Intrinsic Tumor Drivers and Immune Escape Mechanisms in CD19 CAR T-Cell Therapy Resistance for Aggressive Large B Cell Lymphoma

INTRODUCTION

CD19-directed chimeric antigen receptor (CAR-19) T cells have significantly improved outcomes for heavily pretreated patients with aggressive large B-cell lymphoma (LBCL). However, a high proportion of patients experience treatment failure. Our previous study (Jain et al., Blood 2022) identified genomic complexity as a factor associated with treatment failure. However, the mechanisms through which genomic complexity promotes resistance to CART are largely unknown.

METHODS

To investigate this important topic, we expanded our cohort and analyzed 61 whole-genome sequencing (WGS) and 54 RNA sequencing samples from 54 LBCL patients treated with CAR-19 therapy. The samples included 39 collected at baseline, 15 at relapse, and 7 both before and after treatment. Longitudinal samples were collected in different anatomical sites. We conducted phylogenetic analysis on the 7 paired samples and somatic driver discovery on baseline samples, integrating single nucleotide variants, small insertions/deletions, copy number variants, and structural variants.

RESULTS

Phylogenetic analysis of 7 patients who relapsed within the first year revealed minimal differences and absence of clonal shift, likely reflecting spatial heterogeneity rather than post-CAR19 selection. These results suggest that tumors likely resist CAR-19 treatment due to genomic drivers already present at baseline.

Implementing the somatic driver discovery analysis in 54 LBCL patients, we identified several new genomic drivers involved in CAR19 resistance. Overall, these driver genes can be divided in two main groups: alterations involving the i) immune-synapse and ii) CD19 pathways. In the first group HLA-A, HLA-B, HLA-C, and B2M genes were recurrently involved by somatic events. While HLA class I monoallelic loss, detectable in 38.9% of patients, did not have any impact on progression free survival (PFS), all 10 patients with biallelic loss of the HLA class I complex progressed (p=0.0056), 90% within the first year. Given that tumor cells losing the HLA class I should be cleared by natural killer (NK) cells, we investigated somatic events involving in gene that regulate the interactions between tumor cells and NK cells. We identified recurrent loss-of-function (LOF) alterations in CD58, CD48, MICA, PVR, SPPL3, and TNFSF9 genes, and gain-of-function (GOF) alterations in MUC1, GMDS and CD44, affecting 39% of responders and 67% of progressors. These genes have been previously found be associated with resistance to NK cell-mediated killing. We found that alterations involving these genes co-occurred with biallelic loss of HLA class I in 8 out of 10 patients who progressed after CAR-19, emphasizing the importance of losing both innate and adaptive defense mechanisms. Additionally, analyzing genes involved in the inhibition of T cell function, we identified one patient (CAR_130) with PD-L1 double minutes and focal multi-gain event (copy-number = 12) experiencing progression after CAR-19 treatment. In addition, we found GOF of CMTM6, associated with maintaining PD-L1 expression in the tumor surface, in 5 patients, 4 of which showed disease progression within 99 days.

Moving our focus to genes involved in the CD19 expression, we identified 4 interesting genes recurrently involved by somatic events and associated with primary refractoriness. 8 patients with LOF alterations in ETV6 had shorter PFS (p=0.0026). PAX5 and EBF1 genes, important in the regulation of CD19 expression, were involved by LOF mutations in 13 patients, with 85% progressing after CAR-19 treatment. Finally, we confirmed the CAR-19 therapy resistance role of RHOA, with 12 out of 13 patients with LOF alterations progressed in the first year (p=0.0008). Functional work showed how reduction of RHOA causes reduction in CD19 protein density and resistance to CAR19 in vitro.

Combining the immune escape mechanism alterations and the intrinsic tumor drivers related to CD19, we found that 16 out of 36 progressors showed at least two alterations in both groups. Having alteration from both groups associated with shorter PFS (p<0.0001), with 88% of these patients progressing within the first 6 months.

CONCLUSION

These findings offer evidence of how genomic instability promotes resistance to CAR19 by permitting the selection of genomic alterations that affect the critical relationship between CAR19 cell activity, the immune system, and tumor cell immunogenicity.