• Citer

Plasma-Signature-Model for End-Stage Liver Disease Score to Predict Survival in Severe Alcoholic Hepatitis

par Fujiwara, Naoto;Trepo, Eric ;Raman, Indu;Li, Quan Zhen;Degré, Delphine ;Gustot, Thierry ;Moreno, Christophe ;Hoshida, Yujin
Référence Clinical gastroenterology and hepatology, 20, 3, page (651-657)
Publication Publié, 2022-03-01

Article révisé par les pairs

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Background & Aims: Severe alcoholic hepatitis (AH) is a highly lethal condition and it is still a challenge to predict the outcome. We previously identified and validated a composite score of hepatic 123-gene prognostic signature and the model for end-stage liver disease (MELD) score: gene signature–MELD. However, the need for liver biopsy limits its clinical application. Therefore, we aimed to identify a plasma protein–based surrogate of the gene signature and independently validate its prognostic capability. Methods: All patients were diagnosed with severe AH at Cliniques universitaires de Bruxelles Hôpital Erasme (Brussels, Belgium), and the plasma samples were collected at admission before any treatment. The primary outcome was death or liver transplantation within 90 days. Using our computational pipeline, named translation of tissue expression to secretome (TexSEC), a hepatic-transcriptome–based prognostic signature was converted to a plasma-based secretome signature, which was optimized in 50 patients by comparing their hepatic molecular dysregulation status and combining it with the MELD score. The composite score was validated independently in 57 patients. Results: The TexSEC and optimization process identified a 6-plasma-protein panel as a surrogate for the 123-gene signature. A composite score with the MELD score, the plasma-signature (ps)-MELD score, was created by using the coefficients of the gene signature–MELD equation. In the validation cohort, the high-risk ps-MELD (n = 23; 40%) was associated significantly with death or liver transplantation within 90 days (adjusted hazard ratio, 4.57; 95% CI, 2.15–9.30; P <.001). The ps-MELD score showed a stable, high prognostic association (time-dependent area under receiver operating characteristics curve, >0.80) and was well calibrated over time; it consistently outperformed existing clinical scores as indicated by various model performance indices. Conclusions: The high-risk ps-MELD score was associated with short-term survival in patients with severe AH.