Steatosis and Interferon Associated with HBsAg Immune Control in Chronic Hepatitis B: A Real-World Propensity Score-Matched Study

Abstract

Background/Objectives: The baseline determinants of functional cure in chronic hepatitis B (CHB) are largely unknown. By applying propensity score matching (PSM) to real-world data, we aimed to identify traits associated with functional cure. Methods: We included CHB cases which achieved a functional cure and randomly selected non-achievers from patients followed from 2000 to 2020. Initial screening of baseline candidate traits was conducted using PSM-balanced cases and controls. Subsequently, through multiple rounds of leave-one-covariate-out on the balanced cohorts, we validated the impact of these traits using survival analysis. Results: In total, 85 cases (mean age: 35.78; female/male: 23/62) were compared with 247 controls (mean age: 37.08; female/male: 80/167, out of 3666), with a median follow-up of 69.56 months. Steatosis and interferon (IFN) treatment were significantly more frequent in the cases, as confirmed by forest plots showing significant hazard ratios. During validation, whether through balancing all covariates or leave-one-covariate-out matching, both steatosis and exposure to IFN resulted in a higher number of functional cures and HBsAg seroconversions. Further comparisons revealed that add-on or monotherapy outperformed switching (from IFN to NUC), while the de novo (IFN + NUC, followed by NUC) approach was not observed. Conclusions: We confirmed that individuals with steatosis at baseline or those who received IFN were more likely to achieve HBsAg immune control, with monotherapy/add-on therapy being emphasized.

Keywords: HBsAg seroconversion; functional cure; interferon; propensity score matching; steatosis.

Figure 1

Schematic view of the analytic flow. The analytic procedures were numbered sequentially from 1 to 5.

Figure 2

Screening stage: forest plot for Cox proportional hazards model before and after debiasing on steatosis. The (left) graph displays the hazard ratios for sex, age, steatosis, IFN, and baseline HBsAg level before matching. The (right) graph presents the same analysis, with steatosis treated as a strong covariate and balanced. Dotted vertical line represents a ratio of 1. The position of a black square is a point estimate of the ratio. **: p < 0.01 (highly significant); ***: p < 0.001 (extremely significant).

Figure 3

Validation stage: exhaustive leave-one-covariate-out analysis of steatosis/IFN on HBsAg outcomes. The left blue plots show survival analysis based on steatosis status, while the right purple plots represent survival outcomes depending on whether IFN was used. Plots with significant levels are highlighted in color. The middle section indicates the covariates being considered, with grayed-out covariates being excluded.

Figure 4

Antiviral strategy comparisons before and after PSM. Each paired bar plot illustrates the proportion of treatment choices between cases and controls. The (upper) two graphs encompass all choices, including NUC monotherapy, while the (lower) two focus solely on comparing IFN strategies.