Spatiotemporal and genetic tracking of long-term persisting CAR-T cells following ciltacabtagene autoleucel treatment for multiple myeloma reveals molecular insights into late-onset toxicities

Abstract

Abstract Introduction Ciltacabtagene autoleucel (cilta-cel) is an anti-BCMA CAR-T cell therapy which led to >70% complete response (CR) rates in relapsed/refractory multiple myeloma (MM) patients. Long-term (LT) persisting CAR-T cells may influence both remission durability and adverse events (AE), yet cellular kinetics are heterogeneous across patients. Here, we investigated the impact of LT persisting CAR-T cells in MM patients following cilta-cel treatment. Methods Flow cytometry-based immune monitoring in peripheral blood (PB) comprising CAR-T cell quantification was performed at days 0 (D0), 4, 7, 14, and 30, and months 3 (M3), 6, 12, 18, and 24 following cilta-cel infusion. In total, 36 patients with maximum expansion (cmax) >101 CAR-T cells/µL and ≥1 immune monitoring ≥M6 (median: 7 time points, range: 5-10) were included. Single-cell RNA/TCR-seq of PB mononuclear cells was available at D30 (n=27) and M3 (n=20) for 29 patients, and of isolated CAR-T cells at D30 (n=3) and ≥M6 (n=6) for 4 patients. For these, whole exome sequencing (WES) of CAR-T cells at ≥M6 was conducted. Spatial transcriptomics of intestinal biopsies from 2 patients was performed using the Visium HD technology. Results CAR-T cells were detected for a median of 66 days (14–729) post-infusion. LT persisters (LTP) were defined by CAR-T detection in PB ≥M6 (median: 389 days, 248-729), which was observed in 13/36 (36%) patients. 23/36 (64%) had earlier CAR-T clearance (median: 29 days, 14-99) and were classified as short-term persisters (STP). CAR-T counts in LTP around M12 demonstrated substantial variability, with a median of 14 (3-960) CAR-T/µL. Initial peak expansion occurred after a median of 14 (7–30) days with a median cmax of 498 (10-8011) CAR-T/µL. There was no difference regarding cmax between LTP and STP (median: 510 vs 404 CAR-T/µL, p=0.13). All LTP and 16/23 (70%) STP achieved CR post-infusion, while the remaining 7/23 (30%) had a very good partial response. After a median follow-up of 12 (6–24) months, median progression-free and overall survival were not reached in both groups. Time to B-cell recovery (>0 B-cells/µL; median: not reached vs 3 months, p<0.001) and T-helper cell recovery (≥200 endogenous CD4+/µL; median: 16 vs 6 months, p<0.001) were significantly longer in LTP vs STP. 6/13 LTP (46%) and 5/23 STP (22%) developed late-onset (≥D30) grade ≥2 AEs (p=0.15): 5/13 LTP had neutropenia and/or thrombocytopenia, which resolved in 3/5 patients at M3. One LTP developed inflammatory bowel disease (IBD)-like enterocolitis. Spatial transcriptomics of intestinal biopsies around M12 of this patient (n=3 biopsies) and another LTP (n=1) who had failure to thrive but no IBD-like symptoms, revealed only sparse CAR-T infiltration (<0.1%) in the terminal ileum in the absence of enterocolitis, whereas CAR-T cells made up 1.6% of cells in a duodenal sample of the first patient. CAR-T cells were enriched in the lamina propria and colocalized with stromal, dendritic cell and macrophage populations. Single-cell data revealed temporal changes in CAR-T cell composition: Overall, CD8+CD45RA+ effector memory (EM) CAR-T cells were more abundant in STP vs LTP at D30 and M3 (p=0.01). At M3, there was an enrichment in CD4+ memory, CD4+ cytotoxic, and CD8+ EM CAR-T cells in LTP (p<0.01). CAR-T lineage tracing in 4 LTP from D30 to ≥M6 revealed that 1/4 had monoclonal (90% of CAR-T), 1/4 oligoclonal (8 CD8+ clones ≥1% of CAR-T), and 2/4 polyclonal TCR rearrangements. Oligoclonal CAR-T cells remained stable in frequency over time. Atypical CD4-CD8- monoclonal CAR-T cells emerged at M3 and re-expanded, making up 66% of CD3+ (510 CAR-T/µL) at M12. WES indicated two TP53 mutations (p.T125M and p.R249S) and one NF1 mutation (c.480-1G>A) with 38%, 40%, and 31% variant allele frequencies (VAF), respectively. The two TP53 mutations were already present at 4% and 2% VAF in whole blood at leukapheresis. Another patient with polyclonal memory CD4+ CAR-T cells at M12 harbored a TP53 (p.R306*) mutation at 10% VAF. Conclusion LTP and STP showed no significant difference in survival, however LTP experienced prolonged B-cell and endogenous T-lineage suppression, and 46% had late-onset AEs. Lineage tracing revealed an unexpected diversity of LT persisting CAR-T phenotypes and clonal trajectories. Two LTP harbored somatic TP53 mutations in CAR-T cells. These findings emphasize the importance of LT immuno-genetic monitoring to identify patients at risk for treatment-related complications.