HBV is a hepatotropic, non-cytopathic DNA virus that generates a covalently closed circular DNA (cccDNA) intermediate in the nucleus of infected cells, which serve as transcription templates for viral proteins. A report published by the World Health Organization (WHO) in 2024 disclosed that approximately 254 million people worldwide were chronically infected with HBV in 2022, greatly increasing their risk of developing liver fibrosis, cirrhosis, and hepatocellular carcinoma, and causing an estimated 1.1 million deaths per year.(p1) cccDNA is the only known template for pregenomic RNA (pgRNA) transcription, which is essential for reverse transcription and viral genome replication.(p2) With a long half-life, cccDNA sustains chronic infection.(p3) Current therapeutic strategies against HBV include a 1-year course of pegylated interferon α and long-term administration of nucleos(t)ide analogs (NAs) targeting HBV DNA polymerases (Pols), such as lamivudine, adefovir dipivoxil, entecavir (ETV), telbivudine, tenofovir disoproxil, and tenofovir alafenamide.(p4) However, these strategies result in low hepatitis B surface antigen (HBsAg) clearance rates and a high barrier to resistance, leading to a very low rate of functional cure.(p5) The capsid protein (Cp) of HBV is a key component of the nucleocapsid and plays crucial roles in multiple stages of the viral life cycle, including viral entry, regulation of cytosolic trafficking, nuclear delivery of rcDNA, cccDNA formation, and intracellular amplification, as well as encapsidation and reverse transcription of pgRNA.(p6),(p7) Over two decades ago, Bayer reported the discovery of Cp as a therapeutic target, demonstrating that drug-induced depletion of nucleocapsids inhibits HBV replication, and pioneered the strategy of disrupting capsid formation by targeting Cp aggregation with heteroaryldihydropyrimidines (HAPs) such as Bay 41-4109 and its congeners Bay 38-7690 and Bay 39-5493.(p8),(p9) Building on this concept, CAMs targeting Cp have emerged as a promising class of antivirals that inhibit the reverse transcription of pgRNA and the synthesis of rcDNA, thereby preventing the amplification of the cccDNA pool.(p10) Interestingly, CAMs effectively inhibit the replication of HBV mutants resistant to NAs and show activity against multiple HBV genotypes,(p11) offering potential for achieving a functional cure.

CAMs are classified into two types according to their distinct mechanisms of action. Type I CAMs, with HAPs as representative structures, induce the assembly of aberrant capsids or non-capsid HBV core protein polymers that form aggregates and are subsequently degraded in hepatocytes (Figure 1a).(p12) The initial generation of CAMs, exemplified by Bay 41-4109, exhibited limited anti-HBV potency and high toxicity,(p9) prompting the development of second-generation inhibitors, such as GLS4, with improved antiviral activity but constrained by cytochrome P450 family 3 subfamily A member 4 (CYP3A4) inhibition.(p13) Subsequently, third- and fourth-generation CAMs, including HAP_R10(p14) and RG7907,(p15) were designed to minimize human ether-à-go-go–related gene (hERG) liability and improve pharmacokinetic (PK) properties. Although type II CAMs induce the assembly of normal-sized capsids lacking the HBV genome by accelerating the packaging process,(p16) they exhibit structural diversity, including phenylpropenamides (PPAs) like AT-130, sulfamoylbenzamides (SBAs) such as AB-423 and NVR 3-778, and glyoxamoylpyrroloxamides (GLPs) like GLP-26 (Figure 1a).(p7)

Despite encouraging progress in clinical development, significant challenges remain for CAMs, including limited efficacy, drug resistance, and safety concerns, which have resulted in the discontinuation of several candidates.(p15),(p17),(p18),(p19) To overcome these obstacles and optimize the efficacy of CAMs, innovative strategies that accelerate drug discovery and enhance therapeutic potential are urgently required. CADD has emerged as an essential tool in the design and optimization of CAMs. CADD facilitates the identification of novel lead compounds, prediction of binding affinities, and analysis of potential off-target effects, all critical for improving the therapeutic efficacy and safety profiles of CAMs.(p20) Moreover, integrating artificial intelligence (AI) into CADD workflows promises to further streamline the drug discovery process by predicting drug-like properties, enhancing virtual screening (VS) accuracy, and identifying novel molecular scaffolds that can overcome current limitations in CAMs development.(p21) Compared with traditional trial-and-error synthesis, the integration of CADD and AI allows early prediction of both on– and off-target effects, as well as in vivo safety profiles before compound synthesis, which offers a greener alternative and reduces development time, cost, and attrition rates.(p22) In this review, we highlight the successful applications of molecular docking and quantitative structure–activity relationship (QSAR) models in the discovery and optimization of CAMs, as well as the role of molecular dynamics (MD) simulations in elucidating the mechanism of CAMs. We hope this review will provide researchers with valuable insights from the CADD perspective to discover potential CAMs and contribute to the collective effort toward achieving functional cure for HBV.