes a spectrum that includes metabolic dysfunction–associated steatotic liver disease and alcohol-associated liver disease, with an overlap phenotype often referred to as MetALD when metabolic dysfunction coexists with alcohol consumption above accepted thresholds. In real-world practice, this distinction is often blurred. Self-reported alcohol intake is vulnerable to recall bias, social desirability bias, and under-reporting, particularly when patients perceive stigma or anticipate therapeutic consequences.
Phosphatidylethanol is one of the most useful direct biomarkers of alcohol use because it forms only in the presence of ethanol and has better specificity for sustained alcohol exposure than conventional liver tests. However, PEth testing is not universally accessible, may be costly, and is not yet integrated into many primary care, hepatology, or population-screening workflows. This creates a practical gap: clinicians need a scalable method to identify which individuals with steatotic liver disease are most likely to have clinically meaningful excess alcohol use and may therefore benefit from confirmatory PEth testing or more intensive alcohol assessment.
The study by Tavaglione and colleagues addresses this gap by developing and validating an indirect biomarker panel derived against a PEth-based reference definition. This is a notable design choice. Prior alcohol indices often relied on self-report or were developed in broader populations rather than specifically in people with steatotic liver disease, where obesity, diabetes, dyslipidemia, and hepatic inflammation can confound laboratory signals.
Study Design
Derivation cohort
The derivation cohort included 503 community-dwelling adults with overweight or obesity and steatotic liver disease living in Southern California. All participants underwent magnetic resonance imaging and magnetic resonance elastography assessment as well as PEth testing. The use of advanced imaging is a strength because it provides a well-characterized liver phenotype rather than relying only on liver enzymes or administrative coding.
Outcome definition
The study outcome was the presence of MetALD-ALD, defined by incorporating PEth. Although the abstract does not detail the exact PEth threshold used, the core methodological point is that the model was anchored to an objective alcohol biomarker rather than to self-reported intake alone. This improves construct validity for a tool intended to detect excessive alcohol use.
Model development
The authors used bidirectional stepwise logistic regression to identify the optimal predictive model. Internal validation was performed with 2000 bootstrap samples, an appropriate method to estimate optimism and reduce overinterpretation of apparent performance in the derivation set. The final model was named the MetALD-ALD Prediction Index, or MAPI.
External validation
External validation was conducted in an independent Swedish population-based cohort of 1777 individuals with available PEth measurements. This is an important aspect of the study because external validation across different health systems and geographic settings is essential before considering broader implementation.
Performance metric
Discrimination was assessed using area under the receiver operating characteristic curve. AUROC is a standard measure of a model’s ability to separate cases from non-cases across thresholds, though it does not on its own establish calibration or clinical utility at specific cutoff values.
What Is in MAPI?
The final MAPI model included five variables: sex, mean corpuscular volume, gamma-glutamyltransferase, high-density lipoprotein cholesterol, and hemoglobin A1c.
These components are clinically plausible. Mean corpuscular volume has long been associated with chronic alcohol exposure, though it is nonspecific and can be influenced by vitamin deficiency, bone marrow disorders, thyroid disease, and medications. Gamma-glutamyltransferase is a classic alcohol-associated laboratory signal, but it also rises in cholestasis, obesity, diabetes, and medication exposure. High-density lipoprotein cholesterol can increase with alcohol intake, although metabolic dysfunction complicates its interpretation. Hemoglobin A1c likely contributes information about the metabolic background and may help distinguish overlap phenotypes within steatotic liver disease. Sex is relevant because alcohol-related biochemical effects and drinking patterns differ between men and women, and guideline thresholds for unhealthy alcohol use are sex-specific in many frameworks.
Importantly, none of these markers is sufficiently specific on its own. The value of MAPI lies in combining several imperfect but complementary variables into a single risk estimate calibrated against PEth.
Key Findings
Overall discrimination
MAPI achieved an AUROC of 0.76 in the derivation cohort and 0.75 in the external validation cohort. This degree of performance is best interpreted as good but not definitive. In practical terms, MAPI is not a replacement for direct biomarker testing or detailed clinical assessment. Rather, it appears useful as a screening and triage instrument, especially in settings where PEth cannot be routinely ordered for everyone with steatotic liver disease.
Comparison with existing indirect alcohol biomarkers
The authors report that MAPI was the top-performing model among commonly used indirect alcohol biomarkers based on AUROC. This is a clinically meaningful result because existing indirect markers, whether used individually or in older composite scores, often lose specificity in patients with obesity, insulin resistance, and fatty liver. A score developed specifically in the steatotic liver disease population and trained against PEth addresses that limitation more directly than traditional alcohol biomarker panels.
