Background aims Autologous chimeric antigen receptor (CAR) T-cell therapies have shown promising clinical outcomes, but T-cell yields have been variable. CD19- and GD2-CAR T-cell manufacturing records were reviewed to identify sources of variability. Methods CD19-CAR T cells were used to treat 43 patients with acute lymphocytic leukemia or lymphoma and GD2-CAR T cells to treat eight patients with osteosarcoma and three with neuroblastoma. Both types of CAR T cells were manufactured using autologous peripheral blood mononuclear cells (PBMC) concentrates and anti-CD3/CD28 beads for T-cell enrichment and simulation. Results A comparison of the first 6 GD2- and the first 22 CD19-CAR T-cell products manufactured revealed that GD2-CAR T-cell products contained fewer transduced cells than CD19-CAR T-cell products (147 ± 102 × 106 vs 1502 ± 1066 × 106; P = 0.0059), and their PBMC concentrates contained more monocytes (31.4 ± 12.4% vs 18.5 ± 13.7%; P = 0.019). Among the first 28 CD19-CAR T-cell products manufactured, four had poor expansion yielding less than 1 × 106 transduced T cells per kilogram. When PBMC concentrates from these four patients were compared with the 24 others, PBMC concentrates of poorly expanding products contained greater quantities of monocytes (39.8 ± 12.9% vs. 15.3 ± 10.8%, P = 0.0014). Among the patients whose CD19-CAR T cells expanded poorly, manufacturing for two patients was repeated using cryopreserved PBMC concentrates but incorporating a monocyte depleting plastic adherence step, and an adequate dose of CAR T cells was produced for both patients. Conclusions Variability in CAR T-cell expansion is due, at least in part, to the contamination of the starting PBMC concentrates with monocytes.

Background Cell selection is an important part of manufacturing cellular therapies. A new highly automated instrument, the CliniMACS Prodigy, was evaluated for the selection of CD34+ cells from mobilized peripheral blood stem cell (PBSC) concentrates using monoclonal antibodies conjugated to para-magnetic particles. Methods PBSCs were collected by apheresis from 36 healthy subjects given G-CSF or G-CSF plus plerixafor. CD34+ cells from 11 PBSC concentrates were isolated with the automated CliniMACS Prodigy and 25 with the semi-automated CliniMACS Plus Instrument. Results The proportion of CD34+ cells in the selected products obtained with the two instruments was similar; 93.6±2.6% for the automated and 95.7±3.3% for the semi-automated instrument (p > 0.05). The recovery of CD34+ cells from PBSC concentrates was less for the automated than the semi-automated instrument (51.4±8.2% versus 65.1±15.7%; p=0.019). The selected products from both instruments contained few and similar quantities of platelets and red blood cells. The depletion of CD3+ cells was less with the automated instrument (4.34±0.2 log depletion versus 5.20±0.35 log depletion; p < 1 × 10−6). Removal of platelets from PBSC concentrates by washing was associated with better CD34+ cell recovery. We explored the reasons for lower CD34+ cell recovery by the Prodigy and found that the non-selected cells for the Prodigy contained more platelets than those for the CliniMACS Plus. Conclusions CD34+ cells can be effectively selected from mobilized PBSC concentrates with the CliniMAC Prodigy, but the recovery of CD34+ cells and depletion of CD3+ cells was lower than with the semi-automated CliniMACS Plus Instrument.