Piccand, C., Timpanaro, A., Anton-Joseph, S., Rössler, J. K., Bernasconi, M. (2024) Abstract 4015: Improvement of CAR T cell potency against low-density targets like L1CAM and FGFR4 in rhabdomyosarcoma. Cancer Res

Rhabdomyosarcoma (RMS) is the most common pediatric soft tissue sarcoma. Overall survival rates are low for metastatic RMS, emphasizing the urgent need of improved therapies. Chimeric Antigen Receptor (CAR) Ts have shown remarkable results against hematological malignancies, but CAR T cell therapy remains challenging in solid tumors, mainly due to a lack of ideal tumor specific targets, and of the immunosuppressive tumor microenvironment (TME). To maximize antitumoral activity against RMS and to address optimal CAR T persistence in vivo, novel CAR constructs were investigated against upregulated RMS targets. By surfaceome profiling, we recently identified L1CAM as promising candidate for RMS CAR T cell therapy, and confirmed CD276 and FGFR4 expression. CD276- (CD276.CD28HD.28TM.28CSD.3z) and Dual-CAR Ts targeting both CD276 and FGFR4 (F8-FR4.CD28HD.28TM.28CSD.3z) showed tumor eradication in an orthotopic RMS model, whereas FGFR4-CAR Ts exhibited only a partial response in vivo. Moreover, CD276-CAR Ts showed a limited activity against RMS tumors expressing lower amounts of CD276, eradicating the tumor in 1/5 mice, underlying the need of improved CAR T cell potency against low-density target-expressing tumors. First, to improve FGFR4-CAR Ts activity, we optimized the recognition of FGFR4 epitope by modifying the CAR moiety “hinge”. We replaced the CD28-derived hinge with various hinges of different length: no hinge, IgG4 (12aa), IgG4CH3 (119aa), and IgG4CH2CH3 (229aa). Luciferase assays showed effective in vitro killing capacity of FGFR4-CAR Ts expressing an IgG4-derived hinge, with a tumor cell survival less than 10% at Effector to Target (E:T) ratio of 5:1. They also released the double amount of IFN-γ (50pg/ml) compared to the other constructs, when co-incubated with Rh4 cells. Next, to verify the potential of L1CAM as CAR T target for RMS, we generated a panel of L1CAM-CARs expressing either a long (IgG4CH2CH3, 229aa) or a short (CD28, 35aa) Hinge, and a CD28- or 41BB-derived costimulatory domain (CSD). L1CAM.IgG4CH2CH3Hinge.28TM.28CSD.3z CAR Ts outperformed the other constructs, killing ~60% of Rh30 at the E:T ratio of 1:1 and more than 90% at the E:T ratio of 5:1. They also exhibited a 3-fold higher IFN-γ release (120 pg/ml) compared to the other constructs. However, although our examination of patient-derived TMA revealed a strong immunoreactivity for L1CAM, more expressed in aggressive RMS subtypes, the selected L1CAM-CAR Ts against Rh30 cells, exhibiting ~10’000 L1CAM copies, showed limited activity in an in vivo pilot study. Based on these results and on the RMS TME investigation, which revealed consistent collagen (COL4A1, COL4A2 and COL18A1) deposition in RMS-derived mice tumor xenografts, we hypothesize that co-expressing a collagenase, able to partially degrade the dense extracellular matrix surrounding tumors, will lead to an improved CAR T cell infiltration and activity in vivo.