Computer-Aided Drug Design for Immunosuppressive Tumor Microenvironment Reprogramming: A Critical Review

How can cellular interactions in solid tumors guide molecular interventions against immune suppression? This critical review examines how computer-aided drug design (CADD) connects immunosuppressive tumor microenvironment (TME) states to testable molecular hypotheses. We distinguish observational artificial intelligence (AI) predictions from molecular intervention hypotheses and examine the experimental evidence required to evaluate them. We reviewed patient-derived models, spatial and multi-omics profiling, molecular design, simulation, and pharmacology using the literature identified through to 15 October 2026. Adenosine-receptor antagonist discovery and generative protein-binder design illustrated how computational methods contributed to experimentally tested molecular candidates. Native-state structural studies, stromal delivery, and protein degradation broadened the approaches considered. We organized these findings into a proposed framework encompassing in vivo and patient-derived systems, multi-omics mapping, phenotyping of TME states, actionable AI integration, combination-therapy prioritization, and translational testing (IMPACT). A retrospective analysis of a phase III trial illustrated how the framework connects clinical outcomes to questions about mechanism, patient selection, and exposure. The synthesis clarifies the evidence required to connect molecular binding, target engagement, and exposure to patient benefit. IMPACT provides a structure for relating biological context to molecular design and specifying hypotheses for prospective evaluation.