Platelet counts modulate the quantitative relationship between hepatitis B viral DNA and surface antigen concentrations: A cross-sectional study of hematological, histological and viral factors

Background: The concentrations of hepatitis B virus (HBV) DNA and surface antigen (HBsAg) are two critical virological variables to be monitored in chronic hepatitis B. HBsAg is derived from the HBV genome. Thus, higher HBV-DNA concentrations should implicate higher HBsAg levels. Nevertheless, the two variables do not manifest a simple linear relationship due to elusive host factor involvements. The aim of this study was to address the discrepancy of HBV DNA and HBsAg levels by a quantitative modeling of HBsAg concentrations. Methods: Pretreatment hematological, histological and virus serological records of 327 chronic hepatitis B patients were reviewed. Two independent patient cohorts were used for validation. Results: Univariate/multivariate analysis showed that ISHAK fibrosis stages, HBV-DNA levels and hepatitis e-antigen status were independently associated with HBsAg concentrations. In agreement with the natural history of chronic hepatitis B, HBsAg concentrations were negatively correlated with ISHAK fibrosis stages (adjusted P = 0.002). Subgroup analysis showed that significant HBsAg-DNA correlation existed in high-viral-titer patients with HBV-DNA > 6 log₁₀ IU/mL (P < 0.001), but not in low-viral-titer patients with HBV-DNA ≤ 6 log10 IU/mL (P = 0.076). A backward stepwise linear regression analysis in the low-viral-titer subgroup revealed a significant correlation between HBsAg levels and a linear combination of HBV-DNA levels and platelet counts. A biphasic model was thus established to accommodate patients with high and low HBV-DNA titers: HBsAg=0.538*HBV-DNA+0.001*platelet*(|6-HBV-DNA|+6-HBVDNA)-0.321. The estimated HBsAg concentrations correlated well with the measured HBsAg levels not only in the model construction cohort (N =327, P < 0.001), but also in two validation cohorts comprising respectively the patients who had received pretreatment liver biopsy assessments (N = 45, P = 0.001), and the treatment-naïve patients who had not received liver biopsy (N = 80, P < 0.001). Conclusion: HBsAg concentrations can be quantitatively estimated by viral DNA concentrations and human platelet counts.