Analysis of immune parameters and effect of immunoglobulin replacement therapy on infection, survival and non-relapse mortality following CD19 CAR T-cell therapy

Abstract

Abstract The relationship between baseline immune status and immune reconstitution post CD19 CAR T-cell therapy with risk of infection, non-relapse mortality (NRM) and survival are matters of much clinical interest. Though immunoglobulin replacement therapy (IGRT) is frequently used in this setting, its effect on infection risk, NRM and survival has not been systematically investigated. We conducted a real world, retrospective, multi-centre study to address this issue. Consecutive patients (n=430) treated between 2019-2024 with CD19 CAR T-cell therapy for large B-cell lymphoma (n=354), mantle cell lymphoma (n=46) or B-acute lymphoblastic leukaemia (n=30) from 4 UK CAR T centres were included. Serum IgG, IgA and IgM levels, numbers of C19+ B-cells, CD3+ T-cells, CD16+56+ NK-cells, CD4+ and CD8+ T-cell subsets were analysed at pre and various time points post CAR T-cell infusion. Detailed data were collected on infections, IGRT and mortality. Early infections up to day 28 post CAR T infusion were analysed separately from delayed infections arising beyond day 28. Infections post progressive disease (PD) were excluded. Median age was 63 years (range: 16-80). CAR-T product was axicabtagene ciloleucel in 326, brexucabtagene autoleucel in 56, and tisagenlecleucel in 48. With a median follow up of 24 months, 2-year overall survival (OS), progression free survival (PFS) and NRM for the cohort were 51.7%, 43.6% and 11.3% respectively. Analysis of immune parameters pre and at 1 and 3 months post infusion showed low IgA was associated with increased NRM at all time points on multivariable analysis (including age, disease, CAR-T product, LDH (pre-lymphodepletion) and CAR Haematotox score). Low IgM at baseline was associated with worse OS and PFS, at 1 month with NRM and at 3 months with worse OS and NRM. Low NK cell count at 3 months was associated with worse OS. None of the immune parameters were associated with outcomes at 6 months post infusion though analysis was limited because fewer samples were available at this timepoint. Early infection data were available for 420 patients with infection reported in 133 (31.7%). Infection was probable in 33 (25.4%) with no pathogen isolated. Of those with a pathogen identified 76.3% were bacterial, 17.5% viral and 6.2% fungal. A large majority were grade ≥3 (93.5%) including 3 (3.3%) grade 5. There was no association between pre-infusion immune parameters and early infection. Delayed infection data were available for 280 patients with 237 infection episodes (in 70 patients) eligible for analysis. Infection was probable in 43 (18.1%). Of episodes with a pathogen identified, 58.3% were viral, 40.7% bacterial, 0.5% fungal and 0.5% other. The primary focus was respiratory in 75%. Infection was grade ≥3 in 44% including grade 5 in 6.8%. There was no significant association between immune parameters at either 1, 3 or 6 months post infusion and delayed infections. Of the 430 patients, 87 (20%) were given IGRT with 3 starting pre-infusion and 10 post-PD. The majority (68%) received IGRT for recurrent infections with IgG<4 g/L. For the 77 given IGRT pre-PD, the median time to initiation was 11 months (IQR: 5.1 – 17.8) post infusion. IGRT was included as a time varying covariate to look at association with NRM; no significant difference was seen (adjusted HR 0.35 (0.04 – 3.12) p=0.35). Effect of IGRT on infection was analysed on 43 patients who started IGRT post infusion and pre-PD with a median follow-up from IGRT initiation of 19 months. Infection rates were significantly lower post IGRT: median incidence (per month) 0.07 (IQR: 0 – 0.18) pre and 0 (IQR: 0 – 0.06) post (p=0.0021).We find low IgA pre and at 1 and 3 months post infusion is consistently associated with increased NRM. Low IgM, but not IgG, is also associated with worse outcomes with some variability in its effect on NRM, OS and PFS at these early timepoints. Both early and delayed infections are frequent with distinct differences in pathogens isolated and no significant correlation between infections and any of the immune parameters either pre or at 1, 3 or 6 months post infusion. Our data suggests IGRT may confer protection from delayed infection, but there is not enough evidence for reduction in NRM. This contrasts with survival benefit of IGRT reported with BCMA-targeting therapies. In the absence of survival benefit, our data would support a selective approach to IGRT, rather than primary prophylaxis, post CD19 CAR T-cell therapy.