Identifying Modifiers of CAR T-Cell Therapeutic Efficacy and Safety: A Systematic Review and Individual Patient Data Meta-Analysis

CAR T-cell therapy is effective in relapsed/refractory hematologic malignancies, but its use has been tempered by heterogeneity in response and safety outcomes. We performed individual patient data meta-analysis (IPDMA) of CAR T-cell therapy in patients with hematologic malignancies to explore whether patient-level factors modify therapeutic efficacy/safety. We searched MEDLINE, Embase, and Cochrane CENTRAL for relevant trials. IPD was collected and pooled from each included trial, and prevalence of outcomes among strata of potential modifiers was explored. Our primary outcome was complete response, and the secondary outcomes were cytokine release syndrome (CRS), and immune effector cell associated neurotoxicity syndrome (ICANS). We identified 89 trials comprising 2,331 patients for the IPDMA. Complete response proportion ranged from 25% to 75% depending on cancer type. Decreased complete response was seen in those that received bridging therapy compared to those that did not (34% vs 58%, RR:0.55, 95% CI:0.30-0.98), as well as with autologous cell sources compared to allogeneic sources (53% vs 67%, RR:0.61, 95% CI:0.43-0.87). Compared to CAR T-cell therapies targeting CD19 alone, therapies that combine CD19 targeting with additional targets such as CD20, CD22, CD30, CD33, LeY, NKG2D, or BCMA were associated with higher complete response rates (72% vs 58%, RR:1.69, 95% CI:1.15-2.50). Autologous cell sources demonstrated increased risk of ICANS relative to allogeneic sources (24% vs 3%, RR:10.48, 95% CI:1.87-58.57). Safety and efficacy of CAR T-cell therapy within specific cancer types was also affected by modifiers including bridging therapy, CAR T-cell source, CAR T-cell target, sex, age, number of cell infusions, co-stimulatory domain, and dose.