Clinical trials of CD19-targeted CAR-modified T cell therapy; a complex and varied landscape

KEYWORDS:

• Chimeric antigen receptor
• immunotherapy
• T cells
• acute lymphoblastic leukemia
• non-Hodgkin lymphoma
• chronic lymphocytic leukemia
• cellular therapy

CD19 chimeric antigen receptors (CARs) comprise a CD19-specific single chain variable fragment (scFv), fused to an extracellular spacer, a transmembrane domain, and one or more intracellular T cell stimulatory or costimulatory sequences [1]. First-generation CARs contain only a T cell-stimulatory sequence (most often CD3ζ), whereas second or third-generation CARs also contain one or two costimulatory sequences (e.g. 4-1BB, CD28) in addition to CD3ζ, respectively. When introduced into T cells by genetic modification, CD19 CARs redirect the specificity of the T cells to target cells expressing CD19. T cells are generally infused after chemotherapy given for lymphodepletion and the robust proliferation and accumulation of CAR-T cells in the first 2–3 weeks appear important for lysis of CD19-expressing tumor cells and clinical efficacy. Patients may develop cytokine release syndrome (CRS), characterized by fever, hypotension, capillary leak, coagulopathy, organ dysfunction and neurologic toxicity, associated with elevated concentrations of serum cytokines such as IL-6 and IFN-γ. CD19 CAR-T cell immunotherapy has shown remarkable efficacy in refractory B-cell acute lymphoblastic leukemia (B-ALL), and encouraging responses have also been observed in B-cell non-Hodgkin lymphoma (NHL) and chronic lymphocytic leukemia (CLL) [2–9].

CD19 CAR-T cell immunotherapy is still in its infancy; however, a number of centers have begun to accumulate and report significant experience in the manufacturing and delivery of effective CAR-T cell immunotherapy to patients with B-cell malignancies [2–9]. Inevitably, because the rates of efficacy and toxicity reported in clinical trials vary, comparisons between studies have been made in an effort to identify critical factors that govern outcomes of CAR-T cell immunotherapy. Although a comprehensive review of the differences in approaches at each center is beyond the scope of this editorial, it is important to note that manufacturing strategies, including procedures for isolation and stimulation of T cells, the design of CAR constructs, CAR gene transfer, and expansion and formulation of the cell product for infusion widely differ between centers. Many of the details of manufacturing are proprietary, rendering comparisons between studies difficult. In addition, study participants and eligibility criteria vary, as do the selections of lymphodepletion regimens, the dose and timing of CAR-T cell infusions, and the approaches to evaluation and restaging. Patterns identified from analyses of correlative laboratory data and clinical outcomes are emerging and beginning to guide the field. However, caution is warranted when comparing data obtained in studies conducted in different centers using different CAR-T cell products.

Preclinical studies suggest, not surprisingly, that the design of CAR constructs can have a profound impact on function. The selection of scFv, extracellular spacer length and composition, and the presence and selection of intracellular costimulatory sequences have been shown in model systems to impact the efficacy of CAR-T cell immunotherapy [10–13]. In a clinical trial, CD19 CAR-T cells incorporating a second-generation CAR with CD28 costimulation appeared more effective compared to their first-generation counterparts [14]; and the recent success of clinical trials incorporating second-generation CD28- or 4-1BB-costimulated CD19 CARs compared to previous studies using first-generation CARs validates this observation [2–6,8,9]. A more difficult question is if there are outcome differences specific to the costimulatory sequence incorporated into the second-generation CAR constructs. Minimal residual disease (MRD)-negative complete response (CR) rates appear higher and CAR-T cell persistence appears longer in B-ALL patients treated in two clinical trials using different CAR-T cell products incorporating 4-1BB-costimulated CARs compared to two other trials in which the CAR constructs incorporated CD28-mediated costimulation [2,5,6,9]. However, differences in other facets of the CAR constructs used in these trials (e.g. scFv, spacer and transmembrane domains), as well as in CAR-T cell manufacturing, clinical eligibility, lymphodepletion chemotherapy, the schedule of treatments, and selection of additional postremission therapy render comparisons between the studies and identification of the optimal approaches difficult. In NHL patients, the data regarding relative efficacy of 4-1BB- or CD28-costimulation is even more limited, and at this stage does not provide evidence of a difference in efficacy of CAR-T cells that incorporate either 4-1BB- or CD28-costimulated CARs. Clearly, in addition to differences in CAR-T cell manufacturing and delivery, a multitude of other factors such as the presence of costimulatory or inhibitory molecules on the tumor cells might impact outcomes, favoring different requirements for costimulation in different circumstances.

Clinical trials of CD19 CAR-T cell therapy that have shown robust clinical activity have used retroviral gene delivery systems to introduce the CAR into T cells, and are evenly split between gamma-retroviral and lentiviral approaches. There is no data that demonstrates definitive superiority of either of these systems in the clinic, in part due to the confounding effect of the use of 4-1BB costimulation in the lentivirally introduced CARs and CD28 costimulation in the gamma-retrovirally introduced CARs. While there are advocates of the economic advantages and ease of adoption of non-viral delivery systems, e.g. mRNA electroporation or transposon-transposase strategies, there is insufficient data to demonstrate in vivo efficacy of these approaches in human CD19 CAR-T cell clinical trials.

