Browse
Log in / Register
Share
A first-in-human CAR T cell therapy designed to target a surface protein that fusion-driven cancers are forced to display has generated measurable clinical activity in a patient with metastatic alveolar soft-part sarcoma (ASPS)—a rare, aggressive solid tumor that typically resists treatment. The findings, published in Nature Cancer on May 15, 2025, by researchers from the University of Texas MD Anderson Cancer Center and other institutions, represent a critical proof-of-concept that a molecular Achilles' heel can be identified in solid tumors that share a common genetic mechanism.
The core challenge for chimeric antigen receptor (CAR) T cell therapy in solid tumors has always been target selection. Most surface proteins on solid cancer cells are also expressed on healthy tissues, risking off-tumor toxicity. Meanwhile, the genetic drivers of these cancers—mutations, fusions, and amplifications—are intracellular, invisible to the engineered T cells. The MD Anderson team turned this problem on its head: instead of looking for universal solid tumor antigens, they searched for surface proteins whose expression is directly commanded by the specific gene fusions that cause certain rare sarcomas and kidney cancers.
Using transcriptomic profiling of tumor samples from patients with ASPS and translocation renal cell carcinoma (tRCC), both driven by fusions involving MiT/TFE-family transcription factors (TFE3 or TFEB), the investigators identified glycoprotein NMB (GPNMB) as a surface molecule that is uniformly and durably expressed across all malignant cells. In relapsed tumor specimens, essentially every cancer cell carried GPNMB, a homogeneity that is unusually favorable for a CAR T target. GPNMB expression in normal adult tissues is low and restricted to a few cell types, suggesting a therapeutic window.
The engineered CAR T product, named GCAR1, incorporates a GPNMB-binding antibody domain fused to intracellular CD28 and CD3ζ signaling domains. In preclinical models, including patient-derived organoids and xenograft mice, GCAR1 cells demonstrated potent, selective killing of GPNMB-positive tumor cells while sparing normal cells. The activity was maintained in three-dimensional structures that mimic the tumor microenvironment.
The pivotal step was the launch of a first-in-human, open-label, single-participant clinical trial (NCT07104682) for relapsed or refractory fusion-driven solid tumors. The initial interim analysis focused on a 34-year-old patient with metastatic ASPS whose disease had progressed after at least two lines of systemic therapy (including pazopanib and chemotherapy). After a standard lymphodepleting regimen of fludarabine and cyclophosphamide, the patient received a single infusion of GCAR1 at a dose of 1×10⁶ cells/kg.
The primary endpoint was safety, and the secondary endpoint was preliminary efficacy. No dose-limiting toxicities were observed. The patient experienced only mild cytokine release syndrome (grade 1) and no neurotoxicity or organ damage. The engineered cells expanded in the peripheral blood, peaking around day 14 and persisting as a polyclonal population for about one month—a pattern consistent with a controlled, self-limiting response.
Imaging at 4 weeks and 8 weeks post-infusion showed stable disease per RECIST v1.1, but more importantly, many of the patient's non-target pulmonary lesions—smaller metastases that are not formally measured as primary endpoints—disappeared entirely. The sustained disease control lasted for approximately three months. For ASPS, a disease with a median survival of about 3 years from diagnosis and few approved systemic options, even stabilization with regression of some lesions is considered a meaningful clinical signal.
One index lesion did not respond, and the team performed spatial transcriptomics on a biopsy of that resistant lesion. The analysis revealed that the tumor microenvironment in that region contained immunosuppressive niches enriched for regulatory T cells, myeloid-derived suppressor cells, and TGF-β signaling—a molecular ecosystem that effectively shielded the tumor from CAR T attack. Notably, the resistant cancer cells still expressed GPNMB, indicating that the resistance was not due to antigen loss but to local immune suppression.
This finding directly informed a rational combination strategy. When GCAR1 was combined with an anti-PD-1 checkpoint inhibitor in a xenograft mouse model of ASPS, the combination produced synergistic tumor control compared to either agent alone. Checkpoint inhibitors such as pembrolizumab are already approved for several cancers, making a clinical trial combining GCAR1 with PD-1 blockade a logical next step.
The broader significance of the study extends beyond ASPS. The MiT/TFE-family fusion proteins are master transcription factors that reprogram the entire gene expression program of the cancer cell. The team demonstrated that the surface expression of GPNMB is a direct consequence of this transcriptional rewiring, and that this "target by transcription" paradigm can be generalized to other fusion-driven malignancies. Dozens of pediatric sarcomas, leukemias, and carcinomas are driven by similar gene fusions, and many currently lack targeted therapies. Instead of trying to pharmacologically inhibit the fusion protein (which is intracellular and often considered undruggable), clinicians can now train the immune system to recognize the surface signature that the fusion imposes.
The study has limitations: the clinical experience currently involves a single patient in an interim analysis. Larger cohorts will be needed to determine response rates, define the optimal dose schedule, and fully characterize the toxicity profile. The three-month duration of disease control is encouraging but will need to be extended into durable remissions, likely through combinations with checkpoint inhibitors or other microenvironment-modulating agents. The possibility of GPNMB loss in some patients or inter-individual variation in normal tissue expression must also be evaluated.
Nevertheless, the trajectory from target discovery through preclinical validation to a first-in-human infusion with a clean safety signal and meaningful clinical activity is a rare achievement. For patients with ASPS, tRCC, and other fusion-driven solid tumors, the message is clear: the wall that has separated CAR T success in liquid cancers from solid tumors is no longer impermeable. The first engineered cells are already through it.
