• Citer

Diagnostic performance of a new algorithm combining simple, non-invasive and inexpensive tests for predicting the presence of advanced liver fibrosis in patients with chronic hepatitis B

par Nana, Jean;Bosson, Jean Luc;Skaare, Kristina;Vermorel, Céline;Leroy, Vincent;Asselah, Tarik;Adler, Michael ;Zarski, Jean-Pierre
Référence BMC gastroenterology, 25, 1, 447
Publication Publié, 2025-12

Article révisé par les pairs

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Background: Non-invasive methods for scoring fibrosis have been developed to overcome the limitations of liver biopsy. These techniques have not been fully validated for assessing liver fibrosis in chronic hepatitis B (CHB). This study evaluated the utility of combining new, simple, non-invasive, and inexpensive tests to predict advanced liver fibrosis in CHB patients, assisting primary care physicians, especially in resource-limited settings. Methods: This prospective cross-sectional study was conducted on consecutive patients from two centers (n = 408 patients, training group) and one other center (n = 194, external validation group) who underwent liver biopsy for CHB. A decision tree was developed using the cohort, with a cost matrix penalizing type II error to predict patients in stages F0-F1, F2 or F3-F4. The final decision tree contains eight leaf nodes and is based on the following five variables: platelet count, alanine aminotransferase (ALT), aspartate aminotransferase (AST) and gamma-glutamyl transferase (GGT) levels, and age. Results: A total of 408 patients were used to develop a decision tree algorithm, which established Decision Rule No. 1 for predicting the risk of advanced fibrosis (F3-F4): if platelets < 185 × 109/L and age ≥ 38 years and GGT ≥ 27 IU/L, then the patient is classified as F3-F4. This rule applies to 60 out of 408 patients (14.7%), of whom 41 were correctly classified as F3-F4 (68.3%) based on liver biopsy findings (Se = 53.9, Sp = 94.3, PPV = 68.3, NPV = 89.9). On the other hand, combining platelets, GGT, patient age, AST, and ALT (Decision Rule No. 2) allows for the classification of certain patients as having no significant fibrosis. This rule, used to predict F0-F1, applies to 260 out of 408 patients and demonstrates good diagnostic performance (Se = 90.1, Sp = 74.7, PPV = 83.8, NPV = 83.8). Conclusion: A new algorithm combining simple, non-invasive, and inexpensive tests demonstrates a good diagnostic value in predicting advanced liver fibrosis in patients with CHB or excluding significant fibrosis. These two rules can aid clinical decision-making by optimizing and limiting the use of more complex complementary and therapeutic examinations, particularly when such resources are rare.