Using single-cell transcriptomics to reveal CD226 upregulation and enhancement of CD19-CAR-T function in the inhibitory CNS microenvironment of refractory CNS lymphoma.

2020 Background: Refractory CNS lymphoma (CNSL) has a poor prognosis, with 5-year survival of ~30%. We conducted a trial of axi-cel for CNSL, where we achieved a CR of 67%. To deeply interrogate the immune mediators of response, we collected daily paired CSF and peripheral blood (PB) samples post-CAR T infusion. This enabled single-cell transcriptional profiling at an unprecedented depth, identifying key drivers of CD19 CAR T responses and compartment-specific mechanisms of CAR T function. Methods: CNSL patients were enrolled in the ‘Axi-cel in CNS Lymphoma’ Trial, NCT04608487. PB and CSF samples were collected daily from Day 0–14 post-infusion, and 5’10x scRNA /TCR-Seq was performed. This analysis focused on peak CAR T expansion (Day 5-10) and included 1,224,178 T cells from 17 patients. Samples were tested for compartment (CSF vs PB) and response-specific (CR vs PD) transcriptomic differences using a mixed-effects model and GSEA. Functional assays were conducted with CD19 CAR Ts and CD19 CAR+ Jurkat-NFAT reporter cells. Results: Transcriptomic analysis identified distinct compartmental differences in CAR Ts, with PB CAR Ts displaying a robust proliferation signature, while CSF CAR Ts were enriched for type I interferon and T-cell dysfunction signatures, including upregulation of inhibitory genes PD-1, TIGIT, TIM3, LAG3. In vitro , CSF-exposed CD19-CAR Ts showed increased expression vs culture-media controls for TIGIT (up 40.1%, SEM 7.4), PD-1 (18.9%, SEM 3.8), and Tim3 (60.6%, SEM 7.5). CAR T NFAT expression was reduced from 18.4, 0.1 SEM (media) to 7.2, 0.2 SEM (CSF) relative to unstimulated controls. Differential expression analysis comparing CSF CD8+ CAR Ts from patients achieving CR (n = 11) or PD (n = 4) showed upregulation of Type I interferon signaling (IFIT1, IFIT3) in PD patients. In contrast, CR patients exhibited increased expression of counter-inhibitory genes (TCF7, CD226) in CSF CAR Ts, suggesting a functional advantage of these CAR Ts in the inhibitory CNS environment. Functional assays of CD19-CAR Ts overexpressing the costimulatory molecule CD226, demonstrated higher lymphoma-cell killing vs WT-CAR Ts (48.6%, SEM 4.1 vs 24%, SEM 2.2, p = 0.01) and greater IFNγ production (63.2%, SEM 1.6 vs 49.4%, SEM 1.7, P = 0.006). CD226 acts as a costimulatory receptor and counteracts TIGIT signaling by competing for its shared ligands, CD112 and CD155. Notably, scRNA-Seq receptor-ligand analysis identified CD112 exclusively expressed in myeloid cells, highlighting a critical myeloid-CAR T interaction that enhances CD226 high CAR T efficacy. Conclusions: ScRNA-Seq suggests tissue-specific CAR T dysfunction in the CNS microenvironment, with CR patients demonstrating upregulation of counter-inhibitory genes, including CD226. This study offers novel insights into axi-cel's mechanism of efficacy and identifies targets to improve CNSL CAR T therapy.