Validation across different populations
The stability of performance between the Southern California derivation cohort and the Swedish validation cohort is encouraging. External validation in a population-based cohort suggests that the model is not merely detecting local practice patterns or idiosyncrasies of one recruitment strategy. At the same time, the two cohorts likely differed in age distribution, genetic background, alcohol culture, metabolic risk burden, and referral patterns, making the preserved AUROC a favorable signal for transportability.
Clinical implications of the AUROC range
An AUROC around 0.75 places MAPI in a useful middle ground. It is stronger than many single laboratory markers and potentially valuable for case finding, but it should not be overextended into high-stakes decisions in isolation. For example, MAPI could help identify patients with steatotic liver disease who should undergo PEth testing, more formal alcohol history assessment, or counseling. It may also help enrich observational cohorts or clinical trials that lack direct alcohol biomarker data, although some misclassification is inevitable.
Why This Study Matters
The most important contribution of this work is methodological. Alcohol exposure is one of the most mismeasured variables in hepatology. When classification depends primarily on self-report, epidemiologic estimates, treatment effects, and prognostic models may all become distorted. By using PEth as the reference standard to develop an indirect panel, the investigators improve the credibility of the resulting score.
This has direct consequences for several areas of practice. First, patients labeled as having purely metabolic fatty liver disease may in fact have clinically relevant alcohol exposure, which could accelerate fibrosis progression and alter the counseling they need. Second, clinical trials in steatotic liver disease often attempt to exclude or stratify alcohol use, but many cohorts lack PEth. A scalable index could help identify participants who warrant closer review. Third, from a public health perspective, a low-cost panel that uses commonly available laboratory data may be more feasible than universal direct biomarker screening.
Expert Commentary
From a hepatology standpoint, MAPI should be viewed as a pragmatic triage tool rather than a diagnostic endpoint. Its strongest use case is probably a two-step strategy: first apply MAPI using routine laboratory data, then perform PEth testing in those with intermediate or high predicted probability of MetALD-ALD. Such an approach could improve resource allocation while reducing dependence on self-report alone.
The selected variables also reflect a biologically coherent overlap between alcohol-related effects and metabolic disease. This is especially important because steatotic liver disease is rarely driven by a single factor. Obesity, insulin resistance, dyslipidemia, sex-related differences, and alcohol exposure interact in a nonlinear way. A model that acknowledges this intersection is more aligned with modern disease concepts than legacy alcohol screens developed outside hepatology.
Nevertheless, clinicians should be cautious. A score composed partly of liver-associated and metabolic variables may perform differently in populations with advanced cirrhosis, active inflammatory liver disease, significant cholestasis, hematologic disorders affecting MCV, or treatment effects from lipid-lowering and glucose-lowering drugs. If MAPI is to become widely used, calibration and threshold analyses in these specific subgroups will matter as much as headline AUROC values.
Strengths
The study has several strengths. It used a prospective derivation cohort with objective liver phenotyping by magnetic resonance methods. It anchored the target condition to PEth, a direct alcohol biomarker with strong clinical relevance. The investigators performed bootstrap internal validation and then tested the model in a separate, large external cohort. The final model uses inexpensive and familiar laboratory variables that are available in routine care, which improves scalability and implementation potential.
Limitations and Remaining Questions
Several limitations deserve attention. First, AUROC alone does not tell clinicians which cutoff should be used in practice, nor does it provide positive and negative predictive values across prevalence settings. Those quantities are essential for real-world adoption, especially because the prevalence of excess alcohol use differs substantially between specialty clinics and community populations.
Second, stepwise regression is common in clinical score development but can be sensitive to sample-specific correlations and may miss more robust shrinkage-based alternatives. Although bootstrap validation reduces optimism, future work comparing MAPI with penalized regression or machine-learning approaches may be worthwhile.
Third, the abstract does not provide detailed calibration metrics, decision-curve analysis, or subgroup performance by fibrosis stage, diabetes status, sex, or ethnicity. These data would help determine whether MAPI is equally reliable across clinically important strata.
Fourth, PEth itself, while highly useful, is not a perfect gold standard. Concentrations vary with drinking patterns, timing, red blood cell turnover, and assay-related factors. Therefore, any PEth-derived model inherits some limitations of the reference biomarker.
Fifth, the derivation cohort consisted of adults with overweight or obesity and steatotic liver disease. The model may not generalize as well to lean steatotic liver disease, advanced decompensated cirrhosis, transplant populations, or settings with markedly different alcohol consumption patterns.
