Multi-omic analysis reveals metabolic fitness of the ARI0002h BCMA CAR T-cell infusion product as a key predictor of clinical outcomes in multiple myeloma

Abstract

Abstract Background BCMA-directed CAR-T therapies have achieved high initial responses in relapsed/refractory (R/R) multiple myeloma (MM). ARI0002h (cesnicabtagene autoleucel), an anti-BCMA CAR-T developed at our institution, has demonstrated a 95% overall response rate in a clinical trial involving 60 patients, and it is currently available under hospital exemption in Spain. Despite preserved BCMA expression, some patients still progress early. This suggests that intrinsic properties of the infused CAR-T cells may influence outcomes. We performed a multi-omic study of ARI0002h CAR-T cell infusion products collected and cryopreserved prior to their administration, aiming to identify transcriptomic signatures, protein markers, and metabolic parameters that may correlate with long-term treatment response in patients with R/R MM. Methods Metabolic profiling was performed on 37 ARI0002h infusion products using Seahorse XF. Progression-free survival (PFS) was assessed using Kaplan–Meier analysis and univariate Cox regression, with optimal cut-offs defined by the Youden index from ROC curve analyses. Transcriptomic and proteomic profiling was conducted on CAR⁺ T cells sorted into CD4⁺ and CD8⁺ subsets from thawed infusion products. Proteomic analysis was performed by DIA-MS (Orbitrap Eclipse) on 14 CD4⁺ and 10 CD8⁺ samples, stratified by long-term responders (LTR; PFS >700 days) and short-term responders (STR; PFS <365 days). Transcriptomic analysis was performed by bulk RNA sequencing on 12 paired CD4⁺ and CD8⁺ samples using the same clinical stratification. Results We performed functional metabolic profiling on infusion products from 37 patients (median follow-up: 19 months) to investigate the impact of metabolic fitness on clinical outcomes. In univariate Cox regression models, all four metabolic parameters (mitochondrial ATP, glycolytic ATP, total ATP output and spare respiratory capacity [SRC]) were significantly associated with PFS, with higher values consistently predicting longer PFS (HR range 0.987 to 0.996; p < 0.005 for all). To establish clinically applicable thresholds, we performed ROC curve analyses using PFS as the endpoint and calculated optimal cut-offs via the Youden index. Patients were subsequently stratified into high and low expression groups according to each threshold, and differences in PFS were assessed by Kaplan-Meier analysis. Higher mitochondrial ATP production (cut-off 385.46 pmol/min/10⁵ cells, AUC 0.779) was associated with significantly prolonged PFS (40.8 vs 10.2 months, p = 0.0031). Glycolytic ATP production (cut-off 673.02, AUC 0.891) also predicted better outcomes (PFS not reached vs 11.4 months, p = 0.003), with no progressions observed in the high group. Similarly, total ATP output (cut-off 729.81, AUC 0.874) and SRC (cut-off 213.68, AUC 0.809) were significantly associated with improved clinical outcomes (total ATP 31.2 vs 10.2 months, p = 0.0049; SRC 26.2 vs 10.2 months, p = 0.0047). Transcriptomic profiling of CAR-T cells supported these findings. In LTR, CD4⁺ cells showed upregulation of gene sets related to cell cycle and DNA replication, while CD8⁺ cells were enriched in pathways linked to mitochondrial respiration and oxidative metabolism, reflecting enhanced proliferative and metabolic capacity. Proteomic analysis further confirmed these results. In CD4⁺ CAR⁺ cells from LTR, proteins linked to mitochondrial metabolism (SLC25A4, ATP5F1D) and oxidative phosphorylation (NDUFA3) were upregulated. In contrast, short-term responders (STR) showed increased expression of inflammatory and immunosuppressive proteins (ARG1, S100A9). In CD8⁺ CAR-T cells, LTR displayed elevated expression of mitochondrial proteins (UQCR10, SDHC, MRM1) whereas STR upregulated markers of oxidative stress and cellular dysfunction (GSTM1, HBA1, DHRS1). Ongoing transcriptomic and proteomic analyses aim to validate these findings and uncover new biological features linked to response. Conclusions Metabolic profiling of ARI0002h CAR-T cell products supports the relevance of cellular bioenergetics as a determinant of treatment efficacy. By integrating metabolic, transcriptomic, and proteomic data, we highlight the association between enhanced metabolism and sustained clinical benefit. These findings suggest that assessing CAR-T metabolism prior to infusion may help predict therapeutic response and guide future product optimization. Ongoing analyses will further validate and expand these insights.