Computational Design of PD-1/PD-L1 Inhibitors Focusing on Affinity andToxicological Safety

Introduction: The PD-1/PD-L1 immune checkpoint is a central mechanism of tumor immune escape, and its inhibition with monoclonal antibodies has reshaped cancer immunotherapy. Despite their clinical success, antibodies present significant drawbacks, including poor oral bioavailability, limited penetration into solid tumors, high production costs, and immune-related toxicities. These limitations highlight the urgent need for small-molecule inhibitors capable of providing safer, more accessible, and therapeutic alternatives with more favorable pharmacokinetic profiles. Materials and Methods: An integrated computational pipeline was employed, combining ligand-based drug design, virtual screening, molecular docking, and in silico ADMET profiling. To enhance reliability beyond conventional docking scores, we prioritized kinetic parameters, dissociation constant (KD), and residence time (τ), as indicators of binding stability. Interaction mapping focused on conserved PD-L1 hotspots (ARG113, GLU58, TYR123), while metabolism- and toxicity-related endpoints, including CYP inhibition, hERG channel blockade, hepatotoxicity, and skin sensitization, were systematically predicted. Site-of-metabolism analyses guided structural refinement to mitigate metabolic liabilities. Results: Several optimized analogues exhibited nanomolar Kd values, prolonged residence times, and strong, consistent interactions with key PD-L1 residues. ADMET predictions revealed favorable pharmacokinetic properties, low probability of CYP inhibition, absence of hERG blockade, and negligible hepatotoxicity. Only a moderate risk of skin sensitization was predicted for a limited subset of compounds. These findings suggest a high therapeutic potential with manageable safety concerns. Discussion: The combination of kinetic stability (KD and τ) with toxicological endpoints provided a more robust prioritization framework than reliance on binding free energy alone. This integrative strategy highlights the value of computational pipelines in guiding early drug discovery, reducing attrition, and generating leads that balance potency with safety. Moreover, the methodological approach is broadly applicable to other protein–protein interaction targets beyond PD-1/PD-L1. Conclusion: Our results support the feasibility of small-molecule PD-1/PD-L1 inhibitors as next-generation immune checkpoint blockers. By integrating docking-based design with kinetic and toxicological profiling, this study demonstrates a rational, cost-effective framework for advancing immunotherapeutic discovery toward preclinical development.