Multicenter Evaluation of Large Language Models Versus Hepatologists for Prognostic Prediction in Drug-Induced Liver Injury.

Initiate the DILI evaluation only for clinically significant liver injury, serum AST or ALT > 5× ULN, or ALP > 2× ULN (or the pretreatment baseline if that value is abnormal) on two separate occasions; total serum bilirubin > 2.5 mg/dL with an elevated AST, ALT or ALP; or INR > 1.5 with an elevated AST, ALT or ALP, not for any minor liver-test abnormality (GS 13). In a patient with new liver test abnormalities after drug, herbal or dietary supplement (HDS) exposure, take a detailed medication and HDS history including intravenous agents such as monoclonal antibodies; most hepatotoxic drugs cause injury within the first 6 months, though some agents have longer latency and hypersensitivity reactions may appear within 24-72 hours (GS 14). Treat DILI as a diagnosis of exclusion: test for hepatitis A IgM, HBsAg, anti-HBc IgM and HCV RNA, obtain autoantibodies (ANA, anti-smooth muscle) and serum immunoglobulins, add antimitochondrial antibody in cholestatic presentations, test anti-HEV IgM in selected cases, and image the biliary tree, usually with abdominal ultrasound (GS 16). Categorize the case by R value at presentation, R = (ALT/ULN)/(ALP/ULN) (GS 15). Liver biopsy is NOT required for diagnosis but may help in severe or protracted cases, when there is diagnostic uncertainty, or when biochemistries/symptoms fail to improve after dechallenge (GS 19-20). For causality, note that the guidance does not endorse a single instrument: it describes three commonly used methods each with strengths and limitations (GS 24), observes that DILIN semiquantitative expert opinion is frequently used in practice and research but requires specialized expertise that limits generalizability (GS 26), and that updated RUCAM has better instructions but retains age, alcohol and pregnancy as risk factors of unclear value (GS 27). Management: discontinuation of the suspect drug plus supportive care, antiemetics, antipruritics, hydration, is the mainstay (GS 39). A short course of intravenous N-acetylcysteine may benefit hospitalized ADULT patients with DILI-related acute liver failure, but NAC is NOT recommended for children (GS 40); in acetaminophen overdose with established liver injury, continue NAC infusion at 6.25 mg/kg/h until ALT is falling and INR < 2. Corticosteroids for 1-3 months may benefit selected patients, severe hypersensitivity features, DRESS, immune checkpoint inhibitor injury with ALT > 5× ULN, or autoimmune-like features on biopsy, but optimal dose and duration are unknown for lack of controlled trials; typical dosing is prednisone 40-60 mg/day, or methylprednisolone-equivalent 1 mg/kg/day for checkpoint inhibitor cases (GS 41). Ursodeoxycholic acid is not established therapy but is presumably safe, and may be used at 10-15 mg/kg/day in divided doses for severe pruritus (GS 42). Because spontaneous survival in idiosyncratic DILI-related acute liver failure is only about 25%, transfer such patients EARLY to a liver transplant center (GS 36); roughly 10% of registry patients suffer ALF, transplant or death within 6 months (GS 35), with higher bilirubin and INR, lower albumin, severe necrosis or fibrosis on biopsy, comorbidity burden, and pre-existing liver disease marking increased risk (GS 38). Rechallenge with the suspect drug should generally be avoided UNLESS the anticipated benefit is high for a severe or life-threatening condition, e.g. isoniazid rechallenge causes recurrent injury in only 10% but should be attempted only in active, drug-resistant TB (GS 44). Follow liver tests to resolution: DILI persisting beyond 6-12 months (chronic DILI) occurs in 10%-20% of patients (GS 37).