One of the major difficulties in optimizing CD19 CAR-T cell immunotherapy in clinical trials has been the lack of uniformity in the infused cell products, making comparisons of outcomes between studies challenging. CAR-T cells manufactured from different and distinct T cell subsets differ in their capacity to eliminate B-cell lymphoma in xenogeneic models [15]; therefore, the composition of the starting T cell population before CAR-modification may have a profound impact on the efficacy of the infused product. This may be particularly important in patients with B cell malignancies in whom the T cell subset composition in blood is highly variable [9]. Investigators at Fred Hutchinson Cancer Research Center (FHCRC) developed methods to manufacture CAR-T cells from defined CD4⁺ and CD8⁺ T cell subsets that are then formulated for infusion in a defined ratio of CD4⁺:CD8⁺ T cells. While T cell subset selection may add complexity to manufacturing, this strategy allowed infusion of a more uniform cell product than would have been possible by engineering unselected T cells and has facilitated identification of correlations between infused CAR-T cell dose and toxicity and biomarkers that predicted severe CRS [9].

Using this approach, we treated patients with B-ALL with CAR-T cells that were manufactured from CD4⁺ T cells formulated with either central memory or bulk CD8⁺ T cells; however, the nearly universal CR rate in both groups in this study prevented comparison of efficacy in patients treated with CD8⁺ central memory-derived CAR-T cells with those who received bulk CD8⁺ T cell-derived CAR-T cells [9]. High CR rates have also been reported in B-ALL patients by groups engineering CAR-T cells from bulk T cell subsets [2,5,6]. Others have recently reported their experience treating NHL patients after autologous peripheral blood stem cell transplant (PBSCT) with central memory-derived CD8⁺ CAR-T cells or with bulk central memory-derived T cells, but the small number of treated patients does not allow comparison of efficacy [16]. Additional study will be required to establish if manufacturing CAR-T cells from defined T cell subsets confers benefit in terms of longer CAR-T cell persistence or reduced relapse.

One of the challenges in delivery of CAR-T cell immunotherapy is the capacity to manufacture sufficient CAR-T cells from lymphopenic patients who have previously received multiple cycles of chemotherapy. At FHCRC, a single stimulation with a CD19⁺ antigen presenting cell after CAR-modification and during in vitro culture increases the number of CAR-T cells during manufacturing and enables treatment of profoundly lymphopenic patients that may have been excluded from participation in other clinical trials [9]. Manufacturing of CAR-T cells from allogeneic donors may also facilitate treatment of severely lymphopenic patients. Although this is a feasible strategy to treat recipients of prior allogeneic PBSCT, the role of allogeneic CAR-T cell therapy in nontransplanted patients remains unclear [17].

Preclinical studies suggest that the efficacy of adoptive T cell transfer might be enhanced by using strategies to limit differentiation in vitro during manufacturing, for example by supplementing distinct cytokines (e.g. IL-7 and IL-15 rather than IL-2) or modifying signaling pathways in culture [18,19]. It is conceivable that other factors that differ between the current clinical manufacturing approaches, such as the method of polyclonal T cell stimulation before transduction (e.g. anti-CD3 alone versus anti-CD3/anti-CD28 beads), the concentration of IL-2 in culture, or the use of in vitro antigen stimulation to expand CAR-T cells could influence the differentiation status of infused CAR-T cells and the clinical outcomes. While these differences may be potentially relevant, clinical trial data is not yet available to demonstrate a causal effect on efficacy or toxicity.

Lastly, the choice of lymphodepletion chemotherapy may impact adoptively transferred T cell expansion and persistence [9]. This is thought to be due to increasing available IL-7 and IL-15, and has been widely employed in most CAR-T cell immunotherapy clinical trials in the current era [20]. Patients with B-ALL who received cyclophosphamide and fludarabine lymphodepletion prior to CAR-T cell infusion had better in vivo CAR-T cell expansion and persistence, whereas a group who received cyclophosphamide without fludarabine had shorter in vivo persistence of CAR-T cells in part due to the development of CD8⁺ T cell-mediated immune responses to epitopes in the murine scFv of the CAR transgene [9]. CD4⁺ T cell-mediated immune responses to the CAR transgene product have also been observed [5]. All of the CARs used in currently published CD19 CAR-T cell clinical trials contain murine scFvs and fusion sites, which may be immunogenic, suggesting that consideration of the lymphodepletion regimens used in these trials is critical prior to comparing CAR-T cell persistence between studies [2–9]. Although the addition of fludarabine to cyclophosphamide enhanced CAR-T cell expansion and persistence, the importance and optimal intensity of lymphodepletion may differ between individual patients. Intensive lymphodepletion might be important in patients that are at risk of developing an anti-CAR immune response; whereas, highly intensive lymphodepletion, for example by myeloablative conditioning prior to autologous PBSCT, might paradoxically impair the response to CAR-T cells due to depletion of available antigen [16], which is also an important determinant of CAR-T cell expansion [2,9].

The recent successes of CD19 CAR-T cell immunotherapy are built on a foundation provided by previous generations of preclinical and clinical studies. Despite the many differences in CAR design, manufacturing strategies, and clinical delivery that make comparisons between clinical trials difficult to interpret, we are beginning to identify patterns that will inform future generations of the optimal design of CAR-T cell immunotherapy trials.

Declaration of Interest

The authors were supported by institutional funding for the time spent working on the Editorial. C Turtle receives research funding from and has served on advisory boards for Juno Therapeutics. C Turtle is an inventor on a patent application related to CAR-T cell therapy. D Maloney receives research funding from and has served on advisory boards for Juno Therapeutics. FHCRC receives research funding from Juno Therapeutics. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.