Engineering CAR-T cells for solid tumors: Overcoming the microenvironment through integrated design and clinical...

Chimeric antigen receptor T-cell (CAR-T) therapy has produced remarkable therapeutic results in blood cancers, while its application to solid malignancies remains limited by a pooled objective response rate of approximately 9%. This gap stems from core biological obstacles: heterogeneous antigen expression, physical inaccessibility within dense stromal architectures, and immunosuppressive microenvironments that drive T-cell exhaustion through epigenetically fixed transcriptional programs. The period spanning 2024-2025 represents a pivotal turning point. GD2-targeting CAR-T cells delivered intracerebroventricularly achieved durable complete responses (including one sustained beyond 30 months) in H3K27M-mutated diffuse midline gliomas. CLDN18.2-targeting satricabtagene autoleucel demonstrated randomized superiority over physician's choice in advanced gastric cancer (progression-free survival HR 0.37). GPC3-targeting CAR-T cells armored with a dominant-negative TGF-β receptor achieved objective response rates of 50-57% in hepatocellular carcinoma, representing a three- to four-fold improvement over unarmored predecessors. These breakthroughs reflect a paradigm shift from potency-driven engineering toward resilience-based design: metabolic armoring via autocrine IL-10 and IL-15, epigenetic protection through DNMT3A disruption and c-Jun overexpression, logic-gated targeting via synNotch circuits, and microenvironmental shielding through dominant-negative receptors. Beyond the local microenvironment, emerging recognition of systemic neuroendocrine-immune dysregulation further informs CAR-T persistence and fitness considerations. This review synthesizes the mechanistic insights, engineering strategies, clinical evidence, and emerging platforms, including in vivo lentiviral CAR-T generation, that define the current landscape, and proposes a tiered framework for next-generation solid tumor CAR-T development, while explicitly acknowledging the limitations and unknowns that persist.