Two autologous CD19-directed chimeric antigen receptor (CAR) T-cell therapies are now available as a standard of care for lymphoid malignancies, with several others expected in 2020.1-3 These therapies use the patient’s own polytypic T cells transduced with a CAR, reprograming them against surface CD19 on malignant B cells. Despite impressive response rates, autologous CAR T-cell therapies have drawbacks including severe toxicities, treatment delays due to capacity limits and an approximately three-week production time, high cost of manufacturing and delivery logistics for each bespoke product, and negative effects of exhausted T cells, typical in cancer patients.4
The development of allogeneic CARs is an area of widespread and intense effort as they could potentially overcome some of these limitations. Time to treatment could be shortened by obtaining an off-the-shelf product manufactured in advance. Costs could be reduced by simplified logistics and streamlined manufacturing. Finally, the negative impact of T cell exhaustion could be bypassed by using immune cells from healthy individuals. To realize this potential, problems not relevant with autologous CARs must be overcome. When allogeneic effector cells are adoptively transferred, they may harm the patient by causing graft versus host disease (GVHD), or they may be rejected. One approach involves the use of conventional allogeneic T cells that are transduced with the CAR along with gene editing machinery to remove the native T-cell receptor that is a primary mediator of GVHD and rejection.5 An alternative approach, as recently reported by Dr. Enli Liu and colleagues relies on CARs transduced into allogeneic natural killer (NK) cells, which do not contain a T-cell receptor, rather than autologous T cells.
This phase I/II cell dose escalation study tested allogeneic CD19-CAR NK cells, derived from previously frozen cord blood, for patients with relapsed or refractory B-cell malignancies. CAR NK cells were administered at a planned dose of 1 × 105, 1 × 106, or 1 × 107 cells per kg of body weight. The CAR construct contained a CD28 and CD3ζ signaling domain, as well as IL-15 expression known to promote NK cell activity, and a rimiducid inducible caspase-9 in case the cells needed to be turned off due to unchecked toxicity.6 Each product was manufactured over 15 days starting from a frozen cord blood unit and infused fresh into the patient following three days of fludarabine 30 mg/m2 and cyclophosphamide 300 mg/m2 conditioning chemotherapy. Postremission therapy was permitted at the discretion of the treating physician following the day 30 assessment.
Eleven refractory patients were treated with CAR NK cells. Patients had chronic lymphoblastic leukemia (n=5), diffuse large B-cell lymphoma (n=2), transformed follicular lymphoma (n=3), or follicular lymphoma (n=1). Nine of 11 patients received a CAR-NK product that was partially human leukocyte antigen (HLA) matched at four of six HLA loci, while two were HLA mismatched. Six of 11 CAR-NK products were selected due to the presence of a killer immunoglobulin-like receptor (KIR) ligand mismatch, which might potentiate activity of the CAR due to the biology of NK cells’ ability to recognize self.7
The treatment was well tolerated without any cases of cytokine release syndrome or neurotoxicity. There were no cases of tumor lysis syndrome and no nonhematologic grade 3 or 4 toxicities. Rimiducid was not used to activate the caspase-9 safety switch in any patients. At a median follow-up of 13.8 months (range, 2.8-20 months), an objective response was seen in eight (73%) of 11 patients, and a complete response was seen in seven (64%) of 11 patients. Responses were rapid and occurred by day 30 in all responders. Five of eight responders underwent some form of postremission therapy. Expansion of CAR NK cells was seen as soon as three days after infusion, and patients achieving a remission had a higher degree of expansion of their CAR NK cells compared to nonresponders, a phenomena described in other CAR T-cell trials.8 The CAR construct was detectable in the blood of patients as long as one year after therapy; however, persistence did not correlate with ongoing response or relapse. In stark contrast to autologous CAR T-cell therapy, inflammatory cytokines such as IL-6 were not elevated in the serum following CAR NK-cell therapy compared to baseline. Similarly, IL-15 levels in the serum were not elevated. Tests for anti-HLA antibodies against the mismatched alleles were not found, though testing for cellular mediated rejection was not conducted.
What have we learned from this trial? Despite a small number of treated patients, allogeneic CD19-directed CAR NK cells, derived from cord blood, can induce remissions with relatively few toxicities in relapsed/refractory B-cell malignancies. Although the safety profile seems highly favorable, it remains unclear if allogeneic CAR NK-cell therapy could overcome limitations of existing autologous CAR T-cell therapies. First, the trial did not demonstrate true off-the-shelf capabilities: Manufacturing was done for each product immediately before infusion, and the authors note that NK cells are difficult to cryopreserve. Second, lower cost for the therapy is uncertain given the one patient to one product manufacturing and logistics of both procuring cord blood and shipping fresh product. Finally, the ability of these cells to induce durable remissions is unknown; only one of eight patients with opportunity for one-year follow-up remained in remission without having received consolidation. Bottom line, this therapy holds great promise, but process optimization making it truly “off-the-shelf,” and multicenter trial results demonstrating durability, are both needed for the therapy to cross the developmental finish line.
Dr. Locke indicated no relevant conflicts of interest.