The mechanistic elegance of CAR-T therapy explains much of its success. Researchers harvest T lymphocytes and genetically equip them with synthetic receptors whose extracellular portions, typically single-chain variable fragments, bind specific antigens on malignant cells. When the receptor engages its target, intracellular signaling domains transmit activation signals through immunoreceptor tyrosine-based activation motifs in the CD3ζ chain, recruiting kinases such as LCK and ZAP70 that ignite three core pathways: MAPK, PI3K-Akt, and NF-κB. Costimulatory domains like CD28 or 4-1BB amplify and sustain these signals, and the activated cells then kill through two parallel strategies: perforin and granzyme punch pores in target cells to trigger apoptosis, while death receptor binding via FAS/FASL and TRAIL activates extrinsic cell death. Inflammatory cytokines such as interferon-gamma and tumor necrosis factor-alpha add a broader anti-tumor push. Five successive generations of CAR designs have built on this foundation, from first-generation constructs with only the CD3ζ domain to fifth-generation cells incorporating cytokine receptor signaling modules and logic-gated architectures.

But the therapy’s power comes with a dangerous edge. Cytokine release syndrome, the best-known toxicity, arises when activated CAR-T cells flood the body with pro-inflammatory cytokines including IL-1, IL-6, and interferon-gamma, potentially progressing to multiorgan dysfunction and death. In one study of CD70-directed allogeneic CAR-T therapy, 67 percent of patients experienced CRS, including one grade 4 dose-limiting toxicity. Immune effector cell-associated neurotoxicity syndrome, or ICANS, ranges from mild confusion to severe seizures, and emerging evidence points to cytokine-mediated endothelial activation and blood-brain barrier disruption. A pediatric cohort study documented five cases of acute quadriparesis and paraparesis with demyelinating lesions on MRI and elevated neurofilament light chain in cerebrospinal fluid, but no leukocytic infiltration, revealing a distinct neurotoxicity phenotype. Standard management relies on the IL-6 receptor blocker tocilizumab and corticosteroids, though these agents can suppress the very T cells doing the therapeutic work, a paradox that has spurred engineering of CAR-T cells with IL-6 or interferon-gamma silenced via shRNA.

Solid tumors present an even more formidable fortress. The tumor microenvironment combines physical barriers, including dense extracellular matrix and disorganized vasculature that impairs T cell extravasation, with an immunosuppressive cellular landscape of myeloid-derived suppressor cells, regulatory T cells, and M2-polarized macrophages. Hypoxia, acidosis, and metabolic competition further drain CAR-T cell fitness. Tumors also deploy cell-intrinsic defenses: spatial and temporal heterogeneity in antigen expression, confirmed by single-cell RNA sequencing across tumor subregions, allows antigen-negative subclones to escape under therapeutic pressure, while checkpoint ligands such as PD-L1 and B7-H3 drive T cell exhaustion. On-target off-tumor toxicity compounds the danger, since many solid tumor antigens also appear on healthy tissue. Researchers are responding with affinity-tuned binders, logic-gated and dual CAR designs, and synthetic biology circuits that demand multiple simultaneous signals before the cells attack, aiming for precision that spares normal tissue.

T cell exhaustion itself has become a molecular battleground. Transcription factors TOX, TOX2, and NR4A family members program the exhausted state, regulated by NFAT acting even without its AP-1 partner. Exhausted cells upregulate inhibitory receptors including PD-1, TIM3, and LAG-3, suffer impaired mitochondrial function and suppressed glycolysis, and lose proliferative capacity. Tumor-derived extracellular vesicles can push CAR-T cells into this dysfunctional state by inducing supraphysiologic inflammation. Countermeasures are emerging: overexpression of c-Jun confers exhaustion resistance and enhances expansion, inhibition of sphingosine 1-phosphate receptor 3 remodels the microenvironment and improves infiltration, and MEK inhibitors downregulate c-Fos and JunB to prevent exhaustion-driven differentiation. These interventions, combined with checkpoint inhibitors, aim to keep engineered cells in a persistent, cytotoxic, stem-like state.

