An Endothelial Activation and Stress Index (EASIX) Based Predictive Model for Neurotoxicity and Cytokine Release Syndrome (CRS) after B-Cell Maturation Antigen (BCMA)-Directed Chimeric Antigen Receptor (CAR) T-Cell Therapy for Relapsed/Refractory Multiple Myeloma (RRMM)

Background: The EASIX score is a measure of endothelial dysfunction and has been validated as a predictor of survival after allogeneic stem cell transplantation. Modifications of the EASIX score (EASIX-F, EASIX-FC, modified [m] EASIX, simplified [s] EASIX) have been used to predict CRS and immune effector cell-associated neurotoxicity syndrome (ICANS) after CAR T-cell therapy (CAR T) for B-acute lymphoblastic leukemia and diffuse large B-cell lymphoma. However, there are limited data on the utility of these scores in patients with RRMM receiving CAR T. In this study, we sought to develop an EASIX-like model to predict CRS and ICANS after BCMA-directed CAR T, specifically for RRMM.

Methods: We retrospectively reviewed patients with RRMM who received standard of care BCMA-directed CAR T between May 1, 2021, and January 31, 2024. We calculated the EASIX score as [lactate dehydrogenase (LDH) (U/L) x creatinine (Cr) (mg/dL) / platelet count (10⁹/L)], mEASIX as [LDH x CRP / platelet count] and sEASIX as [LDH / platelet count]. Based on the EASIX/mEASIX/sEASIX scores on the day of CAR T infusion, we classified patients into three groups: low risk (score <quartile (Q) 1), intermediate risk (score in Q2-Q3), and high risk (score >Q4). Based on the EASIX-F score, we classified patients as low risk (EASIX <Q4 and ferritin (ng/mL) <Q1), intermediate risk (EASIX <Q4 and ferritin >Q1), and high risk (EASIX >Q4). Based on the EASIX-FC score, we classified patients as low risk (EASIX < median and CRP < median), intermediate risk (not low risk or high risk), and high risk (ferritin >Q4).

We first assessed the utility of these scores in predicting CRS and ICANS in the first 30 days after CAR T. We compared the performance of models using the concordance (C) statistic. We then assessed the prognostic significance of these scores in addition to RRMM characteristics in predicting CRS and ICANS. Variables with a p value <0.1 on univariate analysis were included in the multivariable model. We employed backward elimination to arrive at the final predictive model.

Results: A total of 193 patients with a median of 6 prior lines of therapy (range 4-15) were included. Overall, 116 (60%) received idecabtagene vicleucel (ide-cel) and 77 (40%) received ciltacabtagene autoleucel (cilta-cel). At CAR T infusion, 26% had active extramedullary disease and 13% had received a prior BCMA-directed therapy. The cumulative incidence (95% CI) of any grade CRS at 30 days was 79% (72-84), ≥ grade 3 CRS was 4.7% (1.6-7.6), any grade ICANS was 25% (19-31) and ≥ grade 3 ICANS was 4.7% (1.6-7.6).

The EASIX, EASIX-F, mEASIX, and sEASIX scores demarcated patients into 3 distinct risk groups for CRS and ICANS. E.g., the incidence (95% CI) of ICANS using EASIX was 44% (28-56) for the high risk, 23% (14-30) for the intermediate risk, and 12% (3-21) for the low risk groups (p<0.001, C=0.64).

Based on these results, we included the EASIX score in our prediction model of ICANS. In the final multivariable model, EASIX > Q4 (HR=2.0, 95% CI: 1.1-3.7), ECOG performance status ≥ 2 on the day of CAR T infusion (HR= 1.99, 95% CI: 0.98-4.02), and high risk cytogenetics [t(4;14), t(14;16), del(17p), gain/amp (1q)] (HR=1.84, 95% CI: 1.02-2.03) were associated with risk of ICANS. Each of these variables was assigned 1 point due to their comparable HRs. The final “EASIX-MM” score was calculated as low risk: 0 points, intermediate risk: 1 point, and high risk: ≥ 2 points. The model for CRS could not be assessed due to lack of significance in the multivariable model.

The cumulative incidence (95% CI) of ICANS according to EASIX-MM was 50% (30-64.3) for high risk, 28.4% (17.3-37.9) for intermediate risk, and 12.7% (5-19.7) for low risk groups (p<0.001, C=0.66). EASIX-MM retained its predictive value when applied to subgroups of patients treated with ide-cel and cilta-cel. Additionally, EASIX-MM predicted progression free survival (PFS) (median PFS: 4.2 months (2.3-12.2) for high risk vs 9.3 months (6.9-12.4) for intermediate risk vs not reached (NR) (16.5-NR) for low risk groups, p<0.001) and overall survival (OS) (median OS: 12.5 months (5.6-NR) for high risk vs 26.5 months (19.4-NR) for intermediate risk vs NR (26.3-NR) months for low risk groups, p<0.001) after CAR T therapy.

Conclusion: Modifications of the EASIX score predict development of CRS and ICANS in patients with RRMM receiving CAR T therapy. EASIX-MM is a novel RRMM specific prediction model that predicts risk of ICANS, PFS, and OS after CAR T for RRMM.