Clinical Profile and Prognostic Predictors in Hepatitis A Virus-Induced Acute Liver Failure

The incidence of hepatitis A virus (HAV)-induced acute liver failure (ALF) is rising in India, presenting a significant public health challenge. Despite this trend, validated prognostic markers specifically tailored for HAV-ALF remain scarce. Existing prognostic tools such as the Model for End-Stage Liver Disease (MELD), Acute Liver Failure Study Group Index (ALFSGI), King's College Criteria (KCC), Hepatitis A-related ALF (ALFA) score, and ALF Early Dynamic (ALFED) model have been applied in Indian patients. However, these models were developed for mixed-etiology ALF cohorts (except the ALFA score) and have not been specifically validated for HAV-ALF. Thus, the need for HAV-specific prognostic indicators remains critical in the Indian context.

This retrospective observational study included all consecutive patients aged ≥14 years with serologically confirmed HAV-induced ALF admitted to Mahatma Gandhi Medical College and Hospital, Jaipur, between March 2023 and March 2025. Patients were grouped based on outcomes: survived with medical management vs. died. Clinical and laboratory parameters, along with prognostic scores (MELD, ALFSGI, ALFA, KCC, and ALFED), were analyzed. Independent predictors of poor outcome were identified using multivariate logistic regression and prognostic accuracy was evaluated using receiver operating characteristic curve analysis.

Among 79 patients (mean age: 20 ± 6.05 years, 81% male) with HAV-ALF, 54 (68.35%) recovered with supportive care, 5 (6.33%) underwent liver transplantation (with 100% 30-day survival), and 20 (25.32%) died. On univariate analysis, higher MELD score, prolonged prior hospitalization, elevated international normalized ratio, lactate, ammonia, and creatinine were significantly associated with poor outcome. Multivariate logistic regression revealed that MELD score ≥32 (P = 0.003), prior hospitalization >3 days (P < 0.001), and elevated serum lactate ≥7.1 (P < 0.004) were independent predictors of mortality. Among prognostic tools, MELD had the highest predictive accuracy (AUC = 0.768). Combining ALFSGI and ALFA scores improved overall accuracy to 68.52%, while KCC demonstrated limited sensitivity and poor standalone predictive value.

Conventional prognostic criteria like KCC have limited utility in HAV-ALF. The MELD score demonstrates the best predictive accuracy and may be a reliable tool in this setting. While ALFA and ALFSGI scores show potential, their combined use offers modest improvement and warrants further validation in HAV-ALF populations.