Implications for Clinical Practice
For clinicians in hepatology, internal medicine, and primary care, MAPI offers a potentially practical way to improve alcohol risk stratification in steatotic liver disease. It may be most useful in three scenarios.
First, in routine clinic workflows where alcohol history is uncertain or potentially underestimated, MAPI could flag patients for PEth confirmation. Second, in systems where PEth is available but constrained by cost or logistics, MAPI could support targeted ordering rather than universal testing. Third, in retrospective cohorts or pragmatic studies without stored specimens for PEth, MAPI may provide a more informed estimate of likely excess alcohol use than self-report alone, though this use should be accompanied by acknowledgment of residual misclassification.
However, MAPI should not substitute for a careful, nonjudgmental alcohol history. Nor should it be used as a standalone basis for stigmatizing labels, denial of therapy, or legal and occupational decisions. Biomarker panels are most useful when integrated with clinical interview, medication review, comorbidity assessment, and, where available, direct testing.
Research and Policy Implications
This work aligns with a broader shift toward biomarker-supported disease classification in hepatology. Future studies should define clinically actionable thresholds, test cost-effectiveness of MAPI-guided PEth strategies, and assess performance longitudinally as alcohol use changes over time. It will also be important to determine whether MAPI predicts liver-related outcomes such as fibrosis progression, steatohepatitis, decompensation, and mortality beyond merely classifying alcohol exposure.
For health systems and policy experts, the study raises a practical question: should alcohol biomarker testing become more routine in steatotic liver disease? If universal PEth testing remains impractical, a validated indirect index like MAPI could serve as a bridge strategy. But implementation should be accompanied by clinician education, transparent threshold selection, and safeguards against punitive misuse.
Conclusion
The MetALD-ALD Prediction Index is a clinically relevant advance in the assessment of alcohol exposure among patients with steatotic liver disease. Built from five routinely available variables and trained against PEth, it achieved consistent discrimination in both derivation and external validation cohorts, with AUROCs of 0.76 and 0.75. Its main value is not to replace PEth, but to extend the reach of alcohol risk screening where direct biomarker testing is limited.
In an era when accurate classification of steatotic liver disease increasingly matters for patient counseling, prognosis, research eligibility, and health system planning, MAPI provides a practical and scalable option. The next step is to define how best to integrate it into real-world care pathways, ideally as part of a staged strategy that combines routine laboratory data, skilled clinical interviewing, and confirmatory direct biomarker testing.
Funding and Trial Registration
The abstract provided does not report ClinicalTrials.gov registration information. Readers should consult the full Gastroenterology article for detailed funding disclosures and any protocol registration statements.
References
1. Tavaglione F, Vaz J, Jamialahmadi O, Díaz LA, Ajmera V, Contrano D, Madamba E, Singh S, Bettencourt R, Richards L, Hagström H, Loomba R. The MetALD-ALD Prediction Index: A Phosphatidylethanol-Driven Biomarker Panel for Identifying Individuals With Steatotic Liver Disease and Excessive Alcohol Use. Gastroenterology. 2026;170(5):1033-1046. PMID: 41739005.
2. European Association for the Study of the Liver. EASL clinical practice guidelines on non-invasive liver tests for evaluation of liver disease severity and prognosis. J Hepatol. 2021;75(3):659-689.
3. Singal AK, Bataller R, Ahn J, Kamath PS, Shah VH. ACG Clinical Guideline: Alcohol-Associated Liver Disease. Am J Gastroenterol. 2024;119(1):30-54.
4. Stewart SH, Koch DG, Burgess DM, Willner IR, Reuben A. Sensitivity and specificity of urinary and blood phosphatidylethanol for detecting moderate-to-heavy alcohol consumption in liver disease patients. Alcohol Clin Exp Res. 2014;38(3):844-850.
5. Isaksson A, Walther L, Hansson T, Andersson A, Alling C. Phosphatidylethanol in blood as a marker of ethanol consumption in healthy volunteers: comparison with other markers. Alcohol Clin Exp Res. 2011;35(10):1832-1837.
6. Hagström H, Nasr P, Ekstedt M, Hammar U, Stål P, Hultcrantz R, Kechagias S. Risk for development of severe liver disease in lean patients with nonalcoholic fatty liver disease: a long-term follow-up study. Hepatol Commun. 2018;2(1):48-57.
7. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, Romero D, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology. 2023;77(4):1335-1347.