4CPS-067 Development and management of ICANS: risk factors following CD19 CAR-T therapy in lymphoproliferative disorders

<h3>Background and Importance</h3> Immune effector cell-associated neurotoxicity (ICANS) remains a significant clinical challenge after anti-CD19 chimeric antigen receptor T-cell (CAR-T) therapy due to the limited availability of reliable predictive factors. <h3>Aim and Objectives</h3> To identify potential risk factors for both any-grade-ICANS and significant-grade-ICANS in patients treated with anti-CD19 CAR-T therapy. <h3>Material and Methods</h3> A sample size of 100 was estimated to detect significant differences between axicabtagene-ciloleucel and tisagenlecleucel. Data were retrospectively collected for patients with haematological diseases treated with commercial anti-CD19 CAR-T. ‘Significant toxicity’ was defined as grades 2–4, as it required specific treatment, and ‘severe toxicity’ as grades 3–4. Risk factors were identified using univariate and multivariate logistic regression, both ‘any-grade-ICANS’ and ‘significant-grade-ICANS’, using STATA16. <h3>Results</h3> A total of 101 patients were studied, of whom 36 (36%) developed any-grade-ICANS, 20 (20%) significant-grade-ICANS, and 12 (12%) severe-ICANS. The majority of ICANS cases resolved after treatment (83%). Various factors were associated with both any-grade and significant ICANS: presence of an autoimmune comorbidity, earlier or more sustained cytokine release syndrome (CRS), number of tocilizumab doses used to treat CRS, high tumour burden (measured by LDH on day 0 or metabolic tumour volume), elevated levels of IL-6 and IL-15 on days 0 and +3 after infusion, and elevated day +3 levels of LDH and D-dimer. Incidence of any-grade-ICANS was higher after treatment with axi-cel (46%) compared to tisa-cel (21%) and was linked to elevated D-dimer on day 0. Significant-grade ICANS was associated with an increased number of prior treatment lines and higher day 0 levels of IL-1, C-reactive protein, GM-CSF, and ferritin. In the multivariate analysis, best model (AUC=0.84) for predicting any-grade-ICANS included the type of CAR-T, time from infusion to CRS onset, pre-infusion D-dimer levels and IL-6 levels on day +3. For significant ICANS (AUC=0.82), the model included the presence of an autoimmune comorbidity, the number of prior therapy lines, development of grade ≥2 CRS, and IL-15 and GM-CSF pre-infusion levels. <h3>Conclusion and Relevance</h3> This study reveals a link between ICANS and patient‘s baseline characteristics, CAR-T type, tumour burden, CRS development (including tocilizumab doses), and pro-inflammatory markers before and after infusion. A deeper understanding of ICANS allows for better anticipation of treatment to improve patient outcomes. <h3>References and/or Acknowledgements</h3> <h3>Conflict of Interest</h3> No conflict of interest