APRI and FIB-4 for Liver Fibrosis Assessment in Chronic Hepatitis B: Evidence from a Hyperendemic Andean Setting

Background/Objectives: The 2024 WHO guidelines incorporated the APRI index into treatment decisions for chronic hepatitis B, with thresholds of >0.5 for significant fibrosis and >1.0 for cirrhosis, because it can be computed when neither biopsy nor elastography is available. The evidence supporting those thresholds derives almost entirely from studies using liver biopsy as reference standard; how serological indices behave when the available comparator is elastography remains unknown, and no data exist for Andean populations. We aimed to quantify the agreement of APRI and FIB-4 with transient elastography, determine whether either discriminates better, and evaluate the 2024 WHO thresholds in this setting. Methods: Retrospective agreement study of patients in the Hepatitis B Elimination Plan of Huanta, Peru. APRI and FIB-4, computed from routine laboratory results, were compared against transient elastography (FibroScan 402, M+ probe) used as a reference standard given the unavailability of biopsy. Outcomes followed the 2024 WHO thresholds: >7.0 kPa for significant fibrosis and >12.5 kPa for cirrhosis. Only examinations meeting the pre-specified quality criterion (at least 10 valid measurements; IQR/median ≤ 30%) were analyzed. Diagnostic properties were estimated with exact confidence intervals and likelihood ratios, areas under the curve with bootstrap intervals, and the between-index comparison with DeLong’s test for paired curves. Reporting followed STARD 2015. Results: Of 92 patients, 73 had a valid elastography examination; APRI was computable in 60 and FIB-4 in 57. Prevalence was 23.3% (95% CI 14.2–34.6) for significant fibrosis and 12.3% (5.8–22.1) for cirrhosis. For significant fibrosis, APRI > 0.5 achieved 61.5% sensitivity (31.6–86.1) and 74.5% specificity (59.7–86.1), with a positive likelihood ratio of 2.41 (1.26–4.62) and a negative one of 0.52 (0.25–1.05). For cirrhosis, APRI > 1.0 yielded 71.4% sensitivity, 92.5% specificity and a positive likelihood ratio of 9.46 (3.30–27.11). Areas under the curve were 0.844 and 0.808 for significant fibrosis and 0.914 and 0.874 for cirrhosis; DeLong’s test detected no statistically significant difference between indices at either level (p = 0.47 and 0.53); however, the limited number of outcome-positive cases (13 and 7, respectively) means that clinically important differences cannot be excluded and equivalence has not been established. Conclusions: APRI and FIB-4 identified cirrhosis with likelihood ratios large enough to act on but were only weakly informative at the threshold that governs treatment eligibility, where the negative likelihood ratios did not exclude unity. Neither index can be used as a stand-alone rule-out test at that threshold. The small number of outcome-positive cases limits precision throughout. Although the program intended both assessments to occur during the same session, individual dates were unavailable, preventing verification of the actual interval between elastography and blood sampling.