AUTHORS
Theresia Akhlaghi, MD, Mateo Mejia Saldarriaga, Assistant Professor2, David Jayabalan, MSc3, Joshua Kassner, MD4, Frank Costanzo5, Kameisha Gulgar, MS, MLS, ASCP6, Christian A Gordillo, MSc7, Sarah Wesley, MD8, Roger Pearse, MD9, Giorgio Inghirami, MD10, Ran Reshef, MD, MSc11, Ruben Niesvizky, MD12, Mark Bustoros, MD13
AFFILIATIONS
1 Division of Hematology and Oncology, Weill Cornell Medicine, New York, United States
2 Division of Hematology and Oncology, Weill Cornell Medicine, New York, United States
3 Division of Hematology and Oncology, Weill Cornell Medicine, New York, United States
4 Division of Hematology and Oncology, Weill Cornell Medicine, New York, United States
5 Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, United States
6 Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, United States
7 Columbia University Irving Medical Center, New York, United States
8 Columbia University Irving Medical Center, New York, United States
9 Division of Hematology and Oncology, Weill Cornell Medicine, New York, United States
10 Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, United States
11 Columbia University Irving Medical Center, New York, United States
12 Division of Hematology and Oncology, Weill Cornell Medicine, New York, United States
13 Division of Hematology and Oncology, Weill Cornell Medicine, New York, United States
ABSTRACT
Background: Movement and neurocognitive toxicity (MNT) is a serious complication of B-cell maturation antigen (BCMA) chimeric antigen receptor T cell (CAR-T) therapy, yet its mechanism is unresolved and predictive pre-treatment biomarkers limited.
Methods: We retrospectively analyzed 105 patients with relapsed/refractory multiple myeloma treated with cilta-cel (n=87) or ide-cel (n=18) at our institutions. CD4 and CD8 cells were measured by flow cytometry at apheresis. Single-cell RNA sequencing (scRNAseq) and proteomics were performed on PBMCs/serum from 29 patients at apheresis and peak expansion (days 9–15 post-infusion [post-tx]).
Results: Thirteen patients (12.4%) developed MNT. Apheresis CD4/CD8 ratio was higher in MNT+ patients; ratio >1.6 predicted MNT (AUC 0.80), similar to post-infusion maximum absolute lymphocyte count (ALCmax; AUC 0.78). In multivariable analysinis adjusting for age, prior lines, extramedullary disease, ALCmax, and response to bridging (≥PR vs <PR or none), ratio >1.6 remained an independent predictor (OR 12.7, p=0.02). In a subset of MNT+ patients with available CSF flow cytometry, infiltrating T cells were predominantly CD4+ (mean CD4:CD8 20:1) in cilta-cel patients, with >50% expressing BCMA CAR. A single ide-cel case showed a markedly different CSF composition (CD4:CD8=0.9:1, 24% BCMA CAR).
We analyzed 495,047 cells from 29 patients across two time points (apheresis and at ALCmax); 264,413 T cells and 141,330 innate cells. scRNA-seq confirmed elevated CD4/CD8 ratio in MNT+ patients (baseline 3.8 vs 1.1, p<0.001; post-tx 2.8 vs 0.8, p=0.005). MNT+ patients also had increased CAR-T expansion at the time of ALCmax (50.5% vs 11.4% CAR+ cells, p=0.004) and higher CD4 CAR+ cells.
Dissecting T cell subsets post-tx revealed expansion of CD4 central memory (38.6% vs 17.6%, p=0.007) and CD4 effector-memory (20.9% vs 10.7%, p=0.027) and decreased CD8 effector-memory (2.4% vs 7.7%, p=0.002) and CD8 TEMRA (1.2% vs 16.2%, p<0.001) subsets in MNT+ patients. Functionally, MNT+ patients showed a post-infusion shift towards a T follicular helper-like signature (Tfh-2 1.7-fold higher, p<0.001, Tfh-1 1.8-fold higher, p=0.003, T peripheral helper, 1.5-fold higher, p=0.011), while MNT- patients exhibited upregulation in cytotoxic programs (4.1-fold higher, p=0.0007). This T-helper-skewed transcriptional signature was seen in both the CAR+ and endogenous CAR− compartments.
In the innate compartment, MNT+ patients showed post-infusion depletion of conventional dendritic cells (cDC2, p=0.005) and non-classical CD16 monocytes (p=0.009). Notably, established monocyte activation signatures did not distinguish MNT groups; the discriminating axis was interferon: MNT- patients upregulated the monocyte type-I interferon and interferon gamma response after infusion (NES +2.92 and +2.89, both q=0.001), whereas MNT+ patients did not (type-I IFN NES -1.63, q=0.03). Across patients, monocyte interferon enrichment correlated positively with the T-cell cytotoxic-effector program that expanded in MNT- patients (Spearman ρ=+0.38, p=0.04) and inversely with the T-helper (Tfh-2) program enriched in MNT+ patients (ρ=−0.46, p=0.01).
MNT+ patients had an increased inflammatory signaling at baseline, with T cell upregulation of TNF-α/NF-κB (q=0.001), type-I interferon response (q=0.003), IL-2/STAT5 signaling (q=0.011), and inflammatory response (q=0.026). This baseline inflammatory signaling was also seen to be increased in monocytes, NK cells, and dendritic cells. At ALCmax, MNT+ T cells showed enrichment of cell-cycle programs (G2-M Checkpoint CD8 q=0.014; E2F Targets q=0.055), consistent with increased CAR-T expansion. Proteomic analysis is ongoing to delineate the cytokine profile.
Conclusion: Apheresis CD4/CD8 ratio >1.6 is an independent pre-infusion predictor of MNT. MNT+ patients showed greater CAR-T expansion, higher CD4 CAR+ cells and CD4-dominant CSF infiltration. The T cell compartment showed helper-T-skewed programming in both CAR+ and bystander T cells, and elevated baseline inflammatory signaling. In the innate compartment, MNT- patients mounted a monocyte interferon response that correlated with cytotoxic-effector T-cell differentiation, whereas MNT+ patients did not, paralleling their T-helper skew. These findings support apheresis CD4/CD8 ratio as a pre-infusion screening biomarker and identify specific T cell phenotypes and functional states that contribute to development of MNT.
DISCLOSURES
Theresia Akhlaghi, MD: No relevant conflicts of interest to declare.