P1380: the composition of infused cd19-car-t product impacts on efficacy

Topic: 24. Gene therapy, cellular immunotherapy and vaccination - Biology & Translational Research Background: The use of CD19-CAR-T has become the standard therapy for several R/R B-cell malignancies. Definition of outcome predictors, such as pts and disease-related risk factors (e.g. lymphocyte phenotype at apheresis, lymphodepletion, tumor burden) are still incomplete. Recently, CD8+CAR+ with a central memory (CM) phenotype in the product have been associated to higher peak of CAR-T at day 10, which was predictive for better outcome (Mofrini et al, 2022). Aims: We investigated the CAR-T product in terms of CD4/CD8 CAR+ ratio, CAR-T-cell subset, cellular activated phenotype. Methods: We analyzed leftovers from CAR-T bags. CAR-T cells were detected with a human CD19 His Tag protein and an anti-His APC-labelled secondary antibody. CD45RA and CD62L were used to define CAR-T subset as: naïve (N), CM, effector memory (EM), and terminal effector memory (TEMRA). The co-expression of CD38 and HLA-DR defines activated cells. Data were acquired on BDFACS Canto II Cytometer and analysed with Diva Software. Results: We included 31 pts with aggressive B-lymphoma, treated with tisa-cel (13), axi-cel (13) or brexu-cel (5). Median age, previous lines of therapy, disease status at infusion and LDH levels were similar among groups. The percentage of CAR+CD3+ was significantly lower in the tisa-cel product when compared to axi-cel and brexu-cel (45% vs 71% vs 79%, p=0.022). The percentage of CAR+CD8+ was higher in Brexu-cel than others (60.5% vs 26-38%, p 0.03). In all products the most representative T subset was EM both for CAR+CD4+ and CAR+CD8+ (around 87%). The vast majority of CAR+CD8+ displayed an activated phenotype with a slightly higher proportion in tisa-cel (80.3%) in comparison to axicel (70.9%) and brexu-cel (75.9%) bags [p=0.052]. Activated CAR+CD8+ were lower in pts with more than 2 previous lines (p=0.025). We next analyzed the impact of CAR-T phenotypes on outcome. We considered as responders pts obtaining CR/PR at one, three and six months (M1 M3 and M6) after CAR-T infusion, and compared them with pts failing to achieve any response at the same time points (non-responders). Pts with early failure at M1 showed extremely low proportion of CAR+CD8+EM cells in infused bags compared to responders (5.1% vs 91.8%, p=0.057). The probability of survival of these pts at M3 was low. Infused bags from responders at M3 and M6 showed a higher percentage of CAR+CD8+ cells than non-responders (56% vs 27% p=0.02 and 55.4% vs 27.3% p 0.034 respectively). As a consequence, we observed significant difference in the ratio CD4+/CD8+ CAR+ in infused bags between responders at M3 vs non-responders (ratio 0.74 vs 2.47, p=0.011) (Figure 1A) We used a ROC analysis to identify an ideal ratio CD4+/CD8+ CAR+ cut-off among infused CAR-T to discriminate responders at M3: we found that a ratio of 1.12 had 100% sensitivity and 70 % specificity in identifying responses at M3 (AUC 82%, p=0.001). (Figure 1B) In a logistic regression analysis, the CD4+/CD8+ CAR+ ratio was confirmed as a variable to predict response at M3 (Odds Ratio 23.3 p=0.012) and M6 (Odds Ratio 10 p=0.028). Summary/Conclusions We found that a higher percentage of CAR+CD8+ T cells and a lower ratio CD4+/CD8+ CAR+ in the infusion bags are significant predictors for response at 3 and 6 months in pts treated with CD19-CAR-T. We observed that a very low percentage of CAR+CD8+EM T cells may be associated with early treatment failure at 1 month. These data support the role of CD8-mediated cytotoxic mechanism of action by CAR-T cells, and may be useful for strategies of potential empowerment of manufactured products in order to raise CAR-T efficacy.Keywords: Flow cytometry, Immunophenotype, CAR-T