Impact of Molecular Features of Diffuse Large B-Cell Lymphoma on Treatment Outcomes with Anti-CD19 Chimeric Antigen Receptor (CAR) T-Cell Therapy

Document Type

Abstract

Publication Date

11-5-2021

Publication Title

Blood

Keywords

washington; swedish; swedish cancer

Abstract

INTRODUCTION

Diffuse large B-cell lymphoma (DLBCL) represents several distinct clinical pathologic entities recently identified by molecular profiling. Treatment with anti-CD19 chimeric antigen receptor (CAR) T-cell therapies is now standard for many patients with relapsed/refractory (R/R) disease. Although antigen loss of CD19 represents a known cause of late relapses, the majority of CAR-T cell treatment failure occurs very soon after treatment at which time the impact of molecular subtype and other somatic mutations of DLBCL is undefined. We sought to determine impact of molecular features of DLBCL tumors on clinical outcomes in a cohort of patients with R/R disease who were treated with axicabtagene ciloleucel (axi-cel) or tisagenlecleucel (tisa-cel) in order to provide insight into the mechanism of response or resistance to CAR-T cell therapy.

METHODS

We collected clinical data and formalin-fixed, paraffin embedded (FFPE) biopsy specimens from 121 DLBCL patients at the time of R/R disease after prior treatment with standard chemoimmunotherapy across 12 US academic medical centers who subsequently received commercial CAR-T cell treatment. Whole exome and transcriptome sequencing was performed on all cases to measure gene expression and gene copy number alterations. Genetic analysis was done on 96 patients with pre-treatment biopsies that passed sequencing quality filters, and expression analysis on 93 patients. Progression-free and overall survival (PFS, OS) measured from the day of CAR-T cell infusion were estimated using the Kaplan-Meier method and compared with the log-rank test.

RESULTS

Baseline demographics and treatment details of the patient population are shown in the Table (Panel A). Best overall response was CR in 43% of patients and PR in 10% patients. PFS and OS were significantly different based on best response to treatment (P<0,001 Figure, Panel B). At the time of R/R disease, the most commonly mutated genes were TP53 (25%), KMT2D (23%), CREBBP (23%), BCL2 (20%), BTG2 (12%), ARID1A (11%), CARD11 (11%), MYD88 (11%) and PIM1 (11%), (Panel C). Molecular subtyping based on the method of Wright, et al. revealed cases to be BN2 (N=16), A53 (N=13), EZB (N=14), MCD (N=13), N1 (N=4), ST2 (N=8) and unclassifiable (UC) in 28 cases. Cluster analysis as described by Chapuy et al. assigned cases to be C0 (N=6), C1 (N=18), C2 (N=14), C3 (N=27), C4 (N=17) and C5 (N=14). The impact of subgroups on of PFS are shown in Panels D and E. While not statistically significant different across all groups, there was a trend towards improved outcomes in C5/MCD as well as the C2/A53 subtypes and a trend towards inferior PFS in the C3/EZB subtypes. Inferior PFS was observed in patients with mutations in BCL2 (P=0.009) and MYC (P<0.001), but not BTG2 (P=0.095), MYD88 (P=0.106), or CD79B (P=0.086). An unbiased model comprising mutations in MYC, BCL2, CDKN2A, and KLHL6 was strongly associated with a lack of response to CAR-T therapy and a poor prognosis (HR=3.55, P<0.001, Panel F). Gene Set Enrichment Analysis (GSEA) identified a gene signature reflecting T-cell activation in the pre-treatment tumor biopsy as being associated with a higher likelihood of response to treatment (Panel G).

CONCLUSIONS

DLBCL patients whose tumors have molecular features that are predictive of inferior response to standard frontline treatment including the high-risk subgroups (C2/A53) and (C5/MCD) have favorable treatment outcomes with CAR-T cell therapy. In contrast, individual driver mutations including MYC and BCL2, CDKN2A, and KLHL6 are associated with inferior PFS with CAR-T cell therapy, while mutations in BTG2, MYD88, and CD79B are associated with a favorable PFS. In addition, gene expression analysis implicates a potential role for the microenvironment in modulating responses to CAR-T therapy. These findings suggest that predictive biomarkers for response to traditional chemoimmunotherapy and cellular immunotherapy are distinct. Our results provide insight into potentially targetable pathways for the development of rational treatment strategies that may augment response CAR-T cell therapy.

Clinical Institute

Cancer

Department

Hematology

Department

Oncology

Link to Full Text


Share

COinS