(OA-04) CD4+ T-Cells Mediate Immune-Related Adverse Events (CirAE) Following BCMA CAR-T Therapy

B-cell maturation antigen chimeric antigen receptor T-cell therapy (BCMA-CART) is standard of care in relapsed/refractory MM, with 2 approved products: idecabtagene vicleucel (ide-cel) and ciltacabtagene autoleucel (cilta-cel). We introduce the term CART immune-related adverse events (CirAE) to describe a spectrum of delayed toxicities, including cranial nerve palsies, parkinsonism, and enterocolitis, that are mechanistically distinct from CRS, ICANS, ICAHT, and IEC-HS. We characterize the incidence, clinical features, and outcomes of CirAE, and perform correlative analyses to identify underlying mechanisms and predictors.

Methods:

149 patients were treated with commercial BCMA-CART (ide-cel: 68; cilta-cel: 81) between June 2021 and April 2024. Flow cytometry and serum proteomics were performed at baseline, peak CART expansion, and CirAE onset. RNAscope for CAR and immunohistochemistry for T-cell phenotypes were performed on CirAE biopsies.



Results:

CirAE occurred in 19 patients, significantly more often with cilta-cel (21%) than ide-cel (2.9%). The most common CirAE after cilta-cel were enterocolitis (7.4%), cranial nerve palsies (6.2%), and parkinsonism (4.9%), with a median onset of 45 days post-infusion. In contrast, both ide-cel-associated CirAE were pneumonitis, occurring ~1 year post-infusion. Importantly, cilta-cel patients who developed CirAE had inferior overall survival (HR: 3.4; p=0.031), driven by higher non-relapse mortality (HR: 15; p=0.016). One patient (Cilta-cel#2) represented an extreme case, developing polyclonal hyperleukocytosis (lymphocyte peak: 197x10³ cells/µL), with marked CD4 skewing of CART, and three distinct CirAE: facial palsy, delayed ICANS, and enterocolitis. Flow cytometry showed a shift from naïve CD4 T-cells at baseline to activated effector memory CD4 CART at peak expansion, progressing to terminally differentiated CD4 CART during delayed ICANS (Day 57), and reverting to a naïve-like profile by month 11. Correspondingly, CirAE patients exhibited superior CART expansion, CD4-skewing, elevated proinflammatory (e.g., GM-CSF, IL-6) and reduced anti-inflammatory cytokines (e.g., IL-10, IFN-α2) post-infusion. Risk factors for CirAE were: cilta-cel (OR: 8.8; p=0.0047), day 7 CD4/CD8 ratio >1 (OR: 11.9; p=0.028), peak absolute lymphocyte count (ALC) ≥ 2.3x10³ cells/µL (OR: 9.6; p< 0.001), and ICANS (OR: 3.9; p=0.012). Intestinal biopsies from 3 patients with enterocolitis confirmed on-target, off-tumor infiltration by predominantly CD4⁺, CAR⁺ T-cells, supporting a pathogenic role for CD4 CART in enterocolitis.



Conclusions:

We corroborate recent studies linking ICANS and high peak ALC with an increased risk of delayed neurotoxicity after cilta-cel, extend these correlates to a broader set of clinically diverse CirAE, contribute new insights by implicating a possible pathogenic role for CD4 T-cells, and identify risk factors that may inform strategies (eg. pre-emptive dexamethasone) to reduce non-relapse mortality and improve overall survival.