A semi‐mechanistic model to characterize the long‐term dynamic of <scp>HBV</scp> markers during treatment with lamivudine and <scp>Peg‐IFN</scp>

Antiviral treatments against Hepatitis B Virus (HBV) suppress viral replication but do not eradicate the virus, and need therefore be taken lifelong to avoid relapse. Mathematical models can be useful to support the development of curative anti-HBV agents, however they mostly focus on short-term HBV DNA data and neglect the complex host/pathogen interaction. This work aimed to characterize the effect of treatment with lamivudine and/or Peg-IFN in 1,300 patients (HBeAg-positive and HBeAg-negative) treated for 1 year. A mathematical model was developed incorporating two populations of infected cells, namely I1 , with a high transcriptional activity, that progressively evolve into I2 , at a rate δtr , representing cells with integrated HBV DNA that have a lower transcriptional activity. Parameters of the model were estimated in patients treated with lamivudine or Peg-IFN alone (N=894), and the model was then validated in patients treated with lamivudine plus Peg-IFN (N=436) to predict the virological response after a year of combination treatment. Lamivudine had a larger effect in blocking viral production than Peg-IFN (99.4-99.9%versus 91.8-95.1%), however Peg-IFN had a significant immunomodulatory effect, leading to an enhancement of the loss rates of I1 (×1.7 in HBeAg-positive patients), I2 (>×7 irrespective of HBeAg status), and δtr (×4.6 and ×2.0 in HBeAg-positive and HBeAg-negative patients, respectively). Using this model, we were able to describe the synergy of the different effects occurring during treatment with combination and predicted an effect of 99.99% on blocking viral production. This framework can therefore support the optimization of combination therapy with new anti-HBV agents.