Engineering innovation is accelerating on every front. Charge density modulation of the CAR antigen-binding domain optimizes tonic signaling and reduces spontaneous activation, while endogenous signaling molecule activating CARs recruit native signaling molecules through their transmembrane domains, showing promise against triple-negative breast cancer with less cytokine release than conventional designs. Armored CAR-T cells secrete payloads directly into the tumor: IL-15 and CCL19-secreting cells show enhanced efficacy in glioblastoma models, Serpin B9-armored cells resist granzyme B-mediated fratricide, FOXP3-coexpressing cells acquire stem-like durability, and TIM-3-Fc decoy secretion improves CD19 CAR-T therapy in B-ALL by neutralizing galectin-9. Synthetic Notch receptors enable conditional CAR expression only when cells encounter tumor vascular markers like P-selectin, creating microenvironment-actuated T cells that improve selectivity while preserving potency. Multivalent ELECTRIC CARs targeting KIT, MPL, and FLT3 simultaneously offer a non-genotoxic conditioning strategy for leukemia, and tri-functional M10 cells designed against HIV-1 combine cytotoxicity, viral neutralization, and B-cell follicle homing.

Perhaps the most disruptive shift involves manufacturing. Autologous CAR-T production remains slow, complex, and costly, often exceeding $400,000 per patient, with a median 108 days from consultation to infusion in community networks and 41 percent of patients unable to access timely therapy. Universal off-the-shelf products derived from healthy donors promise immediate availability and consistent quality, with CRISPR/Cas9 knockout of TCR and HLA genes mitigating graft-versus-host disease and host-versus-graft rejection. Mucosal-associated invariant T cells, which do not mediate alloreactivity, offer an allogeneic source requiring minimal genetic modification. Even more radical is in vivo CAR-T generation: delivering viral vectors or targeted lipid nanoparticles encoding the CAR construct directly into the patient, reprogramming circulating T cells in situ and eliminating ex vivo manipulation entirely. Recent work has demonstrated in vivo generation of functional CAR-T cells for cancer and autoimmune disease, though vector immunity, insertional mutagenesis risk, and dosing control remain unsolved.

Clinical results are now extending well beyond blood cancers. Claudin 18.2-targeted CAR-T cells achieved a 38.8 percent overall response rate and 91.8 percent disease control rate in gastrointestinal cancers, with 96.9 percent of patients experiencing only grade 1-2 CRS and no treatment-related deaths. GD2-directed cells delivered a 63 percent response rate in high-risk neuroblastoma with three-year overall survival reaching 60 percent, while CD70-targeted allogeneic cells produced an 81.3 percent disease control rate in clear cell renal cell carcinoma. In multiple myeloma, the dual BCMA/CD19 construct GC012F achieved a 100 percent response rate with 95.5 percent complete remission. Most strikingly, CD19 CAR-T therapy is rewriting the playbook for autoimmune disease: a German cohort achieved 100 percent drug-free remission in 15 refractory autoimmune patients, and universal CD19 CAR-T cells reversed skin fibrosis in systemic sclerosis patients, with one regaining finger mobility within days of infusion, challenging the assumption that fibrotic damage is irreversible.

The review’s authors frame the future around three strategic priorities: optimized CAR design, combination therapies, and scalable manufacturing. Gene editing with CRISPR enzymes such as PcoCas12a can knock out negative regulators like DGKα to boost anti-tumor function, while deletion of NR4A factors, CTLA-4, or the adenosine A2A receptor enhances persistence in hypoxic tumor niches. Combinations with pembrolizumab, ibrutinib, oncolytic viruses, and STING agonists aim to convert immunologically cold tumors into hot ones, and automated bioreactors, non-viral gene transfer, and point-of-care manufacturing promise to slash costs and timelines. Applications in infectious disease, where bNAb-derived CAR-T cells have reduced HIV reservoirs, and in fibrosis and senescence, where senolytic CAR-T cells target age-related dysfunction, suggest the platform may ultimately transcend oncology altogether. What began as a last-resort therapy for dying leukemia patients is evolving into a versatile system of precision immune reprogramming, one whose full scope researchers are only beginning to map.

Subject of Research: Advances, challenges, and clinical breakthroughs of CAR-T cell therapy in refractory cancers and beyond

Article Title: Advances in CAR-T cell therapy for refractory diseases: challenges, innovations, clinical breakthroughs, and future prospects

Article References: Liao, Q., Mao, Y., Feng, M., Zheng, N., Ding, X., Zhang, X., Wang, Z., & Xu, J. (2025). Advances in CAR-T cell therapy for refractory diseases: challenges, innovations, clinical breakthroughs, and future prospects. Clinical Cancer Bulletin, 4(1), Article 21. https://doi.org/10.1007/s44272-025-00050-2

Image Credits: AI Generated

DOI: 10.1007/s44272-025-00050-2

Keywords: CAR-T cell therapy, immunotherapy, solid tumors, cytokine release syndrome, CRISPR gene editing, autoimmune disease, hematological malignancies, tumor microenvironment, off-the-shelf CAR-T, in vivo CAR-T, T cell exhaustion, cell therapy manufacturing