Unparalleled remission rates in patients with chemorefractory B-ALL treated with CD19-CAR T cells illustrate the potential for immunotherapy to eradicate chemoresistant cancer. CD19-CAR therapy is poised to fundamentally alter the clinical approach to relapsed B-ALL and ultimately may be incorporated into frontline therapy. Despite these successes, as clinical experience with this novel modality has increased, so has understanding of factors that limit success of CD19-CAR T cells for leukemia. These insights have implications for the future of cell based immunotherapy for leukemia and provide a glimpse of more global challenges likely to face the emerging field of cancer immunotherapy. Five challenges limiting the overall effectiveness of CD19-CAR therapy will be discussed: 1) T cell exhaustion is a differentiation pathway that occurs in T cells subjected to excessive T cell receptor signaling. A progressive functional decline occurs, manifest first by diminished proliferative potential and cytokine production, following by diminished cytolytic function and ultimately cell death. High leukemic burdens predispose CD19-CAR T cells to exhaustion as does the presence of a CD28 costimulatory signal, while a 4-1BB costimulatory signal diminishes the susceptibility to exhaustion. This biology is likely responsible for limited CD19-CAR persistence observed in clinical trials using a CD19-zeta-28 CAR compared to that observed using a CD19-zeta-BB CAR. 2) Leukemia resistance occurs in approximately 20% of patients treated with CD19-CAR and is associated with selection of B-ALL cells lacking CD19 targeted by the chimeric receptor. Emerging data demonstrates two distinct biologies associated with CD19-epitope loss. Isoform switch is characterized by an increase in CD19 isoforms specifically lacking exon 2, which binds the scFvs incorporated into CD19-CARs currently in clinical trials. Lineage switch is characterized by a global change in leukemia cell phenotype, and is associated with dedifferentiation toward a more stem-like, or myeloid leukemia in the setting of CD19-CAR for B-ALL. These insights raise the prospect that effectiveness of immunotherapy for leukemia may be significantly enhanced by targeting of more than one leukemia antigen. 3) CAR immunogenicity describes immune responses induced in the host that can lead to rejection of the CD19-CAR transduced T cells. Anti-CAR immune responses have been observed by several groups, and mapping is underway to identify the most immunogenic regions of the CAR, as a first step toward preventing this complication. 4) The most common toxicities associated with CD19-CAR therapy are cytokine release syndrome, neurotoxicity and B cell aplasia. Cytokine release syndrome is primarily observed in the setting of high disease burdens and efforts are underway to standardize grading and treatment algorithms to diminish morbidity. Increased information is needed to better understand the neurotoxicity observed in the context of this therapy. Although clinical data is limited, B cell aplasia appears to be adequately treated with IVIG replacement therapy. 5) Technical graft failure (e.g. inadequate expansion/transduction) is a challenge that has received limited attention, primarily since many trials have not reported the percentage of patients in whom adequate products could not be generated. We have observed that technical graft failure is often associated with a high frequency of contaminating myeloid populations in the lymphocyte product and selection approaches designed to eradicate myeloid populations have resulted in improved T cell expansion and transduction. These results suggest that optimization of lymphocyte selection may diminish the incidence of technical graft failure. Disclosures Mackall: Juno: Patents & Royalties: CD22-CAR. Off Label Use: cyclophosphamide.